Functional Pavement Design
Functional Pavement Design
Editors
Sandra Erkens, Xueyan Liu & Kumar Anupam
Delft University of Technology, The Netherlands
Yiqiu Tan
Harbin Institute of Technology, China
CRC Press/Balkema is an imprint of the Taylor & Francis Group, an informa business
All rights reserved. No part of this publication or the information contained herein may
be reproduced, stored in a retrieval system, or transmitted in any form or by any means,
electronic, mechanical, by photocopying, recording or otherwise, without written prior
permission from the publisher.
Although all care is taken to ensure integrity and the quality of this publication and the
information herein, no responsibility is assumed by the publishers nor the author for any
damage to the property or persons as a result of operation or use of this publication and/or
the information contained herein.
Table of contents
Preface xvii
Committees xix
Organizers xxiii
Flexible Pavements
A multi-scale diagnosis model for asphalt pavement cracking in China 3
D. Wang, A. Cannone Falchetto, M. Goeke, M.P. Wistuba & Y. Tsai
A new look at the UK flexible pavement design method 13
I. Artamendi, B. Allen & P. Phillips
A numerical definition of functional pavement 23
R. Guo, Y. Zhou, S. Chen & Y. Zhao
Adhesion properties of warm recycled mixtures produced with different WMA additives 29
F. Frigio, A. Stimilli, M. Bocci & F. Canestrari
Applications of FEMOL and EEP self-adaptive method in static analysis of
asphalt pavement structures 39
Y. Dong, Q. Xing, N. Fang & S. Yuan
Comparative study on dynamic modulus of asphalt mix with two-point bend
and uniaxial compression 51
Y. Huang, Z. Liu, X. Wang & S. Li
Compressible behaviour of bituminous mixtures in creep recovery test in confinement 59
A. Mathruswamy, P.K. Athira & S.P. Atul Narayan
Economically optimal structural design methodology for flexible pavements 69
A. Tsiknas & A. Athanasopoulou
Effect of different variables on asphalt mechanical properties 79
H.M. Al-Mosawe, N.H. Thom, G.D. Airey & A.H. Al-Bayati
Evaluating the feasibility of Freeze-Thaw cycling test on durability of asphalt
mixtures: A pilot study 89
C.-H. Ho, J.Y. Shan, F.Y. Wang & Y.Y. Chen
Evaluation of asphalt pavement bearing capacity using SAFEM 99
P. Liu, D. Wang, F. Otto, J. Hu & M. Oeser
Formulation and application of Burger Model Based Representation function in
characterization of asphalt mixtures 109
C.-H. Ho
Issues related to measurement of dynamic modulus of bituminous mixtures 119
S. Deepa, U. Saravanan & J. Murali Krishnan
v
Low-temperature properties evaluation index of Rubber Asphalt 129
Y.K. Fu, L. Zhang, Y.Q. Tan & D.Y. Meng
Mechanical performance of pavement using porous asphalt as overlay in maintenance 139
M. Li, B. Xu, D. Cao, S. Ping, Y. Fan & J. Li
Multiscale modeling to predict Hot Mix Asphalt stiffness behavior 149
L. Eberhardsteiner, B. Hofko & R. Blab
Prediction of dynamic tensile strain at the bottom of asphalt concrete layer
under FWD load 157
C. Ai, C. Xiao, D. Ren & Y. Qiu
Regularities of creep and long-term strength of hot asphalt concrete under tensile 169
B.B. Teltayev, A. Iskakbayev & C. Oliviero Rossi
Research on nonlinear fatigue damage properties of asphalt mixture based
on the decay of strength 179
H. Liu, Z. Ouyang, C. Liu & G. Qian
Review of Mechanistic-Empirical design approach for asphalt pavements 191
A. Gupta & S. Adhikari
Simulation of three point bending test of asphalt mixture based on Cohesive Zone Model 203
T.H. Yan, X.M. Huang & T. Ma
The effect of mixing sequence on the workability and indirect tensile strength
of asphalt concrete 213
J. Voskuilen, B. Gaarkeuken, D. van Vliet & M. Poot
The impact of dynamic modulus of various HMA mixes on fatigue
cracking prediction 221
K. Georgouli, C. Plati & A. Loizos
The use of steel mesh reinforcement for the cracking control in flexible
pavements: FE analysis in static and dynamic conditions 229
M. Agostinacchio, D. Ciampa, S. Olita & M. Simonetti
Use of dissipated energy to quantify fatigue life of Warm Mix Asphalt 239
K. Lakshmi Roja & J. Murali Krishnan
Use of Mechanistic-Empirical Pavement Design Guide for bituminous pavements in India 249
B. Dokku & J. Murali Krishnan
vi
Effect of spindle diameter and plate gap on the rheological properties of asphalt binders 293
B. Singh, N. Saboo & P. Kumar
Effect of the Xanthan gum biopolymer on rheological and aging properties of bitumens 301
L.-l. Tu, S.-p. Wu, G. Liu & Q. Liu
Effects of the warm-mix surfactant content on the rheological properties of asphalt 311
Y. Yang, H. Liu, E. Li, Y. Dong & L. Yang
Estimation of viscous and fatigue dissipation of bituminous concrete in repeated loading tests 319
R. Gupta & S.P. Atul Narayan
Evaluation of intermolecular interaction between virgin and aged asphalt binders using GPC 329
Y. Ding, B. Huang & X. Shu
Evaluation of oxidation products generated in long term field-aged bitumen 339
K. Zhao, Y. Wang, Y. Wen & D. Chong
Experimental study on warm mix technology of rubber modified asphalt mixture 349
A.H. Kang, M.B. Yao, X.Y. Jiu & P. Xiao
How to characterize the mechanical response of Bitumen Stabilized Material? 355
M.K. Nivedya, A. Veeraragavan & J. Murali Krishnan
Investigation on asphalt’s wax-induced phase separation behavior and its impact
on rheological properties 365
M. Gong, H. Yao, T. Pauli, Z. Yao & J. Yang
Investigation on fatigue properties of basalt fiber asphalt mixture 375
P. Xiao, Y. Lv, J. Jiang & Z.G. Wu
Laboratory evaluation of rejuvenated asphalt binders 383
F. Giustozzi, E. Toraldo & M. Crispino
Laboratory investigations of DBM (Grade 1) mix using different types of additives 393
K. Gupta, T. Chopra & M. Kumar
Low temperature cracking of modified asphalt mixtures as related to binder characteristics 403
W. Huang, B. Li & P. Wang
Material optimal design of binder course between new and existing pavement in Alpine regions 411
W. Wei, N. Li & C. Han
Microscopic evaluation method of Polymer Modified Bitumen performance 419
C. Kou, A. Kang, P. Xiao & Z. Wu
Performance of asphalt binders modified with Re-refined Engine Oil Bottoms (REOB) 427
T. Bennert, C. Ericson, R. Corun & F. Fee
Physical and rheological characteristics of unaged and aged binders 435
B.A. Feyissa
Research on the adhesive properties between asphalt and aggregates in high
temperatures based on the theory of surface free energy 445
Y.Z. Wang, X.D. Wang, Y. Zhang & Y.Z. Zhang
Study on the permanent deformation of HR rubber powder composite modified
asphalt mixture based on dynamic creep test 453
Z. Wu, R. Lu, D. Cai & P. Xiao
The “false positive” on the antiaging properties of asphalt fines investigated
by RTFO laboratory aging of mastics 463
M. Makowska & T.K. Pellinen
vii
The application of Polyphosphoric Acid for preparation of polymer-modified
bituminous blends 473
O. Shulga, R. Maldonado, L. Degao & Z. Yan
The high temperature performance and microstructure of TLA modified asphalt 479
F.Q. Dong, X. Yu, S.J. Liu, B. Xu & J. Chen
The importance of wetting in healing of bitumen 489
G.A. Leegwater, A. Scarpas & S.M.J.G. Erkens
The rheological properties and microstructure of Crumb Rubber Modified Asphalt 499
G.Y. Ding, X. Yu, F.Q. Dong & F. Gu
Thermal behaviour of extracted rubber from hot and warm rubberized asphalt binders 509
H.Y. Yu, Z. Leng & Z.M. Gao
Using molecular dynamics to investigate interfacial adhesion between asphalt
binder and mineral aggregate 517
M. Guo, Y. Tan, A. Bhasin, J. Wei, X. Yang & Y. Hou
Viscoelastic solid-fluid transition in bitumen using rheology and FTIR spectroscopy 529
M.R. Nivitha & J. Murali Krishnan
viii
Pavement performance evaluation and creep properties study on asphalt mixture
modified by anti-rutting agent PCF 645
P. Zhang, H. Li, N.Y. Yin & D.H. Ma
Performance evaluation of Preventive Maintenance Treatments for asphalt pavements
in Beijing 655
C. Wang, S. Zhao, H. Zhang & J. Zhang
Preliminary assessment of functional life of anti-icing asphalt mixture 665
L. Hao, D. Yuming & L. Gen
Study on fatigue properties of recycled asphalt 673
Y. Tan, Y. Fu & Y. Li
Study on preventive maintenance technology for porous asphalt in China 683
B. Xu, M. Li, D. Cao, C. Lu, J. Chen & W. Wang
Sustainable and long life flexible pavements 693
A.F. Nikolaides
Towards sustainable horizontal asphalt recycling 705
J. Qiu, M. Huurman, M. Jacobs, M. Woldekidan & M. Frunt
Pavement Structures
Adaptability of pavement structures on expansive soil embankment with
differential deformations 717
Y. Cheng, X. Huang & C. Li
Analytical solution and distribution characteristics of mechanical response
for elastic half-space body under hyperbolic paraboloid vertical load 729
S. Chen, D. Wang, J. Yi & D. Feng
Design inputs variabilities influence on pavement performance reliability 741
Y.H. Dinegdae & B. Birgisson
Evaluation of permanent deformation of mixtures under intersection vehicle loading 751
X. Wang, X. Gu, M. Yang, H. Ma & H. Shen
Experimental and numerically supported determination of the interlayer bond
shear stiffness 763
B. Hristov, F. Wellner & G. Canon Falla
Experimental research on service performance of fiber-reinforced stress
absorption interlayer 773
H. Liu, Z. Li & L. Yang
Field investigations of the crack pattern in porous asphalt—Continuously Reinforced
Concrete Pavements in The Netherlands 781
D. Ren, L.J.M. Houben, W. Nijssen, R. van der Aa & B. Sluer
Filler bitumen ratio measurement of asphalt mixture based on histogram matching 793
C.F. Wang, A.M. Sha & Z.Y. Sun
Frost heave of saturated silty clay under thermal-hydro-mechanical loading 801
H. Ma, F. Zhang, R. Jing & D. Feng
Impact of dual gauge railway tracks on traffic load induced Permanent
Deformation of low embankments 813
A.D. Mwanza, P. Hao, M. Muya & H. Zhang
ix
Influence of principal stresses distribution on damage of asphalt pavement
on semi rigid bases—case study on Ji-Qing highway 821
G. Yuan & X. Li
Influencing factors on bonding layer performance effect between the ultra-thin surface 829
N. Zhang, N.-X. Zheng & Z.-H. Cong
Investigation on base course of cement stabilizing Phosphorous Slag and gravel 837
G. Qian, Q. Xiang & J. Bo
Investigation on fatigue properties of asphalt mixtures with different compaction
levels using microstructural analysis 845
J. Hu, P. Liu, D. Wang & M. Oeser
Mechanical deformation behaviour of road base Unbound Granular Materials
considering the effect of temperature 855
J. Patzak & F. Wellner
Mechanical properties and durability of Composite Soil Stabilizer Stabilized
Gravel Soil compared with Cement Stabilized Macadam 867
Y. Zhao, Y. Gao & Y. Zhang
Numerical analysis of factors affecting the fatigue life of asphalt pavement
with semi-rigid base 879
T.Y. Zhu, S.Q. Wang & X.M. Huang
Pull-out tests on bituminous specimens with steel wire mesh reinforcements 889
F. Mazzotta, A. Simone, V. Vignali, C. Lantieri, C. Sangiorgi & G. Dondi
Research on the influence of light truck on pavement Permanent Deformation 899
X. Dai, Y. Gao & X. Hu
Review and study of mechanical vibrations for pavement condition assessment 907
C.-H. Ho & A.-M. Davidescu
Road pavement density evaluation—a case study 915
A. Hartikainen, T.K. Pellinen, J. Eskelinen, E. Huuskonen-Snicker & P. Eskelinen
Statistical analysis of axle load data and pavement damage for a few National Highways in India 923
D. Savio, P. Paul & J. Murali Krishnan
The mechanistic-empirical pavement design: An Egyptian perspective 933
A.A. Aguib & S. Khedr
Thermal effect analysis of steel bridge during Guss Asphalt pavement paving 943
Y. Liu, Z.D. Qian, J. Hu & L. Jin
Thermal regime analysis and protective measure evaluation for wide embankment
in permafrost regions of Qinghai-Tibet plateau 955
T. Ma, T. Tang, H. Wang, X.M. Huang & S.Q. Wang
Use of repeated load CBR test to characterize pavement granular materials 965
N. Sparsha, R.G. Robinson & J. Murali Krishnan
x
Comparative analysis of force amplified piezoelectric transducer used for asphalt
pavement energy harvesting 993
H. Zhao, Z. Lin & L. Qin
Comparing in-lab and on-site measurement of pavement acoustic absorption 1001
F.G. Praticò & D. Vizzari
Components changes of bitumen aging and recycling with waste vegetable oils 1011
Z. Suo, L. Chai, J. Ji, F. Guo & L. Sun
Effects of sodium acetate on asphalt mixture comparing to chloride-based deicers 1019
S. Liu, N.X. Zheng & J. Yang
Evaluation of emulsifier optimization based on early performances of a mixture
for cold recycling 1027
H. Liu, Z. Wang & L.Y. Yang
Examining the effects of contributory factors on curing of Cold Bitumen
Emulsion Mixtures 1037
A.I. Nassar, N.H. Thom & T. Parry
Intelligent pavement materials with ecological active dust control technology 1049
Z. Suo, F. Guo, J. Ji, S.S. Jin & A. Zhang
Low noise gradation design for micro-surfacing mix based on indoor noise test method 1059
L.P. Cao, S.Q. Liu, X.S. Hou & R. Zhang
Phase change heat storage properties of PEG/SiO2 Shape-Stabilized Phase
Change Materials in asphalt 1069
L.H. He, B.M. Tang, H.Z. Zhu & X. Cao
Preparation of nano TiO2 dispersed in silica sol and study on NO degradation
rate used in cement-based specimens 1077
X.J. Cao, C. Chen, X.Y. Yang & B.M. Tang
Quantification and modelling of tyre-pavement interaction noise: Effect of vehicular,
pavement, and environmental characteristics 1087
A. Kamineni & V. Chowdary
Stiffness and fatigue performance of recycled mixtures with 50% RAP and different
types of virgin bitumen 1097
E. Manthos & A.F. Nikolaides
Study on the effect of warm mix asphalt mixture performance caused
by the dosage of recycled asphalt pavement 1109
P. Guo, W. Wei, B. Tang & C. Shi
The materials selection and treatment effect of the filtration layer in porous pavement 1115
D. Feng, X. Xie, H. Wang, R. Jing & J. Liu
The preparation, characterization and application of modified nano-TiO2
in automobile exhaust purification 1123
R. Li, J. Cui, B. Xue, D. Kuang & J. Pei
Tire-pavement contact modeling and noise prediction with different tire
configurations and surface characteristics 1129
Y.M. Ding & H. Wang
Treatment effect test on road runoff by different combinations of gravel and activated carbon 1139
G. Qian, J. Zhu & J. He
Warm Mix Asphalt (WMA) technologies: Benefits and drawbacks—a literature review 1145
A. Diab, C. Sangiorgi, R. Ghabchi, M. Zaman & A.M. Wahaballa
xi
Pavements & Innovation
Analysis and application research on volume parameters of porous asphalt mixture 1157
Z. Chen, D. Zhang, P. Hao & J. Liang
Characterization of polymer capsules used for self-healing asphalt pavements 1167
T. Al-Mansoori, R. Micaelo & A. Garcia
Comparative study of two different polymer systems for the application
in flexible pavements 1175
M. Jasso, D. Bakos, J. Stastna & L. Zanzotto
Comparison of fatigue test methods on asphalt mastic 1185
C. Riccardi, A. Cannone Falchetto & M.P. Wistuba
Design and challenges of an innovative load-bearing permeable concrete pavement 1193
Y.-H. Lee, N.N.S. Chou & J.-W. Chen
Design and performance investigation on dense graded anti-sliding ultra-thin
wearing course material 1203
Y.S. Guan, L. Geng, S.L. Zhang, W. Zhou & G.L. Jin
Evaluation of chloride induced damage in stone mastic asphalt mixes suitable
for induction heating 1213
P. Apostolidis, X. Liu, A. Scarpas, M.F.C. van de Ven & G. van Bochove
Evaluation of laboratory performance on half-warm asphalt mixture 1223
H. Liu, Y. Tong & Y. Dong
Evaluation of low temperature performance of warm mix asphalt by glass transition
temperature 1231
X. Yu, F. Gu, F.Q. Dong, G.Y. Ding & T.Y. Wang
Experimental study on damage evolution of asphalt mixture using Digital Image
Correlation 1237
C. Xing, T. Tan, L. Zhang & Y. Tan
Framework for in situ soil stabilization based on electrokinetic principles 1247
N.G. Pavlatos & A. Scarpas
Innovative pavement structures for urban express corridors 1257
C. Yin, Y. Pan & Z. Zhang
Mitigation of asphalt pavement temperatures by means of natural air convection 1269
A. Chiarelli, A. Dawson & A. García
Mixing and compaction properties of asphalt mixture modified with silane 1279
S. Büchler, A. Cannone Falchetto & M.P. Wistuba
Numerical simulation of dynamic responses of basalt fiber reinforced asphalt
pavement under moving loads 1287
P.Q. Wang, F.J. Wang, M.Q. Ke, Z.G. Lu, H.Q. Zhang & Z.Y. Zeng
On the use of low-cost PVA fiber to develop ECC material for pavement use 1297
Z.G. Zhang, Q. Zhang, S.Z. Qian & V.C. Li
Research into the material properties of prefabricated rollable asphalt pavement 1309
Y.-s. Dong, Y. Hou, D.-w. Cao & Y.-j. Zhang
Research on application of semi-flexible pavement base on trunk highway 1317
B.F. Zheng & W. Liu
xii
Research on ice-melting technology of high elasticity pavement surface 1327
L. Wang, M. Li, X. Wang & X. Sun
Research on material and method of microwave ice-melting technology 1337
M. Li, X. Wang & L. Yuan
Research on the coarse aggregate configuration of asphalt mixture with
skeleton dense structure based on center of mass synthesis 1347
H. Guo & Y.L. Zhao
Rutting prediction of nanomaterial modified asphalt concrete using
FEM simulation 1355
L. Gu & L. Sun
Self-powered sensors for road pavements 1365
F.G. Praticò, F.G. Della Corte & M. Merenda
Strain sensing capacity of multifunctional cementitious material 1375
Z. Ge, W. Hao, X.S. Wu & R. Sun
The influence of road condition on the shelf-life of tomatoes 1381
C.J. Pretorius & W.J.vdM. Steyn
Rigid Pavements
A fast and durable concrete repair material: High-Early-Strength Engineered
Cementitious Composites 1393
H. Deng & S. Qian
Analysis of surface texture character of Exposed-Aggregate Concrete Pavement
using digital image technology 1401
X. Cheng, S. Han & X. Wang
Characteristics of water-cement ratio for grouting cement asphalt emulsion paste
with different asphalt-cement ratios based on viscosity study 1409
J. Ouyang & Y. Tan
Comparative analysis of short paneled pavement using 3-D Finite Element analysis 1421
A. Gupta & T. Gandhi
Expansive reactions in cement concrete pavement structure and the way how
to analyse the risk of their occurrence 1431
J. Stryk, Z. Nevosad, M. Gregerova, A. Frybort & V. Chupik
Optimum design for sustainable, ‘green’ concrete overlays. Part I: (a) mix-design,
(b) controlling flexural failure 1441
Y. Lin, J.N. Karadelis, A.O. Olubanwo, Y. Xu & P. Phillips
Optimum design for sustainable, ‘green’ concrete overlays. Part II: Shear failure
at cracks and inadequate resistance to reflection cracking 1453
Y. Xu, J.N. Karadelis, Y. Lin, A.O. Olubanwo & P. Phillips
Optimum design for sustainable, ‘green’ concrete overlays. Part III: Failure due
to interfacial delamination 1465
A.O. Olubanwo, J.N. Karadelis, Y. Xu, Y. Lin & P. Phillips
Recommendation on noise control standard for concrete pavements in China 1477
L. Li, B. Li, Z.W. Zhang & X.L. Yang
xiii
Research of cement concrete pavement using thermal imaging method 1489
I. Gameliak, V. Vyrozhemskyi, I. Voloshyna & M. Dmitriev
Static and dynamic response of rigid concrete pavements: Parametric study 1499
J. Ramaekers, L. Vandewalle & H. Degée
Safety
Effect of compaction on skid resistance of asphalt pavements 1513
E. Kassem, A. Awed & E. Masad
Evaluating wet-weather driving safety risks of pavement ruts 1523
L. Chu & T.F. Fwa
Optimisation of photoluminescent painting treatments on different surface layers 1533
F.G. Praticò, S. Noto & A. Moro
Preparation and evaluation of haze-eliminating coating materials for
asphalt pavement 1543
B. Xue, Y. Wang, L. He, J. Zhang & J. Pei
Road safety research about influence of crosswind on highway parallel twin-deck
bridges using numerical simulation 1553
Z.W. Yu, M. Xu, F. Chen, X.D. Pan & W.J. Zhu
Rockfall protection of road infrastructures: Design approach and
numerical simulations 1563
M. Agostinacchio, D. Ciampa & S. Olita
Study on the decay law of skid resistance of High Friction Surfaces 1571
Z. Li, L. Dong & P. Hao
Study on the influence of heavy load on pavement structure and load control 1577
X. Ma & X. Wang
Study on working conditions of expressway asphalt pavement by monitoring test road 1585
Z.J. Xi, J. Chao, H.D. Ding & W. Chao
Traffic Engineering
A review of pedestrian safety models for urban areas in Low and Middle Income Countries 1597
R.A. Kraidi & H. Evdorides
An investigation for an all-encompassing iRAP road Star Rating index 1607
A.K. Jameel & H. Evdorides
Analyzing and modeling of the lane-crossing behavior and traffic conflicts
when riders of non-motorized vehicles drive on curb-parking sections 1617
Y. Ci, Z. Liu, S. Zheng, L. Wu, Z. Bian & X. Li
Perceived differences in travel speeds and traffic states 1627
H. Li, H. Tu, H. Liu & H. Shi
Real-time freeway crash prediction model by using single ultrasonic detector
lane-level data 1635
J.M. You, J.H. Wang & S.E. Fang
xiv
Real-time monitoring of the extended road network by utilising telematics technology 1645
I. Wessels & W.J.vdM. Steyn
Recognising risk factors associated with crash frequency on rural four lane highways 1657
C. Naveen Kumar, M. Parida & S.S. Jain
Rural two-lane highway crash occurrence formulation and injury-mitigation strategy
development in China 1669
Y. Ci, L. Wu, J. Zhao, Z. Liu & G. Zhang
The consistency of horizontal alignment at tunnel entrance and exit zone 1679
Z. Yang, H.F. Han, Z.Y. Xiong & D.L. Lei
The gradation test prediction model of GAC-20 modified asphalt mixture 1691
Z. Sun, T. Wang, Z. Wu & G. Zhu
Other Topics
A fast simulation tool for the undergraduate course of pavement engineering 1701
X. Chen
Bearing capacity of bored pile in refreezing metaphase and telophase
in permafrost region 1711
X. Liu, C. Zhou, D. Feng, Y. Cao, P. Cao & R. Jing
Dynamic evolution trajectory of defects and aggregates and its relation
to asphalt mixtures’ cracking propagation behavior 1721
X. Zhu, L. Li, D. Sun, Y. Yuan & H. Wang
Reuse of mining waste into innovative alkali-activated-based materials
for road pavement applications 1735
C. Sangiorgi, C. Lantieri, P. Tataranni, J. Castro-Gomes & M. Gabriel
Safety and reliability of a road: A prototypical study based on the directive 2008/96/EC 1745
F.G. Praticò, D. Catalfamo & C. Lanciano
Sound barriers landscape design: Dynamic aesthetics factors and visually
compensation method 1755
X. Qin, Q. Tan & H. Han
Strength size effect on small asphalt mixture specimens at low temperature 1767
A. Cannone Falchetto, K.H. Moon & M.P. Wistuba
The effect assessment for urban flooding under different curb inlets with the
2D hydrodynamic model 1777
Y. Geng, X. Chen & Z. Wang
xv
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Preface
Following the great success of the three earlier Chinese-European Workshops (CEW) on
Functional Pavement Design, the 4th Workshop was held at Delft University of Technology,
in Delft, the Netherlands, on June 29 – July 1, 2016. Over the years participation increased
with delegates not only from China and Europe but also from other parts of the world.
The focus of the CEW series is on field tests, laboratory test methods and advanced anal-
ysis techniques. The Workshops are meant to provide a forum for the exchange of ideas,
information and knowledge amongst experts involved in analysis, material development and
production, experimental characterization, design and construction of pavements.
All submitted contributions were subjected to exhaustive peer review by at least three
members of the Scientific Committee and the Editors. On the basis of their recommenda-
tions, 184 contributions, originating from 27 different countries, were chosen for presentation
and inclusion in the Proceedings.
The Editors would like to thank the members of the Scientific Committee who took the
responsibility of reviewing the manuscripts and ensuring the excellent quality of the accepted
papers, all sponsors and in particular our platinum sponsors the Dutch Ministry of Infra-
structure and the Environment, Lucobit AG and Fahe China for their contribution to the
successful organization of the event.
We hope that the Workshop will contribute to the establishment of a new generation of
pavement design methodologies in which rational mechanics principles, advanced constitu-
tive models and advanced material characterization techniques shall constitute the backbone
of the design process.
The Editors
Yiqiu Tan
Harbin Institute of Technology, China
xvii
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Committees
EXECUTIVE COMMITTEE
Chair
A.(Tom) Scarpas, Delft University of Technology, The Netherlands
Vice chair
Markus Oeser, RWTH Aachen University, Germany
Secretary
Anupam Kumar, Delft University of Technology, The Netherlands
Members
Imad L. Al-Qadi, University of Illinois at Urbana-Champaign, USA
Björn Birgisson, Aston University, UK
Fwa Tien Fang, National University of Singapore, Singapore
Bernhard Steinauer, RWTH Aachen University, Germany
Yiqiu Tan, Harbin Institute of Technology, China
Jun Yang, Southeast University, China
SCIENTIFIC COMMITTEE
xix
Alvaro Garcia Hernandez, University of Nottingham, UK
Baoshan Huang, University of Tennessee, USA
Xiaoming Huang, Southeast University, China
Michael Kaliske, Dresden University of Technology, Germany
John N Karadelis, Coventry University, UK
Rebekka Kienle, University of Stuttgart, Germany
J. Murali Krishnan, Indian Institute of Technology Madras, India
Anupam Kumar, Delft University of Technology, The Netherlands
Zhen Leng, Hong Kong Polytechnic University, China
Hui Li, Tongji University, China
Jianming Ling, Tongji University, China
Tianqing Ling, Chongqing Jiaotong University, China
Chaohui Liu, Changsha University of Science and Technology, China
Gang Liu, Wuhan University of Technology, China
Pengfei Liu, RWTH Aachen University, Germany
Quantao Liu, Wuhan University of Technology, China
Xueyan Liu, Delft University of Technology, The Netherlands
Yu Liu, Chang’an University, China
Andreas Loizos, National Technical University of Athens, Greece
Bert Jan Lommerts, Latexfalt B.V., The Netherlands
Eyad Masad, Texas A&M University at Qatar, Qatar
Athanasios Nikolaidis, Aristotle University of Thessaloniki, Greece
Markus Oeser, RWTH Aachen University, Germany
Laszlo Petho, ARRB Group Ltd, Australia
Jianzhong Pei, Chang’an University, China
Terhi Pellinen, Aalto University, Finland
Filippo Pratico, Mediterranea University of Reggio Calabria, Italy
Guoping Qian, Changsha University of Science and Technology, China
Yaning Qiao, Mott MacDonald Group, UK
Jian Qiu, BAM Infra Asfalt, The Netherlands
Ruibo Ren, Shandong Jianzhu University, China
Wolfram Ressel, University of Stuttgart, Germany
Irmgard Rucker, University of Stuttgart, Germany
Chantal Rudman, Stellenbosch University, South Africa
A.(Tom) Scarpas, Delft University of Technology, The Netherlands
Aimin Sha, Chang’an University, China
Shihui Shen, Tongji University, China
Lijun Sun, Tongji University, China
Zhaohui Sun, Shenyang Jianzhu University, China
Boming Tang, Chongqing Jiaotong University, China
Yiqiu Tan, Harbin Institute of Technology, China
Huizhao Tu, Tongji University, China
Waheed Uddin, University of Mississippi, USA
Andreas Ueckermann, RWTH Aachen University, Germany
Dawei Wang, RWTH Aachen University, Germany
Duanyi Wang, South China University of Technology, China
Hainan Wang, Chang’an University, China
Linbing Wang, Virginia Tech, USA
Yuhong Wang, Hong Kong Polytechnic University, China
Daniel Wegener, RWTH Aachen University, Germany
Frohmut Wellner, Dresden University of Technology, Germany
Tobias Winkler, RWTH Aachen University, Germany
Ines Wollny, Dresden University of Technology, Germany
xx
Chunying Wu, Jiangsu Transportation Institute, China
Shaopeng Wu, Wuhan University of Technology, China
Martin van de Ven, Delft University of Technology, The Netherlands
Yue Xiao, Wuhan University of Technology, China
Xiaoguang Xie, Harbin Institute of Technology, China
Huining Xu, Harbin Institute of Technology, China
Jian Xu, Research Institute of Highway, China
Jun Yang, Southeast University, China
Junyan Yi, Harbin institute of Technology, China
Zhanping You, Michigan Technological University, USA
Xin Yu, Hohai University, China
Hongchao Zhang, Tongji University, China
Jinxi Zhang, Beijing University of Technology, China
Jiupeng Zhang, Chang’an University, China
Ji Zhe Zhang, University of Nottingham, UK
Junhui Zhang, Changsha University of Science and Technology, China
Xiaoning Zhang, South China University of Technology, China
Hongduo Zhao, Tongji University, China
Yongli Zhao, Southeast University, China
Jianlong Zheng, Changsha University of Science and Technology, China
Chunxiu Zhou, Dalian Maritime University, China
Hongzhou Zhu, Chongqing Jiaotong University, China
Xingyi Zhu, Tongji University, China
xxi
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Organizers
Rijkswaterstaat
Ministry of Infrastructure and Environment,
The Netherlands
xxiii
Flexible Pavements
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Di Wang
Technische Universität Braunschweig – ISBS, Braunschweig, Germany
Chang’an University, Xi’an, China
Yichang Tsai
Georgia Technology, Atlanta, USA
ABSTRACT: Identifying the causes of asphalt pavement distress is crucial for selecting
adequate management strategies and rehabilitation treatments. However, it is difficult for
engineers to determine the causes associated to the specific distress based on simple field
observation. This paper presents a multi-scale approach for diagnosing the causes of asphalt
pavement cracking. This deleterious phenomenon is addressed based on a number of multi-
scale characteristics, including crack shape, extent, location, patterns, and width. Then, sev-
eral causal factors linked to the cracks formation are categorized. Finally, a diagnosis model
is developed by establishing the relationship between the multi-scale crack characteristics and
the causal factors. Four real examples illustrate the use of the proposed pavement crack diag-
nosis model. The proposed model has the potential of being expanded into a deterministic-
stochastic model when a wider data base (knowledge base data set) is established.
1 INTRODUCTION
Thin asphalt surface with thick semi-rigid base is widely used for highway pavements in
China, due to its economic and high quality combination. However, this structure may be
prone to cracking, followed by water penetration leading to additional distress and ultimately
to premature failure. Hence, it is crucial to identify the accurate cause of early cracking, so
that adequate maintenance and rehabilitation (M&R) methods can be selected for limiting
and mitigating deleterious cracking phenomena.
Currently, pavement engineers in China diagnose the origin of a distress mainly relying on
their own experience and on national standards (MOC, 2001; MOC, 2007; MOC, 2009), while
most of the younger engineers refer to the work of experienced practitioner, leading to low
M&R decision effectiveness. Therefore, there is an urgent need for establishing a judgment
process and method to help training young engineers to diagnose the causes of asphalt pave-
ment distresses. Among pavement distresses, the causes at the origin of cracking are the most
complex to identify (SHRP, 1993; Shen, 2004; ODOT, 2010). The characteristics of cracks are
different; the same origin may leads to various distresses, meanwhile, similar characteristics
are most likely associated to the mutual interaction of several concurrent phenomena. There-
fore, correctly identifying the main causes of cracking formation and propagation is crucial
for improving the quality and the effectiveness of M&R activities (Mao, 2004).In China, the
national standards (MOC, 2007) divide asphalt pavement crack types by geometric charac-
teristics: longitudinal crack, transverse crack, alligator cracking, block crack/irregular crack.
On this basis, pavement engineers record certain quantitative characteristic indicators such as
3
width, length and extension of cracks during the survey process. In addition, the asphalt pave-
ment crack can also be divided according to their origin (SHRP, 1993; Sha, 2001; Shen, 2004).
For example, in a former research (Zheng, 2003), it was found that the main causes of asphalt
pavement crack in China can be divided into two categories: traffic load induced (load crack)
and temperature induced (non-load crack) cracks. In addition it was also observed that the two
types of causes can be identified according to the different cracking orientation. In this study, a
cracking cause diagnosis model is proposed. This model is based on the preliminary work con-
ducted in a previous research (Tsai, 1998) and integrates field observation and the results from
interviews with expert engineers. The objective is to assist pavement engineers in effectively
evaluating the causes of asphalt pavement distresses and determining the proper rehabilitation
methods. For such a purpose, the research approach is divided into two major steps:
1. Knowledge acquisition and representation of crack multi-scale characteristics and causes;
2. Identification of the distress origin by establishing the connection between the multi-scale
characteristics and the potential causal factors.
The ultimate purpose of diagnosing the causes of cracking is determining the relevant per-
tinence maintenance rehabilitation scheme, which depends on the dominant distress cause
(Evdorides H., 1994). For traffic safety, the isolated distress have less influence than the entire
project distresses, therefore, in this paper, the judgment of entire project level cracking is a
priority in the model.
The basic statistical interval length of pavement damage survey is 100 m, which is pro-
posed in the Chinese national standard (MOC, 2001; MOC, 2007). This section length
was selected as the minimum statistical unit in this study. If one similar cracking type
appears continuously more than 100 m, then it can be defined as an entire project dis-
tress, in this case, the maintenance scheme must consider the distress type and its severity
through the whole extending area. Meanwhile, the characteristics of an isolate cracking
is also considered, hence, the proposed model is suitable for both the extended and isolate
distresses.
The diagnosis model for asphalt pavement cracking was developed mainly based on the
knowledge acquired from the pavement experts’ interview from SRAB; the extracted infor-
mation are from field observations and literature review. The use of the Knowledge Based
System (KBS) starts by identifying the type of the cracks (longitudinal, transverse, poly-
gon, block and slippage) and then follows with the secondary attributes such as straight-
ness, locations, extent of cracks and accompanying distress. While conflict illation exists,
the model will ask for further information (e.g. take a core) to resolve the conflict. The
prototype flow diagram for determining the causes of asphalt pavement distresses is shown
in Figure 1 (Tsai, 1998).
5
Figure 1. Continued.
6
Figure 1. Prototype flow diagram for determining the causes of asphalt pavement distresses.
4 CASE STUDY
The field observation used for the present investigation was chosen in the middle of Shaanxi
province, in China. Four real distress examples among sixteen cases were selected. For the iden-
tification of the causes of asphalt pavement cracking, extended and isolate level of cracking
were used. In additionally, pictures of the specific distress are used to illustrate the problem.
7
Figure 2. The illustration of longitudinal cracking.
Type longitudinal crack, the extension direction parallel the driving direction
Straightness Not exists
Location beyond the wheel path, and near the pavement edge, the length is about
12 m, the widest point is about 5 cm, it has been treated by sealing
Embankment yes
Judgment load and base problem (embankment uneven settlement)
(Fig. 3b) which are located on the top of a culvert. The pavement engineers declared that the
structure subsides between the two parallel cracks, and the cracks repeatedly occurred after
joint sealing treatment. Hence, as shown in Table 2, the result is structure problem based on
the judgment process shown in Fig. 1 c).
8
Table 2. The judgment process of the transverse cracking.
9
heavy traffic load experienced by this section (Fig.4 c), the pavement was significantly dam-
aged within one year after rehabilitation.
Inference procedure: alligator cracking mainly appear on the right side (Fig. 4 a), and over
the entire lane. The extent is more than one kilometer with rutting and potholes overlapping
to the main distress phenomenon (Fig. 4 b); therefore, this has to be intended as a rehabilita-
tion project for an extended distress. Hence, as shown in Table 3, the result is load induced
problem according to the judgment process shown in Fig. 1 d).
The purpose in this paper was to assist pavement engineers evaluating the causes of asphalt
pavement cracking and suggesting the rehabilitation methods. A prototype multi-scale char-
acteristic diagnosis model was proposed through knowledge acquisition and representation.
Then the model was validated by four real case studies on expressway and provincial high-
way in Shaanxi province, in China. On this basis, the prototype model can be preferable
developed as follows. For example, the time perspective can be established to incorporate the
deterioration rates (e.g. length, width, density, pattern etc.) at different locations (e.g. wheel
path, pavement edge) and scales (e.g. entire project or isolate distress) to determine the causal
factors. In addition, with the application of 2D imaging and 3D laser technologies, a detailed
level of crack properties can be further characterized at multi-scale spatial clustering levels to
generate more crack properties at multi-scale levels (e.g. 0.1 km, 1 m or 1 cm) to assist in the
determination of causal distress factors.
ACKNOWLEDGMENTS
The authors will like to thank Shaanxi Road Administration Bureau (SRAB) for its support.
The authors will also like to thank Mr. Sen Shu, and many other SRAB engineers for their
technical assistance during the data collection process. We will also like to thank Dr. Lian-zhu
Wang of Shaanxi Expressway Testing & Measuring Limited Company for his assistance. The
authors thank the China Postdoctoral Science Foundation (2014M562287) and National Natu-
ral Science Foundation of China (51408083, 51508064) for vigorously supporting this study.
10
REFERENCES
Evdorides H., 1994. A Prototype Knowledge-Based System for Pavement Analysis. Doctoral dissertation,
University of Birmingham.
MOC, 1997. Technical Standard of Highway Engineering (JTJ 001–1997), Beijing, China Communica-
tions Press: 3.
MOC, 2001. Technical Specifications for Maintenance of Highway Asphalt Pavement (JTG 073.2–2001)
Beijing, China Communications Press: 10–11.
MOC, 2003. Technical Standard of Highway Engineering (JTG B01–2003), Beijing, China Communica-
tions Press: 1.
MOC, 2007. Highway Performance Assessment Standards (JTG H20–2007). Zhejiang, China Com-
munications Press: 3–4.
MOC, 2009. Technical Specifications for Maintenance of Highway (JTG H10–2001) Beijing, China
Communications Press: 34–36.
Mao Cheng, 2004. Asphalt Pavement Crack Formation Mechanism and Propagation Behavior Research.
Chengdu: Southwest Jiaotong University: 24–42.
Myers, L.A., R. Roque, and B. Birgisson, 2001. Propagation Mechanisms for Surface—Initiated Lon-
gitudinal Wheel Path Cracks. In 80“Annual Meeting (CD-ROM), Transportation Research Board,
National Research Council, Washington, D.C., Paper No. 01–0433.
ODOT, 2010. Pavement Distress Survey Manual. Oregon Department of Transportation, Distress Survey
manual
SHRP-P-338,1993. Distress Identification Manual for the Long-Term Pavement Performance Project.
Strategic Highway Research Program (SHRP), National Research Council, Washington, D.C..
Sha Qing-lin, 2001. Premature Damage and Its Preservative Measure of Bituminous Pavement on
Expressway. Beijing, China Communications Press: 105–139.
Shen Jin-an, Li Fu-pu, Chen Jing, 2004. Analysis and Preventive Techniques of Premature Damage of
Asphalt Pavement in Expressway. Beijing, China Communications Press: 218–251.
Yi-Chang Tsai, Chenglong Jiang, Yuchun Huang, 2012. A Multi-scale Crack Fundamental Element
Model for Real-world Pavement Crack Classification. Journal of Computing in Civil Engineering,
11.
Yi-Chang Tsai, James S. Lai, Lixiang Sun, 1998. Developing a Prototype KBS for Diagnosing Causes of
Asphalt Pavement Distressses. Proceedings of the 1st Conference on New Information Technologies
for Decision Making in Civil Engineering, Montreal, Canada, Vol. 1, pp. 515–526.
Zheng Jian-long, Zhou Zhi-gang, Zhang Qi-sen, 2003. Asphalt Pavement Cracking Design Theory and
Methods. Beijing, China Communications Press: 1–5.
11
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
The UK pavement design method has its origins on the experimental pavements constructed
and monitored during a 30 year period starting in the early 1950s. This early work led to a
design approach based on the interpretation of the structural performance of these roads
and supported by mathematical structural models and laboratory testing of the materials
used at the time (Powel et al. 1984).
The mechanistic-empirical method developed by Powel et al. (1984) uses a multi-layer,
linear elastic response model of the pavement to calculate the strains induced by a standard
wheel load. Horizontal tensile strains at the underside of the asphalt base layer and vertical
compressive strain at the top of the subgrade are then used to predict the life of the pavement
in terms of the cumulative number of standard axles.
Moreover, the design curve that led to the fatigue criterion was derived using labora-
tory fatigue tests as an aid to interpreting the results from the experimental road sections.
These tests were carried out on a standard mixture used at the time. Large shift factors,
however, had to be used to correlate the laboratory fatigue results with the experimental
data.
A more versatile design approach which allowed the use of new materials and introduced
the concept of foundation stiffness classes was developed by Nunn (2004). Furthermore,
with the switch to using foundation classes the subgrade strain criterion became redundant.
The current analytical design method relies on a single criterion, or flexural criterion, that
limits the strain at the underside of the base asphalt layer to a permissible level to achieve the
required pavement life. This flexural criterion is still based on the original fatigue criterion by
Powel et al. (1984). Nevertheless, the versatile approach forms the basis of the current UK
pavement design method (DMRB 2006).
The UK analytical design method employs an effective in-service stiffness modulus
that is measured at a frequency of 5 Hz and at a temperature of 20ºC. This effective
in-service stiffness modulus is the stiffness attained by the material after approximately
one to two years in service. The standard design method is, however, very restricted
and allows only a limited number of asphalt mixtures and design stiffness to be used
(DMRB 2006).
13
The response of a pavement to traffic loading is affected among other factors by the traf-
fic speed and the thickness of the pavement layers. The speed of traffic determines vehicu-
lar loading time. Thus, as the traffic speed increases, the loading time is reduced and the
equivalent frequency of loading is increased. Also, increasing pavement thickness results in
larger effective loading time and, consequently, lower frequency. Therefore, since the stiff-
ness modulus of asphalt depends on loading frequency, both parameters, traffic speed and
asphalt thickness, should be considered when designing flexible pavements (Watlther & Isail-
ovic 2015).
In the US, The Mechanistic-Empirical Pavement Design Guide (MEPDG) uses a sim-
plified method to calculate the frequency of the applied load as a function of the vehicle
speed and the cross-section of the pavement structure. The applicable complex modulus is
then determined at the pertinent temperature and loading frequency. Work by Al Qadi et
al. (2008) indicated, however, that fitting functions currently used to predict the stress pulse
duration and simplified time-frequency conversion relations may result in significant errors
in the design calculations.
It has been estimated that an increase in loading frequency of 1 Hz may equate approxi-
mately to an increase in traffic speed of 10 km/h, which suggests that most design methods
use design traffic speeds in the range 50 to 100 km/h (Nunn & Merrill 1997). Mollenhauer
et al. (2009) reported loading frequencies between 0.5 and 8.9 Hz for vehicle speeds between
2.9 km/h and 31.4 km/h and asphalt thickness between 140 mm and 340 mm. Hence, chang-
ing the design traffic speed and asphalt thickness will imply using different design stiffness
depending on the frequency of loading.
In this work stiffness modulus and fatigue properties of asphalt mixtures currently used
in the UK road network have been determined in the laboratory using a four-point bend-
ing machine. The current UK design criterion was then modified to account for the fatigue
performance of the different mixtures. Furthermore, the effect of traffic speed and pavement
thickness was introduced in the proposed method by selecting the appropriate loading fre-
quency to determine the design stiffness.
2 EXPERIMENTAL
2.1 Materials
Granite aggregates and limestone filler were used to produce AC 20 and AC 14 base and
binder course mixtures complying with EN 13108-1 (CEN 2006a). The AC 20 was manufac-
tured using a 40/60 penetration grade binder and the designed binder content was 4.8%. This
pen grade bitumen is the most common grade in the UK for base and binder course mixtures.
AC 14 mixtures, on the other hand, were produced with 10/20 pen hard grade bitumen, and
two different polymer modified binders, PMB1 and PMB2. Binder content for these mix-
tures was 5.5%. Empirical properties of the binders declared by the bitumen suppliers are
presented in Table 1.
Aggregates and binder were mixed in the laboratory using a mechanical mixer and then
compacted to slabs using a laboratory roller compactor. Target air void content of the slabs
was 4.0%. Four identical slabs were manufactured per mixture.
14
2.2 Stiffness test
Stiffness modulus was determined using the four-point bending (4PB) stiffness test (CEN
2012a). Beam specimens 400 × 50 × 50 mm3 cut from slabs were used for testing. Five speci-
mens per mixture were tested. Tests were carried out under controlled strain mode at a target
strain amplitude of 100 microstrain. Tests were carried out at 20, 15 and 10ºC and at frequen-
cies of 0.5, 1.0, 2.5, 5.0, 8.0 and 10.0 Hz. At each temperature and frequency the number of
loading cycles applied was 100 and the stiffness of the specimen was defined as that at 100
cycles.
Stiffness values determined using the 4PB stiffness tests are presented in Figure 1. It can be
seen that the stiffness values depended on the type of mixture and, bitumen and PMB grade.
Also, the stiffness increased as the loading frequency increased (see Fig. 1a) and the tem-
perature decreased (see Fig. 1b). Interestingly, the rate of increase in stiffness with frequency
depended on the mixture type, i.e. AC 14 or AC 20, but not on the grade of the binder.
Furthermore, this change in stiffness with frequency was more pronounced for the AC 20
mixture, particularly at low frequencies (see Fig. 1a).
N50 Aε b (1)
where N50 = No. of cycles to failure, ε = applied strain (microstrain) and A and b are regres-
sion constants.
Fatigue lives are shown in Figure 2. Regression constants A and b, and R2 values are pre-
sented in Table 2. Fatigue resistance of AC mixtures is given by the microstrain at 106 cycles
(ε6) (CEN 2006a). Microstrain at 106 cycles values (ε6) were determined from the plots of the
strain vs number of cycles to failure, and are shown in Table 2. The higher the ε6 value the
better the resistance to fatigue.
Results indicated better fatigue resistance of the AC 14 mixtures with the polymer modi-
fied binders, PMB 1 and PMB2, compared with the same mixture with the hard grade binder.
15
Figure 2. 4PB fatigue test data at 20ºC and 30 Hz.
ε6
Material A b R2 No. of specimens microstrain
Also, the fatigue resistance of the AC 20 mixture was worse than those of the AC14 mixtures.
This could be attributed to the volume of binder in the mixtures, the rheology of the binders
and the maximum aggregate nominal size.
3 PAVEMENT RESPONSE
The response of a pavement to traffic loading is affected by many factors including axle load,
traffic speed, layer thickness, temperature and materials’ mechanical properties. Typically,
Accelerated Pavement Test (APT) facilities are used to determine pavement responses, i.e.
stress and strains, under controlled loading and environmental conditions, on instrumented
test sections. These test sections can provide information like the amplitude of the stresses
and strains and the duration of the load at different locations within the pavement. The mag-
nitude of these strains/stresses and their frequency can then be used in numerical models and
laboratory testing to estimate the life of the pavement (COST 2005).
Traffic speed and pavement thickness are two parameters that affect pavement response.
For instance, the speed of traffic determines vehicular loading time. Thus, as the traffic speed
increases, the effective loading time at the bottom of the asphalt layer is reduced and the peak
strain is reduced, as seen schematically in Figure 3. Loading time can then be converted to an
equivalent design loading frequency. This equivalent frequency increases as the traffic speed
increases. Similarly, increasing pavement thickness results in larger effective loading time and
a reduction of the peak strain at the bottom of the asphalt layer (see Fig. 3). Consequently,
the equivalent frequency for a thicker layer is reduced.
Mollenhauer at al. (2009) reported the results of an APT program consisting of 8 different
pavement test sections with asphalt thickness varying from 120 mm to 340 mm and loaded
16
Figure 3. Schematic representation of the strain at different traffic speeds and pavement thicknesses.
Figure 4. Relationship between frequency of loading and: a) traffic speed and b) asphalt layer
thickness.
at various vehicle speeds. Measured strains at the bottom of the asphalt base layers were
then used to determine an equivalent frequency by fitting a sinusoidal signal through the
experimental data. Figure 4a shows the results from the test sections. It can be seen that the
equivalent frequency at the bottom of the base asphalt layer increased with vehicle speed and
decreased with increasing asphalt layer thickness. Furthermore, Figure 4b shows extrapolated
data for vehicle speeds of 20, 50 and 80 km/h. It can be seen that the equivalent frequency at
the bottom of the asphalt layers decreased with increasing asphalt thickness.
17
Table 3. Loading frequencies (Hz).
where εr = calculated tensile strain at the underside of the asphalt layer, KFlex = 1.089 × E−0.172,
E = Design stiffness (GPa) and KSafety = 1 (typical value)
The design stiffness used in the calculations is an effective in-service stiffness modulus that
is measured at a frequency of 5 Hz and at a temperature of 20ºC. This effective in-service
stiffness modulus is the stiffness attained by the material after approximately one to two years
in service. Furthermore, the frequency of 5 Hz is assumed to correspond to a design traffic
speed of 50 km/h. Design stiffness values for standard materials are given in the UK pave-
ment design manual HD26/06 (DMRB 2006) and include for instance 8000 MPa for EME2
and 4700 MPa for DBM50. The design stiffness values given in the design manual, however,
do not take into consideration traffic speed or layer thickness. Moreover, it has been shown
that these parameters affect the frequency of loading and therefore the stiffness of the mate-
rial. Thus, variations in design traffic speeds and asphalt thicknesses will result in different
design stiffness depending on the frequency of loading.
In this work, three traffic speeds, 20, 50 and 80 km/h, and three asphalt thicknesses, 290,
320 and 360 mm, were selected. The frequencies were then estimated from the results of the
test sections given in Figure 4 and are presented in Table 3. 4PB stiffness test data presented
in Figure 1 was then used to determine the stiffness of the mixtures at these frequencies. Stiff-
ness values were then rounded down to the nearest 500 MPa. Table 4 shows, for example, the
stiffness values obtained for the AC14 PMB1 mixture. It can be seen that the design stiffness
increased as the traffic speed increased and the asphalt thickness reduced.
18
frequencies depending on traffic speed and layer thickness. Total asphalt thickness selected
were 290, 320 and 360 mm. It should be noted that these thickness are typical design thick-
nesses for a standard material (DBM50) on a FC2 for design traffic lives of 10, 20 and 80 msa
according to HD26 design manual.
Regarding the foundation, the layer stiffness of the sub-base was 150 MPa and the thick-
ness 225 mm. The sub-grade was 1.5 m thick and the stiffness was 50 MPa. Finally, the
ground below subgrade was assumed to have a stiffness of 10000 MPa.
Critical strains induced by a standard wheel load (40 kN) represented by a circular area
(0.151 m radius) with a uniform vertical stress were calculated using a linear elastic, multi-
layer pavement model. Calculations were carried out at a temperature of 20ºC and at loading
frequencies given in Table 4. Tensile strains calculated at the underside of the base/binder
asphalt layer were then used to determine the design lives of the proposed pavement struc-
tures using Equation 2.
Figure 5. Design lives: a) effect of traffic speed (290 mm thickness), b) effect of asphalt thickness
(50 km/h).
19
Figure 6. Design lives: a) same fatigue properties (Eq. 2), b) different fatigue properties (Eq. 3).
n
N ⎛ εr ⎞
=⎜ ⎟
( )
(3)
106 ⎝ K Flex K SSafety
f t × 201 × 10 6 × 6 New
New 6 DBM ⎠
where: ε6(New) = strain at 106 cycles of the new mixture measured in the new fatigue test, ε6(DBM)
= strain at 106 cycles of the DBM mixture measured in a new fatigue test, n = exponent of
the fatigue relationship for the new mixture determined from the new fatigue test. (Note: this
criterion assumes that the exponent for the DBM mixture determined from the new method
is the same as that determined with the original method, i.e. −4.16).
In this work, fatigue properties were determined at 20ºC and 30 Hz using the 4PB test. The
strain at 106 cycles (ε6) and the slope of the fatigue relationship (b) of the mixtures investigated
are presented in Table 1. The regression constant “b” in Table 1 is the exponent of the fatigue
relationship “n” in Equation 3. Also, the exponent of the AC 20 40/60 mixture determined
with the 4PB fatigue is −4.57,which is similar to the value determined for the reference DBM50
mixture using the original uniaxial tension/compression fatigue test method, i.e. −4.16.
Figure 6 shows the design lives calculated using Equation 2 (Fig. 6a) and Equation 3
(Fig. 6b). These values correspond to an asphalt thickness of 290 mm. It can be seen that
when the fatigue properties of all the mixtures were assumed to be the same, the life of the
pavement was controlled by the stiffness of the mixture. In other words, the higher the stiff-
ness of the asphalt, the longer the design life of the pavement. For instance, the design life of
the pavement with the stiffer mixture, i.e. AC14 10/20, was superior to those pavements with
a softer grade or PMBs.
When the fatigue performance of the mixtures was introduced in the design method both,
asphalt stiffness and fatigue properties controlled the life of the pavement. In this case, mix-
tures with higher stiffness and better fatigue showed longer pavement lives. For instance, the
design life of the pavement with a relatively stiff mixture with very good fatigue characteris-
tics like the AC 14 PMB2 was significantly longer than that for a pavement with a very stiff
mixture like AC 14 10/20 but with inferior fatigue resistance. Also, the design life of the pave-
ment with a standard material like AC 20 40/60 was the shortest as the fatigue properties of
this mixture were inferior to those of the AC 14 materials.
5 DICUSSIONS
The work presented here has shown that both traffic speed and asphalt thickness have a
considerable effect on the design life of flexible pavements and should not be ignored within
any design process. Instrumented test sections on APT facilities have shown that these two
parameters influence the magnitude and frequency of the strains at the underside of the
asphalt layers. Strains measured on the asphalt layers are also dependent on other factors like
the temperature and the mechanical properties of the asphalt and foundation layers.
20
In this work, the frequencies used to determine asphalt stiffness for different traffic speeds
and asphalt layer thicknesses were estimated from APT sections. The temperature of these
sections were, however, lower than the design temperature of 20ºC used in the analytical
design (Mollenhauer et al. 2009). Thus, at a design temperature of 20ºC, differences in the
magnitude and frequency of the strains will be expected.
Similarly, structural properties of the foundation and asphalt layers also affect the response
of the pavement. Surface stiffness of the foundations layers of the test section reported by
Mollenhauer et al. (2009) in which this study is based varied between 100 and 150 MPa.
These foundation surface stiffness values correspond to a UK Foundation Class 2 (FC2).
This FC2 was used in the pavement model for the analytical design calculations. The asphalt
mixtures used in the APT sections were also different to those used in the current study. Thus,
differences in pavement responses will be expected between the different types of mixtures.
Thus, although the frequencies determined on the pavements test sections are specific for
that particular set of conditions and materials, similar frequencies could be expected on a
pavement of similar characteristics. For a more detailed determination of the frequencies
with traffic speed and layer thickness, however, the test sections should be constructed with
materials with similar structural properties to those of the pavement model.
6 CONCLUSIONS
Based on the laboratory work and on the analytical pavement design calculations the follow-
ing conclusions can be drawn:
• Stiffness values depended on the type of mixture and bitumen and PMB grade. Also, the
stiffness increased as the loading frequency increased and the temperature decreased. Fur-
thermore, the change in stiffness with frequency depended on the mixture type but not on
the binder grade.
• Fatigue resistance of the AC 14 mixtures was better than that for the AC 20 mixture. This
could be attributed to the volume of binder and the rheology of the binders. Best fatigue
performance was obtained for the AC 14 mixtures with the polymer modified binders fol-
lowed by that with the hard grade bitumen.
• Results from accelerated loading test section elsewhere have shown that as the traffic speed
increases the peak tensile strains at the underside of the asphalt layer decreased and the
equivalent loading frequency increased. Similarly, increasing pavement thickness resulted
in smaller tensile strains and lower frequencies
• Loading frequencies estimated from pavement test sections were used to determine the
design stiffness of the asphalt layers at various traffic speeds and layers thicknesses. Design
stiffness values increased with traffic speed and decreasing layer thickness.
• Analytical pavement design results on a typical pavement structure showed that the design
life of the pavement shortened when the design traffic speed was reduced.
• When the fatigue properties of the asphalt mixtures were assumed to be the same, the life
of the pavement was controlled by the stiffness of the asphalt mixture. Thus, the higher
the stiffness of the asphalt, the longer the design life of the pavement. However, when the
fatigue performance of the mixtures was incorporated in the design method both asphalt
stiffness and fatigue properties controlled the life of the pavement. Thus, a combination of
high stiffness and good fatigue properties resulted in longer pavement lives.
REFERENCES
Al-Qadi, I.L, Xie, W. & Elseifi, M. 2008. Frequency determination from vehicular loading time pulse
to predict appropriate complex modulus in MEPDG, Journal of the Association of Asphalt Paving
Technologists, 77: 739–772.
COST Action 347, 2005. Improvements in pavement research with accelerate load testing. Final Report
COST Action 347. European Commission, Brussels.
21
Comité Européen de Normalisation (CEN), 2006a. EN 13108-1. Bituminous mixtures. Materials speci-
fications. Part 1: Asphalt Concrete.
Comité Européen de Normalisation (CEN), 2006b. EN 13108-20. Bituminous mixtures. Materials speci-
fications. Part 20: Type testing.
Comité Européen de Normalisation (CEN), 2012a. EN 12697-26. Bituminous mixtures. Test methods
for hot mix asphalt. Part 26: Stiffness.
Comité Européen de Normalisation (CEN), 2012b. EN 12697-24. Bituminous mixtures. Test methods
for hot mix asphalt. Part 24: Resistance to fatigue.
Design Manual for Roads and Bridges (DMRB). 2006. Volume 7 Pavement Design and Maintenance,
Section 2 Pavement Design and Construction, Part 3: HD 26/06 Pavement design (DMRB 7.2.3).
London, The Stationary Office.
Mollenhauer, K. Wistiba, M. & Rabe, R. 2009. Loading frequency and fatigue: in situ condition &
impact on test results, 2nd Workshop on Four Point Bending, Pais (ed.)
Nunn, M & Merrill, D. 1997. Review of flexible and composite pavement design methods. Paper and
Articles PA3298, Crowthorne: TRL Limited
Nunn, M. 2004. Development of a more versatile approach to flexible and flexible composite pavement
design, TRL Report TRL615, Transport Research Laboratory, Crowthorne, England.
Powel, W.D., Potter, J.F., Mayhew, H.C. & Nunn, M.E. 1984. The structural design of bituminous roads,
TRRL Laboratory Report 1132, Transport and Road Research Laboratory, Crowthorne, England.
Walther, A. & Isailović. 2015. Effects of low vehicle speed on the service life time of asphalt pavements.
Proc. 6th Inter. Conf. Bituminous Mixtures and Pavements, Thessaloniki, 10–12 June, 2015.
22
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Functional pavement has been widely used in the preventive maintenance of
road for its distinguished improvement in road performance like noise reduction or driving
comfortability upgrade. However, the confirmation of its thickness tends to be empirical-
based instead of theoretical-based. In this paper, a numerical definition of functional pave-
ment was presented. Abaqus 6.14 was used in the modeling process in order to evaluate the
largest tensile stress level at the bottom of base course, where different elastic modulus and
thickness of functional pavement were set on given substratum parameters. Subsequently,
Matlab 7.11.0 was employed to fit the relation of the aforementioned variables in reference
to that of asphalt concrete. Finally, a numerical definition of functional pavement have been
implemented.
1 INTRODUCTION
Functional pavement, generally considered as thin asphalt layer, has a growing aspect on the
application of road engineering in recent years. As a renewable ‘skin’ of the stable road, func-
tional pavement is mainly used to improve road performance rather than bearing capacity. It
can effectively reduce noise, improve skid resistance and forward visibility during wet condi-
tion and it is durable compared with other maintenance methods (Sandberg et al. 2011). As
more attention is paid on functional improvements than structural enhancements of asphalt
pavements, Asphalt Pavement Alliance (APA) defined perpetual pavement as ‘an asphalt
pavement designed and built to last longer than 50 years without requiring major structural
rehabilitation or reconstruction, and needing only periodic surface renewal in response to
distresses confined to the top of the pavement’ (APA 2002). By restraining distresses to the
functional layer on the top, traffic congestion can be reduced with fewer rehabilitation activi-
ties, lower consumption of non-renewable resources and lower life-cycle cost of the pavement
system (Timm and Newcomb 2006).
In fact, there are many kinds of definitions for the thickness of functional pavement in
different countries, most of which are empirical-based rather than theoretical-based. As for
specific standard for its thickness, European Standard stipulates that Asphalt Concrete for
Very Thin Layer (BBTM) is 20 mm to 30 mm thick (BSI 2006a), while Asphalt for Ultra-thin
Layer (AUTL) used for surface courses with a thickness of 10 mm to 20 mm (BSI 2006b).
Chinese Specifications for Design of Highway Asphalt Pavement requires that coarse func-
tional asphalt concrete is 20∼25 mm thick with large surface deepness and good skid-resisting
capacity (JTG D50 2006). In America, on the other hand, the thickness of thin overlay varies
from one state to another. Some may specify thin overlay to be about 2 inch (50 mm) while
others consider it no more than 1 inch (25 mm) (Watson and Heitzman 2014).
Although all criterions give the thickness range, no pointed definition based on mechani-
cal theory is presented. On the accepted premise that functional pavement makes little contri-
bution to bearing capacity, this paper put forward an innovative method to define functional
pavement. The method is built on theoretical analysis and data fitting, without experimental
verification.
23
2 MODELING ANALYSIS
The data of the six parts of the road structure refer to Liao (2008) except functional layer.
24
Figure 2. Load schematic diagram.
with 0.7 MPa tire pressure. According to the principle of equivalent stress (Huang 1998),
load area is converted into rectangle with 227.7 mm length and 156.8 mm width. Two rectan-
gular centers space 319.5 mm shown in Figure 2.
As for the boundary conditions, in view of the symmetry of pavement structure and loads,
the bottom side adopts full consolidation constraints while other four sides adopt symmetry
constraints.
25
3 NUMERICAL FITTING
Given the lower pavement structure’s data, the modulus of elasticity and thickness of the
functional pavement are changed to get a series of computational data. After preliminary
test, we chooe 200, 600, 1000, 2000, 3000, 5000 and 7000 (MPa) as the representative values
of elasticity modulus and the representative values of thickness are 0.5,1,2,3,4,5,7,9(cm).
Through the combination of two, 56 sets of data could be calculated.
When analyzing the pavement structure, there are 5 important data: surface deflection,
the maximum tensile stress at the bottom of surface course, the maximum tensile stress at
the bottom of base course, the maximum compressive stress at the top of base course and
the maximum compressive stress at the top of subgrade. As for asphalt concrete, when using
semi-rigid base, the base course emerge fatigue cracks firstly (Yao 2003). So the most impor-
tant factor for bearing capacity is the maximum tensile stress at the bottom of base course.
This factor is used to classify functional pavement and structural pavement.
In the field of road engineering, when the change of the stress is less than 5%, it can be
ignored. With the help of this view, we can define the functional pavement. When laying a
new top layer on the original pavement structure, if the change range of the maximum tensile
stress at the bottom of base course is less than 5%, then the new top layer can be defined as
functional pavement, otherwise, it belongs to structural pavement.
Matlab7.11.0 was used to fit the stress distribution surface chart based on the 56 repre-
sentative values. The method is cubic interpolation based on the triangle. According to the
result of interpolation, draw the three-dimensional surface gird graph. The result is shown in
Figure-4. In the graph, x-axis represents the thickness of the new top layer, y-axis represents
the elasticity modulus of the new top layer and z-axis represents the maximum tensile stress
value at the bottom of base course after laying the new top layer.
Considering that the role of functional pavement is to improve traffic performance of road
rather than to contribute to bearing capacity, so functional pavement and structural pavement
can be classified as follow. If the decrease of the maximum tensile stress at the bottom of base
course is less than 5% after laying the new top layer, it can be defined as functional pavement,
otherwise, it belongs to structural pavement. The most important thing is to find critical value.
For the original road structure without new top layer, the maximum tensile stress value at the
bottom of base course is 69147 Pa, so the critical value should be 65689.65 Pa which is 95%
of the original value without laying a new layer. Then a series of data can be got from Fig-
ure 4 which the maximum tensile stress is 65689.65 Pa with different thickness and elasticity
modulus shown in Figure 5. In consideration of the distribution of data, we used exponential
function to fit it. The result showed that the matching effect was perfect.
Finally, we could get the numerical definition of the functional pavement. In this case,
if, then the top layer might belong to functional pavement and would only improve road
Figure 4. Distribution diagram of maximum tensile stress at the bottom of base course.
26
Figure 5. Contour fitting results.
performance. If E×h1.6 > 3000, the top layer might be structural pavement and would make a
contribution to bearing capacity. (E: modulus of elasticity/MPa; h: thickness/cm) The result
from the study is agreed with the common sense. If the top layer is not thin with a low elastic-
ity modulus, it might make little contribution to bearing capacity and could be classified into
functional pavement. And a thin layer with high elasticity modulus is also important to load
sharing. Besides, if we fix the elasticity modulus in the range of 1000 to 2000 MPa ( in which
many top layer materials’ modulus belong to this range), then the result of thickness range
is 1.42 cm to 2.58 cm which fit the definition of functional pavement by most of the current
specifications and standards from country to country. By the way, the format E×h1.16 is simi-
lar to the definition of the radius of relative rigidity used in the design of cement concrete
pavement. So we might define a new variable for the design of asphalt concrete pavement in
the future. However, there is still a lot of work to do.
1. Combining with an example, this paper gave a numerical method to define functional pave-
ment, which based on the maximum tensile stress value at the bottom of base course.
2. The function which defines functional pavement is an exponential form only associated
with elasticity modulus and thickness of the layer.
3. This paper just gave a new thought to define functional based on numerical calculation,
and the universality of the method needed to be testified by more field tests.
REFERENCES
Asphalt Pavement Alliance (APA). (2002). Perpetual Pavements: A Synthesis. APA 101, Lanham, Maryland.
BSI (2006). BS EN 13108-2. Bituminous Mixtures-Material Specification-very Thin Layer Asphalt
Concrete (BBTM).
BSI (2006). BS EN 13108-2. Bituminous Mixtures-Material Specifications–Asphalt for Ultra-Thin
Layer (AUTL).
Huang Yangxian (1998). Analysis and Design of Pavement:20-21. Beijing:China Communications Press.
JTG D50(2006). Specifications for Design of Highway Asphalt Pavement [S].
Liao Gongyun & Huang Xiaoming. (2008). Abaqus Finite Element Software Application in Road
Engineering:51. Nanjing:Southeast University Press.
27
Sandberg, U., Kragh, J., Goubert, L., et al. (2011), Optimization of thin asphalt layers: state-of-the-
art review.
Watson, D. E. & Heitzman, M. (2014). Thin Asphalt Concrete Overlays (No. Project 20–05, Topic 44–07).
Timm, D. H. & Newcomb, D. E. (2006) Perpetual pavement design for flexible pavements in the US,
International Journal of Pavement Engineering, Vol. 7 No. 2, 111–119.
Yao Zukang (2003). Review on the Current Design Index of Asphalt Pavement in China. Highway:
43–49.
28
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
In road construction, eco-friendly materials and sustainable production processes are not
anymore a luxury of few developed countries and careful construction companies.
Environmental and economical requirements force the identification of innovative techno-
logical solutions able to provide environmental benefits and cost savings as well as acceptable
or even improved mechanical performance.
In that sense, one of the major promising “green” techniques employed for the production
of bituminous mixtures is given by the combination of warm mix technologies and recycling.
Warm Mix Asphalt (WMA) mixtures are produced at temperatures approximately 25–30°C
less than traditional Hot Mix Asphalt (HMA) mixtures thank to physico-chemical bitumen
composition changes during the mixing process (D’Angelo et al. 2008). Reclaimed Asphalt
Pavement (RAP) is the primary product coming from the milling of old asphalt pavements.
Its adding generally causes stiffer and more brittle mixtures due to the presence of aged bitu-
men and, for this reason, the total amount of RAP has to be kept lower than a maximum
limit. The combination of the two techniques (use of WMA additives for producing mixtures
containing RAP at reduced temperatures) guarantees a lower aging rate of the virgin bitu-
29
men, allowing the incorporation of higher amounts of RAP in warm mixtures without any
need to select softer virgin bitumen grade (Prowell & Hurley 2007). As a consequence, it
should be taken into account that most of the new produced warm asphalt mixtures include
RAP.
Therefore, further investigations are needed to efficiently exploit possible advantages
related to the combination of the abovementioned techniques as well as to optimize the mix
design of recycled WMA mixtures. In particular, due to the lack of long-term performance
data from the field, laboratory performance tests continue to fulfill a fundamental role in the
design and deployment of existing and innovative WMA technologies.
Warm recycled mixture performance are strictly related to a number of factors (e.g. type
of warm technology, bitumen, production temperatures, environmental conditions, RAP
amount).
Among various parameters, the adhesion at the bitumen-aggregates interface represents
one of the most critical aspects associated to the failure of warm mixtures. In fact, reduced
production temperatures could cause poor bond strength due to incomplete and not homo-
geneous coating of aggregates by bitumen as well as a not proper drying process of wet
aggregate particles. Low adhesion properties mean higher water susceptibility which in turn
implies stripping of bitumen from the aggregate surface or cohesive failure within the bitu-
men leading to a rapid deterioration of overall pavement performance and durability.
Among other possible advantages (decrease in material costs, energy savings, higher rut-
ting resistance) (Chiu et al. 2008, Canestrari et al. 2014), the use of reclaimed material could
be helpful to partially reduce the detrimental effect of water since RAP aggregates can be
seen as a water-resistant material due to the presence of the thin film of aged bitumen.
Nonetheless, two aspects related to RAP could counteract and compromise the potential
benefits. First, milled material stockpiles tend to retain high moisture (often higher than
virgin aggregates). Since usually RAP aggregates are not pre-heated (or pre-heated at very
low temperature) prior to being mixed with the other material components, great amount of
undesired water remains entrapped during mixing, negatively affecting bitumen-aggregate
interactions. Second, the reduced production temperatures can alter the degree of reactiva-
tion of RAP bitumen with consequences on the total bitumen amount. The effective “work-
ing” bitumen is lower than what expected for traditional HMAs prepared at standard high
temperatures with same mix design and equal amount of RAP. Without taking into account
this aspect, the resulting overestimation of the effective “working” bitumen leads to lower
aggregate coating and so higher risk of water damage mostly resulting in raveling and strip-
ping. This explains the reason that often leads to the adding of more anti-stripping agents
in WMA mixtures.
Several studies have addressed the evaluation of water susceptibility of recycled WMA
mixtures (Mallick et al. 2008, Doyle et al. 2011, Zhao et al. 2012), but analyses were usually
based on mechanical test results or analytical methods, rather than conducting direct meas-
urements of the interactions between bitumen and aggregates. This latter aspect is directly
linked to the adhesion properties that represent a key point to properly identify the best mate-
rials combination and to consciously design mixtures with enhanced moisture resistance.
This paper faces the problem with a double analysis aimed at evaluating the interaction
between the component materials (bitumen and aggregates), measured in terms of funda-
mental adhesion properties and the water susceptibility of the corresponding asphalt mix-
tures. Since material response significantly varies depending on WMA technology as well as
mixture type, the investigation involved both open and dense graded mixtures and the three
main categories of WMA additives (organic, chemical and zeolite). HMA mixtures were also
prepared for comparison purposes. The raveling resistance of mixtures was analyzed through
Cantabro tests and correlated, both in dry and wet conditions, to the adhesion measurements
evaluated by means of Bitumen Bond Strength (BBS) tests.
Moreover, in order to assess the effects of RAP aggregates, all mixtures were prepared
including a certain amount of milled material and a specific BBS testing protocol was
adopted to account for the presence of a pre-coated substrate in the interaction between
virgin bitumen and reclaimed aggregates (Canestrari et al. 2014).
30
2 EXPERIMENTAL STUDY
2.1 Materials
BBS test
Bitumen-aggregate adhesion properties were evaluated by means of Bond Bitumen Strength
(BBS) tests. Two aggregate types (basalt and limestone) were selected as they represent the
typical materials used to produce open and dense graded mixtures. Aggregate plates were
prepared with adequate dimensions (10 × 10 cm2) in order to allow the positioning of five
specimens (Fig. 1a). For each aggregate type, two different surface treatments were con-
sidered: untreated surface (i.e. uncoated) were used to simulate virgin aggregates, whereas
a specific protocol (Canestrari et al. 2014) was adopted to reproduce the surface of RAP
aggregates (i.e. coated). In particular, a small amount of bitumen (0.2÷0.3 g) was uniformly
distributed on heated aggregates plates. Afterwards, in order to simulate the aging process
that RAP aggregates experienced in the field, the plates were aged in a forced-draft oven at
135°±3°C for 4±0.5 hours (short term aging) and then at 85°±3°C for 120±0.5 hours (long
term aging), according to the AASHTO R 30 specification.
A Styrene-Butadiene-Styrene (SBS) polymer modified bitumen was used to coat the
substrates since the majority of RAP for motorway pavements currently available in Italy
includes SBS polymer modified bitumens (Fig. 1b).
The aggregate-bitumen system was prepared by placing a small amount of virgin bitumen
(0.08 g) onto the surface of a pull-stub pre-heated at 170°C for a minimum of 30 minutes
to simulate the in plant production temperature for bitumen. Afterwards, the pull-stub was
pressed onto the pre-heated aggregate surface avoiding twisting stresses that could entrap air
between the bitumen and the aggregate. In order to simulate the aggregate-bitumen adhesion
properties that characterized HMA and WMA mixtures, aggregates surfaces were pre-heated
at different temperatures, equal to 170°C to simulate the HMA condition and 130°C to simu-
late the WMA condition.
All the HMA and WMA conditions were simulated by using the same SBS polymer modi-
fied bitumen; the main characteristics of the bitumen are reported in Table 1. For the WMA
condition three different technologies were adopted as representative of the three main catego-
ries nowadays available in the market: a chemical additive (dosed at 0.5% by bitumen weight),
an organic additive (dosed at 3.0% by bitumen weight) and a zeolite (dosed at 6.0% by bitu-
men weight). Moreover, an additional WMA condition was taken into account by using the
same reduced production temperature (130°C) without adding any WMA additive to the bitu-
men, in order to differentiate the effects due only to the temperature reduction from the ones
related to the adding of WMA additives. The identification codes are summarized in Table 2.
Cantabro test
HMA and WMA mixtures were prepared in laboratory and tested by means of Cantabro tests
in order to evaluate the influence of reduced production temperatures on raveling resistance.
The experimental program included both open graded (hereafter named OG) and dense graded
Figure 1. Pull-stub on an aggregate plate (a). Coated and uncoated aggregates substrate (b). Schematic
representation of BBS equipment (c).
31
Table 1. Bitumen main characteristics.
Production temperature
(hereafter named OD) mixtures; both mixtures type included recycled aggregates (dosed at
different amount as prescribed by the Italian technical specification for motorways).
The open graded mixtures were prepared including 15% of selected RAP (coarse fraction
8/16 mm) from milled porous asphalt surface layers. Basalt virgin aggregates, RAP (8/16 mm)
and filler were combined obtaining the final grading curve reported in Figure 2a. Moreover,
70% cellolose-30% glass fibres dosed at 0.3% by the aggregates weight were added to the mix-
tures. According to the mix design optimized in previous studies (Frigio et al. 2013, Frigio
et al. 2015), the total bitumen content was equal to 5.25% (by aggregate weight) given by the
virgin bitumen and the bitumen within RAP (4.0% by RAP weight).
The dense graded mixtures were prepared including 25% of un-fractioned RAP (0/16
mm). Limestone virgin aggregates and filler were combined obtaining the final grading curve
showed in Figure 2b. The total bitumen content was equal to 4.80% (by aggregate weight)
including the virgin bitumen and the bitumen within RAP (4.68% by RAP weight).
The virgin bitumen used to produce both OG and DG mixtures was a SBS modified bitu-
men, the same used to prepare the BBS specimens (Table 1).
For each type of mixture, one HMA reference mixture and three WMA mixtures (pre-
pared with different WMA additives) were investigated in this research study; moreover, an
additional WMA mixture was prepared at reduced production temperatures without adding
any WMA additives for comparison purposes.
The reference mixtures, hereafter named OG_H and DG_H, were mixed and compacted
at 170°C and 160°C respectively whereas all the other WMA mixtures were mixed at 130°C
and compacted at 120°C. In particular, virgin aggregates were heated at 130°C for about 3
hours whereas the bitumen was heated at 170°C in order to achieve proper fluidity, follow-
ing the in plant procedure.
The same type and dosage of WMA additives used to produce the BBS specimens were
used to prepare the three WMA mixtures. The chemical and the organic additive were added
to the hot virgin bitumen and mixed for 15 minutes right before the mixture preparation
whereas the zeolite was added directly to the asphalt mixture in the mixing chamber, accord-
ing to the producer recommendations. The identification codes are summarized in Table 3.
32
2.2 Test program and protocols
The experimental program includes BBS tests and Cantabro tests with the aim at evaluating
the influence of reduced production temperature and warm technologies on adhesion prop-
erties and moisture susceptibility. To this purpose, BBS and Cantabro tests were performed
at 25°C on specimens after both dry conditioning (in air for 24 h at test temperature) or wet
conditioning (in water at 40°C for 72 h, according to EN 12697-12 Method A).
BBS test (AASTHO TP 91) quantifies the cohesion/adhesion properties between aggregate
and bitumen within asphalt mixtures. The bond strength is evaluated by measuring the tensile
force required to break the bond between the bitumen and the aggregate substrate. The equip-
ment is a portable pneumatic adhesion tester (Fig. 1c) that includes a pressure hose, a piston,
a reaction plate and a metal pull-stub (diameter = 12.7 mm, surrounding edge = 0.3 mm).
The test consists in increasing the pulling force applied to the stub through the pneumatic
system with a loading rate of 315 kPa/s. The pressure value necessary to reach the failure of
the aggregate/bitumen bond is taken into account and converted into the Pull-Off Tensile
Strength (POTS) as a function of the bonding surface area. Then, the failure type is visually
identified as two main mechanisms may occur: failure at the interface (adhesive failure code
A), defined as loss of adhesive bond strength between bitumen and aggregate; failure within
the bitumen (cohesive failure code C), characterized by loss of cohesive strength within the
bituminous component caused by the rupture of bonds in the asphalt film. In the case of this
study, five specimens were prepared for each test configuration.
Cantabro tests (EN 12697-17) allow the estimation of the raveling resistance of asphalt mix-
tures. The test consists in subjecting an asphalt specimen to 300 revolutions inside the Los Ange-
les machine drum without any metal balls; the particle loss (PL) at the end of the tests is taken
into account as a measure of the internal cohesion between particles. For each mixture type
(open and dense graded mixtures) and for each test condition (dry and wet), four specimens were
prepared and tested by means of Cantabro tests. All specimens were compacted using the gyra-
tory compactor at a fixed height (63.5 mm) in order to achieve similar air voids contents for each
mixture type (selected equal to 20% for OG and 4% for DG) since it is recognized that particle
loss values are strongly dependent on the air void content of the specimen (Frigio et al. 2013).
Figure 2. Aggregates grading curves: (a) open-graded mixtures; (b) dense-graded mixtures.
Production temperature
33
3 RESULTS AND ANALYSIS
Figure 3. BBS results. Comparison between dry and wet condition for different substrates: (a) lime-
stone uncoated; (b) limestone coated; (c) basalt uncoated; (d) basalt coated.
34
gates is able to provide an adhesive substrate which guarantees better interaction with the virgin
bitumen, although reduced production temperatures. This finding suggests that RAP aggre-
gates are helpful for the production of warm mixtures since they reduce the water sensitivity of
the material as well as the negative effect related to the use of warm technologies.
Regarding the comparison between the various warm additives, the overall results suggests
that, for both limestone and basalt aggregates (coated or uncoated), the organic was the one
characterized by the lowest performance, whereas the chemical additive provided acceptable
performance in both dry and wet condition, comparable with the conventional HMA.
Figure 4. Cantabro test results: (a) open-graded mixtures; (b) dense-graded mixtures.
35
3.2 Cantabro test results
Cantabro tests results in terms of particle loss average values are shown in Figure 4 for open-
graded mixtures and for dense-graded mixtures in both dry and wet conditions. The corre-
sponding mean air voids content of each mixture are shown in the figure along with the error
bars reporting standard deviation values for each testing condition. It is important to under-
line that, for both mixture types, the air voids contents of mixtures were similar, meaning that
the Cantabro tests results can be considered fully comparable as previously highlighted.
In case of open-graded mixtures, similar particle loss values were obtained in dry conditions
for all materials, demonstrating that the raveling resistance is not significantly influenced by
reduced production temperatures when the effect of water is not considered. Moreover, results
were entirely consistent with international requirements for high traffic highways corresponds to
a maximum of 20% loss (Alvarez et al. 2010, Kline & Putman 2011), with the exception of WMA
mixtures including organic additive. The effect of water is quantified for OG mixtures by means
of the particle loss increase (PLI) parameter, evaluated as the difference between PL values meas-
ured in dry and wet condition with respect to the PL value after dry conditioning (Fig. 4a).
Results in wet conditions as well as PLI values demonstrated that all WMA mixtures (with
and without additive) were significantly susceptible to the detrimental effect of water in terms
of raveling resistance and their behavior was considerably worse than the one shown by the
reference HMA. In particular, the presence of the organic additive and the zeolite did not
allow any improvement in terms of water susceptibility as their performance in wet condi-
tions were even worst that the WMA mixture prepared without any WMA additive. Only the
open-graded mixture including the chemical additive (OG_WC) showed limited particle loss
values also after the wet conditioning ensuring adequate water resistance, although it does
not guaranteed comparable performance than the HMA. Such outcomes confirm previous
results (Hamzah et al. 2011, Frigio et al. 2015) suggesting that raveling resistance and water
susceptibility are major concerns that affect WMA mixtures. Anti-stripping agents could be
added in order to improve performance, especially when organic additives are used.
In case of dense-graded mixtures, the reduced production temperatures did not significantly
affect the performance since the particle loss values were found to be comparable between
HMA and WMA mixtures in both dry and wet conditions. The use of the organic additive and
the zeolite led to a slight increase in particle loss values, whereas the chemical additive ensured
even better raveling resistance than the HMA mixture in both dry and wet conditions.
Overall Cantabro test results showed that the effect of the different WMA additives on
the raveling resistance performance was similar in case of open and dense graded mixtures.
In particular, the chemical additive was the only one that ensured good performance in both
dry and wet conditions.
36
mixtures) was taken into account. In this sense, BBS test results in terms of POTS obtained
for the uncoated substrate (virgin aggregate) and the corresponding coated configuration
(RAP aggregate) were proportionally combined in order to consider the presence of a certain
amount of RAP within a given asphalt mixture as following specified:
Figure 5. BBS vs Cantabro tests results: (a) basalt/OG mixtures; (b) limestone/DG mixtures.
37
4 CONCLUSIONS
The present experimental study showed that the production of WMA mixtures at reduced
temperature may affect adhesion properties between aggregates and bitumen with conse-
quences on raveling resistance as well as water susceptibility. However, other factors (i.e.
mineralogical nature of aggregates, volumetric properties, internal aggregate structure) can
play a fundamental role on the overall mechanical response. In particular, RAP material is
able to partially compensate this loss in performance thank to the presence of pre-coated
aggregates which adhere more easily to the virgin bitumen even at reduced production tem-
peratures. Moreover, RAP aggregates provide a water-resistant surface less prone to undergo
water damage. Such an effect was found to be significant especially when the chemical WMA
additive was used with major benefits on water susceptibility.
Good correlation between adhesion properties and raveling resistance was found mainly
in case of open graded mixtures: a very severe decrease in raveling performance with the loss
in adhesion properties was detected when these mixtures were subjected to wet condition.
Thus, such kinds of WMA technologies are not suggested for open graded WMA mixtures
when good water resistance is needed. Contrarily, the aggregate structure of dense graded
mixtures prevents from significant water susceptibility and reduces the impact of different
WMA technologies on the overall results.
REFERENCES
Alvarez, A.E., Martin, A.E., Estakhri, C. & Izzo, R. 2010. Evaluation of durability tests for permeable
friction course mixtures. International Journal of Pavement Engineering, 11(1), 49–60.
Canestrari, F., Ferrotti, G., Cardone, F. & Stimilli, A. 2014. Innovative testing protocol for the evalu-
ation of binder-reclaimed aggregate bond strength. Transportation Research Record: Journal of the
Transportation Research Board, ISSN:0361-1981, 2444:63–70. doi:10.3141/2444.07
Chiu, C., Hsu, T. & Yang, W. 2008. Life Cycle Assessment on using Recycled Materials for Rehabilitat-
ing Asphalt Pavements. Resources Conservation and Recycling, 52(3):545–556.
Doyle, J.D., Mejias-Santiago, M., Brown, E.R. & Howard, I.L. 2011. Performance of High RAP-WMA
Surface Mixtures. Journal of the Association of Asphalt Paving Technologists, 80:403–437.
D’Angelo, J., Harm, E., Bartoszek, J., Baumgardner, G., Corrigan, M., Cowsert, J., Harman, T., Jam-
shidi, M., Jones, W., Newcomb, D., Prowell, B., Sines, R. & Yeaton, B. 2008. Warm-Mix Asphalt:
European Practice. Report FHWA/PL-08/007.
Frigio, F., Pasquini, E., Ferrotti, G. & Canestrari, F. 2013. Improved durability of recycled porous
asphalt. Construction and Building Materials, 48: 755–763.
Frigio, F., Pasquini, E., Partl, M. & Canestrari, F. 2015. Use of reclaimed asphalt in porous asphalt
mixtures: laboratory and field evaluations. Journal of Materials in Civil Engineering, 27(7).
Hamzah, M.O., Aman, M.Y. & Shahsdan, Z. 2011. Resistance to disintegration of Warm Porous Asphalt
Incorporating Sasobit®. Australian Journal of Basic and Applied Science, 5(9), 113–121.
Kline, L. & Putman, B.J. 2011. Comparison of open graded friction course (OGFC) mix design pro-
cedures in the United States. Proceeding of the 90th Annual Meeting of the transportation Research
Board, Washington D.C.
Little, N.D. & Jones, D.R. 2003. Chemical and Mechanical Processes of Moisture Damage in Hot-Mix
Asphalt Pavements. In Moisture Sensitivity of Asphalt Pavements: A National Seminar, Transporta-
tion Research Board of the National Academies, pp. 37–70.
Mallick, R.B., Kandhal, P.S. & Bradbury, R.L. 2008. Using warm-mix asphalt technology to incor-
porate high percentage of reclaimed asphalt pavement material in asphalt mixtures. Transportation
Research Record: Journal of the Transportation Research Board, 2051:71–79.
Prowell, B. & Hurley, G.C. 2007. Warm-Mix Asphalt: Best Practices. National Asphalt Pavement Asso-
ciation Quality Improvement Series 125. Lanham, MD.
Zhao, S., Huang, B., Shu, X., Jia, X. & Woods, M. 2012 Laboratory Performance Evaluation of Warm
Mix Asphalt containing High Percentages of RAP, Transportation Research Record: Journal of the
Transportation Research Board, 2294:98–105.
38
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Finite Element Method of Lines (FEMOL) is a general and powerful semi-
analytical and semi-discretized method. Based on the newly-developed Energy Element
Projection (EEP) super-convergent method, the self-adaptive analysis, as a modern goal of
numerical computation, has been succeeded in application to FEMOL. The FEMOL and
EEP self-adaptive method are applied to structural analysis of asphalt pavements for the
first time, with only the static and elastic state being considered. Taking the 3D axisym-
metric problem simplified from the asphalt pavement structure under a single axle load as
the model problem, an introduction of self-adaptive FEMOL with EEP strategy was given
firstly. Then numerical calculations were implemented to three mechanical models of pave-
ment. The results show that the adaptive FEMOL can obtain displacements satisfying the
specified tolerance, stresses with super-convergent accuracy, and optimum meshes adaptive
to the mechanical properties of problems, and it is feasible, reliable and superior for analysis
of pavement.
1 INTRODUCTION
As the rapid development of pavement structure design, better numerical computation meth-
ods with high accuracy, fast speed and easy implementation are much required. As a general
numerical method, Finite Element Method (FEM) has been widely used for analysis of pave-
ment, and the commercial software based on FEM like ABAQUS has been used as the main
tool in engineering. However, since the accuracy and convergence of FEM results mainly
depend on the quality of the mesh partition, users are required to have enough knowledge on
mechanics and may spend much time on the adjustment of the mesh. Recently the semi-an-
alytical finite element method (SAFEM) was applied to the analysis of pavement structures
and has obtained superior results (Liu et al. 2014, 2015).
In 1990s, Yuan proposed a new semi-analytical and semi-discretized method called Finite
Element Method of Lines (FEMOL) (Yuan, 1993). In FEMOL, by using finite element
technique and variational principle, partial differential equations (PDEs) governing physical
problems are semi-discretized into a system of ordinary differential equations (ODEs) which
are defined on discrete mesh lines, and then the ODEs can be solved by existing ODE solver,
for example, COLSYS (Ascher et al. 1981). FEMOL has been successfully applied in analysis
of a series of linear and nonlinear problems (Yuan 1993, Zhang & Yuan 1993, Xu & Yuan
1999, Huang et al. 2010).
Currently, a new and modern way of numerical computation has been noticed by more
and more researchers, that is the adaptive analysis, in which users only need to pre-specify
an error tolerance instead of a mesh in conventional solution, and then an adaptive mesh
is automatically obtained by the algorithm, on which the conventional solution satisfies the
specified error tolerance in certain norm. A number of researchers have got achievements on
this subject (Zienkiewicz & Zhu 1992, Babuska et al. 1995) including Yuan and his group. A
set of general, efficient and reliable self-adaptive strategies based on Energy Element Projec-
39
tion (EEP) super-convergent method has been successful for various types of 1D, 2D and 3D
linear and nonlinear problems (Yuan et al. 2006, 2008, 2011, 2014).
The semi-discrete technology of FEMOL, the EEP super-convergent approach, and the
self-adaptive strategy based on EEP method can naturally meet the computational require-
ment of road structure analysis. In the present paper, they were applied to the asphalt pave-
ment structure for the first time. Taking the 3D axisymmetric problem governing a common
asphalt pavement structure under a single axle load as the model problem, an introduction
of self-adaptive FEMOL based on EEP method was given. Then numerical calculations were
implemented to the following mechanical models of pavement structures under the static and
elastic state: the axisymmetric layered problems with finite region and with infinite region,
and the 3D problems with finite region. The numerical results show that FEMOL has per-
formed very well on mechanical analysis of asphalt pavement structures.
40
Figure 2. The solution domain and a possible FEMOL mesh of the model problem.
u(ξ ,η ) Nd e (1)
with
1
2 ∫Ω
Π( ) ( ) D( ) − ∫ f T u dA − ∫ q T u r ds (3)
Ω Sσ
41
where LS is the differential operator matrix in the geometric equation of 3D axisymmetric
problem, D is the elastic matrix of 3D axisymmetric problem; q = [qr qz]T is the distributed
force loaded on the boundary of the domain which is denoted by Sσ, f = [fr fz]T is the body
force vector which equals to zero for the model problem and is still kept without loss of
generality.
Substituting Equation (1) into Equation (3) and taking the element e as the integration
domain, the element energy functional ∏e(de) can be obtained, by assembling which the global
energy functional can be calculated, i.e. ∏(d) = ∑∏e(de). Based on the variational principle, let-
ting δII(d) = 0 will lead to a system of second-order ODEs with respect to the global nodal
line displacement vector as follows:
Ad ′′ + Gd ′ + Hd + F 0 (4)
with the corresponding boundary conditions. The detailed forms of all the matrice and vec-
tor in Equation (4) and the boundary conditions can be found in Fang’s thesis (Fang, 2011).
Finally Equation (4) is solved by COLSYS (Ascher et al. 1981), which uses the adaptive spline
collocation method to solve ODEs and finally gives a solution of d which satisfies the error
tolerance pre-specified by users. And then the FEMOL displacement solution at any point
of any element can be computed with Equation (1), which is denoted as uh in the following
sections. For other types of problems, FEMOL applies the similar idea and algorithm (Yuan,
1993).
( *
La ((u h )L a + 2 ( ) 1
(N 2aa ∫
ξ
ξa
N r h Jr d N a∫
ξa
ξ1
N 2 r h J dξ ) (5a)
( p* ) L a ( p h )L a
1
(c c )
ξ2
+ ⎡ I + 2( T 1
) 1 1
− b1T b1−1 ( b1η + c1 )c 2 1 + c1T c 2−1 N 2 a ⎤⎦ ∫ N1r h J dξ
rJ L a ⎣ 1 1 2 2 ξ ξa
1
(c c )
ξa
− ⎡ I − 2( I + b1T b1 1 ) 1 1
b1T b1 1 ( b1η + c1 )c 2−11 + c1T c 2−1 N1a ⎤⎦ ∫ J dξ
N 2 r h Jr
rJ L a ⎣ 2 2 ξ ξ1
(5b)
where the subscript La means that the value is taken at the points on La, the superscripts ‘*’
and ‘h’ represent EEP super-convergent solution and FEMOL solution respectively; p = [pr
pz]T is the stresses corresponding to u. The meanings of other notifications and the detailed
derivation process are omitted for simplification. Since Equation (5) is derived based on an
arbitrary element, it can be directly used for the elastic layered model problem.
42
The super-convergent solution of EEP method has many superior features (Yuan et al.
2007): (i) High accuracy. The convergence order of EEP solutions are at least one order
higher than that of FEMOL solutions for both displacements and stresses, (ii) Continuous
stresses. The two components of stress vector p are continuous on the common mesh line and
side of two adjacent elements, (iii) Simple implementation. The super-convergent calculation
is carried out in the post-processing stage of FEMOL, with only an integration term being
added to the conventional FEMOL solution. Therefore EEP method has shown great advan-
tages among the super-convergence algorithms.
max u u h T a w − wh
max T (6)
Since u is practically unavailable, and the EEP solution u* gains higher order convergence
than uh, u* is used to replace u to control errors, leading to the following criteria:
max u* u h ≤ T a w* − w h
max T (7)
e e
If certain element does not satisfy Equation (7), it will be subdivided into two elements
using the error-averaging method (Fang, 2011). At last Equation (7) holds true for all the ele-
ments on the current mesh. This is the basic idea of EEP self-adaptive method. It should be
noticed that for the layered system shown in Figure 2, a line going through all the layers from
z = 0 to z H should be inserted to refine the mesh. That is to say, if element <2> did not
satisfy Equation (7) and should be subdivided, element <4> and <6> would be subdivided as
well. Therefore, the proposed adaptive FEMOL based on EEP method can briefly be sum-
marized as the following triple-step procedure (Fang, 2011):
1. FEMOL solution. On the current mesh (the initial mesh is given by the user and a quite
rough one is all right in general), the FEMOL solution uh is obtained.
2. Super-convergent solution. The EEP solution u* is calculated using Equation (5a), and the
maximum error of each displacement component is calculated on each element.
3. Mesh refinement. For those elements that Equation (7) is not satisfied, the error-averaging
method is used to insert the new nodal lines, and consequently a new refined mesh is
obtained. Then the first step is repeated again until all elements satisfy Equation (7).
It can be seen that EEP self-adaptive technology shows many advantages in application: (i)
The optimal mesh which is adaptive to the physical and mechanical properties of problems
are automatically obtained by the algorithm program with little redundancy, (ii) The tradi-
tional FEMOL solution of displacements satisfies the error tolerance T pointwise on the
final mesh, and super-convergent stresses with high accuracy can be obtained using Equation
(5b), (iii) The EEP self-adaptive analysis can be easily implemented based on the simple and
clear idea which is also suitable for the adaptive analysis of FEM.
4 NUMERICAL EXAMPLES
In this section, four numerical examples are given corresponding to the following three
mechanical models of asphalt pavement structure respectively, i.e. the 3D axisymmetric lay-
ered problem with finite region, the 3D axisymmetric layered problem with infinite region
and the 3D problem with finite region. For the first three examples conventional FEMOL is
43
Table 1. Geometrical data and material properties of the pavement.
Figure 5. Comparisons of stresses along I (upper) and II (lower) between FEMOL and ABAQUS on π0.
used, and for the fourth one the efficiency of EEP adaptivity is demonstrated. The geometri-
cal data and material properties of the asphalt pavement (Liu et al. 2014, 2015) are given in
Table 1, and the totally contacted case was considered. Computations were carried out with
a set of Fortran90 code of FEMOL and the results of ABAQUS on certain fine mesh were
used for comparison and examination. The error tolerance for COLSYS was always set to
be 10–5. In addition, for convenience of presentation, denote the top of the asphalt surface
course as I and the op of the road base course as II.
44
Figure 6. The mesh on infinite solution domain.
1 0.1805
2 0.2157
3 0.2205
4 0.2209
5 0.2210
Figure 9. Comparisons of the amount of settlement and stress σx between FEMOL and ABAQUS.
The solution mesh in FEMOL is shown in Figure 6 with totally 33 elements where the
arrowed lines are nodal lines and the others are end-sides. Table 2 givens the amount of set-
tlement at the location directly below the loading center (r z = 0) calculated with different
degrees of elements in FEMOL. It shows that with the increasing of degree of elements,
FEMOL solution approaches to 0.22mm. Setting the depth of sub-grade to be 100m in order
to approach the infinite model, ABAQUS resulted a settlement of 0.215mm at the same loca-
tion. Obviously quadratic elements in FEMOL obtained results of nearly the same accuracy
as ABAQUS. Comparisons of the horizontal stress σr and the vertical stress σz along I from
FEMOL with cubic elements and ABAQUS are shown in Figure 7. Both displacements and
stresses of FEMOL are satisfactory.
46
Figure 11. The distributions of errors of uh (left) and wh (right) with respect to ABAQUS (mm).
Figure 12. Comparisons of stresses resulted from super-convergent FEMOL and ABAQUS along I.
4.4 The 3D axisymmetric model with finite region (using EEP adaptivity)
Using π0 in Figure 4 as the initial mesh and setting the error tolerance to be T = 10 −4 , EEP self-
adaptive solution finally obtained the adaptive mesh π of 90 elements shown in Figure 10.
Actually only 13 lines were inserted in r-direction which went through the six layers. The
smallest element is of 0.0194m in width and is right on the left of the line r = 0.15m, i.e. the
adaptive mesh is automatically ‘adaptive’ to the physical property of the problem itself.
Computed on the above mesh of 82,500 elements, displacement results u A = [ A A]
T
from ABAQUS were used to check the accuracy of adaptive FEMOL solution. Figure 11
47
Figure 13. Comparisons of stresses resulted from super-convergent FEMOL and ABAQUS along II.
shows the distributions of errors with respect to ABAQUS solutions. It is obvious that the
adaptive displacement solution of FEMOL pointwise possesses an accuracy satisfying the
tolerance T = 10−4. On the adaptive mesh π super-convergent stresses were obtained from
Equation (5b). And all the four stress components from EEP method were compared with
ABAQUS along I and II, which are illustrated in Figure 12 and Figure 13. It can be easily
found that both FEMOL and adaptive FEMOL are efficient and effective.
5 CONCLUSION
The present paper reported the first try to apply FEMOL and EEP self-adaptive solution in
analysis of pavement structures in engineering and of the layered elastic system in mechanics.
No matter EEP adaptive method is used for FEMOL or not, superior results can be obtained.
It is worth mentioning that the adaptive calculation may not be essential for each problem.
When model problems are of similar mechanical properties, the final adaptive meshes may
have similar distribution, among which any one could meet the requirement of engineering
practice. Further research results will be published in other papers in future.
ACKNOWLEDGEMENTS
This work is part of the three projects supported by the National Natural Science Foundation
of China (Nos. 51378293, 51508305 and 51078199).
REFERENCES
Ascher, U. & Christiansen, J. & Russell, R.D. 1981. Collocation software for boundary-value ODEs.
ACM Transactions on Mathematical Software (TOMS), 7(2): 209–222.
Babuska, I. & Strouboulis, T. & Upadhyay, C.S., et al. 1995. A posteriori estimation and adaptive control
of the pollution error in the h-version of the finite element method. International Journal for Numeri-
cal Methods in Engineering, 38(24): 4207–4235.
Fang, N. 2011. Adaptive FEMOL Analysis of Elastic Problems Based on EEP Super-Convergent
Method. Ph. D. dissertation, Tsinghua University, Beijing (in Chinese).
Huang, Q. & Dai, Y. & Gao, J., et al. 2010. The application of axisymmetric element of finite element
method of lines in the analysis of thermal field. Engineering Mechanics, 27(S1): 1–5 (in Chinese).
Liu, P. & Chen, X. & Oeser, M., et al. 2014. Application of semi-analytical finite element method in
analysis of asphalt pavement structural response. In: Proceedings of 12th International Conference on
Asphalt Pavements, Raleigh.
48
Liu, P. & Wang, D. & Oeser, M. 2015. Application of semi-analytical finite element method coupled
with infinite element for analysis of asphalt pavement structural response. Journal of Traffic and
Transportation Engineering (English Edition), 2(1): 48–58.
Pang, Z. 1993. An error estimate of the finite element method of lines. Computational Mathematics,
15(1): 110–120.
Xu, Y. & Yuan, S. 1999. Calculation of the stress intensity factor based on the singular line mapping
technique of finite element method of lines. Engineering Mechanics, 16(S1): 85–91 (in Chinese).
Yuan, S. & Du, Y. & Xing, Q., et al. 2014. Self-adaptive one-dimensional nonlinear finite element
method based on element energy projection method. Applied Mathematics and Mechanics (English
Edition), 35(10): 1223–1232.
Yuan, S. & Fang, N. & Wang, X., et al. 2011. New Progress in self-adaptive analysis of two-dimensional
finite element method of lines. Engineering Mechanics, 28(3): 1–8 (in Chinese).
Yuan, S. & He, X. 2006. A self-adaptive strategy for one-dimensional FEM based on EEP method.
Applied Mathematics and Mechanics (English Edition), 27(11):1461–1474.
Yuan, S. & Wang, M. & Wang, X. 2007. An element-energy-projection method for super-convergence
solutions in two-dimensional finite element method of lines. Engineering Mechanics, 24(1): 1–10 (in
Chinese).
Yuan, S. & Xing, Q. & Wang X. & Ye, K. 2008. Self-adaptive strategy for one-dimensional FEM based
on EEP method with super-convergence order. Applied Mathematics and Mechanics (English Edi-
tion), 29(5):591–602.
Yuan, S. 1991. Finite element method of lines for 3D elasticity, Technical report to British Aero Space.
Sir George Ceylay Institute, London.
Yuan, S. 1993. The Finite Element Method of Lines. Beijing-New York: Science Press.
Zhang, Y. & Yuan, S. 1993. Analysis of nonlinear model problems by the finite element method of lines-
III. free vibration of membranes. Engineering Mechanics, 10(3):1–8 (in Chinese).
Zienkiewicz, O.C. & Zhu, J. 1992. The super-convergence patch recovery (SPR) and a posteriori error
estimates, Part 1: The recovery technique, Part 2: error estimates and adaptivity, International Journal
for Numerical Methods in Engineering, 33(7): 1331–1382.
49
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Xudong Wang
Research Institute of Highway Ministry of Transport, Beijing, P.R. China
Sheng Li
Changsha University of Science and Technology, Changsha, P.R. China
1 INTRODUCTION
Asphalt mix is a typical viscoelastic material, whose mechanical property is affected by tem-
perature, frequency and load level (ZHANG 2006, SHEN 2001). Pavement structure is sub-
jected to alternative processes of traffic load and environment change, and static modulus is
insufficient to describe its mechanical behaviour. Static modulus is still in use in China, which
is part of the reason why some roads dysfunction at early stage even if they are designed and
constructed according to the norms and regulations. Dynamic modulus can better represent
the dynamic mechanical properties of asphalt mix, which is adopted by most foreign design-
ing methods as one of the most important input, (i.e. MEPDG (AASHTO 2008)) (WEI
2008, LUO 2010).
There are many different ways to conduct dynamic modulus tests, including uniaxial com-
pression, indirect tension, four-point bend, and semi-circle bend (Yao 2011, MA 2008, Liu
2009, UNDERWOOD 2011). Doctor WANG et al found that dynamic modulus is affected
by test methods (WANG 2002, Zhang 2011, LI 2014, MA 2009). Two point bend is a typical
method to conduct dynamic modulus test, and is widely used in the EU, but rarely used in
China. It employs trapezoid beam specimen, the bottom of which is fixed on the base, and
the top is subjected to continuous sinusoid horizontal load, under stress control or strain
51
control mode. This method not only investigates the affects of temperature or frequency on
modulus, but also the stress level or strain level on it. In the test, the specimen is subjected
to combined action of bend and shear, and the stress condition is close to real situation. On
the other hand, SHRP has carried out projects to explore appropriate dynamic modulus
method since 1995, and SPT (Simple Performance Tester) was developed. SPT uses cylindri-
cal specimen to conduct uniaxial compression. The diameter-height ratio is about 1:1.5 or
1:2. Haversine or partial sine wave is loaded at different temperatures or frequencies. Uniaxial
compression is one of the most widely used methods for its simplicity, stability and effective-
ness (A Guide, 2010).
This paper used two-point bend and uniaxial compression methods to conduct dynamic
modulus on four asphalt mixes. Master curves of dynamic modulus and phase angle were
plotted to analyse the dynamic mechanical property of asphalt mix, and the two methods
were compared and evaluated.
Asphalt mixes 16 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075 Mineral powder
SAC16-35 100 95.0 68.9 35.0 26.7 20.4 15.7 12.0 9.2 7.0 —
SAC16-55 100 95.0 79.7 55.0 38.9 27.5 19.7 13.9 9.9 7.0 —
52
Figure 2. Sketch of shaping specimen.
Two-point bend 0, 5, 10, 15, 20, 25, 30, 35, 40 100 40, 35, 30, 25, 20, 15, 10
Uniaxial compression 5, 10, 15, 20, 25, 30, 40, 50 90∼110 25, 20, 10, 5, 2, 1, 0.5, 0.2, 0.1, 0.01
For viscoelastic material, the same mechanical property can be obtained at high temperature
∼ high frequency, or low temperature ∼ low frequency, which is called Time-Temperature
Superposition Principle (TTSP). Based on TTSP, master curve of asphalt mix could be plot-
ted. Master curve enlarges test results within limited time and temperature to a larger scale,
and reflects the dynamic mechanical behaviour and viscoelastic property of asphalt mix more
comprehensively (ZHAO 2006, MA 2008).
where y is the logarithm of dynamic modulus, f is the logarithm of frequency, A1, A2, x0, and
dx are parameters describing the shape of the curve.
Every master curve of dynamic modulus determines a unique shift factor lg(αT), through
which could shift the curve of phase angle to form the master curve of phase angle.
As shown in Figures 3∼6, although distinctions exist between them, the trends of dynamic
modulus using two-point bend and uniaxial compression are similar: the value of modulus
decreases as frequency decreases (or temperature increases). The value of phase angle increases
first and then declines as the frequency decreases (or temperature increases). Phase angle is
a parameter represents the viscoelasticity of materials. Generally speaking, smaller value of
phase angle means that material is more elastic. Viscoelasticity of asphalt mix mainly comes
from asphalt material. When temperature is within a particular range, viscidity of asphalt mix
increases as temperature increases, just like the asphalt material behave. But when tempera-
ture exceeds a particular range, asphalt material softens and weakens its bond effect, in which
condition asphalt mix relies more on the aggregate frame, and the viscoelasticity of asphalt
mix is less affected by that of asphalt material. So, the phase angle declined after reaching a
vertex when temperature keeps going up. In both test methods, the affection of gradation is
so puny that it can be ignored. SBS can modify the high-low temperature property of asphalt
and improve the pavement performance of asphalt mix, as the asphalt mix with SBS modified
asphalt shows larger dynamic modulus and smaller phase angle (LIU 2014).
When compared with uniaxial compression, the scope of the master curves of dynamic
modulus and phase angle with two-point bend is smaller. This is mainly because the frequency
54
Figure 5. Master curve of dynamic modulus and phase angle of SBS-SAC16-35.
of uniaxial compression can reach as low as 0.01 Hz, while the lowest frequency of two-point
bend is just 10 Hz.
The value of dynamic modulus with uniaxial compression is slightly larger than that with
two-point bend, while value of phase angle with uniaxial compression is generally smaller
than that with two-point bend. The discrepancy increases as frequency decreases (or tem-
perature increases). For both dynamic modulus and phase angle, the slope of master curve
with two-point bend is larger, meaning that two-point bend test is more sensitive to fre-
quency or temperature. These differences of master curve mainly come from three aspects:
the load mode differences of the two tests, heterogeneity and anisotropy of asphalt mix.
Uniaxial compression hires uniaxial compression load mode, which is beneficial to the speci-
men, as the compressing process will densify the asphalt mix. The interlock of aggregates and
cohesion of asphalt material are strengthened, and stress inside the specimen is transferred
properly. As a result, the anti-transformation ability is stronger, and the dynamic modulus
is larger and more stable. At the same time, as the tightened structure of asphalt mix under
uniaxial compress could effectively transfers and disperses stress through aggregate skeleton,
the viscoelasticity of asphalt mix is less affected by asphalt material, so the phase angle with
uniaxial compression is less influenced by frequency or temperature. In contrast, specimen
in two-point bend test is mainly under bend-shear mode. On the tensile side of the asphalt
mix, the aggregates tend to be separated, weakening the interlock of the aggregate skeleton.
Stresses inside the asphalt mix relies more on the cohesion of asphalt to transfer and disperse,
so the mechanical property is more affected by the asphalt material. However, anti-transfor-
mation ability and thermal stability of asphalt is relatively weak. Therefore, the asphalt mix
55
with two-point bend test shows smaller dynamic modulus and larger phase angle, and is more
sensitive to frequency or temperature changes.
Shift factor is the function of temperature, and its value reflects the temperature sensitiv-
ity of material. As shown in Table 3, the shift factors of asphalt mixes with SBS modified
asphalt are generally smaller than those of asphalt mix using matrix asphalt. It indicates that
SBS improved the temperature sensitivity of asphalt material, and enhances the pavement
performance of asphalt mix.
Temperature (°C)
Table 4. Relationship of dynamic modulus between uniaxial compression and two point bend.
Advantages 1. Combination of bend and shear is closer 1. The process is simple, and easy to learn.
to the real load situation of pavement 2. Frequency scope is large, and low
under traffic load. frequency can be achieved.
2. Control strain level strictly.
3. Two or four specimens can be tested at
a time, saving time.
Disadvantages 1. The preparation for trapezoid beam 1. Unable to control strain level.
specimens and the remove of them 2. Insufficient to guide real projects.
are complicated.
2. Unable to carry out low frequency test.
3. Strain control is manual, increasing the
complexity and time of the test process.
56
correlation coefficients are over 0.999, showing that good relationship exists between the
dynamic modulus with the two test methods.
4 DISCUSSION
Comparing the uniaxial compression and two-point bend dynamic modulus test, it is obvi-
ous that different load modes will affect the value of dynamic modulus and phase angle of
asphalt mix.
The trends of the dynamic modulus master curves of the two test methods are similar, but
the values of dynamic modulus of uniaxial compression are slightly larger than those of two-
point bend. Two-point bend is unable to undertake low frequency test, so the scope of the
master curve is narrow. This deficiency can be remedied by extending the curve through curve
fitting. Dynamic modulus with two-point bend is more sensitive to frequency or temperature,
declining faster as frequency drops. As frequency decreases or temperature increases, the
influence of load modes on dynamic modulus becomes more significant, as the discrepancy
between the values gets larger. Phase angle of the two test methods has similar characteris-
tics, except that the value of phase angle with uniaxial compression is smaller than that with
two-point bend.
Stress states at pavement layers are different in different pavement structures and at differ-
ent depth. Generally speaking, the upper layers are subjected to compression, while the lower
layers are subjected to bend and tension. Uniaxial compression and two-point bend have
their own features, and should be selected properly for different parts of pavement according
to their stress state. It is recommended that uniaxial compression should be used for structure
design and material evaluation for upper layers, while two-point bend should be used for
structure design and material evaluation for lower layers.
5 CONCLUSIONS
Dynamic modulus tests of four asphalt mixes were conducted by two test methods. Master
curves, as well as equipment capacity and test process were systematically compared and
evaluated. Main conclusions are so follows:
The frequency of two-point bend is generally high, from 10 ∼ 40 Hz. The scope of master
curve is narrow because its disability at low frequency. Strain level can be strictly controlled.
Master curve of uniaxial compression is more comprehensive, as the frequency can be as low
as 0.01 Hz.
The trends of master curves for the two methods are similar, indicating that both methods
can rightly reflect the dynamic mechanical characteristic of asphalt mix.
Values of dynamic modulus of two-point bend are smaller than those of uniaxial compres-
sion, while values of phase angles of two-point bend are larger. Dynamic modulus and phase
angle of two-point bend are more sensitive to frequency or temperature. Load modes influ-
ence the dynamic modulus of asphalt mix. But the influence changes with frequency or tem-
perature: when frequency is low or temperature is high, the difference is more significant.
Good correlation relationship exists between the two test methods, and numerical conver-
sion is well achieved by second-order polynomial equation.
SBS can improve the viscoelasticity of asphalt material, and enhance the pavement per-
formance of asphalt mix.
57
This paper is expected to provide an important reference for dynamic mechanical study of
asphalt mix and asphalt pavement design.
ACKNOWLEDGEMENT
The authors acknowledge the assistance of National Natural Science Foundation of China
(51178062); Key Basic Research Program sponsored by Ministry of Transportation of PRC
(2012 319 223 010); Hunan Provincial Natural Science Foundation (14JJ7041); and Higher
School Science Research Project in Hunan Province (14B001); Open Found of Key Lab-
oratory of Road Structure and Material of Ministry of Transport (Changsha University
of Science & Technology) (kfj150304); and Open Found of State Engineering Laboratory
of Highway Maintenance Technology( Changsha University of Science & Technology)
(kfj150102).
REFERENCES
A Guide of France Asphalt Mixtures Design, 2010. Nanjing: Work Group of France RST Asphalt Mix-
tures Design, translated by Project Group of Comparative Research of Asphalt Pavement Technol-
ogy of PRC, US and Fr. (in Chinese).
Li Qiang, LI Guofen, Wang Hongchang, 2014. Effect of Loading modes on Dynamic Moduli of
Asphalt Mixtures. Journal of Building Materials, 17(5): 816–822. (in Chinese).
Liu Yu, 2009. Research on Dynamic Response and Fracture Performance Based of Asphalt Mixture on
Semi-circular Bend Test. Dissertation of Harbin Institute of Technology, Haribin. (in Chinese).
Liu Zhaohui, Zhang Jingyi, Zhou Ting, Li Sheng, 2014. Development and Performance Evaluation of
Composite Modified Asphalt Material for Tact Coat. Materials Review, 28(2): 134–139. (in Chinese).
Luo Sang, Qian Zhendong, Harvey J., 2010. Research on Dynamic Modulus for Epoxy Asphalt Mix-
tures and Its Master Curve. China Journal of Highway and Transport, 23(6): 16–20. (in Chinese).
MA Lin, Zhang Xiaoning, 2009. Comparison of HMA Dynamic Moduli between Indirect Tension and
Uniaxial Compression Test Modes. Journal of Highway and Transportation Research and Develop-
ment, 26(10): 11–17. (in Chinese).
MA Lin, Zhang Xiaoning, Chen Shaoxing, 2008. Dynamic Modulus of Asphalt Mixtures Based on
Indirect Tension Mode. Journal of South China University of Technology: Natural Science, 36(10):
86–91. (in Chinese).
MA Xiang, NI Fujian, Chen Rongsheng, 2008. Dynamic modulus test of asphalt mixture and predic-
tion model. China Journal of Highway and Transport, 21(3): 35–39. (in Chinese).
Simple Performance Tests: Summary of Recommended Methods and Database. Washington D.C.
NCHRP Report 547.
Shen Jinan, 2001. Asphalt and Asphalt Mixture Pavement Performance. Beijing: China Communication
Press. (in Chinese).
Standard Test Methods of Bitumen and Bituminous Mixtures for Highway Engineering. Beijing: China
Communication Publication, 2011. (in Chinese).
Underwood S B, Kim R Y., 2011. Viscoelastoplastic of Continuum Damage Mode for Asphalt Concrete
in Tension. Journal of Engineering Mechanics, 137(11): 732–739.
Wang Xu-dong, ShA Ai-min, Xu Zhi-hong, 2002. Dynamics Characteristics and Dynamic Parameters of
Asphalt Pavement Materials. Beijing: China Communications Press. (in Chinese).
Wei Jin-cheng, CUI Shi-ping, HUJia-bo, 2008. Research on Dynamic Modulus of Asphalt Mixtures.
Journal of Building Materials, 11(6): 657–661. (in Chinese).
Yao Bo, Cheng Gang, Wang Xiao, 2011. Dynamic modulus of epoxy asphalt mixture based on bend
test mode. Journal of South East University: Natural Science, 41(3): 597–600. (in Chinese).
Zhang Xiaoning, 2006. Viscoelastic Mechanics Principle and Application of Asphalt and Asphalt Mix-
tures. Beijing: China Communication Press. (in Chinese).
Zhang Lei, Wang Xudong, Zhou Xingye et al. Laboratory Evaluation of Dynamic Complex Modulus
for HMA with Varying Asphalt Content. in: Geotechnical Special Publication No. 212: pavements
and materials, ASCE, 2011.
Zhao Yanqing, Wu Jian, Wen Jian, 2006. Determination and Analysis of Dynamic Modulus of Asphalt
Mixture and Its Master Curve [J]. Highway, 8: 163–166. (in Chinese).
58
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
A. Mathruswamy
Gitam University, Telangana, India
1 INTRODUCTION
59
simple shear tests (Sousa et al. 1991). There also exists a specification for the measurement
of shear modulus using a Superpave Shear Tester (AASHTO TP7). Poisson’s ratio of bitu-
minous concrete has not been investigated to the extent of Young’s modulus or even shear
modulus. There have been some techniques proposed for its measurement, but the Poissons’
ratio thus measured has been observed to vary over a wide range depending on the technique
used (Di Benedetto et al. 2007).
Bulk modulus of bituminous concrete has received the least attention in the literature
amongst the above stated material functions. But for a few studies that were conducted as
a part of Superpave performance evaluation (Di Benedetto et al. 2001), hardly any experi-
mental investigation has addressed the compressibility of bituminous concrete. A rigorous
experimental investigation of the time-dependent compressible behavior of bituminous con-
crete would thus be valuable and this is the objective of this study. Constant hydrostatic pres-
sure was applied in the absence of any deviatoric stress to observe the creep of volumetric
strain and the pressure was suddenly removed to study the recovery. The results could be
used to determine the bulk modulus of bituminous concrete if it is an applicable or relevant
parameter, or it could be used to determine the limits of compressibility as observed by some
researchers.
2 MATERIALS
2.1 Aggregates
The aggregate used in the study is crushed granite obtained from a quarry in Chennai, India.
Aggregate gradation conforming to Bituminous Concrete (BC) Grade 2 as per the MORT&H
specifications (Ministry of Road Transport and Highways, 2013) with a nominal maximum
size of 13.2 mm, was chosen for the bituminous mixtures. The limits of BC Grade 2 grada-
tions are presented in Table 1. Figure 1 shows the gradation of the mix.
Sieve size (mm) 19 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075
60
2.2 Binders
An unmodified binder of Viscosity Grade 30 (VG-30), a styrene-butadiene-styrene modified
binder (PMB-E), a polyolefin modified binder (PMB-P) and a crumb rubber modified binder
(CRMB), were used for the investigation. All the binders were manufactured by M/s Hindus-
tan Petroleum Corporation. The dosage of modifier used for each of the modified binder is
presented in Table 2. The properties of the unmodified and the modified binders tested in
accordance with the Indian specifications IS:73–2013 and IS 15462:2004, respectively, are
presented in Table 3 and Table 4.
Properties
61
samples of diameter 93 ± 0.5 mm were first cored from each beam. From each cylindrical
specimen, 10 mm was sawed off using a thin-saw arrangement so as to ensure that the end
horizontality was within the tolerance limit of ≤ 0.5 mm and end perpendicularity was ≤
1.0 mm. Each cylindrical sample was 93 ± 0.5 mm in diameter and 150 ± 0.5 mm high. Speci-
mens were prepared with an air void content of 6 ± 0.5%.
3 EXPERIMENTAL METHODOLOGY
Experiments were conducted using an Asphalt Mixture Performance Tester (AMPT) manu-
factured by IPC, Australia. The equipment is capable of conducting triaxial tests at isothermal
conditions. It can apply confinement pressure up to 300 kPa, an axial load up to 15 kN. The
equipment can maintain the temperature within the triaxial test chamber between 0 to 60° C
with an accuracy of 0.5°C. Figure 2 shows the experimental set-up. Metal studs were glued
on to the sample at three different locations that are 120° apart for fixing three linear vari-
able differential transducers (LVDTs). The specimen was kept sandwiched between two steel
plates and was covered by a tight polymer membrane for the purpose of applying confinement
pressure. The membrane was fixed to the top and bottom plates with two O-rings. The three
LVDTs were fixed on top of the membrane using the metal studs that were earlier glued to the
sample. All the LVDTs were oriented along the axial direction as shown in Figure 2.
The tests were conducted at a temperature of 60° C. This temperature was so chosen so
that there is significant volumetric strain observed during the test. Volumetric strain in creep
recovery tests is expected to be higher at higher temperature and 60°C is the highest tempera-
ture that can be applied with this particular equipment. The specimens were conditioned at
the test temperature for a minimum of 4 hours inside the triaxial testing chamber, prior to
loading. The tests were so designed to observe the decrease in volumetric strain with time
when a constant confinement pressure is applied and the increase in volumetric strain when
the confinement pressure is removed. In each such creep-recovery test, the confinement pres-
sure was ramped up to the desired magnitude in the first 40 seconds and was retained for
3000 seconds and then ramped down to 0 kPa in 40 seconds. The strain recovery in the
specimen was observed over the next 3000 seconds. The loading history corresponding to
this test is presented in Figure 3. The axial deformation of the three LVDTs and the applied
confinement pressure were recorded every one-thousandth of a second by the data acquisi-
Figure 2. Experimental set up for pure confinement creep and recovery experiment.
62
Figure 3. Schematic representation of the confinement pressure creep-recovery test.
Figure 4. Axial strain history of bituminous mixture with plastomer at 100 kPa pressure.
tion system. The axial deformation divided by the gauge length would give the axial strain in
each location. The average of the axial strain values obtained at the three locations was taken
as the axial strain of the specimen.
Creep-recovery tests in confinement were conducted at 100 kPa and 200 kPa confinement
pressures for all binders. Test at each confinement level was conducted twice to ensure repeata-
bility. As an example the axial strain history obtained when the bituminous concrete with plas-
tomer modified binder was subjected to creep-recovery test at 100 kPa confinement pressure is
shown in Figure 4. The repeatability of the creep-recovery tests were found to be satisfactory.
The coefficient of variation between the three LVDT’s is found to be less than 30% and this is
acceptable since for such similar tests the acceptable range is 30% (AASHTO TP 79–10).
where ∈v is the volumetric strain and ∈a is the axial strain. The creep and recovery of the
volumetric strain of each binder at different levels of confinement pressure are presented in
Figure 5. At 200 kPa confinement pressure, the maximum creep volumetric strain ranged
from about 4000 microstrain to 7000 microstrain which corresponds to a change in air-
void content of about 0.4% to 0.6%. The permanent change in the air void content ranged
between 0.1% and 0.4%. There have been observations in the literature that there is a limit
to the permanent volume change above which all the permanent deformation is completely
due to shear flow. Considering that the air void content is known to reduce by at least 2%
in highways, and the permanent decrease in air void content is not more than 0.4% in these
tests, it is reasonable to regard that the existence of a maximum limit on air void reduction
would not cause a nonlinearity in the volumetric response observed in this study.
A common trend found among all the bituminous concrete specimen is that during creep
a significant portion of the increase in volumetric strain occurred during the 40 second ram-
ping period. At 200 kPa confinement pressure, almost 68% of the total volumetric creep
Figure 5. Comparison of volumetric strain of different binders at 100 kPa and 200 kPa pressure.
64
strain of the bituminous concrete with crumb rubber modified binder occurred during the
ramping time. Similarly, almost 74% of the increase in volumetric strain of the bituminous
concrete with the unmodified binder occurred during the ramping time. This can be regarded
as the instantaneous elastic compressibility of the specimens. The decrease in volumetric
strain in the 40 seconds of time when the confinement pressure is ramped down to zero is
only between 32–60% of the increase in the volumetric strain during the ramping up of con-
finement pressure for all the bituminous concrete specimens. Had the mechanical behavior
been viscoelastic, the entire instantaneous increase in strain at the start of the creep loading
should be recovered instantaneously during unloading. This is because the accruement of
permanent deformation in viscoelastic materials is a time dependent phenomenon. Thus, the
difference between the instantaneous increase in strain and instantaneous recovery observed
for these bituminous mixtures illustrates the existence of time-independent mechanisms of
permanent deformation such as plastic behavior, in addition to the viscoelastic time-depend-
ent mechanisms. The mechanical behavior of bituminous mixtures should therefore be not
regarded as viscoelastic but possibly as visco-elastic-plastic. Results at 100 kPa and 200 kPa
confinement pressures are as presented in Table 5 and Table 6 respectively.
For all the bituminous mixtures, the rate of increase in strain appears to reach a constant
rate. This constant rate of increase of volumetric strain was measured at both confinement
pressures and are shown in Table 5 and 6. At both the confinement conditions the bituminous
concrete specimens with the plastomer modified binder was found to possess the highest rate of
increase of strain. For the bituminous mixtures with binders modified by crumb rubber, elas-
tomer and plastomer, the slope of the creep curve measured towards the end of the creep part
of the test was found to increase with the increase in confinement pressure. But for the bitumi-
nous mixture with the VG-30 binder the slope was higher at the lower confinement pressure.
When the percentage recovery of strain at the end of the creep recovery test was analyzed, it
was found to vary with the applied confinement pressure. At the lower confinement pressure,
for the bituminous mixtures prepared with all the binders except that with CRMB, more than
65
Figure 6. Check for linearity of creep-recovery response.
50% of the maximum strain reached by the sample during loading remained as permanent
strain at the end of the recovery period. Bituminous mixtures with CRMB had 34% of the
maximum strain left unrecovered at the end of recovery. At higher confinement, for bituminous
mixtures with CRMB and plastomer modified binder, at least 50% of maximum creep strain
remained as permanent strain whereas for mixtures with unmodified binder and elastomer
modified binder the permanent strain was between 15–45% of the maximum creep strain.
The bituminous concrete specimens were prepared with the same content of bitumen and
compacted to the same air-voids content for all the four bitumens used. But the volumetric
strain response in the creep-recovery tests was found to vary considerably from binder to
binder. Linearity of the volumetric strain response was examined by normalizing the strain
response of each binder at each confinement level by dividing it by the applied confinement
pressure. These were plotted for each binder as shown in Figure 6. As is evident from the
figures, only the bituminous concrete with the crumb rubber modified binder exhibited a
response that is close to linearity. Bituminous concrete with all the other modified binders
are seen to exhibit nonlinear response. Particularly, for both the bituminous concrete with
66
the unmodified binder and that with the elastomer modified binder, the creep of volumetric
strain appears to be nearly independent of the applied confinement pressure.
On closer inspection of the volumetric strain response of the bituminous concrete with
unmodified binder and that with elastomer modified binder, it can be seen that there is a
small but gradual increase in volumetric strain with time. Thus, the near independence of the
volumetric strain response on the applied confinement pressure is not because of the volu-
metric strain reaching the maximum compressibility but due to some other reason.
Since the binder content and the air void content of all the bituminous concrete specimen
are the same, it is possible that the dependence of volumetric strain response on the con-
finement pressure has similar functional form for all the binders. In other words, although
the volumetric strain response of the bituminous concrete with the crumb rubber modified
binder is proportional to the confinement pressure in the tested range of confinement pres-
sures, at higher confinement pressure the volumetric strain response may become nonlinearly
dependent on the confinement and eventually become independent of the confinement pres-
sure. Similarly, the response of the bituminous concrete with the unmodified binder could be
nearly independent of applied confinement pressure over the 100 kPa to 200 kPa range but
may linearly depend on the confinement pressure at lower confinement levels. The markedly
different response shown by different binders may simply be due to each binder possessing a
different threshold above which the response is nonlinear and another threshold above which
the response is independent of stress.
5 CONCLUSIONS
In this study, pure constant hydrostatic pressure was applied on various bituminous con-
crete mixtures for a fixed period of time and then removed in order to observe the creep
and recovery of volumetric strain in the specimen. Bituminous concrete with four different
bitumen were studied with this test, and considerable variation was observed in the volumet-
ric strain response from one bitumen to another. The bituminous concrete with the crumb
rubber modified binder was found to exhibit nearly linear response, while the volumetric
strain response of the bituminous concrete with the plastomer modified binder was found to
nonlinearly depend on the applied confinement pressure, and the bituminous concrete made
with the unmodified bitumen and the elastomer modified binder exhibited volumetric strain
response that is nearly independent of confinement pressure. It was found that a major part
of the permanent volume change during the test was due to a time-independent inelastic
response such as a plastic response rather than a viscoelastic response. However, a part of the
increase in volumetric strain during the creep period was also time-dependent.
The results illustrate that the three-dimensional mechanical behavior of bituminous con-
crete cannot be regarded as simply viscoelastic. At least a part of the three-dimensional
response comprises of time-independent inelasticity similar to granular materials. Poisson’s
ratio or bulk modulus therefore may not be appropriate parameters for characterizing the
compressible behavior of bituminous concrete.
ACKNOWLEDGEMENT
The authors thank the Department of Science and Technology (India) for funding this inves-
tigation. The grant number is DST/TSG/STS/2011/46.
REFERENCES
AASHTO R30. 2002. Standard practice for mixture conditioning of Hot Mix Asphalt. American Asso-
ciation of State Highway Transportation Officials. Washington DC.
67
AASHTO TP 7. 2001. Standard test method for determining the permanent deformation and fatigue
cracking characteristics of Hot Mix Asphalt (HMA) using the Simple Shear Tester (SST) device.
American Association of State Highway Transportation Officials. Washington DC.
AASHTO TP 79–10. 2010. Determining the Dynamic Modulus and Flow Number of the Hot Mix
Asphalt (HMA) using the Asphalt Mixture Performance Tester (AMPT). American Association of
State Highway Transportation Officials. Washington DC.
ASTM D7981–15. 2015. Standard practice for compaction of prismatic asphalt specimens by means of
the shear box compactor. ASTM International. Washington DC.
Di Benedetto, H., Delaporte, B. & Sauzeat, C. 2007. Three dimensional linear behavior of bituminous
materials. International journal of Geomechanics. 7:149–157.
Di Benedetto, H., Partl, M.N., Francken, L. & De La Roche Saint Andre. 2001. Stiffness testing for
bituminous materials. Materials and Structures. 34: 66–70.
Gibson, N., Sean, X. Li, & Kutay, M.E. 2010. Rutting resistance of laboratory-prepared and field-
compacted asphalt mixtures. Journal of the Transportation Research Board. Transportation Research
Record. 2181.1: 109–116.
Graziani, A., Maurizio B., & Francesco C. 2013. Complex Poisson’s ratio of bituminous mixtures.
Materials and Structures. 47:1131–1148.
IS 73:2013. Indian standard paving bitumen-specification.2013.Bureau of Indian Standard. New Delhi
IS 15462:2004. Indian standard polymer and rubber modified bitumen-specification.2004. Bureau of
Indian Standard. New Delhi.
Kassem, E., Grasley, Z.C. & Masad, E. 2011. Viscoelastic Poisson’s ratio of asphalt mixtures. Interna-
tional Journal of Geomechanics. 13(2):162–169.
Kim, J., Lee, H.S. & Kim, N. 2010. Determination of shear and bulk moduli of viscoelastic solids from
indirect tension creep test. Journal of Engineering. 136: 1067–1075.
Lee, H.S. & Kim, J. 2009. Determination of viscoelastic Poisson’s ratio and creep compliance from the
indirect tension test. Journal of Materials in Civil Engineering. 21: 416–425.
MoRTH. 2013. Specifications for roads and bridge works. Indian Roads Congress. New Delhi.
Sousa, J.B., Craus J. & Monismith, C.L., 1991, Summary report on permanent deformation in asphalt
concrete, SHRP-A/IR-91–104, Strategic Highway Research Program. National Research Coucil.
Washington D.C.
Sousa, J.B. & Weissman, S.L. 1994. Modeling permanent deformation of asphalt—aggregate mixes.
Proceedings of Association of Asphalt Pavement Technologists. 63: 224–245.
Witczak, M.W., Kaloush, K., Pellinen, T., El Basyouny, M., & Quintus, H.V. 2002. NCHRP Report
465: Simple performance tests for super pave mixture design. National Cooperative Highway Research
Program. Transportation Research Board. Washington DC.
Zak, J., Stastna, J., Vavricka, J., Milackova, K., Kasek, L. & Zanzotto, L. 2015. Poisson’s ratio of hot
asphalt mixtures determined by relaxation and small amplitude oscillation test. Journal of Testing
and Evaluation. 43:2.
68
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Anastasios Tsiknas
Hellenic Open University, Patras, Greece
Adamantia Athanasopoulou
Hellenic Open University, Patras, Greece
Athens Metropolitan College (AMC), Athens, Greece
ABSTRACT: This paper aims to present guidelines towards the economically optimal
design method for flexible pavements based on specific design parameters and in compliance
to Greek national and European technical regulations concerning road works. The study
considers different design methods for flexible pavements which are most commonly used in
Greece, namely the Asphalt Institute methodology, the British design method and the meth-
odology of Egnatia Odos. Design parameters examined are the volume of traffic, the soil
subgrade quality and environmental conditions. The authors have concluded that the Asphalt
Institute methodology for flexible pavements offers the economically optimal design solution
for the pavement thickness in the majority of cases. However, the British design method
offers the lowest cost concerning pavement thickness when traffic load is heavy or very heavy
and subgrade soil quality is poor. Lastly, the Egnatia Odos methodology suggests the lowest
cost for flexible pavement thickness for roads of heavy traffic load and good subgrade soil
quality.
1 INTRODUCTION
Flexible pavements are the most common type used in Greece in the national and regional
road network. Rigid pavements are rarely used due to increased cost of maintenance with the
exception of airport pavements and several parts of the recently constructed “Egnatia Odos”
national highway. Flexible pavements support loads through bearing rather than flexural
action. They comprise of several layers of carefully selected materials designed to gradually
distribute loads from the pavement surface to the layers underneath. The design ensures the
load transmitted to each successive layer does not exceed the layer’s load-bearing capacity.
Thus, the primary goal of the pavement structural design is to dimension it (i.e. provide the
necessary thickness) so as to withstand all the traffic loading anticipated throughout the
selected design life without any structural failure.
The structural design and dimensioning process for flexible pavements is considered by
designers as a complicated, mechanistically intensive and demanding process based on
various design parameters. When designing pavements in terms of material mixture design
and structural design, some of the fundamental design parameters to consider are: the
mechanical characteristics of the subgrade upon which the pavement is placed (i.e. soil-type
quality expressed usually through it bearing capacity), the applied loads (volume of traffic)
and the climate (e.g. air temperature).
Typical design methods for flexible pavements are either based on practical experience
(empirical and mechanistic-empirical methods of design) or are mechanistic methods using
analytical calculations based on structural mechanics and elasticity theory (analytical and
semi-analytical methods). As a general perception, the analytical design methods require
69
intensive and cumbersome calculations when compared with the empirical and semi-ana-
lytical design methods. The various design methods may give different results (i.e. pavement
thickness) for the same design scenario due to variations in the design approach. Thus, the
resultant pavement thickness is different when using different design methods and as such the
cost for the pavement construction varies.
The absence of legislative or regulatory framework in Greece as to which flexible pavement
design method should be applied in particular with respect to a set of criteria against which
to propose the most appropriate and consequently the most economical design method of
flexible pavement was the primary motivation for this study.
The present study focuses on the structural design of flexible pavements and seeks the opti-
mal methodology of calculating the thickness of the different layers that form the pavement
so that the resulting pavement cost can be minimized. The authors aim to present guidelines
towards the economically optimal design method for flexible pavements based on specific
design parameters and in compliance to Greek national and European technical regulations
concerning road works.
Primary objective of this study is to investigate the design methodologies of flexible pave-
ments. There are various methodologies that can be applied for the dimensioning of flexible
pavements giving different pavement thicknesses and thus cost, due to different philosophy
of the design parameters considered (e.g. traffic volume). Therefore the objective of the study
is to select different design methods for flexible pavements suitable for use in Greece and
provide guidelines for the most effective method according to the design parameters. Thus,
designers can benefit from reduced analysis time since they can consider from the beginning
the optimal design method leading to reduced construction cost.
The structural design of pavements, either rigid or flexible, mainly refers to the design mix-
ture of materials to be used (i.e. concrete, asphalt and aggregates) and the dimensioning of
the different layers of the pavement (i.e. thickness of layers). The most important factors
responsible for the pavement design are: (i) climate conditions (e.g. rainfall, temperature,
frost action), (ii) environment and geometry (height of embankment, foundation cutting),
(iii) pavement materials (concrete versus asphalt), (iv) subgrade soil and (v) volume of traf-
fic. The design process for the pavement thickness is considered to be a rather complicated
procedure and several methods of diverse design philosophy exist.
Traditional pavement design methods were essentially empirical having evolved from the
experience gained from constructing pavements, observing their behavior and rejecting struc-
tural forms and materials which did not perform satisfactorily. Modern analytical design
method have evolved during the past 25 years, namely the University of Nottingham method
(Burton, 1983; Brown & Dawson, 1992) and the one from the University of Illinois (Thomp-
son, 1992). Most analytical methods have not been used extensively in practice as it is agreed
that these methods require lengthy and cumbersome calculations and thus designers tend to
prefer the empirical or semi-empirical design methods.
Presently, designers apply modern theoretical-analytical methods and semi-analytical or
semi-empirical methods for the design of flexible pavements. All such methods are based on
the theory of elasticity in parallel with the experience and knowledge gained from studies
on the actual behaviour of road surfaces and experimental investigations. Such methods are
easier to be applied compared to the analytical ones as the thickness of the different layers of
the pavement can be calculated using design charts and nomograms without complicated and
lengthy calculations. Thus the design procedure is relatively easy resulting in a comparative
advantage in relation to the analytical methods.
70
Mechanistic-empirical methods are based on the analysis of stresses and strains in the
pavement structure due to traffic loads or environmental effects using multi layer theory. In
order to be able to make such analysis, assumptions need to be made on the behavior of the
materials. For example, it could be assumed that the materials in the different layers of the
pavement are linear elastic, isotropic and homogeneous. This however is a serious simplifi-
cation of reality because most pavement materials exhibit a stress dependent, elasto-visco-
plastic behavior. In any case the calculated stresses and strains are then used as input in a
transfer function (e.g. a fatigue relation) that allows to calculate how many times a particular
tensile strain can occur before failure occurs. This part of the design method is the empirical
part. The reason for this is that although these transfer functions are usually based on exten-
sive material testing in the laboratory, most of these tests give a specimen property rather
than a material property. This makes them dependent on the local conditions and type of
pavement structure involved and empirical knowledge is needed to derive and use these shift
factors successfully. One of the most used semi-analytical mechanistic method is the method
by Shell (Shell International, 1985). Further, one example of an empirical method is the 1993
AASHTO design method that is based method on the theory of multi-layer elastic systems.
Input design parameters vary within each method but traffic volume, subgrade bearing
capacity and climatic conditions are necessary data to be used in all design methodologies.
Fundamental requirement in all design methods is that the thickness for all layers considered
should be such that the natural subgrade bearing capacity is not exceeded. In theoretical
methods, the pavement thickness is obtained by analytical calculations using complicated
models based on elasticity theory. Analytical design specifically involves the calculation of
stress and strain in an assumed pavement structure to ensure that they do not suffer excessive
distress before the pavement has performed satisfactorily for its design life. In the semi-empir-
ical / semi-analytical methods the layers’ thickness results from diagrams and nomograms (i.e.
design charts). In general, selection and implementation of a specific design method remains
the discretion of the designer.
One flexible pavement design methodology that is widely used is the Asphalt Institute
method. It was originally proposed in 1981 and after several revisions and updates was com-
pleted in 1991 and it is applicable up to present (Asphalt Institute, 1999). The method is semi-
analytical and it is primarily based on elastic theory of multi-layer systems and makes use of
design chards for the calculation of the pavement thickness (in terms of all its layers). These
design charts were recommended based on studies on the actual behaviour of road surfaces
and experimental investigations in conjunction with application of the software DAMA. The
method of the Asphalt Institute has application for the design of flexible pavements either
with the conventional structure (sub-base and base made of aggregates and surface of bitu-
minous layer) or flexible pavements where all layers are bituminous (Nikolaidis, 2002). This
design method finds wide application due to its simplicity for the calculation of the pavement
layer thickness.
Another semi-analytical method for the design of flexible pavements is the British design
method that has been proposed by the U.K. Department of Environment (1970) and in its
present form is provided by the Highways Agency (1998), denoted a HD 26/94 method. The
method can be applied in sistuations where the pavement is structured with three layers (either
typical or mixed type) and is also applicable for the stuctural design of rigid pavements. In the
the British methodology, the flexible pavement consists of three layers—namely, the asphalt
surface layer, the asphalt base layer and finally the subbase layer consisting of loose or sta-
bilized aggregates. Therefore, the British method refers to a different flexible pavement layer
structure compared with flexible pavement structure in other design methodologies where the
base is made of aggregates.
During the mid-90s’, a committee of greek experts on behalf of Egnatia Odos developed
a flexible pavement design methodology that would be suitable for the Egnatia road (Egnatia
Odos, 2001). The Egnatia road is the national highway that connects the city of Igoumenitsa
with the greek-turkish border. Despite the fact that the methodology was developed to meet
the specific needs of this highway, due to subsequent amendments and revisions, the method
acquired broader scope across Greece and even in countries with similar climatic conditions
71
with Greece. The design criteria of this method are similar to those of the method of the
Asphalt Institute with some additional requirements (e.g. total residual deformation) and
special care for the maintenance of the wearing course layer (Nikolaides et al., 1997). The
design methodology of Egnatia Odos presently tends to be applied with great frequency on
the design of public works in Greece.
Several other empirical or semi-empirical design methods were developed by other coun-
tries and an extensive literature review can be found elsewhere (Tsiknas, 2015). Examples
include the CRR method in Belgium which gives long design life pavements of over 25 years,
with low construction and maintenance costs and the method LPC in France that gives thin
pavements suitable for heavy traffic. It is noted that these methods take into account local
climatic and traffic conditions prevailing in the country that developed the corresponding
method, thus limiting their scope. Thus is has been observed a replacement from 1980 and
onward of the empirical methods from analytical and semi-analytical methodologies (Bar-
bars, 2013).
Current Greek standards and legislative framework conceding flexible pavement design
lack of an in-depth analysis and justification on the determination of the layer thickness of
a pavement and basically do not enforce the use of a specific design method. Flexible pave-
ments are designed primarily on personal practical experience with the help of an official
circular issued back in 1961 by the Ministry of Public Works based on technical knowledge
and experience of the early 50’s. Significant developments have taken place, ever since, in
other countries in connection to the pavement design. The traffic volume and its composi-
tion has changed dramatically ever since, new improved materials and techniques have been
developed and by no means the recommendations of this circular can be applied to highways
or major state roads.
The absence of greek legislative or regulatory framework as to which flexible pavement
design should be applied in particular with respect to a set of criteria against which to pro-
pose the most appropriate and consequently the most economical design method of flexible
pavement, where appropriate, has motivated the authors to investigate different flexible pave-
ments design methods for typical road construction in Greece. One of the main benefits of
choosing a modern method for the design of flexible pavements is to address effectively the
specific conditions of each road construction project satisfying the increased quality demands
placed nowadays (skid resistance, low noise, water drainage, etc.) and using new materials of
improved performance. Naturally, the choice of the most effective and economically optimal
design method should not be made in any case at the expense of safety.
Primary objective of this study is to investigate the economically optimal design methodol-
ogy for flexible pavements mainly for road construction works in Greece and countries with
similar design requirements. Thus, the authors aim to recommend the most effective, in terms
of cost, design methodology for several different design scenarios. It should be noted that a
low-cost design solution does not necessarily mean that the pavement has the smallest thick-
ness as each layer corresponds to a different cost and thus the total pavement cost results on
the combination of cost and thickness of each layer comprising the pavement.
Several types of flexible pavement design methodologies were considered and it was
decided to focus the investigation on the semi-analytical methods that are considerably easier
to use due to the use of design charts and nomograms. In the past years, the design meth-
ods that find wide application in Greece by engineers are the Asphalt Institute method, the
British design method and the one proposed by Egnatia Odos. These methods find wide
application also in other European countries and have the advantage that cover most sat-
isfactorily the environmental and climatic conditions in Greece. The design scenario was
chosen using a common case for a national highway in Greece: length 10km and total width
of 24m (typical width of highway with two lanes per direction of traffic), design life consid-
ered 25 years. Thus, the geometric characteristics of the highway (length and width) were
72
pre-decided by the authors and as such they do not affect the resulting pavement thickness
as design parameters.
It is worth mentioning that studies in the literature often implement analytical flexible
pavement design methodologies in the context of academic studies. But analytical design
methodologies are based on a number of assumptions that are extremely doubtful that simul-
taneously apply in their entirety in the design of a new road. Therefore, the implementation
of the analytical design method is only possible under certain conditions and for this reason
is basically practiced in theory. Also pavement design with the analytical methods requires
as input characteristics the mechanical properties of soil materials such as indicatively the
modulus of elasticity (E). This requirement creates limitations to the generalization of the
results obtained by applying a specific analytical method and thus the investigation of such
methods was not justified considering the objectives of this study.
Among the several design parameters that affect the pavement thickness, the authors have
focused on the following three basic and critical parameters: volume of traffic, subgrade soil
bearing capacity and climatic conditions. It is noted that the volume of traffic is expressed in
terms of Equivalent Standard Load Axles (ESLAs), the subgrade bearing capacity is expressed
based on the California Bearing Ratio (CBR) and as the fundamental mechanic property of
the layers is considered the Young Modulus. For each group of design values, the thickness of
the flexible pavement was calculated using the three chosen design methodologies. Using the
resulting thickness of each pavement layer, the overall cost was estimated in accordance with
the approved road projects uniform tariffs as issued by the relevant greek ministry (Ministry of
Public Works) in 2015. Thus, for each design scenario the overall pavement thickness and esti-
mated cost was obtained for all three design methods investigated. Details regarding the meth-
odology and all analysis results regarding the cost can be found elsewhere (Tsiknas, 2015).
It should be noted that it was not the objective of this study to assess all other construction
related costs (labor, transfer of equipment, construction site operation etc.) Additionally, the
effect of the asphalt type used in the bituminous layers was not investigated even though it is
a parameter that could differentiate the overall cost of the pavement. This choice was due to
the main aim of the study towards the investigation of the cost incurred in flexible pavements
only by using different design methodology rather than exploring the effect of the quality
and type of the construction materials. However, the cost implications of road surfaces due
to changes in the quality and type of the applied material could be the subject of a future
study as proposed at the end of this paper. Further, it should be assumed that the materials
considered (aggregates and bituminous mixtures) follow all relevant regulations (i.e. Greek
Technical Specifications) for flexible pavements. Details on such specifications are available
in detail elsewhere (see Tsiknas, 2015).
Table 1 presents the range of design values considered. The volume of traffic was expressed
as equivalent standard load axles (ESLAs) and three distinct values were considered that
describe conditions of low volume of traffic, heavy and very heavy so that the effect of this
parameter can be adequately investigated. The chosen values are based on typical design
values for greek territory (Kolias & Vlachos, 2000). It is noted that the volume of traffic is
considered for the entire design life of the flexible pavement (i.e. 20 years) with an average
annual increase rate of approximately 3% that is the standard one for road design (Gian-
nopoulos, 2002).
Further, it was decided to study the effect of the subgrade soil quality using two different
cases based on different CBR values and AASHTO requirements for soil classification cat-
egories. The chosen values for CBR reflect usual cases for soils in Greece and have been based
on geological & geotechnical site investigations by Egnatia Odos (Egnatia Odos, 2010) and
laboratory tests conducted at Aristotle University of Thessaloniki (Nikolaidis, 2012).
Last, all design scenarios and pavement thickness calculations were considered for cold
and warm average air temperature. Using the guidelines by Egnatia Odos (2000) for the
73
Table 1. Design parameters for flexible pavement.
Volume of traffic per Traffic load expressed in terms Light volume of traffic
direction (106 ESLAs) of Equivalent Standard ≤0.1 × 106 ESLAs
Load Axles Medium & heavy volume of traffic
≤10 × 106 ESLAs
Very heavy volume of traffic
≤60 × 106 ESLAs
California Bearing Subgrade bearing capacity Bad/medium quality soil
Ratio CBR CBR ≤4.5%
Good quality soil CBR ≤13.6%
Average annual air Environmental conditions Region with low air temperature 13°C
temperature Region with high air temperature 16°C
6 ANALYSIS RESULTS
In many design cases examined, there was not a significant cost difference between the design
methods. In those cases, the authors have chosen to propose the method that could best
adapt to the climatic conditions of Greece and as a result provide a pavement structure and
thickness that would require the minimal maintenance works.
Table 2 presents the results of the analysis for the cases of light volume of traffic and cold
regions (13°C average air temperature). The British design method recommends a minimum
total pavement thickness of 450mm so that there is no frost action effect—however, if the
total pavement thickness is less than the recommended, then a frost protection layer may be
used as an extension of the sub-base layer. When the soil conditions are poor (i.e. a CBR less
than 5%), it is generally recommended to construct a capping layer to act as an improvement
layer for the sub-base (Nikolaidis, 2005). However, this layer is only a demand in the method
of Egnatia Odos resulting in a pavement of increased cost compared with the other two
methods. In the case of good subgrade soil (CBR > 13.16%), the Asphalt Institute method
provides the option to the designer to omit the sub-base. However, for the benefit of safety
with respect to the stresses transmitted to the subgrade, it was decided to use a sub-base for
the design case scenario of this study. With respect to the effect of the soil quality, it can be
concluded that for the design criteria considered, only the British design method results in
different thickness for the pavement layers and thus reduced cost.
Table 3 presents pavement thickness considering heavy volume of traffic and cold regions
(13°C average air temperature). As expected, the increased traffic load results in increased
layer thickness compared with the previous cases examined. The benefit of good soil qual-
ity is reflected in reduced total pavement thickness in all design methods considered. It is
interesting to comment on the Egnatia Odos methods that results in the pavement with the
highest thickness but with lowest cost (for good soil conditions) due to the resulting thinner
asphalt layer.
74
Table 2. Pavement thickness—light volume of traffic & 13°C average air temperature.
Design method
Surface course 25 25 50 50 40 40
Binder course 30 30 60 50 0 0
Asphalt course 45 45 80 90 60 60
Base course 150 150 240 165 400 400
Subbase course 150 150
Total (mm) 400 400 430 355 500+300* 500
Total cost (€) 4.063.200 4.063.200 6.993.600 6.814.200 5.301.600** 4.617.600
Design method
Surface course 50 40 50 50 40 40
Binder course 80 40 100 100 50 50
Asphalt course 150 100 110 110 160 70
Base course 150 150 240 165 400 400
Subbase course 150 150
Total thickness 580 480 500 425 650 + 300* 560
Total cost (€) 9.768.000 6.672.000 9.038.400 8.849.400 9.717.600** 6.355.200
Subsequently, the case for very heavy traffic load and 13°C average air temperature is pre-
sented in Table 4. It is obvious that for the case of weak soil conditions, the British method
results in the lowest cost solution for the pavement design. Contrary, when the soil condition
are good, the optimal cost solution results from the Asphalt Institute design method.
As it has been discussed in the previous section, one design parameter considered by this
study is the average air temperature of the region where the highway is to be constructed.
All the design cases presented and discussed above where also considered for average air
temperature of 16°C. Though it may seem that the difference in the temperature might be
insignificant for the resulting pavement thickness, it is of great importance for the mechanic
behavior of bituminous mixtures. The British method does not consider the air temperature
as an input parameter that affects the pavement design and thus the results presented above
are irrespective of the temperature. The method of the Asphalt Institute considers the effect
of the air temperature in the asphalt mixture modulus of elasticity and as such the observed
differences are in the type of asphalt to be used and not in the pavement thickness. As the
temperature variation affects indirectly the pavement thickness through the materials to be
used, this was not considered in the present study. Contrary, the Egnatia Odos method gave
increased thickness for the asphalt layer when considering air temperature of 16°C. However,
the increase in the resulting thickness of the asphalt layer (and then total pavement thickness)
was less than 10% in all design cases with small effect in the total pavement cost.
Tables 5 and 6 presents the summary of the analysis results concerning the recommended
design method for flexible pavement design according to specific set of design requirement. It
75
Table 4. Pavement thickness—very heavy traffic & 13°C average air temperature.
Design method
Surface course 40 50 50 50 40 40
Binder course 100 100 100 100 100 50
Asphalt course 230 150 190 190 190 160
Base course 150 150 240 165 400 400
Subbase course 150 150
Total thickness 670 600 580 505 730 + 300* 650
Total cost (€) 12.268.800 8.112.000 11.419.200 11.230.200 12.050.400** 9.033.600
Table 5. Recommendation for economically optimal design methodology for flexible pavements (cold
air temperature).
Volume of traffic
Table 6. Recommendation for economically optimal design methodology for flexible pavements (warm
air temperature).
Volume of traffic
was shown that the method of the Asphalt Institute gives the economically optimal solution
for flexible pavement thickness in cases of pavements with light traffic load and irrespectively
of soil and environmental conditions and also in the case of pavement with heavy traffic load
(good soil conditions and warm regions) or very heavy traffic load (good soil conditions,
irrespective of the temperature).
For the cases of heavy and very heavy volume of traffic, the British method provided
flexible pavements design with the lowest cost when the subgrade soil conditions are weak
and irrespectively of air temperature. Contrary, for good soil subgrade conditions, it was
concluded that the design method of Egnatia Odos method gives the economically optimal
solution for the pavement layer thickness when the air temperature is low.
Among the design parameters examined in the present study, it was concluded that the most
critical one that significantly affects the thickness of the asphalt layers is the subgrade bear-
ing capacity expressed by the California Bearing Ratio (CBR) index, in particular when using
the Asphalt Institute design method and the Egnatia Odos method. Subsequently, the second
76
most critical parameter is the volume of traffic. Contrary, the former was not concluded when
using the British design method as this method only takes into account traffic load.
7 CONCLUSIONS
Overall, the analysis has shown that the Asphalt Institute design method gives the economi-
cally optimal flexible pavement section in most cases that were examined in the present study
compared to the British method and the Egnatia Odos method. Therefore, it is suggested for
designers to prefer this method for the structural design of flexible pavements, considering
the ease of application and fast extraction of results.
Additionally, among the design parameters investigate, it was concluded that the bearing
capacity of the soil subgrade appears to be the most critical design parameter for the result-
ing pavement thickness and cost. The effect of the environmental conditions (i.e. air tem-
perature) mainly affect the resulting pavement thickness when using the method of Egnatia
Odos.
It should be noted that in practice, designers select the appropriate pavement thickness
taking into account other criteria beyond the resulting construction cost. Such criteria are
the type and cost of pavement surface maintenance work, the availability of suitable con-
struction material for the pavement layers, environmental conditions (frost action, annual
rainfall etc.). Future work will include that investigation of other important design param-
eters on the resulting pavement design in relation with the economically optimal type of
asphalt for wearing cost. Thus, the authors aim to expand the objectives of the current study
and provide guidelines for the economically optimal design method with respect to type of
asphalt, construction.
As it has been noted, it was not the scope of the study the investigation of design methods
for rigid pavements that do not find wide application in Greece. However, it would be of great
interest to conduct an economic evaluation or comparison between flexible and rigid pave-
ments and previous studies have demonstrated potential economic benefit from increasing
the application of rigid pavements on greek road works.
REFERENCES
American Association of State Highways and Transportation Officials. (1993). Guide for design of pave-
ment structures. AASHTO.
Barbars, J. (2013). Development of Pavement Design Catalogue. Proceedings of International Baltic
Road Conference, 1–7.
British Highways Agency. (1998). Design Manual for Roads and Bridges, Vol.7: Pavement design and
maintenance, HD 26/94. TSO.
Brown, S.F., & Dawson, A.R. (1992). Two-stage mechanistic approach to asphalt pavement design. 7th
International Conference on Asphalt Pavements, 1, p. 16.
Burton, J.M. (1983). Developments in the analytical design of asphalt pavements using computers. Ph.D
Thesis. The University of Nottingham, Department of Civil Engineering.
Egnatia Odos (2000) Guidelines for pavements design (in greek: Εγνατíα Οδóς Α.Ε. (2000). Οδηγóς
διαστασιολóγησης οδοστρωμα´των.)
Egnatia Odos (2001), Guidelines for design projects for road works (in greek: Εγνατíα Οδóς Α.Ε. (2001).
Οδηγ ες Σ νταξης Μελετ ν ργων Οδοποι ας (ΟΣΜΕΟ), κε .3: Μελ τη Οδ ν, Παρ γρα ος 3.8:
Οδοστρ ματα.).
Egnatia Odos (2010). Requirements for pavements (in greek: Εγνατ α Οδóς Α.Ε. (2010). Προδιαγρα
ς Οδο .).
Giannopoulos (2002). ISBN 960-88681-0-6 (in greek: Γιανν πουλος, Γ. Α. (2002). Σχεδιασμóς των
μετα ορ ν. Η διαδικασ α πρ βλεψης των μελλοντικ ν αναγκ ν μετακιν σεων. Επαν κδοση
Παρατηρητ ς 2002, ISBN 960-88681-0-6).
Kolias S. & Vlachos C. (2000). Critical Appreciation of the Axle Load Equivalencies in Road Pave-
ment Damage.. Techica Chronica Scientific Journal of the Techical Champer of Greece, Vol. 20, No.
2, p. 43–55. (in greek: Κóλιας, Σ., Βλ χος, Κ. Και Αμπακουκμιν Κ. (2000). Κριτικ θε ρηση των
77
μεθóδων υπολογισμο των συντελεστ ν ισοδυναμ ας ορτ ων των εμπορικ ν οχημ των. Τεχν.
Χρον.Επιστ. κδ. ΤΕΕ(2), 43–55.)
Laboratoire Central des Ponts et Chaussees, LPC Bituminous Mixture Design Guide. The RST Working
Group, September 2007.
Nikolaidis (2002). Highway Engineering. Pavements—Materials. Quality Control. (in greek: Νικολα δης,
Α. Φ. (2002). Οδοποι α, Οδοστρ ματα-Υλικ , λεγχος ποιóτητας. Μ. Τριαντα λλου & Σ α.)
Nikolaidis (2005). Flexible pavements. Design Methods—Bituminous mixtures—Antiskdding layers. (in
greek: Νικολα δης, Α. Φ. (2005). Ε καμπτα οδοστρ ματα, Μ θοδος διαστασιολ γησης-Ασ αλτικ
μ γματα-Αντιολισθηρ ς στρ σεις. Αθ. Φ. Νικολα δης.)
Nikolaidis 2012. A modern approach towards bituminous surface pavement layers for antiskidding pur-
poses. Proceedings of 3rd International Conference on Asphalt Pavements. (in greek: Νικολα δης, Α.
Φ. (2012). Σ γχρονη προσ γγιση ασ αλτικ ν μιγμ των για αντιολισθηρ ς στρ σεις-Ε αρμογ ς
ασ αλτικο σκυροδ ματος για πολ λεπτ ς επι ανειακ ς στρ σεις. Πρακτικ 3ου Πανελλ νιου
Συνεδρ ου οδοποι ας, 1–10.).
Nikolaidis, Tsochos and Papavasileiou (1997). Design guide for pavements. (in greek: Νικολα δης, Α. Φ.,
Τσ χος, Γ., & Παπαβασιλε ου, Α. (1997). Οδηγ ς διστασιολ γησης οδοστρωμ των. Εγνατ α Οδ
ς Α.Ε.).
Recommendation of Belgian Road research Center CRR 23/61. Code de bonne pratique pour les betons
hydrocarbones, betons asphaltique et betons de goudron.
Shell International Petroleum Co. Ltd. (1985). Shell Pavement Design Manual (Addendum). Shell Inter-
national Petroleum Co. Ltd.
The Asphalt Institute. (1999). The thickness Design, Asphalt pavements for highways & streets. Manual
Series No.1 (MS-1). Asphalt Institute, Inc.
The Asphalt Institute. (2005). Asphalt Pavement Thickness Design Manual for Highways. Asphalt Insti-
tute, Inc.
Thompson, M.R. (1992). ILLI-PAVE based convetional flexible pavement design procedure. Proceed-
ings of 7th International Conference on Asphalt Pavements, 1, 318.
Tsiknas (2015). Economically optimal structural design methodology for flexible pavements using specific
design scenarios. MSc Thesis. Hellenic Open University. Engineering Project Management MSc Pro-
gramme. (in greek).
78
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
A.H. Al-Bayati
University of Baghdad, Baghdad, Iraq
1 INTRODUCTION
Aggregate gradation plays a very important role in the performance of an asphaltic mixture.
Many researchers have examined the packing of particles, in order to understand how the aggre-
gate can be structured. Starting with the maximum density of a single sized aggregate, stud-
ies have been made into the effect of particle gradation and calculation of maximum density.
Regarding asphalt mix design, gradation will affect all mechanical properties. However, perma-
nent deformation is critically affected by gradation and this paper will therefore focus on the
effect of changing aggregate gradation on the permanent deformation of an asphalt mixture.
According to Miranda (2012), a good gradation for an asphalt mixture is one that gives the
densest packing of particles, increasing stability by increasing the number of contact points
between particles and by reducing air voids, and many attempts have been made in the past
to understand the best gradation of particles to achieve the densest possible packing. Mix-
ing variations of stones and sand particles to give the densest concrete was first examined by
Fuller and Thompson (1907). The conclusion of their study was to relate the sand particle size
to the coarse particle size and they proposed that the sand particle diameter should not exceed
one-tenth of the coarse particle diameter to achieve a dense concrete mixture. They also con-
cluded that the coarser the aggregate used, the coarser the sand particle that was required.
Following their work, Talbot and Richart (1923) studied the gradation (size distribution) of
aggregate to achieve the maximum density of a mix. A slope of 0.5 on a straight line when
plotting percent passing a sieve versus the particle size on logarithmic scales was considered to
give maximum density. The formula for obtaining the maximum density line was:
n
⎛d⎞
P = 100 ⎜ ⎟ (1)
⎝ D⎠
where P is the percentage of material by weight passing a specific sieve with an opening size
d; D is the maximum particle size; and n is a parameter affecting the coarseness or fineness
of the gradation.
79
Nijboer (1948) studied the effect of particle size and shape in asphalt mixtures. He con-
firmed the idea of the maximum density line given by a straight line on logarithmic scales of
percent passing a sieve versus sieve size. He empirically proved that with typical stone parti-
cles the maximum density line has an approximate slope of 0.45.
In 2006, Kim et al. [5] developed a conceptual and theoretical approach to evaluate the struc-
ture of coarse aggregate in the gradation. Their research indicated a method to quantify the
size range of aggregate that carries the main load named the Dominant Aggregate Size Range
(DASR). The DASR principle divides the gradation into 3 active sections: Primary Structure,
PS, Secondary Structure, SS and fines. Oversize particles are considered non active.
2 BAILEY METHOD
The Bailey method was first developed by Mr. Robert Bailey from Illinois Department of Trans-
portation. It is an approach of blending aggregate to provide good aggregate interlock. Bailey
proposed ratios linked to the performance of a mixture. The Bailey method considers two prin-
ciples in evaluating a mixture: aggregate packing and the definition of coarse and fine aggregate.
In terms of aggregate packing, it is not possible to pack the aggregate particles to completely fill
a certain volume. There are certain factors affecting the degree of packing, for example, grada-
tion, particle strength, particle surface texture and shape, and finally the compaction effort. With
regard to coarse and fine aggregate, there are several methods for defining the coarse and fine
fractions. However, Bailey proposed a specific and meaningful way to define them. The coarse
aggregate particles are defined as those particles retained on a specific sieve size (explained
below) and the fine aggregate consists of particles that can fill the voids in the coarse aggregate.
where PCS = PCS for the overall blend; NMPS = NMPS for the overall blend. NMPS is
defined by Superpave as one sieve larger than the first sieve on which more than 10% of the
aggregate is retained. The value of 0.22 has been proposed by several researchers, Kim et al.
(2006), who made use of packing theories. The Bailey method suggests this value as an aver-
age of particle diameter ratios that theoretically range from 0.15 (round particles) to 0.29
(flat particles), Olard and Perraton (2010). The fine fraction (passing the PCS) is then split
into two further fractions using the Secondary Control Sieve (SCS). The value of 0.22 is used
again to create the SCS, see Eq. (3)
CS PCS × 0.22
PCS (3)
The fine material is then further divided with reference to a Tertiary Control Sieve (TCS)
which is again determined using the 0.22 factor applied to the SCS, see Eq. (4).
TCS
C SCS × 0.22
SCS (4)
80
ratios were suggested by Bailey: Coarse Aggregate Ratio (CA Ratio), Fine Aggregate Coarse
Ratio (FAc), and the Fine Aggregate Fine Ratio (FAf).
CA ratio
The CA Ratio is used to evaluate the interlock of the coarse aggregate and also to understand
the void structure in the coarse part of the gradation. For this purpose a Half Sieve is intro-
duced, Vavrik (2002). The Half Sieve (HS) is defined as half of the NMPS. Those particles
smaller than the HS and larger than the PCS are termed ‘interceptors’. Interceptors play an
important role in mixture volumetric properties. By changing the proportions of interceptors
the VMA will be changed. Interceptor particles do not fill the voids created by the larger par-
ticles because of their relatively large size compared to the voids created. However, the shape,
elongation and orientation of the particles are all factors affecting the filling of the voids
between coarse particles and interceptors. The CA ratio is given by Eq. (5) below:
i g HS − % Passing
% Passing i PCS
C
CA = (5)
100% − % Passing
i HS
Vavrik et al. (2001) related the CA ratio to the ability of a mixture to compact. They sug-
gested that in mixtures that have low CA ratios (below ∼ 1.0); the compaction of the fine
aggregate will improve relative to those with higher CA. However, this suggested value of
CA ratio (1.0) means there are equal amounts of interceptors and coarser aggregate and the
authors suggested that there will therefore be interlock between them and this will increase
the packing density. Higher values of CA ratio imply excessive amounts of interceptor par-
ticles which they suggest might lead to segregation among the particles larger than the PCS.
This would lead to a non-compactible mix. Such mixes with high CA ratio generally have a
pronounced S-shaped gradation (as will be seen later).
Fac ratio
The aggregate passing the PCS includes a wide blend of particle sizes. Similar to the coarse
part of the blend, the coarser part of the fine fraction will induce voids between the particles
and these voids will be filled by the finer particles. FAC describes the ratio of these fractions
and Eq. (6) below represents the definition of this ratio.
% Passing
i SCS
C
FAc = (6)
% Passing
i PCS
C
It is recommended that FAC should not be too high (no more than 0.5) as this will lead to
excessive amounts of fine material in the mix. On the other hand, very low values of this ratio
mean there is a low content of fine material.
Faf ratio
This ratio represents the content of the very finest material (including filler) in the mix and
is important for certain aspects of mixture behaviour (e.g. voids in fine fraction, binder
demand). However, this ratio will not be considered here as this study has focused on coarse
aggregate effects on permanent deformation.
New ratios
After analysis of early test results, it was found that the two Bailey ratios (CA and FAc) are
not sufficient on their own to give a full understanding of aggregate structure and interlock.
One poorly described part of the gradation consisted of the voids among the interceptor
particles and the interlock between particles in the fine fraction and the interceptors. Also
there was no clear representation of the dominant particle size transferring load. These two
omissions were filled by introducing two new ratios: Cf / Fc and F/C. The first is a representa-
tion of how much the coarser particles in the fine fraction interact with and fill the voids in
the interceptor fraction. The definition of this ratio is given by Eq. (7).
81
Cf % Passing
i g PCS
C % Passing SCS
C
Fc = % Passing (7)
i g HS − % Passing
i PCSC
The F/C ratio gives an indication of the relative overall percentages of the fine and coarse
materials which will help, with the aid of other ratios, in understanding which size range of
aggregate is most active in transferring the applied load. The F/C formula is simply the ratio
of fine to coarse materials as they are defined by the Bailey method, see Eq. (8).
F % Passing
i PCS
C
C = % Passing (8)
M − % Passing
i g NMPS i PCS
C
Finally, Bailey ratios do not show the percentage of the large particles that are supported
by the interceptors. In this case a mix may have a high CA ratio but a small amount of large
particles (such as Mix 4 as shown later). For this reason the CC ratio was introduced; it rep-
resents the proportions of aggregate retained on the HS and larger.
3.1 Materal
Six different aggregate gradations have been chosen, most of them at random, within the lim-
its of a 14 mm maximum aggregate size Asphalt Concrete according to BS EN 4987-1:2005.
The aggregate used was granite; the aggregate blend for each gradation was mixed with a
5.1% binder content of 40/60 penetration grade at 160°C. The aggregate gradation details are
shown in the table and graphs below:
82
The aggregate and binder were heated up to 160°C for more than 8 hours for aggregate and
3–5 hours for binder. The material was then mixed together at 160 ± 5°C and then compacted
by gyratory compactor at 150 ± 5°C with target air voids of 5%. After compaction, the sam-
ples were cut from both faces to a height of 60 mm and diameter of 100 mm.
4 TESTING PROGRAM
Obviously, the six mixes have different aggregate packing according to the ratios of each
mixture presented in Table 1. Some of the mixes, such as Mix A, have a good interaction
between the interceptors and the coarse aggregate (CA = 0.82), however in Mix A the inter-
ceptors are not supported well by the finer particles (Cf/Fc = 0.35). This combination of ratios
makes the mixture voided and influences negatively its performance. On the other hand, the
Sieve mm % Passing
83
interceptors in Mix F are well supported by finer material (Cf/Fc = 1.81) and in their turn they
support a reasonable amount of coarse aggregate. This combination makes Mix F denser
than Mix A and better performance is expected. To evaluate the asphalt mixtures in the light
of these packing ratios, three variables were considered: testing temperature, binder content,
and compaction effort.
84
the stiffness of the mix. It can be noticed that Mix D has much more increase in the per-
manent strain than Mix B because it has fewer large particles and more fine material.
c. Some of the mixes are always poor. Mix C suffered from poor deformation resistance
over the range of binder content from 4.1% to 6.1% because the aggregate interlock in the
mixture is poor. Mix C has a very high proportion of interceptors supporting a very low
proportion of large aggregate particles (CA and CC respectively). Moreover, it has a low
amount of fine materials, so this mixture is made up mainly by the interceptors with poor
support by the fine material.
d. Some mixes have a gradual increase in permanent strain as the binder content increases. Mix
A has good interlock between interceptors and large particles but the amount of smaller par-
ticles is small. However the percentage of filler is high, so that the increase in binder content
increases the percentage of mastic which will increase the strain and decrease the stiffness.
85
Figure 5. Effect of number of gyrations on air voids.
300. The reason behind this is that Mix A has high proportion of coarse aggregate (CC) and
these are well supported by interceptors (CA) but the interceptors are not supported well by
the finer aggregate (Cf/Fc).
This makes the increase in number of gyrations cause a high contact pressure between inter-
ceptors and large particles and this might result in particle breakdown. In Mix C, there is also a
low amount of fine material to support interceptors (Cf/Fc); however, it also has a low amount
of large particles. As a result the main aggregate size forming Mix C is the interceptors and
because they are no large particles, the probability of breakdown is not as high as in Mix A.
6 CONCLUSION
Aggregate gradation is one of the most important parameters that affect the mechanical
properties of asphalt mixtures. It was extensively studied by using the theory of packing and
introducing new ratios to define it and by making six asphalt mixtures with different aggre-
gate gradations. Different variables were then considered to understand how these packing
ratios affect the performance. The conclusions of this study can be summarized as:
1. The packing ratios gave a good definition of the aggregate gradation and they could be
used to understand and predict the performance of the mixture.
2. The effect of binder content on performance varied between different mixtures. Some
mixtures had more mastic when the binder content increased to 6.1% which affected nega-
tively on the performance, especially the stiffness.
86
Figure 6. Effect of number of gyrations on (a) Permanent strain (b) Stiffness
NOTE: Numbers shown in (a) are the number of cycles reached to the corresponding strain.
3. The effect of temperature was studied and it has been found that at relatively low tempera-
ture (30°C), the effect of aggregate packing was negligible. At higher temperature (50°C)
most of the samples fully deformed except for one which was defined by the packing ratios
to have good aggregate interlock (Mix F).
4. The high number of gyrations may have caused aggregate breakdown for some mixes
which had a negative effect on the performance. Other mixes also failed to show improve-
ment in performance when the highest number of gyrations was used.
REFERENCES
Fuller, W. & Thompson, S. 1907. The laws of proportioning concrete. Trans Am Soc Civil Eng 59:67.
Kim, S., Roque, R., Guarin, A. & Birgisson, B. 2006. Identification and Assessment of the Dominant
Aggregate Size Range (DASR) of Asphalt Mixture (With Discussion). Journal of the Association of
Asphalt Paving Technologists, 75.
Miranda, B.L. 2012. Gradation-Based Framework for Asphalt Mixtures. KTH, Royal Institute of Technology.
Nijboer, L.W. 1948. Plasticity as a factor in the design of dense bituminous road carpets, Elsevier Pub. Co.
Olard, F. & Perraton, D. 2010. On the optimization of the aggregate packing characteristics for the
design of high-performance asphalt concretes. Road Materials and Pavement Design, 11, 145–169.
Talbot, A.N. & Richart, F.E. 1923. The Strength of Concrete-ITS Relation to the Cement, Aggregates
and Water. Illinois Univ Eng Exp Sta Bulletin.
Vavrik, W.R. 2002. Bailey method for gradation selection in hot-mix asphalt mixture design, Transporta-
tion Research Board, National Research Council.
Vavrik, W.R., Pine, W.J., Huber, G., Carpenter, S.H. & Bailey, R. 2001. The bailey method of gradation
evaluation: the influence of aggregate gradation and packing characteristics on voids in the mineral
aggregate (with discussion). Journal of the Association of Asphalt Paving Technologists, 70.
87
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: This paper aims at evaluating the feasibility of using an ASTM C666 apparatus
in freeze-thaw (F-T) characterizations of asphalt mixtures. Currently, there is no standard availa-
ble that provides guidelines of how to prepare asphalt specimens for ASTM C666 F-T cycle tests,
so this research serves as a pilot study to bring together an innovative material testing methodol-
ogy to characterize the F-T resistance of asphalt mixtures. Asphalt specimens were tested at six
different cycles which were determined to be 0, 100, 150, 200, 250, and 300. After a desired cycle
is completed, specimens removed from the apparatus and were tested for their thermal resist-
ance. Based on test results, stiffness values of asphalt mixtures dropped as the number of freeze-
thaw cycle increased. The relationship between the asphalt mixtures and freeze-thaw cycles can be
expressed as an exponentially empirical equation. The findings presented in the paper provides a
positive trend for future research focusing on the long term durability of asphalt mixtures.
1 INTRODUCTION
89
1.2 Background
Northern Arizona University located in Flagstaff, Arizona lies near the southwestern edge of
the Colorado Plateau with an elevation of 2,250 meters. In the winter months, the tempera-
tures with high attitudes are extremely cold, and the daily temperature changes dramatically.
In addition, a report by Hershfield (1974) indicated northern Arizona has been recorded
with a mean annual frequency of 250 freeze-thaw cycles. The extreme temperatures along
with high frequency of freeze and thaw cycles have been a significant impact on the freezing
and thawing durability of pervious concrete. The effect of freezing and thawing process on
the performance of the asphalt material is significant, particularly with moisture damage
influencing the aggregate skeleton within asphalt structure. Huang et al. (2005) studied mois-
ture resistance using a strategic highway research program asphalt and limestone and granite
aggregates by freeze–thaw cycling of the mixtures in water. In addition to their research, a
number of research projects were performed to investigate the performance of asphalt mix-
tures under freeze-thaw cycles (Jacobsen & Sellevold 1997, Sebaaly et al. 2003, Kettil et al.,
2005, Feng et al. 2010, Wang et al. 2013). However, the F-T cycles used for those studies var-
ied depending on individual need and the fact that there is no testing standard available for
systematical estimation of tensile strength and viscoelastic characteristics of asphalt mixtures
subjected to freeze-thaw cyclies in cold regions. This paper aims at evaluating the feasibility
of using an ASTM C666 apparatus in freeze-thaw (F-T) characterizations of asphalt mix-
tures. The ASTM C666 standard is exclusively used for pervious concrete materials and it has
never been used for asphalt materials. This research serves as a pilot study to bring together
an innovative material testing methodology to characterize the F-T resistance of asphalt
mixtures under a series of freeze-thaw cycles.
Figure 1. Layout of asphalt overlay project at NAU. Note: hatching area represents the paving loca-
tion of fiber reinforced asphalt pavements.
90
Table 1. Mix design information.
samples from both asphalt mixtures were collected at the back of a paver from both lances at
the time of laydown. All samples collected were shipped back to the Materials Lab of NAU
where asphalt mixture samples were reheated and compacted to specimens using a Superpave
gyratory compactor (SGC) with appropriate dimensions (150 mm in diameter and 110 mm in
height) suitable for freeze-thaw cycling tests.
91
the volume of the specimen increases as a result of frozen asphalt structure. This type of vol-
umetric expansion would cause the asphalt pavement to crack. One of goals of the paper is to
evaluate the effect of repeated F-T cycles on the tensile strength/stiffness of asphalt mixtures.
It would be interesting to observe if the frozen pore moisture of a mixture specimen would
affect the performance of asphalt mixtures due to volumetric expansion. In a thin mixture
beam used for BBR tests, its void ratio plays a significant role in controlling the stiffness of
the specimen. The BBR testing method has been used by several researchers at institutions
for the prediction of thermal cracking resistance of asphalt concrete Zofka et al. at (2005,
2008a, 2008b), Velásquez et al. at University of Wisconsin (2009, 2010). More recently, Ho
and Romero (2011, 2012, 2013) at Northern Arizona University and the University of Utah
respectively have worked together to address the concern of aggregate sizes played in the
asphalt mixture thin beams to represent global properties of asphalt mixtures using both
numerical analyses and laboratory experiments. Their research results have been in support
of using thin beams in a BBR to characterize the thermal aspects of asphalt mixtures in the
laboratories and fields. Thus, it is feasible and reasonable to obtain creep compliance data of
asphalt thin beams from the BBR to predict the stiffness of both PMA and FRA mixtures.
Each block specimen could be trimmed in approximately 15 thin beams. All thin beams
were categorized into three groups tested at -12°C, -18°C, and -24°C respectively. The
testing process and the number of replicates tested for a valid result can be referred to Ho
and Romero (2011, 2012, 2013). Stiffness values measured at 60 seconds were used as a
representative stiffness of a thin beam in accordance with the AASHTO T313 standard
(2009). All creep compliance data were recorded and the results will be discussed in the
following section.
92
Figure 4. Relationship between stiffness and Freeze-Thaw cycles-fiber reinforced asphalts.
RMA specimens 150–1 150–2 200–1 200–2 250–1 250–2 300–1 300–2
Initial void ratio 7.48% 7.37% 6.17% 6.87% 3.54% 3.98% 3.51% 6.08%
Void measured at a 7.29% 8.14% 7.69% 8.22% 4.16% 4.09% 4.44% 7.03%
completed FT cycle
Change,% −0.19% 0.77% 1.52% 1.35% 0.62% 0.11% 0.93% 0.95%
FRA Specimens 150−1 150−2 200−1 200−2 250−1 250−2 300−1 300−2
Initial void ratio 2.54% 1.67% 3.33% 1.93% 1.67% 2.34% 2.01% 2.28%
Void measured at a 2.54% 0.58% 4.81% 3.28% 1.86% 3.45% 2.52% 2.18%
completed FT cycle
Change,% 0.00% −1.09% 1.48% 1.35% 0.19% 1.11% 0.51% −0.10%
Note: specimen 150–1 represents a specimen was removed from the freeze-thaw cycle apparatus after
completion of 150 cycles, and the rest of specimens are labeled based on the same definition.
3.2 Air void analysis of specimens before and after Freeze-Thaw cycles
Before in a freeze-thaw cycle test, the air void of all specimens were measured. After comple-
tion of a designed freeze-thaw cycle (i.e., 100, 150, 200, 250, 300), two labeled specimens were
removed from the freeze-thaw cycle apparatus. Before the specimens were tested for their
creep compliance/stiffness using a BBR instrument, their void ratio was measured using the
ASTM C127 standard. The measures are shown in Table 2.
Table 2 explains how air voids change as the number of freeze-thaw cycles increases. Except
for specimens 150–1 (RMA), 150–2 (FRA), and 300–2 (FRA) with a slight air void decrease,
the air void of all specimens increases ranging from 0.11% to 1.52% in the RMA group and
from 0.51% to 1.48% in the FRA group, respectively. The repeated freezing and thawing
cycling not only could it introduce the moisture intrusion into the structure, and cause fatigue
cracking to the pavement, but it also gradually leads to chip out find aggregates resulting in
an increase of void ratio as well as lowering the stiffness of the pavement. The findings in
Table 2 have agreement with the stiffness drop as shown in Figure 3 and Figure 4.
It is noticed that the 200 cycles at both RMA and FRA mixtures have the highest air void
than others. Specimens made of FRA mixtures have less air void ratios than others com-
posed of RMA mixtures. This is due to the binding property of the fibers melted in the FRA
mix that hold aggregates and binder together.
94
T
ξ= (1)
aT (T )
Ea ⎛ 1 1 ⎞
log [ T (T )] .303 ⋅ − (2)
R ⎜⎝ TR T ⎟⎠
where, Ea = the activation energy for flow below TR, 261 kJ/mol.
R = the ideal gas constant, 8.34J/mol-°K
TR = reference temperature, °C or °K
T = selected temperature, °C or °K
• Individual creep compliance curves with corresponding temperatures from the three tem-
peratures were shifted along a log time scale to superimpose to a master creep compliance
curve as shown in Figure 5 (fiber reinforced mixtures) and Figure 6 (rubberized modified
mixtures).
As noticed in both Figure 5 and Figure 6, stiffness drops apparently from 0 cycle to 100
cycles and 150 cycles. After specimens experienced 150 freeze-thaw cycled, the slopes of stiff-
ness mater curves measured at 200, 250, and 300 cycles tend to slow. Figure 5 and Figure 6
also confirm the findings from Figure 3 and Figure 4 as well as show agreement in the expres-
sion of relationship between the stiffness and number of freeze-thaw cycles as an exponential
equation.
With all findings and results received from BBR tests, it is confident to conclude the mate-
rial testing method using a freeze-thaw apparatus is capable of measuring stiffness data under
repeated freeze-thaw cycles. The ASTM C666 standard not only can it use for pervious con-
crete, but it also can be applied in the prediction of thermal performance of asphalt mixtures
subjected to the number of freeze-thaw cycles. This project serves as a pilot study for evaluat-
ing the feasibility of using a freeze-thaw apparatus in the durability of asphalt mixtures. The
F-T tests presented in the paper can be used in predicting the effect of freeze-thaw cycles on
the performance of asphalt mixtures in cold regions.
95
Figure 6. Master curves of rubberized modified mixtures.
4 CONCLUSIONS
The feasibility of using a freeze-thaw apparatus compatible with the ASTM C666 standard
is presented and the effect of repeated freeze-thaw cycles on the performance of asphalt mix-
tures is evaluated. The conclusions are described as follow:
1. Based on findings and results from BBR tests, the material testing method presented in
the paper is capable of measuring stiffness of asphalt mixtures subjected to a number of
freeze-thaw cycles.
2. The stiffness values drop as the number of freeze-thaw cycles increased. The relationship
between the stiffness and the number of freeze-thaw cycles can be expressed as an expo-
nential function. This finding tied the numerical analysis on the thermal performance of
both fiber reinforced asphalt and rubber modified asphalt as shown in the master curves
of asphalt mixtures obtained from the BBR tests.
3. The air void of specimens increases as the number of freeze-thaw cycles increases. The
volumetric expansion allows the excess moisture to intrude into the structure of the pave-
ment resulting in lower stiffness values and damage to the integrity of the pavement.
4. Given the results of ANOVA analysis, the stiffness difference of asphalt mixtures from
both fiber reinforced asphalt and rubber modified asphalt among all freeze-thaw cycles is
significant.
5. The findings presented in the paper provides a positive trend for future research focusing
on the long term durability of asphalt mixtures
REFERENCES
American Association of State Highway and Transportation Officials. (2009) Determining the Flexural
Creep Stiffness of Asphalt Binder Using the Bending Beam Rheometer (BBR). Standard Specifica-
tions for Transportation Materials and Methods of Sampling and Testing T 313. AASHTO 29th
edition.
Christensen, D.W. & Anderson, D.A. (1992) Interpretation of dynamic mechanical test data for paving
grade asphalt cements. Journal of the Association of Asphalt Paving Technologists. (61):67–116.
Compliance at Low Temperatures. Journal of Road Materials and Pavement Design. (9)269–286.
96
Feng, D., Yi, J., Wang, D., Chen, L. (2010). Impact of salt and freeze–thaw cycles on performance of
asphalt mixtures in coastal frozen region of China, Journal of Cold Regions Science and Technology.
(62) 34–41.
Guo, S.W & You, Z. (2012). Evaluation of Hot-Mix Asphalt Distress under Rapid Freeze-Thaw Cycles
Using Image Processing Technique. Proceedings of the Twelfth COTA International Conference of
Transportation Professionals, August 3–6, 2012 in Beijing, China.
Hershfield, D.M. (1974). The Frequency of freeze thaw cycles. Journal of Applied Metrology.
(13)348–354.
Ho, C.H. and Romero, P. (2011). Using Asphalt Mixture Beams in the Bending Beam Rheometer: Exper-
imental and Numerical Approach. Journal of Road Materials and Pavement Design. 12(2):293–314.
Ho, C.H. and Romero, P. (2012). Asphalt Mixture Beams Used in the Bending Beam Rheometer for
Quality Control: Utah Experience” Journal of Transportation Research Board. (2268):92–97.
Ho, C.H. and Romero, P. (2013). Using Linear Viscoelastic Modeling to Evaluate the Low Temperature
Properties of Asphalt Mixtures Prepared With Aggregates of Different Sizes. Journal of Advances in
Civil Engineering Materials, American Society for Testing and Materials. 2(1):122–139.
Huang, S., Robertson, R., Branthaver, J., and Claine Petersen, J. (2005). Impact of Lime Modification
of Asphalt and Freeze–Thaw Cycling on the Asphalt–Aggregate Interaction and Moisture Resist-
ance to Moisture Damage. J. Mater. Civ. Eng., 17(6):711–718.
Jacobsen, S., Sellevold, E.J. 1997. Frost/salt scaling and ice formation of concrete: effect of curing tem-
perature and silica fume on normal and high strength concrete. Proceedings of International Union
of Testing and Research Laboratories for Materials and Structures in Sainte-Foy, Québec, Canada.
93–106.
Kettil, P., Engström, G., Wiberg, N.E., 2005. Coupled hydro-mechanical wave propagation in road
structures. Comput. Struct. 83, 1719–1729.
Schwarzl, F., & Staverman, A. (1952). Time-temperature dependence of linear viscoelastic behavior.
Journal of Applied Physics. 23(8):838–843.
Sebaaly, P., Hitti, E., Weitzel, D. (2003). Effectiveness of lime in hot-mix asphalt pavements. Journal of
Transportation Board, (1832)34–41.
Stiffness of Asphalt Mixtures at Low Temperatures. Advanced Testing and Characterization of
Bituminous Materials. (1): 313–322.
Velásquez, R., Marasteanu, M. Turos, M, and Labuz, J. (2009). Effect of Beam Size on the Creep
Velásquez, R., Marasteanu, M., Labuz, J.F., Turos, M. (2010). Evaluation of Bending Beam Rheometer
for Characterization of Asphalt Mixtures. Journal of the Association of Asphalt Paving Technolo-
gists. (79)295–324.
Wang, Y., Ye, J., Liu, Y. Qiang., X., Feng, L. 2013. Influence of freeze–thaw cycles on proper-
ties of asphalt-modified epoxy repair materials. Journal of Construction and Building Materials.
(41)580–585.
Zofka, A., Marasteanu, M.O., Li, Xinjun, Clyne, T.R., McGraw, J. (2005). Simple Method to Obtain
Asphalt Binders Low Temperature Properties from Asphalt Mixtures Properties. Journal of the
Association of Asphalt Paving Technologists. (74) 255–282.
Zofka, A., Marasteanu, M., and Turos, M. (2008). Investigation of Asphalt Mixture Creep
Zofka, A., Marasteanu, M., and Turos, M. (2008). Determination of Asphalt Mixture Creep Compli-
ance at Low Temperatures Using Thin Beam Specimens. Journal of the Transportation Research
Board. (2057):134–139.
97
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
P. Liu
Institute of Highway Engineering, RWTH Aachen University, Aachen, Germany
D. Wang
Institute of Highway Engineering, RWTH Aachen University, Aachen, Germany
Sino-European Research Center for Advanced Transportation Infrastructure Technology,
Würselen, Germany
F. Otto & J. Hu
Institute of Highway Engineering, RWTH Aachen University, Aachen, Germany
M. Oeser
Institute of Highway Engineering, RWTH Aachen University, Aachen, Germany
Sino-European Research Center for Advanced Transportation Infrastructure Technology,
Würselen, Germany
1 INTRODUCTION
The bearing capacity is of significant importance for pavement design and also generally pro-
vides necessary information for estimation of the remaining pavement life and the selection
of a feasible maintenance and rehabilitation strategy. Large-scale investigations on the index
of the pavement bearing capacity have been carried out worldwide. Under the premise of
avoiding damage to the road, non-destructive testing (NDT) and back-calculating pavement
layer moduli are well-accepted procedures for evaluation of the material properties (Saltan
et al. 2011).
Deflection measurements are currently the most popular method of determining the bear-
ing capacity of asphalt pavements but influenced by numerous factors, such as the loading
type, temperature, moisture, etc. (Ferne et al. 2009a, Saltan et al. 2011). There are two broad
approaches to surface deflection measurement: measuring the vertical displacement at a fixed
point on the ground resulting from a real or simulated wheel load; measuring the transient
vertical displacement due to a rolling wheel using a test apparatus that moves along with the
wheel (Moffatt & Martin 2013). Falling weight deflectometer (FWD) is a representative of the
first approach which requires traffic control and measures the deflections at a stationary loca-
tion at discrete locations along the pavement structure. In the past decade, the development of
99
high speed deflection measuring devices which belong to the second approach has been found
to be the most economical technique at a network level. Some examples are quest/dynatest
rolling weight deflectometer (RWD), Swedish road deflection tester, Texas rolling dynamic
deflectometer and traffic speed deflectometer (TSD) (Ferne et al. 2009b).
In order to determine the bearing capacity of the asphalt pavement using measured deflec-
tions, it is necessary to characterize the inverse mapping of theoretical pavement response
model, which is generally referred to as the pavement back-calculation technique (Zhou 2000).
There are generally three basic categories of back-calculation methods: adaptive, static, and
dynamic (Goktepe et al. 2006). Recently, some mixed back-calculation methods have been
developed, such as static and adaptive combinative method (Gopalakrishnan et al. 2007,
Pekcan et al. 2010). Except for the purely adaptive one, the back-calculation methods normally
contain a computational kernel, which computes deflections from given loads and pavement
structures. Most of the computational kernels are currently based on elastic layered theory
or finite element (FE) method. They are used for analysis of conventional stationary loading
condition such as FWD (Bilodeau & Doré 2014, Khattak et al. 2012, Loizos & Boukovalas
2005, Loizos & Scarpas 2005). If the measurements are carried out by TSD, the evaluation
method should be modified, i.e. the traditional computation using stationary loads does not
accord with the traffic loading conditions. The deflection caused by the traffic loads will not
be as symmetric as for stationary loads (Oeser 2010). However, the computational kernels used
for measuring devices at traffic speed are rarely covered extensively in the literature.
In this paper, a computational program SAFEM is utilized as a computational kernel in
a back-calculation program which can compute the material properties such as E-moduli of
the test track layers. The SAFEM is based on semi-analytical finite element method; it is a
three-dimensional FE program that requires only a two-dimensional mesh by incorporating
the semi-analytical method using Fourier series in the third dimension. The parallel comput-
ing technology can be applied easily in this code; as a result, the computation time could be
significantly reduced. The deflection of the asphalt pavement was measured by geophones
from a full-scale and fully enclosed test track loaded by a passing truck. Compared with the
measurement of TSD in a real pavement, the method used in this study can avoid the influ-
ence of environmental factors and derive data points more economically. With the help of
the accelerated loading machine Mobile Load Simulator (MLS10), three service states of the
test track with different bearing capacity were generated. For the initial, intermediate and
the final states, the material parameters were computed and its relationship with the number
of loading cycles of the MLS10 was studied. The results show the back-calculated bearing
capacity has a quite good negative linear relation with the number of loading cycles in the
experiment, which proves reliability of the proposed back-calculation program.
2 METHODOLOGY
100
Figure 1. a: The test track in BASt (Gohl 2006); b: The field setup of passing track and geophones.
101
direction. As a result, the pavement structure problem could not be simplified into a 2D
plane strain case. However, by assuming that the displacements in the geometrical z-direction
can be represented using a Fourier series and exploiting its orthogonal properties, the prob-
lem of such a class can be numerically solved by a series of 2D FE-meshes (Fritz 2002, Hu et
al. 2008, Kim et al. 1994, Liu et al. 2013, 2014, 2015, Zienkiewicz & Taylor 2000, 2005) and
the computational time can be reduced significantly.
The pavement is assumed to be held at z = 0 and z = a in a manner preventing all displace-
ments in the XY plane but permitting unrestricted motion in the z-direction, as shown in
Figure 3b (Fritz 2002, Hu et al. 2008, Kim et al. 1994, Liu et al. 2013, Liu et al. 2015, Liu et
al. 2014, Zienkiewicz & Taylor 2000, Zienkiewicz & Taylor 2005). The general form of the
shape functions used in SAFEM defining the variation of displacements can be written as a
Fourier series (Zienkiewicz & Taylor 2005):
⎡ lπ z ⎤
⎢sin a 0 0 ⎥ l
⎧u ⎫ L 6 ⎢ ⎥ ⎧ uk ⎫ L
⎪ ⎪ l z ⎪ ⎪
U = ⎨ v ⎬ = ∑ ∑ Nk ⎢ 0 si 0 ⎥ ⎨ ⎬ = ∑ N l ⋅U l (1)
⎢ a ⎥
⎪w ⎪ l =1 k =11 ⎢ ⎥ ⎪w l ⎪
l =1
⎩ ⎭ lπ z ⎥ ⎩ k ⎭
⎢ 0 0 cos
⎢⎣ a ⎥⎦
where ukl , vkl and wkl are the displacements of the node at the term of the Fourier series along
x-, y- and z-directions, respectively.
The loading function for the pavement analysis can be written as (Hu et al. 2008):
L
l z L
f ∑p x y
l =1 a
= ∑{ p}l
l =1
(2)
n
⎛ 2 Pt ⎞ ⎡ lπ lπ ⎤
p( x, y ) ∑ ⎝ lπ ⎠ ⎣cos a Z
t =1
t1 − cos
a
Zt 2 ⎥
⎦
(3)
where Pt is the tire load pressure; Zt1 is the z coordinate where the tire load starts; Zt2 is the
z coordinate where the tire load ends.
After determining the element displacement, geometrical and physical equations can be
used to obtain the strain and stress of one element. The strain-displacement matrix Bkl is
defined as follows:
⎡ ∂N k lπ z ⎤
⎢ ∂x sin a 0 0 ⎥
⎢ ⎥
⎢ ∂N k lπ z ⎥
0 si
sin 0
⎢ ∂y a ⎥
⎢ ⎥
⎢ lπ lπ z ⎥
⎢ 0 0 − N k sini
a a ⎥
Bk = ⎢
l
⎥ (4)
⎢ ∂N k sin lπ z ∂N k
sin
lπ z
0 ⎥
⎢ ∂y a ∂x a ⎥
⎢ ⎥
⎢ lπ lπ z ∂N k lπ z ⎥
0 N k cos cos
⎢ a a ∂y a ⎥
⎢ ⎥
⎢ lπ N cos lπ z 0
∂N k
c s
co
lπ z ⎥
⎢⎣ a k a ∂x a ⎥⎦
By using the principle of minimum potential energy, a typical sub-matrix of the element
stiffness matrix (Klm)e is (Zienkiewicz & Taylor 2005):
102
Figure 3. a: Pavement geometry and load mode; b: Schematic representation of an SAFEM situation.
( l
) ∫∫∫ ( B )
vvol
l T
DB m dxdydz (5)
From Equations 4 and 5, the stiffness matrix of one element includes (Zienkiewicz & Tay-
lor 2005):
a lπ z mπ z
I1 = ∫ sin ⋅ cos ⋅ dz
0 a a
a lπ z mπ z
I 2 = ∫ sin ⋅ siin ⋅ dz (7)
0 a a
a lπ z mπ z
I 3 = ∫ cos ⋅ cos ⋅ dz
0 a a
The integrals exhibit orthogonal properties which ensure that:
⎧1
⎪ a ffor l m
I2 I3 = ⎨ 2 (8)
⎪⎩ 0, for l ≠ m
Only when l and m are both odd or even numbers, the first integral I1 is zero. Due to the
special structure of the Bl matrix, all terms that include I1 become zero. This means that
the matrix (Klm)e becomes diagonal. Thus, the stiffness matrix can be reduced and the final
assembled equations have the following form:
⎡ K 11 ⎤⎧ U1 ⎫ ⎧ F1 ⎫
⎢ ⎥ ⎪ U 2 ⎪ ⎪F 2 ⎪
⎢ K 22 ⎥ ⎪⎨ ⎪ ⎪ ⎪
⎬+ ⎨ ⎬ = 0 (9)
⎢ ⎥⎪ ⎪ ⎪ ⎪
⎢ ⎥
⎣ K LL ⎦ ⎪⎩U LL ⎪⎭ ⎪⎩F L ⎪⎭
The Equation 9 shows that the large system of equations splits up into L separate problems.
The equilibrium equations are fully decoupled for each harmonic of Fourier series, which can
be well adapted to the individual processors of a parallel computer, and thus reduce the com-
putational time significantly compared to the sequential solving procedure (Kim et al. 1994).
The Equation 9 is used for the time independent finite element analysis. In order to analyse
the dynamic response of the asphalt pavement, the time coordinates should be introduced to
the finite element algorithm. Therefore, the analysed objects turn into four dimensions (x, y,
z, t), and the Equation 9 should be rewritten. This deducing process can be found in many
literatures such as (Zienkiewicz & Taylor 2000) thus it will not be covered in this paper.
103
2.4 The algorithm of back-calculation
A back-calculation program combining SAFEM and artificial neural networks (ANN) is
developed to evaluate the bearing capacity of asphalt pavements. The details about the ANN
tool can be found in (Wang 2011). The other input parameters such as boundary condi-
tions, geometrical parameters and material parameters except for E-moduli are known. The
E-moduli of different pavement layers are generated as a collection within a reasonable
range to have all possible combinations encountered in the pavement. The series of input
parameters are imported into the SAFEM and the corresponding deflections are computed.
The database including the pairs of E-modulus of each pavement layer and correspond-
ing surface deflection of the pavement is then fed into the ANN which is trained to set up
the relationship between the E-moduli and the surface deflection. Hereafter, the deflections
from field measurement are input into the trained ANN and the corresponding E-moduli are
back-calculated.
In order to simplify the computation the five real layers of the test track were merged
whereby the three asphalt layers were merged to an equivalent asphalt layer and the two
bottom layers to an equivalent unbound layer. The interlayer behaviour was defined as
being partially bound. The parameters of the simplified test track structure are listed in
Table 1. A reasonable range of the E-modulus was given and the combinations of the input
parameters were 392.
The surface deflections derived from the three measurements when the second axle of the
truck was passing the fourth geophone was included into ANN as the input, as shown in
Figure 4.
Following the procedure in the section 2.4, the E-moduli of the equivalent asphalt and
unbound layers at the three states were back-calculated and shown in Figure 5a and b, respec-
tively. Both the E-moduli of the equivalent asphalt and the unbound layers at state 2 of the
Table 1. Thicknesses and material properties of the test track used in back-calculation.
104
Figure 5. The relationship between the loading cycles by MLS10 and a: E-modulus of the equivalent
asphalt layers; b: E-modulus of the unbound layers.
test track are smaller than those at state 1, as a result, the maximum of the measured deflec-
tion at state 2 is larger than that at state 1, which is shown in Figure 4. It's interesting that
the E-modulus of the equivalent asphalt layer at the state 3 is smaller than that at the state
1, but the E-modulus of the equivalent unbound layer at the state 3 is larger, which leads the
maximum deflection at state 3 to be smaller than that at state 1. The increase of E-modulus
of the unbound layer may be caused by the post compaction that occurred during loading
which is not unusual.
With the back-calculated E-moduli, the tensile strains εel at the bottom of the equivalent
asphalt layer were computed by the SAFEM under the loading conditions of MLS10. In
order to verify the reliability of the prediction of the back-calculation, drill cores of some
specimens were extracted from the asphalt base layer of the test track to test the fatigue
resistance with the indirect tensile test based on German rules and regulations (FGSV 2009),
which was at the same temperature and frequency as applied for MLS10. The relationship
between the number of loading cycles to the failure of the indirect tensile test Nind and ten-
sile strain εel was determined:
Nind = 4 221⋅ ( )
−3.474
eel ⋅ (10)
The corresponding numbers of loading cycles resulting in the failure of the indirect tensile
test were calculated according to the Equation 10. Their relationships with the loading cycles
by MLS10 are shown in Figure 6.
The number of loading cycles to the failure of the indirect tensile test decreases with the
increase of loading cycles applied by the MLS10, which confirms the decrease of the bearing
capacity. The relationship between the back-calculated number of loading cycles to the fail-
ure of the indirect tensile test and the loading cycles by MLS10 on the test track is established
as a linear dependency with very good correlation, which further proves the reliability of the
SAFEM and the algorithm of the back-calculation.
and the corresponding tensile strains at the bottom of asphalt base course are in accord-
ance with the evolution of the bearing capacity. A relationship between the back-calculated
number of loading cycles to the failure of the indirect tensile test and the number of loading
cycles loaded by MLS10 is established as a linear dependency with very good correlation,
which provides significantly positive meanings on the evaluation of the remaining service life
of the asphalt pavements.
On the whole, with the SAFEM as the computational kernel of the back-calculation pro-
gram the proposed assessment system has great potential to offer a fast and reliable evalu-
ation of the bearing capacity of asphalt pavement under traffic loads at the network level.
For further investigation, SAFEM will be developed with the application of various material
properties, such as viscoelasticity for asphalt and nonlinear elasticity for the sub-base of the
pavement. The influence of the temperature and moisture will be considered, which can be
realized by back-calculating the E-moduli of the test track at different controlled environmen-
tal condition. The ultimate aim is to extensively apply this system to on-site measurement.
ACKNOWLEDGEMENT
This paper is based on parts of the research projects carried out at the request of the Federal
Ministry of Transport and Digital Infrastructure, requested by the Federal Highway Research
Institute, under research projects No. 04.0259/2012/NGB and FE 88.0137/FE88.0138, as well
as parts of the research project carried out at the request of the German Research Foundation,
under research projects No. FOR 2089. The authors are solely responsible for the content.
REFERENCES
Bilodeau, J. & Doré, G. 2014. Stress distribution experienced under a portable light-weight deflectom-
eter loading plate, International Journal of Pavement Engineering, 15(6): 564–575.
Ferne, B.W. & Langdale, P. & Round, N. & Fairclough, R. 2009a. Development of a Calibration Proce-
dure for the U.K. Highways Agency Traffic-Speed Deflectometer, Transportation Research Record,
2093(1): 111–117.
Ferne, B.W. & Langdale, P. & Round, N. & Fairclough, R. 2009b. Development of the UK Highways
Agency Traffic Speed Deflectometer, Bearing Capacity of Roads, Railways and Airfields, London,
Champaign, IL: Taylor & Francis Group: 409–418.
FGSV 2009. Arbeitsanleitung zur Bestimmung des Steifigkeits- und Ermüdungsverhaltens von
Asphalten mit dem Spaltzug-Schwellversuch als Eingangsgröße in die Dimensionierung, AL Sp -
Asphalt 09. FGSV Verlag, Köln. (In German)
106
Fritz, J.J. 2002. Flexible Pavement Response Evaluation Using the Semi-Analytical Finite Element
Method, International Journal of Materials and Pavement Design, 3(2): 211–225.
Gohl, S. 2006. Vergleich der gemessenen mechanischen Beanspruchungen der Modellstraße der BASt
mit den Berechnungsergebnissen ausgewählter Programme, Diplomarbeit, TU Dresden, Professur
für Straßenbau. (In German)
Goktepe, A.B. & Agar, E. & Lav, A.H. 2006. Advances in Backcalculating the Mechanical Properties of
Flexible Pavements, Advances in Engineering Software, 37: 421–431.
Gopalakrishnan, K. & Thompson, M.R. & Manik, A. 2007. Rapid Finite-Element Based Airport Pave-
ment Moduli Solutions using Neural Networks, International Journal of Information and Math-
ematical Sciences, 3(1): 63–71.
Hu, S. & Hu, X. & Zhou, F. 2008. Using Semi-Analytical Finite Element Method to Evaluate Stress
Intensity Factors in Pavement Structure, Pavement Cracking: 637–646.
Khattak, M. & Mohammad, L. & Yuan, F. & Abadie, C. 2012. Variability of in-situ HMA volumetric
and mechanistic characteristics using non-destructive test (NDT): case study, International Journal
of Pavement Engineering, 13(2): 110–125.
Kim, J.R. & Kim, W.D. & Kim, S.J. 1994. Parallel Computing Using Semianalytical Finite Element
Method, AIAA Journal, 32(5).
Liu, P. & Wang, D. & Oeser, M. 2013. Leistungsfähige semi-analytische Methoden zur Berechnung von
Asphaltbefestigungen, Tangungsband, 3. Dresdner Asphalttage. (in German)
Liu, P. & Wang, D. & Oeser, M. 2015. The Application of Semi-Analytical Finite Element Method Cou-
pled with Infinite Element for Analysis of Asphalt Pavement Structural Response, Journal of Traffic
and Transportation Engineering (English Edition), 2(1): 48–58.
Liu, P. & Wang, D. & Oeser, M. & Chen, X. 2014. Einsatz der Semi-Analytischen Finite-Elemente-
Methode zur Beanspruchungszustände von Asphaltbefestigungen, Bauingenieur, 89(7/8): 333–339.
(in German)
Loizos, A. & Boukovalas, G. 2005. Pavement soil characterization using a dynamic stiffness model,
International Journal of Pavement Engineering, 6(1): 5–15.
Loizos, A. & Scarpas, A.T. 2005. Verification of falling weight deflectometer backanalysis using a
dynamic finite elements simulation, International Journal of Pavement Engineering, 6(2): 115–123.
Moffatt, M. & Martin, T. 2013. State-of-the-art Traffic Speed Deflectometer Practice, Austroads Pub-
lication. AP-T246–13, ARRB.
Oeser, M. 2010. Nichtlineare numerische Simulationsmodelle für Verkehrswegebefestigungen unter
Berücksichtigung von mechanischen, thermischen und hydraulischen Einwirkungen, Habilitation,
TU Dresden, Veröffentlichungen des Lehrstuhls für Statik, H. 18. (In German)
Pekcan, O. & Tutumluer, E. & Ghaboussi, J. 2010. Soft Computing Methodology to Determine Pave-
ment Thickness from Falling Weight Deflectometer Testing, Advances in Analysis, Modeling &
Design: 2621–2630, GeoFlorida.
Rabe, R. 2004. Bau einer instrumentierten Modellstraße in Asphaltbauweise zur messtechnischen Erfas-
sung der Beanspruchungssituation im Straßenaufbau, AP 03 342, interner Bericht, Bundesanstalt für
Straßenwesen, Bergisch Gladbach. (In German)
Rabe, R. 2007. Messtechnische Erfassung der Beanspruchungen im Straßenaufbau infolge LKW-Über-
fahrten über eine Modellstraße in Asphaltbauweise, AP 04 342, interner Bericht, Bundesanstalt für
Straßenwesen, Bergisch Gladbach. (In German)
Rabe, R. 2014. Angaben zum Aufbau der Modellstraße und Angabe von ausgewählten Ergebnissen und
Materialkennwerten, Bundesanstalt für Straßenwesen, Bergisch Gladbach. (In German)
Saltan, M. & Terzi, S. & Kücüksillle, E.U. 2011. Backcalculation of pavement layer moduli and Pois-
son’s ratio using data mining, Expert Systems with Applications 38: 2600–2608.
Wacker, B. 2015. Accelerated pavement testing program with the mobile load simulator MLS10 – tem-
perature analysis, Bituminous Mixtures and Pavements VI, Taylor & Francis Group: 687–692.
Wang, D. 2011. Schaffung des Bewertungshintergrunds zur Charakterisierung des Polierverhaltens der
einzelnen, gesteinsbildenden Minerale und zur Untersuchung des Griffigkeitsverhaltens der Min-
eralaggregate in Abhängigkeit von den Polierbedingungen, Aachener Mitteilungen Straßenwesen,
Erd- und Tunnelbau, Band 58. (In German)
Zhou, H. 2000. Comparison of backcalculated and laboratory measured moduli on AC and granular
base layer materials, NDT of pavements and backcalculation of moduli, 3. Special technical publica-
tion; STP 1375, Vol. 3–37: 161–172.
Zienkiewicz, O.C. & Taylor, R.L. 2000. The Finite Element Method: 468–492, Volume 1 The Basis, 5th
edition, Elsevier Butterworth-Heinemann, Oxford.
Zienkiewicz, O.C. & Taylor, R.L. 2005. The finite element method for solid and structural mechanics:
498–516, 6th edition, Elsevier Butterworth-Heinemann, Oxford.
107
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Chun-Hsing Ho
Department of Civil Engineering, Construction Management and Environmental Engineering,
Northern Arizona University, Arizona, USA
ABSTRACT: This paper presents a numerical modeling mechanism using a burger model
to formulate a representation function that can be used to characterize relaxing properties
and thermal cracking of asphalt mixtures. The objective of the paper is to (1) present a Burger
Model Based Representation (BMBR) function and validate its feasibility and application
using an asphalt paving project, and (2) advance characterizations of asphalt mixtures for
future research work. The representation function presented in the paper is structured with
a Maxwell model connecting with a Kelvin element known as a Burger model. An asphalt
paving project was used to validate the applicability of the BMBR function and compare its
prediction with the generalized Maxwell model. Based on numerical modeling results, the
BMBR function is capable of predicting thermal cracking properties of asphalt mixtures
within a reasonable accuracy as compared with the results from the generalized Maxwell
model.
1 INTRODUCTION
Hot mix asphalt concrete has been seen as a composite material made by asphalt binders,
aggregates, and air voids from which its mechanical performance is much more complicated
for prediction. It is known that at low temperature temperatures, asphalt pavement is seemed
to be a Linear Viscoelastic (LVE) material that representation functions must be used to
simulate its mechanical behaviors. In the past decades, a number of mathematical models
have been developed to predict LVE responses of asphalt mixtures using numerical methods
and computer programming techniques. However, these methods require mathematics skills
and substantial computer operating power which might be too complicated for prediction of
the LVE behavior of asphalt mixtures. One can argue that there are computer programs that
can be used by practitioners to facilitate their understanding on LVE performance of asphalt
pavements. Unfortunately, as of today, these computer programs are not commercially avail-
able in the industry, particularly in state agencies, that can be adopted by practitioners to run
LVE analysis.
The main complexity of using numerical analysis to predict LVE responses of asphalt
mixtures is the processes of Laplace transform and Laplace transform-inversion converting
in between creep compliance and relaxation modulus of the viscoelastic materials. As cur-
rently used, a Prony series function consisting of one Maxwell model linked with six Kelvin
elements was formed to produce a generalized Maxwell model. However, to generate a gener-
alized Maxwell model, advanced mathematical models associated with substantial computer
operating power must be used to perform the sophisticated processes of Laplace transform
and Laplace transform-inversion. This approach might not be practical for some practition-
ers to evaluate mechanical behavior of asphalt mixtures during pavement construction.
This paper presents a Burger Model Based Representation (BMBR) function with a simple
structural component aimed at helping practitioners predict relaxation modulus of asphalt
109
mixtures with less complicated analysis processes. The objective of the paper is to provide the
asphalt industry with an option for use in LVE analysis and to advance characterizations of
asphalt mixtures for future research work.
( )
N
1 t
D(t ) = D0 t + ∑ Di e i (1)
η i =1
where D0, Di, and η = Prony series parameters; and τi = retardation times.
In general, the Prony series function (Eq. 1) is composed by one Maxwell model (a spring
and a dashpot connected in series) connected with several Kelvin elements in parallel combina-
tions as shown in Figure 1. As can be seen from Eq. 1 and Figure 1, the number of Kelvin ele-
ments determines the complexity of the calculation of LVE analysis. The more Kelvin elements
involved in Eq. 1, the more sophisticated mathematical equations must be solved. As currently
110
Figure 1. Generalized Maxwell model. Figure 2. Components of a Burger model.
used, one Maxwell model and six Kelvin elements have been included in Eq. 1 in which there
are 14 unknown parameters (Do, η , Di and τi components) that need to be solved. Even if one
decade increments in τi components can be assumed to reduce the number of unknown param-
eters, the Prony series function (Eq. 1) still has 8 unknown parameters to be generated. Moreo-
ver, when converting creep compliance D( ˆ ( ) to relaxation modulus E(
ˆ ( ) of asphalt mixtures
in the Laplace domain, one of difficult issues encountered is the Laplace transform-inversion
process of the Prony series function because a direct Laplace transform-inversion of E(ˆ ( ) is not
available (Christenson 1998). To address this issue, numerous computer programs or numerical
analysis methods have been used to provide approximate approaches to invert Laplace trans-
form such that LVE responses of asphalt mixtures can be predicted. For example, Christensen
p
(2003) proposed an approximate method called “direct method”, which allows the inverse of
relaxation modulus E( ˆ ( ) to be approximately determined. Another approach by Findley et al.
(1989) is to simplify the Laplace transform-inversion process of Eq. 1 so as to reduce the com-
plexity of the LVE analysis. Their method results in a formula of relaxation modulus known as
the generalized Maxwell model (Eq. 2). Apparently, this model is an approximately superposi-
tion-oriented solution in the prediction of relaxation modulus.
N
E (t ) = ∑ Ci e −t / λi (2)
i =1
where Ci, and λi are constants that satisfy the partial fractions
More recently, a research by Ho and Romero (2012) provided an alternative function to
directly invert the Laplace transform of the Prony series function. Their function is struc-
tured with a Maxwell model connecting with a Kelvin element (Figure 2) known as a Burger
model. As can be seen, only three parameters (i.e., D0, D1, η) in the Burger model need to
be solved while the value of τ1 is assumed to be 0.01 (17). From the calculation point of
view, this Burger model based representation (BMBR) function significantly reduces the
complexity of parameter generations and does not involve with the sophisticated processes
of Laplace transform and Laplace transform-inversion, thus making the LVE analysis less
complicated. Their work was implemented in the laboratory and has not yet been applied in
the field. Thus, it would be of interest to evaluate the applications of the alternative function
in the prediction of relaxation modulus using field data. Since the alternative function is not
as complicated as the Prony series function, it would be feasible to provide highway staff,
engineers, and researchers with an option dedicated to facilitate the understanding of LVE
analysis and further research work.
111
3 BURGER MODEL BASED REPRESENTATION FUNCTION
⎛ 1 ⎞
D0 1 N
⎜ τi ⎟
Dˆ ( s )
η × s2
∑ Di ⎜ ⎟ (3)
s i =1
⎜ s ⎛ s + 1τ ⎞ ⎟
⎝ ⎝ i⎠ ⎠
where a caret (^) over the symbols shows that the quantity is now a function of Laplace
g model is applied, the creep
transform and s is a Laplace transform parameter. When a Burger
ˆ ( ) as:
compliance (Eq. 3) can be converted to a relaxation modulus E(
η × s + ητ
ˆ( ) =
E( 1
(4)
As 2 Bs C
where: A = D0 × η (5)
η η
B = D0 × + D1 × + 1
τ1 τ1 (6)
1
C=
τ1 (7)
Based on the factorization principle, Eq. 4 can be rearranged to result in a two-term
function:
⎛ B⎞ η η×B B2 C
⎜⎝ s + ⎟ − −
ˆ( ) = η × A⎠
22A τ1 2A 4 A2 A
E( 2
+ × 2
(8)
A ⎛ B⎞ ⎛ B2 C⎞ B2 C B ⎞ ⎛ B2 C ⎞
⎜⎝ s ⎟ A ⎜⎝ s + ⎟⎠ − ⎜⎝ 2 − ⎟⎠
2 ⎠ ⎝4 2 ⎠ 4 2 2 4A A
B
Let β , and (9)
2A
B2 C
α= − (10)
4 A2 A
⎛1 ⎞
η× − β⎟
η τ
⎝ 1 ⎠
E (t ) = ×e βt
cosh( t ) × u(t ) + ×e βt
× sinh(α ⋅ t ) × u(t ) (12)
A A×α
112
⎧1, if t > 0 ⎫
u(t ) = ⎨ ⎬ (13)
⎩0, if t ≤ 0⎭
Replacing hyperbolic sine and cosine functions with an exponential function, Eq. 12 (where
t > 0) is therefore formulated as:
E (t ) = C1 [e ( )t
+e ( )
] C2 × [ e ( )t
e ( )t
] (14)
where α, β parameters are referred to Eq. 9 and 10, and C1, C2 represent coefficients stated as:
η
C1 = (15)
2A
η
−η×β
τ1
C2 = (16)
2A × α
The difference between the generalized Maxwell model (Eq. 2) and the BMBR function
(Eq. 14) is their structural components. The BMBR is a hyperbolic function that provides
a theoretical solution to directly invert the Laplace transform of E( ˆ ( ) . While there is some
physical meaning attached to each parameter (springs and dashpots), the fact is that the gen-
eralized Maxwell model is an expansion series. The relaxation modulus of asphalt mixtures is
solved by superimposing a structural element (spring and dashpot) on the top of each element
until an approximate relaxation modulus is achieved. The question is that the more Kelvin
elements included in the components, the more sophisticated a constitutive equation must be
solved. With the composition of one Maxwell model and six Kelvin elements involving in a
Prony series function, 8 unknown Prony parameters need to be generated. This might appear
to be quite difficult for predicting relaxation modulus of asphalt mixtures unless advanced
computer programming techniques and numerical analyses are used. However, the BMBR
function (Eq. 14) has only three unknown variables (i.e., D0, D1, η) and does not involve in the
processes of Laplace transform and Laplace transform-inversion of a Prony series function,
so making the prediction of relaxation modulus less complicated than the generalized Max-
well model. Understanding the advantage of using the BMBR function in a LVE analysis, the
paper is therefore to demonstrate the applications of the BMBR function in the prediction of
relaxation modulus of asphalt mixtures using field data from a construction project.
NMAS 12.5-mm
Gradation Sieve Size, mm %, passing
19 100
12.5 93
9.5 83
4.75 60
2.36 38
1.18 26
0.300 14
0.075 6.2
Binder Grade PG 64–34
Design gyrations Ndes 75
Binder Content,% Pb 5.2
Design Air Voids,% Va 3.6
VMA,% 16.6
VFA,% 78.3
Dust Proportion 1.3
Aggregate Absorption 0.43
Max. Specific gravity Gmm 2.431
prediction of relaxation modulus of asphalt mixtures and to compare the LVE results with
the generalized Maxwell model.
1M + 6K 1 Maxwell and Generalized 6 (D0, D1, D2, D3, D4, D5, D6, and η)
6 Kelvin elements Maxwell model
1M + 1K 1 Maxwell and Generalized 3 (D0, D1, and η)
1 Kelvin element Maxwell model
BMBR 1 Maxwell and Alternative 3 (D0, D1, and η)
1 Kelvin element function
Model η
(Paving day) (sec./Mpa) D0 D1 D2 D3 D4 D5 D6
115
Figure 3. Relaxation moduli of three models on August 3 (L) and August 4 (R).
As described earlier, the two Prony series function (1M + 6K and 1M + 1K model) are
formulated based on an expansions series, which is, of course, a mathematical representation
of the observed behaviors of the materials. This type of structure is a superposition-oriented
model. As can be seen in Figure 3, the number of structural elements in the generalized
Maxwell model influences the accuracy of relaxation moduli which reflects on the significant
differences between the 1M + 6K and 1M + 1K model.
In contrast, the proposed model (BMBR function) shows better match with the 1M + 6K
model in Figure 3. It is observed that, upon loading, both models start with a slight dif-
ferent relaxation modulus. On August 3 the proposed model begins with 15.3 Mpa and
1M + 6K starts with 18.0 Mpa while the proposed model begins with 14.6 Mpa and the
1M + 6K model starts with 17.4 Mpa on August 4. However, on two paving days, the two
models diverge between 20 to 30 seconds and tend to merge toward the end of time, so the
relaxation modulus curves from both models, while not identical, are close to each other.
From the quality control point of view, the information shown in Figure 3 also clarify the
fact that pavement materials placed on these two days come from the same population,
meaning that there is no anomaly on the construction quality during the two paving days.
To further evaluate the accuracy between the proposed model (BMBR function) and the
generalized Maxwell model, differences of relaxation moduli among the three models were
calculated and compared. Two compared groups (proposed BMBR function versus 1M + 6K
and 1M + 1K versus 1M + 6K) are made with the intent to specifically quantify the difference
of relaxation modulus along the time domain. The results are shown in Figure 4. Speaking
for the first group (1M + 1K versus 1M + 6K), the differences between the two generalized
Maxwell models are significant due to the number of structural elements superimposed in an
expansions series. This phenomenon is, of course, expected. Given the comparisons between
the proposed model (BMBR) and the 1M + 6K model on two paving days, the accuracy curve
vibrates ranging from 0% to 20% along the time domain. The differences in relaxation moduli
of asphalt mixtures can be attributed to their structures of constitutive equations, hyperbolic
functions versus an expansions series. It should be noted that a strain-controlled relaxation
test was not used to validate the prediction accuracy provided by both models. The authors
do not know which model is correct to represent an accurate mechanical behavior of asphalt
materials. Since the generalized Maxwell model (1M + 6K) is one of commonly used mod-
els, we can assume the relaxation moduli determined by the 1M + 6K are reasonable and so
does the BMBR function. Both models are capable of representing LVE responses, but the
only difference is that the BMBR function is simple and less complicated than the 1M + 6K
model. This efficient computational procedure would give the proposed model (BMBR func-
tion) a favorable option for practitioners to predict relaxation moduli of asphalt materials.
It also brought attention that regardless of relaxation moduli predicted by proposed or
1M + 6K models, the materials from two paving days exhibit similar mechanical behaviors
which mean that the materials lasted in the filed have the same abilities in relaxing thermal-
induced stresses in pavements. While the main topic of the paper is not on the quality control
for asphalt construction, the BMBR function does show a future direction for research in the
quality control during pavement construction.
116
Figure 4. Differences between two compared groups on August 3 (L) and August 4 (R).
4.4 Discussions
Using asphalt mixture beams in the BBR has been studied in depth to characterize the effect
of aggregates in pavement performance. However, the stiffness and m-values provided by the
BBR tests only reveal the properties of asphalt materials with limited information based on
the fact that those stiffness and m-vales data do not specifically indicate the mechanical behav-
ior of asphalt materials. Nevertheless, if relaxation modulus analysis presented in the paper
can be associated with the BBR test results (stiffness and m-values), the integrated approach
(stiffness, m-values, and the alternative function) would advance a feasible implementation
plan for day-to-day mix design and quality control/quality assurance. For example, stiffness
and m-values data can be used to track construction quality and capture any anomaly in the
early stage during asphalt pavements. Whenever needed, predictions of relaxation modulus
of asphalt mixtures using the BMBR function can be performed shortly to evaluate the LVE
responses of the asphalt materials. This would give practitioners a comprehensive under-
standing on pavement performance during construction.
While this is an exploratory research work, it will be interesting to obtain creep compliance
data from IDT and DSR tests to determine relaxation moduli using the BMBR function. This
would further strengthen and calibrate the applications of the BMBR function in many ways.
5 CONCLUSIONS
The paper presents the BMBR function that provides a theoretical solution to predict relaxa-
tion modulus without going through the sophisticated processes of Laplace transform and
Laplace transform inversion. Throughout the comparison of relaxation moduli and accu-
racy among three models (1M+1K, 1M+6K, and BMBR function), the following conclu-
sions have been drawn:
i. The BMBR function is structured with one Maxwell model connecting with one Kelvin
element which requires only three unknown parameters to be solved. In comparison with
the generalized Maxwell model (eight unknown parameters), the BMBR function is sim-
ple and less complicated.
ii. Based on LVE analyses, the BMBR function is capable of predicting relaxation modulus
of asphalt mixtures with reasonable accuracy as compared with the generalized Maxwell
model.
iii. The difference of relaxation moduli between the 1M+6K model and the BMBR function
is their structural components, hyperbolic functions versus an expansion series. Given
that the expansions series function (i.e., 1M+6K and 1M+1K model) is a superposition-
oriented formula (approximate solution), the BMBR function provides a theoretical
approach to represent LVE responses of asphalt mixtures.
iv. The BMBR function shows good agreement with the 1M+6K model in predicting relaxa-
tion modulus of the asphalt materials. The relaxation modulus curves provided by both
models, while not identical, are very close to each other on two paving days.
117
v. For future research work, it will be interesting to obtain creep compliance data from IDT and
DSR tests to determine relaxation modulus of the asphalt materials using the BMBR func-
tion. This would further strengthen and calibrate the applications of the BMBR function.
REFERENCES
Abbas, A. R. & Papagiannakis, A. T. and Masad, E. A. 2004. Linear and Nonlinear Viscoelastic Analysis
of the Microstructure of Asphalt Concretes. Journal of Materials in Civil Engineering: 16(2)133–139.
American Association of State Highway and Transportation Officials. 2009. Determining the Flexural
Creep Stiffness of Asphalt Binder Using the Bending Beam Rheometer (BBR). Standard Specifications
for Transportation Materials and Methods of Sampling and Testing T 313. AASHTO 29th edition.
American Society of Testing and Materials. 2008. Determining the Flexural Creep Stiffness of Asphalt
Binder Using the Bending Beam Rheometer (BBR). D6648-08, ASTM.
Christensen, R.M. Theory of viscoelasticity. 2nd Edition, Dover Publications Inc., New York, 2003. M.
Christensen, D. W. and Anderson, D. A. 1992. Interpretation of Dynamic Mechanical Test Data for
Paving Grade Asphalt Cements. Journal of the Association of Asphalt Paving Technologist: 67–116.
Christensen, D. W. 1998. Analysis of Creep data from Indirect tension Test on Asphalt Concrete. Jour-
nal of the Association of Asphalt Paving Technologist: 458–492
Chehab, G. R., and Kim, Y. R. 2008. Interrelationships among asphalt concrete stiffness. Kim, Y. R.
Eds, Modeling of Asphalt Concrete, McGraw-Hill Construction:139–159.
Elseifi, M. A., Al-Qadi, I. L., and Yoo, P. J. 2006. Viscoelastic Modeling and Field Validation of Flexible
Pavements. Journal of Engineering Mechanics: 132(2)172–178.
Findley, W. N., Lai, J. S., and Onaran, K. 1989. Creep and relaxation of nonlinear viscoelastic materials.
Dover Publications Inc., New York.
Ho, C.H., and Romero, P. 2011. Using Asphalt Mixture Beams in the Bending Beam Rheometer: Exper-
imental and Numerical Approach. Journal of Road Materials and Pavement Design: 12(2) 293–314.
Ho, C.H., and Romero, P. 2012. An Alternative Function to Represent Relaxation Modulus of Viscoe-
lastic Materials. Journal of Materials in Civil Engineering: 24(24)152–158.
Ho, C.H., and Romero, P. 2012. Using Asphalt Mixture Beams in the Bending Beam Rheometer for
Quality Control: Utah Experience. Journal of Transportation Research Board: (2268)92–97.
Kim, Y.R. and Little, D. N. 2004. Linear Viscoelastic Analysis of Asphalt Mastics. Journal of Materials
in Civil Engineering:16(2)122–132.
Kim, J. S., Scholar, G. A., and Kim, S. G. 2008. Determination of Accurate Creep Compliance and
Relaxation Modulus at a Single Temperature for Viscoelastic Solids. Journal of Materials in Civil
Engineering: 20(2)147–156.
Kim, J. S., Roque, R., and Byron, T. 2009. Viscoelastic Analysis of Flexible Pavements and its Effects on
Top-Down Cracking. Journal of Materials in Civil Engineering: 21(7)324–332.
Malkin, A. Y., and Masalova, I. 2001. From Dynamic Modulus via Different Relaxation Spectra to
Relaxation and Creep Functions. Journal of Rheological Acta: (40)261–271.
Marasteanu, M. O., and Anderson, D. A. 2000. Establishing linear viscoelastic conditions for asphalt
binders. Journal of Transportation Research Record: (1728)1–6.
Nikonov, A., Davies, A.R., and Emri, I. 2005. The determination of creep and relaxation modulus func-
tions from a single experiment. Journal of Rheology: 49(6)1193–1211.
Park, S.W., and Kim, Y.R. 1999. Interconversion between Relaxation Modulus and Creep Compliance
for Viscoelastic Solids. Journal of Materials in Civil Engineering: 11(1)76–82.
Park, S. W., and Schapery, R. A. 1999. Methods of Interconversion between Linear Viscoelastic Mate-
rial Functions. Part I—A numerical method based on Prony series. International Journal of Solids
and Structures: (36)1653–1675.
Romero, P., Ho, C.H., and VanFrank, K. 2011. Development of Methods to Control Cold Temperature
and Fatigue Cracking for Asphalt Mixtures. Final Report, Contract No. UT-10.08, Utah Depart-
ment of Transportation.
Zofka, A., Marasteanu, M. O., Li, Xinjun, Clyne, T. R., McGraw, J. 2005. Simple Method to Obtain
Asphalt Binders Low Temperature Properties from Asphalt Mixtures Properties. Journal of the
Association of Asphalt Paving Technologists: (74)255–282.
Zofka, A., Marasteanu, M., and Turos, M. 2008. Determination of Asphalt Mixture Creep Compliance
at Low Temperatures Using Thin Beam Specimens. Journal of the Transportation Research Board:
(2057)134–139.
Velásquez, R., Marasteanu, M. Turos, M, and Labuz, J. 2009. Effect of Beam Size on the Creep Stiff-
ness of Asphalt Mixtures at Low Temperatures. Advanced Testing and Characterization of Bituminous
Materials: (1)313–322.
118
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The AASHTO: TP 79-10 standard stipulates the test procedure for the meas-
urement of dynamic modulus of bituminous mixtures. In this investigation, dynamic modu-
lus data of two types of bituminous concrete mixtures (an unmodified binder and crumb
rubber modified binder) are studied using TP-79-10 protocol. The experiments showed that
the total strain decreased with time at higher temperatures and at lower frequencies. Also,
for some mixtures, the phase lag decreased with decreasing frequency. While many studies
reported such observations, these anomalous results are normally ascribed to the errors in
measurement mechanisms and have not been explained within the context of the material
behavior. Here it is shown that Burgers’ viscoelastic model exhibits these phenomena. Also,
the need for data processing techniques using the complete loading history and a viscoelastic
model is emphasized.
1 INTRODUCTION
Dynamic Modulus test is one of the tests used for the mechanical characterization of the
bituminous material. Pavement design as per M-EPDG (NCHRP: 1-37A 2004) uses dynamic
modulus as an input parameter in stress analysis. The Asphalt Mixture Performance Tester
(AMPT) supports the bituminous mix design procedures by providing the dynamic modulus
119
values to be used in the structural design of bituminous pavements. Thus, accurate determi-
nation of dynamic modulus seems to be essential for adequate structural design of bitumi-
nous pavements.
AASHTO: TP-79 (2010) deals with the methodology to be adopted for finding dynamic
modulus. As per the AASHTO: TP-79 (2010) protocol, a specimen, at a specific test tem-
perature, is subjected to a haversine compression uniaxial load at ten frequencies (ranging
from 25 Hz to 0.01 Hz) continuously without any rest period between the frequencies. The
load levels are chosen such that the average strain range is 75 to 125 micro-strains for any
frequency of testing. This is based on the assumption that the material response is linear
viscoelastic within this range. For each frequency, ten conditioning cycles are applied fol-
lowed by ten test cycles. The test starts from the highest frequency and proceeds towards
the lowest. An initial modulus (seed modulus) value should be specified to the equipment
software and this is used to determine the load level to be applied for the first test frequency.
Since the average strain range is set between 75 to 125 micro-strains, the software calculates
the required load and applies five initial conditioning cycles. After the fifth conditioning
cycle, the calculated dynamic modulus value is used to readjust the load that is to be applied
to maintain the required target micro-strain and then further five conditioning cycles are
applied followed by ten test cycles. For the next frequency, 75% of the dynamic modulus of
the preceding frequency is chosen as the seed modulus. This process will continue for all the
selected frequencies. For the test cycles, the applied load and the resulting axial deformation
from three on-sample transducers are recorded as a function of time. However the data for
the ten preconditioning cycles cannot be retrieved from the equipment. The data from the
test cycles is used to calculate the dynamic modulus, phase lag and a number of data quality
measurements.
The post-processing of data as per the AASHTO: TP-79 (2010) protocol involves curve
fitting the stress and strain data to a function of time of the form: a0 + a1t + a2cos (2πωt)
+ a3sin (2πωt), where ω is the frequency of loading, ai’s are to be determined constants.
Then, the dynamic modulus and phase lag are computed from the determined constants,
ai. This method closely follows the post-processing method adopted for an earlier protocol:
AASHTO: TP-62 (2009), with the addition of a drift term (a1t). The experimental procedure
associated with AASHTO: TP-79 (2010) and the prescribed post-processing method are not
compatible. The curve fitting approach followed for determination of dynamic modulus and
phase lag does not take into account the viscoelastic nature of the material response and
this leads to inconsistencies in data interpretation. This will be explained in detail in this
investigation.
2 EXPERIMENTAL STUDY
The bituminous mixture employed for the study is Bituminous Concrete (BC—Grade 2)
with a nominal maximum aggregate size of 13.2 mm. The BC mix used in the current study
had mid-gradation with 5% binder content. Two binders were used in this study, an unmodi-
fied bitumen and this base bitumen modified with crumb rubber. The base bitumen used
for modification is an air blown bitumen of VG30 grade as per IS: 73-2013 (2013). The
crumb rubber modified bitumen is classified as CRMB-60 (IS: 15462-04 2004). In this manu-
script, the mixes with VG30 and CRMB-60 are further referred to as just VG30-BC and
CRMB-BC. The summary of binder properties is tabulated in Table 1. The performance
grade test was conducted as per ASTM: D6373-07 (2007) and the critical temperatures for
VG30 and CRMB are tabulated in Table 2.
The bituminous mixtures were short term aged for 4 hours at mixing temperature and
half an hour at compaction temperature before compaction in order to account for the aging
that happens in field during mixing and compaction process. VG30-BC was mixed at 165ºC
and compacted at 150ºC while CRMB-BC was mixed at 178ºC and compacted at 159ºC.
Compaction was carried out in the shear compactor (ASTM: D7981-15 2015) by applying
a constant vertical load of 600 kPa and a constant shear angle of 4°. The fully cast beam
120
Table 1. Summary of binder properties.
VG30
43 33 70 72 10.2 20.8 75 69
was 450 mm long, 150 mm wide and the height varied between 165–185 mm based on the
required air voids. Three cylindrical samples of 100 mm diameter and 150 mm height were
cored from the beam. The present study used cored samples with 4% air voids.
Dynamic modulus test was conducted as per AASHTO: TP-79 (2010) protocol and the
test was performed in AMPT. The test matrix for dynamic modulus test is given in Table 3.
The test was carried out at six temperatures from 5 to 55ºC in steps of 10ºC.
122
Figure 2. Phase lag vs frequency.
5 – –
15 – –
25 0.01 0.1
35 0.2 1
45 2 5
55 10 10
The decreasing trend in phase lag at higher temperatures has been quoted as an anomaly in
many previous literatures including the NCHRP-629 report. As part of the ruggedness test-
ing, experiments were carried out on a 9.5 mm dense-graded mixture with a PG64-22 binder
and a 12.5 mm stone matrix asphalt mixture with a PG76-22 binder (Bonaquist 2008). When
such data was analyzed, and when such decreasing trend in phase lag at higher temperature
was seen, it was ascribed to the LVDT drift. In their study Lee et al. (2007) also obtained
similar trend of higher phase angles at higher frequencies when the mixes were tested at
high temperatures. However for low temperatures phase angle showed a decreasing trend
with increasing frequency. The study by Lee et al. (2007) used six different mixtures with two
different binders namely PG 58-22 and PG 64-16. Their study also showed that at the same
loading frequency the phase angle increased with the increase in test temperature from −10
to 21ºC while at 40 and 55ºC, the phase angle decreased with the increase in test temperature.
For the current study the frequency at which the so called ‘anomaly’ occurs for both the mixes
is tabulated in Table 4.
It is expected that the trends observed for phase lag is a manifestation of similar trends
in the experimental strains and hence the time lag calculations. Hence an attempt is made
to relate the observed phenomenon in the phase lag computations to the time lag between
peak stress and strain and in turn to the experimental strains. Figure 3 presents the time lag
between peak stress and strain at 0.01 Hz for VG30-BC for all the test temperatures. It is
expected that as temperature increases the time lag between the occurrence of peak stress and
peak strain increase. However in this study it is seen that at a frequency of 0.01 Hz, for both
VG30-BC and CRMB-BC, the time lag increases with temperature up to 25ºC and further
decreases.
123
Figure 3. Time lag between stress and strain for VG30-BC at 0.01 Hz.
Frequency (Hz)
Data quality
indicators at 55ºC 25 20 10 5 2 1 0.5 0.2 0.1 0.01
Load standard 4.9 5.1 4.8 4.6 10.7 11.1 1.8 1.3 1.3 1.9
error (%)
Average 4.6 11.2 5.9 2.5 7.2 9.1 5 5.2 5.6 6.8
deformation
standard error
(%)
Deformation 11.7 12.6 13.8 15.6 17.6 18 18.9 20.4 21.8 24.8
uniformity (%)
Phase uniformity 1.6 1.7 1.8 2.1 3 2.7 3.5 4.4 5.1 6.9
(Degrees)
haversine load to the specimen by the equipment and for a good quality data, the maximum
limit specified is 10%. Deformation standard error is a measure of the proximity of the meas-
ured strains to a haversine, and for a good quality data the maximum limit specified is 10%.
Deformation uniformity indicates the proximity of individual deformation measurements
made on a sample to each other with a maximum accepted value of 20%. Phase uniform-
ity indicates the proximity of individual phase lag measurements made on a sample to each
other with a maximum accepted value of 3%.
The data quality indicators for VG30-BC at 55ºC are tabulated as Table 5. It can be
observed that even when the data quality indicators are well within the limits, the phase lag
decrease can happen (for instance 10 and 5 Hz). Since the data quality indicators clearly cap-
ture the required data accuracy, the phase lag decrease cannot be claimed as an anomaly or as
drift of LVDT. For viscoelastic materials, especially when there is a solid to fluid like behavior
transition, the observed variations in phase and time lags are known to occur. Findley et al.
(1989) show the observed variations in the phase angle to occur when one uses the Burgers’
model.
124
Figure 4. Strain vs time plot for VG30 at 55ºC.
Since the post-processing as per the AASHTO: TP-79 (2010) protocol follows a curve fit-
ting approach, the viscoelastic nature of the deformation is not taken into consideration. To
capture the true material response to continuous loading across frequencies, the experimental
strains have to be predicted using an appropriate viscoelastic model.
As per the AASHTO: TP-79 (2010) protocol, the material is loaded continuously across
ten frequencies. However, in order to maintain the strain between 75 and 125 micro strains to
ensure linear viscoelastic response, the maximum magnitude of the load has to be changed
between successive frequencies. Since, dynamic modulus of the tested material as a func-
tion of frequency is not known, how this change in the maximum magnitude of the load
needs to be made is not known a priori. Therefore, the first five cycles is utilized to estimate
this dynamic modulus at the tested frequency. This is achieved by testing the specimen to a
maximum load magnitude needed to cause a 75 micro strain arrived at by assuming a modu-
lus value obtained by reducing the dynamic modulus value from that measured during the
previous test at a higher frequency by 25 percent or as the prescribed modulus value for the
first frequency of testing. Using this determined value of the dynamic modulus, the maxi-
mum magnitude of the load required to cause 75 micro-strain is reassessed and 15 cycles of
haversine load with this as the peak load is applied. The load and strain time history from the
125
last 10 cycles is only used for further processing. This approach has the following drawbacks:
(a) since, bituminous concrete behaves like a viscoelastic material, the strain realized for a
particular load at a particular time depends on the loading history. For the same magnitude
of applied load the measured strains would be different if the loading histories were differ-
ent. Hence, as demonstrated next, one cannot ignore the loading history, specifically the
first ten cycles where the magnitude of the applied load is adjusted to ensure that only 75 to
125 micro-strain is experienced by the specimen. (b) The dynamic modulus in a nonlinear
viscoelastic material depends even on the applied load magnitude apart from the frequency.
Consequently, if the prescribed modulus value is inappropriate, high magnitude of strains
would occur during the first 5 cycles. Therefore, the appropriateness of the dynamic modulus
obtained from this 5 cycle data is not evident. Further, the time taken by the specimen to
recover its initial state after experiencing this high strain depends on its retardation time(s).
In light of this, any subsequent strain measured after initial high strains would have a recov-
ery component added to its current response and this causes the total strain to decrease.
t ∂J(( − τ )
ε ( ) J ( )σ ( ) + ∫ σ ( )dτ , (1)
0 ∂τ
1 t −τ 1 −(t
(t ) /τ
J (t )= + + (1 − e c ), (2)
E η E
1 1 2
E1, E2 are the spring constants; η1 is the viscosity of the dashpot; τc is the retardation time
and σ is the applied stress. Here E1 = 160 MPa, E2 = 15 MPa, η1 = 300 GPa⋅s, ηc = 20 s.
Since, the interest here is to demonstrate the need for considering the entire load history
in the computation of the strain, for a given value of the material parameters, the strain is
computed using two approaches. In both the approaches, the strain time history correspond-
ing to 1 Hz frequency for a given load time history is computed using the Burgers’ model
with the value of its material parameters as given above. In approach 1, the load time his-
tory corresponding to the last 10 cycles of 1 Hz frequency is alone used to predict the strain
time history. Specifically, the stress and strain at the start of the 10 loading cycles of 1 Hz
frequency is taken as zero. Thus, in this approach the load time history prior to this loading is
not considered. In approach 2, the entire load time history from the start of the experiment is
considered to predict the current strain. In particular, the loading cycles used to estimate the
dynamic modulus for each frequencies tested is also considered. The stress and strain only at
the beginning of the experiment is taken as zero.
Figure 5 shows the comparison between the model predictions using the two approaches
and experimentally observed strains for VG30-BC at 55ºC. In the experiment, as well as in
approach 2 the strain at the start of recording of the response at 1 Hz frequency (in figure,
time = 0s) would not be zero. For convenience in plotting, the strain at a given time, t is taken
as the difference in the actual strain at time t and the strain at the start of the last 10 cycles of
1 Hz haversine loading. Hence, the strain at the beginning of the 1 Hz loading cycle is zero.
It can be observed from figure 5 that the strains predicted using the first approach increase
while that from the second approach decrease over cycles. This demonstrates the need to
consider the entire load history. It can also be seen from figure 5 that the second approach
126
Figure 5. Comparison of normalised experimental vs predicted strains at 55ºC, 1 Hz for VG30-BC.
qualitatively predicts the trends in the experimental strains. One needs to go through a formal
model development for the tested material to achieve quantitative comparisons between the
model prediction and the experimentally measured strains.
4 CONCLUSIONS
From the current study it is inferred that the trend observed for the strains which is generally
termed as anomaly is not really an experimental anomaly. The observations are consistent
with the response of the viscoelastic material. Further, it is shown that to capture the material
response to continuous loading across frequencies as per the AASHTO: TP-79 (2010) pro-
tocol, the entire load history has to be considered. In summary, when the material is tested
as per AASHTO: TP-79 (2010) protocol, the post-processing of data should be carried out
using appropriate viscoelastic models rather than using the existing viscoelastic model inde-
pendent data processing methods.
ACKNOWLEDGEMENT
The authors thank Department of Science and Technology, Govt. of India for funding this
investigation. The grant number is DST/TSG/STS/2011/46. The authors acknowledge the
technical assistance provided by M/s IPC Controls, Australia during the conduct of the
experiments.
REFERENCES
AASHTO: TP-62. 2009. Standard method of test for determining the dynamic modulus of hot mix
asphalt concrete mixtures. American Association of State Highway and Transportation Officials.
Washington D.C., USA.
AASHTO: TP-79. 2010. Standard method of test for determining the dynamic modulus and flow
number for hot mix asphalt using the asphalt mixture performance tester (AMPT). American Asso-
ciation of State Highway and Transportation Officials. Washington D.C., USA.
ASTM: D6373-07. 2007. Standard specification for performance graded asphalt binder. ASTM Interna-
tional. West Conshohocken, Pennsylvania, USA.
127
ASTM: D7981-15. 2015. Standard practice for compaction of prismatic asphalt specimens by means of
the shear box compactor. ASTM International. West Conshohocken, Pennsylvania, USA.
Bonaquist, R. 2008. NCHRP Report-629, Ruggedness testing of the dynamic modulus and flow number
tests with the simple performance tester. Transportation Research Board. Washington D.C., USA.
Findley, W.N., Lai, J.S., & Onaran, K. 1989. Creep and relaxation of nonlinear viscolastic materials
-with an introduction to linear viscoelasticity. New York: Dover Publications Inc.
Gibson, N., Qi, X., Shenoy, A., Al-Khateeb, G., Kutay, M.E., Andriescu, A., Stuart, K., Youtcheff, J. &
Harman, T. 2012. Full-scale accelerated performance testing for superpave and structural validation.
Technical report:FHWA-HRT-11-045. Federal Highway Administration.
IS: 15462-04. 2004. Indian standard specification for polymer and rubber modified bitumen. Bureau of
Indian Standards. New Delhi.
IS: 73-2013. 2013. Specification for paving bitumen. Bureau of Indian Standards. New Delhi.
Lee, K., Kim, H., Kim, N. & Kim, Y. 2007. Dynamic modulus of asphalt mixtures for development of
Korean pavement design guide. Journal of Testing and Evaluation, 35 (2).
NCHRP: 1-37A. 2004. Guidelines for the mechanistic-empirical design for new and rehabilitated pave-
ment structures. Transportation Research Board. Washington, D.C., USA.
128
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
Nowadays, the environmental problems caused by waste tires growing are becoming severer,
and large quantities of low-temperature cracking on pavement are appearing. The technolo-
gies that adding scrap tire rubber powder into asphalt and its mixture to produce Rubber
Asphalt (RA) and RA mixture is arousing researchers’ interest. The technology can relieved
environmental problems caused by waste tires, also in addition, the Rubber Asphalt (RA)
mixture pavement has better performance such as high-temperature rutting resistance, low
temperature cracking resistance, fatigue resistance, etc. However, limitations still exact, espe-
cially on the low-temperature cracking resistance of RA.
Up to today, the conventional test methods for original asphalt specifications are not suf-
ficient in characterizing modified asphalt, especially crumb rubber modified asphalt. Speci-
fication used in china recommends five-degree ductility as the low-temperature evaluation
index. Relevant literature suggested (Wang et al. 2008; Sun et al. 2010; Cong et al. 2013) that
there were some limitations for five-degree ductility to evaluate the low-temperature proper-
ties of modified asphalt especially for RA. Reasons are as follows:
1. Five-degree ductility is an empirical index, it can characterize the low-temperature per-
formance of pure asphalt, but for RA with more complex structure and more complex
performance, it lacks theoretical basis and needs more theoretical studies.
2. In north of china, the actual service temperature of the pavement in winter was far lower
than five degree. It means that taking five-degree ductility test couldn't describe the low-
temperature properties of asphalt at this area. Meanwhile the testing material was used to
be taken original asphalt instead of asphalt aged by RTFOT and PAV, and this couldn’t
reflect actual performance of RA mixture pavement at low temperature.
129
3. Due to the existence of rubber powder, the stress concentration phenomenon happened
around rubber powder which existed in RA when taking ductility test, so As a result, a
poor reproducibility of test results would be happened; on the other hand, stress concen-
tration phenomenon would lead to a smaller result.
4. The level of the operation difficulty in ductility test at 5 °C is much harder than the test at
10 C or 15°C.
Based on the aforementioned limitations and shortcomings, conducting the study on the
low-temperature properties evaluation indexes of RA was necessary.
According to the aforementioned research background (Anderson et al. 1999; Feng et al. 2008;
He et al. 2010; Bahia et al. 2013), considering the low-temperature performance evaluation
index of RA, the study which carried on was aimed to find better low-temperature indexes
to evaluate the low-temperature properties of RA so that a more consummate and effective
low-temperature properties evaluation system could be built to evaluate the low-temperature
properties of RA reasonably and exactly. It would make sense to provide meaningful and posi-
tive reference for application of RA and to promote RA into utilize in cold areas.
130
Table 2. Codes of different RA.
4 TEST METHODS
In this paper, ductility test, dynamic frequency sweep test, Bending Beam Rheometer test
(BBR) and Differential Scanning Calorimetry test (DSC) were used to investigate the low-
temperature properties of RA.
132
Figure 2. Complex modulus-frequency curve Figure 3. Loss modulus-frequency curve under
under −15°C. −15°C.
certain extent, phase angle (δ ) index considered complex modulus and loss modulus at the
same time. Bigger phase angle at low temperature meant better low temperature rheological
behavior. Phase angle-frequency curves of RAs were shown in Figure 4.
Figure 4 showed that differences existed between different curves, RAs showed bigger
phase angle compared to neat asphalts which meant a better low temperature rheological
behavior after adding rubber powder into asphalt. However, some δ values were almost the
same and the differences were too small to distinguish them. What’s more, phase angle index
didn’t completely take elastic component into consideration on rheological properties of RA.
Recommending phase angle index as the low-temperature evaluation index of RA needed
further analyzing and studying.
On the contrary, the test results at −24°C and −30°C differed from each other. Bitumen at
very low temperature characterized breakable and existed hardening phenomenon. The
hardening of asphalts had a negative impact on low-temperature properties of bitumen,
and it led to a larger S and a smaller m which beyond the recommended requirements of
ASTM D 6648. Besides, it was more difficult to take BBR test at lower temperature and
an adverse effect on long-term use of BBR instrument existed which led to restrictions
on its application and promotion. S and m at −18°C not only met the requirement of
SHRP, but also could clearly distinguish the low-temperature properties of RA and neat
asphalt. At the same time, the difference between different RAs was obvious. Based on the
time temperature equivalence principle, test temperature at −18°C was actually equivalent
with actual temperature of asphalt pavement at −28°C which was consistent with pavement
temperature in winter in the northeast of china. As a consequence, S and m measured at
−18°C was recommended as low-temperature evaluation indexes for a correlation analysis
with mixture indicator.
Simultaneously, in accordance with the definition of Performance Grade temperature, PG
temperature of each binder was calculated according to Figure 5 and Figure 6. PG tempera-
ture of RAs improved one level compared with neat asphalts. PG grade of each RAs were all
belong to PG-28, while PG temperature of neat asphalts were graded as PG-22. Obviously,
PG temperature didn’t apply to evaluate the low-temperature properties of RAs.
In order to distinguish low-temperature properties differences among RAs, Continuous
Grade temperature (ST) is used to characterize low-temperature properties of RAs. Gen-
erally, Smaller ST meant a better low-temperature performance of RAs. ST indicator had
some advantages. It had definite physical meaning and better precision, it also took flow
deformation ability and stress relaxation ability of bitumen into consideration which avoid
one-sidedness by a single S or m. According to definition of ST. ST values were calculated
shown in Figure 7.
There were lower ST values for RA from Figure 7. Compared to PG temperature, ST made
big differences among different RAs. Asphalts which labeled as 110–40, 110–60, 90–40, 90–60
had lower ST values which meant preferable low-temperature anti-cracking performance,
while RAs modified with No.70 neat asphalt had larger ST values which meant a poor low-
temperature anti-cracking performance at low temperature. At the same time, asphalts modi-
fied by 40 mesh rubber powder had lower ST values.
the perspective of micro-structure, it also had definite physical interpretation and high meas-
urement accuracy. Generally speaking, lower Tg meant a better low-temperature property of
RA (Bahia 2001). Results of DSC test were shown in Figure 8 as follows. From Figure 8, Tg
reduced significantly after rubber powder added into asphalt.
In order to analyze the sensitivity of Tg index, two-factor variance analysis method with
interaction was taken to analyze the sensitivity on Tg of different RAs. The analysis results
are shown in Table 3. From Table 3, it could be seen that the mark number of asphalt and
the fineness of rubber powder and the interaction had highly significant impacts on the glass
transition temperature under the level of 0.01. That indicates that Tg index was sensitive
between different kinds of RAs. Considering the physical interpretation of the Tg index and
the sensitivity to different RAs, Tg index was recommended as an evaluation index to evaluate
the low-temperature properties of RAs.
Indexes X1 X2 X3 X4 X5 X6
From Table 5, Tg index had strongest relevance with flexural strain energy density which the
coefficient reached to 0.93 showing a better degree of association with flexural strain energy
density and validated the applicability and accuracy to appreciate the low-temperature proper-
ties of RA. Compared to m and S indexes based on BBR test, ST index had a stronger relevance
with mixtures which reached 0.88, it was found that ST index which takes the flow deformation
ability and the stress relaxation ability of bitumen into consideration avoids one-sidedness of
a single S or m. If we couldn’t conduct DSC test because of limiting by conditions, ST index
could be completely recommended as an evaluation index to evaluate the low-temperature
cracking resistance of RA. At the same time, as δ index obtained by dynamic frequency scan-
ning test also had strong relevance with flexural strain energy density, it meant a positive corre-
lation between distortion resistant ability and the low-temperature cracking resistance of RA.
However, a poorer relevance of the five-degree ductility further pointed out the limitations of
five-degree ductility index on evaluating the low-temperature properties of RA.
6 CONCLUSIONS
In the view of limitations of the low-temperature properties evaluation index of RA, dif-
ferent kinds of RA was prepared in the laboratory, a number of standard test methods for
the low-temperature properties evaluation of asphalt rubber have investigated and the low-
temperature properties evaluation index of RA was recommended. The following main con-
clusions were made:
1. The low-temperature properties of RAs improved a lot with the rubber powder adding
into asphalt especially for 40 mesh rubber powder;
2. Compared to Ductility test, the low-temperature rheological properties test and DSC test
are more suitable to determine the difference between different RAs;
3. Through grey relational analysis with flexural strain energy density of mixtures, the appli-
cability and accuracy for Tg index and ST index to evaluate the low-temperature properties
of RA were validated. And finally, Tg index and ST index were recommended as the low-
temperature properties evaluation indexes of RA.
Future research will consider effects of physical hardening on the low-temperature per-
formance of RAs and mixture validation will be going on for building a perfect low-temper-
ature performance evaluation system of RAs. Finally, the authors are grateful to the financial
support by National Science Fund for Distinguished Young Scholars of China (51225803)
and the Fundamental Research Funds for the Central Universities (HIT.NSRIF.201662).
REFERENCES
Anderson DA & Marasteanu MO 1999. Physical hardening of asphalt binders relative to their glass
transition temperatures. Transportation Research Record: Paper No. 99–1547.
Bahia Hu & Hanson DI & Zeng M, et al. 2001. Characterization of Modified Asphalt Binders in Super-
pave Mix Design. Transportation Research Board, 459: 176–179.
Baha VK & Mehmet Y & Alaaddin G 2013. Evaluation of Low-Temperature and Elastic Properties of
Crumb Rubber and SBS Modified Bitumen and Mixtures. Journal of Materials in Civil Engineering,
25(02): 257–265.
Cong PL & Xun PJ & Xing ML & Chen SF 2013. Investigation of asphalt binder containing various
crumb rubbers and asphalts. Construction and Building Materials, Volume 40: 632–641.
137
Deng JL 1984. Theory and method of social economic grey system. Social Sciences in China, Volume
06: 47–49.
Feng ZL & Cao RJ & Jia Y 2009. The study using characteristic temperature to evaluate the
low-temperature performance of bitumen. Journal of China & Foreign Highway, 29(01): 245–247.
Geng H & Clopotel CS & Bahia HU 2013. Effects of high modulus asphalt binders on performance of
typical asphalt pavement structures. Construction and Building Materials, 44 (01): 207–213.
He ZY & Lu ZF & Zhang WW 2010. Performance Study on Rubber Modified Asphalt of Waste Tire.
ICCTP: 3272–3278.
Nejad F & Aghajani P & Modarres A & Firoozifar H 2012. Investigating the properties of crumb rubber
modified bitumen using classic and SHRP testing methods. J. Hazard. Mater, 26(01): 481–489.
Sengoz B & Isikyakar G 2008. Analysis of styrene-butadiene-styrene polymer modified bitumen using
fluorescent microscopy and conventional test methods. J. Hazard. Mater, 150(01): 24–32.
Sun XW & Tang S & Wang JJ 2010. Research and analysis of influence factors on ductility of RA at low
temperature. Journal of China & Foreign, 20(05): 310–313.
Tan YQ & Zhang L & Xu HN 2012. Evaluation of low-temperature performance of asphalt paving
mixtures. Cold Regions Science and Technology, Volume 70: 107–112.
Tan YQ & Guo M 2013. Study on the phase behavior of asphalt mastic. Construction and Building
Materials, Volume 47: 311–317.
Wang XD & Li MJ & Lu KJ 2008. The Applied Technology of the Crumb Rubber in the Asphalt and
Mixture. China Communications Press, Beijing.
Zhang XY & Hu GY & Tan YQ 2005. Study of Low Temperature Anti-cracking Performance of
Diatomite Modified Asphalt Mixture. Journal of Highway and Transportation Research and
Development, 23(04): 11–13.
138
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Mingliang Li
Research Institute of Highway Ministry of Transport, Beijing, P.R. China
Bin Xu
Dalian University of Technology, Dalian, Liaoning, P.R. China
1 INTRODUCTION
Porous asphalt is defined as a wearing course with a high stone content (typically 81–85%)
and a high air voids content (usually > 20%). Because of the special structure, porous asphalt
has the advantages of noise reduction, improving anti-slipping performance and reducing
splash and spray in rainy days. In Europe, porous asphalt was developed by the Transport
Research Laboratory (TRL) in the UK in the late 1950s for use on airport runways. It has
been used on highways since the beginning of the 1980s in a number of countries. In the
Netherlands, it was regulated as the standard road surface for highways since the end of the
1980s. At present, more than 90% of the Dutch primary road network has this type of sur-
face layer for the purpose of noise reduction (COWI 2006, Li 2013, Molenaar et al. 2006).
In China, trial sections of porous asphalt were constructed in 1980s but not applied in
practical engineering at that time due to the lower structural strength (Cao et al. 2009). In
21st century, high-viscosity modified bitumen was introduced to be used as binder of porous
asphalt. The problem of low strength and raveling was effectively solved. In this way, the
porous asphalt was promoted and started to construct on expressways as well as on municipal
roads (Liu & Cao 2009). There are around 300 km porous asphalt sections on expressways in
China in total at the moment.
Recently, an increasing number of existing pavements need to be repaired or reconstructed,
as the service life of these roads has been close to or exceeded ten years. There is require-
ment for using porous asphalt as overlay of old pavement in maintenance engineering to
improve the function of noise reduction and drainage. However, before using porous asphalt
in the practical engineering, it is necessary to study the mechanical property of the existing
139
pavement structure and learn whether it is suitable to pave the porous asphalt overlay in the
maintenance or reconstruction engineering.
In this research, laboratory measurements as well as mechanical analysis were carried
out for investigating the bearing capacity of the old pavement and those overlaid by porous
asphalt. Old pavements in different conditions are investigated, including pavement with
various rutting depth and hot in-place recycled surface. This study provides suggestion for
engineers to design the pavement structure with porous asphalt overlay in maintenance or
reconstruction. The measurement and analysis methods in this research can also be used in
evaluation of the performance of existing old pavement.
2.1 Materials
2.1.1 Existing pavement and core samples
The existing pavement taken into account is the S29 Expressway in Jiangsu province in China.
The pavement was firstly constructed in 2001, and the service periods of sections to be inves-
tigated in this paper are all over ten years. The pavement structure consists of a 40 mm upper
layer, a 60 mm middle layer and a 70 mm bottom layer.
The road sections are in different state, and certain sections have been maintained during
the service. Sections with three typical states are studied: (1) normal section: section with
relatively small rutting depth (<10 mm); (2) section with large rutting depth (around 20 mm);
(3) section with hot in-place recycled upper layer.
Cores with different sizes were drilled from the existing sections. Φ320 mm cores are used
in full-thickness continuous loading test, while Φ100 mm cores are used in rotary-loaded
wheel test and resilience modulus measurement.
the pavement and determine whether the existing pavement structure is suitable for adding a
porous asphalt overlay.
The Φ320 mm full-thickness core samples together with the test molds are shown in
Figure 1.The standard rutting test device was used for the measurement. In the test, 0.7 MPa
stress was applied to the sample and acted reciprocally. The rutting values were automatically
recorded until the pavement surface structure was obviously broken. Generally, the load-
ing can be applied more than 10,000 times (to and fro). In certain cases, it was more than
100,000 times. The testing temperature was 67ºC. The temperature is higher than that for the
standard rutting test for accelerating the destruction of the sample. The testing process is
presented in Figure 2. The tests were carried out on original pavement structure as well as on
structure with the porous asphalt overlay.
perpendicularly on the surface of the test sample. It simulates the way how traveling vehicle
load is applied on the real pavement. The deflection of the pavement and loading times are
automatically measured and recorded. The loading is conducted at 60°C in the water bath.
The measurements were also carried out on both the original cores and cores with porous
asphalt overlay.
Figure 5. Cores drilled from Φ320 mm samples after continuous loading test. (left: original pavement;
right: pavement with porous asphalt overlay).
143
Figure 6. Marshall samples after rotary-loaded wheel test.
4 MECHANICAL CALCULATION
The bearing capacity of the pavement structure was calculated by finite element analysis.
The Ansys FE software was employed in this study. 0.7 MPa standard axle load was applied
to the pavement structure. The top deflection of the pavement in the pavement structural lay-
ers and the bottom deflection of the different structural layers were recorded.
In this study, laboratory measurements and mechanical analyses were carried out on the old
pavement S29 to investigate the feasibility of using porous asphalt overlay in road recon-
struction. Conclusions drawn from the research are:
Generally, using porous asphalt overlay in pavement maintenance and reconstruction can
effectively reduce the rutting deformation and improve bearing capacity of the original pave-
146
ment. The life expectancy of the pavement is thus considered to be extended. For sections
with deep rutting, paving the porous asphalt overlay on a rebuilt dense layer results in a better
performance of the pavement structure. As to hot in-place recycled sections in S29, there is
no need to remove the whole recycled layer. The porous asphalt layer can be laid after the
rutting on original upper layer is eliminated.
The measurement methods used in this study, such as full-thickness continuous loading
test and rotary-loaded wheel test are proved applicable to evaluate the performance of old
pavement and used in the porous asphalt overlay design. Standards of these test methods
need to be formulated for guiding the old reconstruction design in the future. The evaluation
of fatigue performance of the existing pavements is also suggested to be considered.
ACKNOWLEDGEMENTS
REFERENCES
Cao, D., Liu, Q. & Tang, G. 2009. Porous Asphalt Pavement. Beijing: China Communications Press.
COWI. 2006. Noise classification of road pavements: Task 1: Technical background information.
European Commission—DG Environment. Brussels, Belgium.
Li, M., 2013. Tyre-road noise, surface characteristics and material properties. Delft University of
Technology. Delft, The Netherlands.
Liu, Q. & Cao, D. 2009. Research on material composition and performance of porous asphalt pavement.
Journal of Materials in Civil Engineering, SPECIAL ISSUE: China: Innovative Use of Materials for
Highway Construction: 135–140.
Molenaar, A.A.A. & Meekerk, A.J.J. 2006. Performance of Porous Asphalt. Journal of the Association
of Asphalt Paving Technologists, CD.
NEN-EN12697-26. 2004. Bituminous mixtures—Test methods for hot mix asphalt—Part 26: Stiffness.
European Committee for Standardisation. Brussels, Belgium.
147
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Fatigue behavior of Hot Mix Asphalt (HMA) is the important factor, when
it comes to evaluate the structural lifetime of bituminous pavement constructions (pavement
design). Thereby, the progress of fatigue is strongly affected by resulting stresses in the bitumi-
nous layers due to traffic and/or temperature change, whereat HMA stiffness has huge impact
on the magnitude of these stresses and, hence, on the design result. It is obvious that a reliable
characterization and prediction of HMA stiffness is crucial. A suitable way for the description
of the viscoelastic response of a material is continuum micromechanics, where the mechanical,
volumetric and morphologic properties of the constituents of the material are considered to
predict the homogenized, “overall” material behavior. Thereby, the material is observed on dif-
ferent, reasonably chosen lengths scales allowing for a description of mechanical effects where
they occur. These model assumptions were validated extensively showing a remarkable accord-
ance between experimental results obtained from 4PBB-PR tests and model predictions with
only the mechanical properties of the constituents (bitumen, aggregate) and the volumetric
composition as model input. The results of this investigation suggest that applying the pre-
sented technique can lead to a significant reduction of experimental efforts in mix design.
1 INTRODUCTION
The fatigue performance of the hot mix asphalt (HMA) used in the bituminous base course,
which is considered as relevant for the design of bituminous pavements, is strongly influenced
by strains induced by traffic loading. As an indicator for the resistance against deformation,
the stiffness behavior of HMA has a huge impact on the magnitude of these strains and,
hence, the design result.
To find a bituminous mixture meeting the demands of a base course, experimental mix
design is used. Thereby, stiffness tests (among others) are performed in the laboratory to
characterize the mechanical behavior of HMA. Although several experimental setups can be
distinguished (ONI, 2004), two configurations are used frequently: (i) a four point bending
test performed on prismatic specimens (4PB-PR), and (ii) an indirect tension test conducted
on cylindrical samples (IT-CY). These laboratory tests require advanced equipment and are
very time consuming.
To achieve more economic mix design, models are used to predict HMA stiffness behavior.
Such material models should account for the binder stiffness as well as the volumetric com-
positions. Two examples of frequently applied semi-empirical models for the estimation of
the stiffness modulus of HMA are the Shell model (Van der Poel, 1954) and the Hirsch model
(Christensen et al., 2003).
To link microstructural characteristics to effective material properties, more sophisticated
material models are needed. To observe a macro-homogenous material as a micro-hetero-
geneous medium (multiscale approach) allows for the physical understanding of material
behavior and, thus, enables the description of mechanical effects like accumulative damage
processes (fatigue) or aging effects.
Continuum micromechanics has proven its ability to provide a good framework for multi-
scale modeling as multi-step homogenization schemes were introduced for several materials
149
like concrete (Scheiner and Hellmich, 2009), wood (Hofstetter et al, 2005) or bone (Fritsch
et al, 2009) in recent years. In terms of hot mix asphalt, Lackner et al (amongst others in
(Lackner et al., 2006)) proposed a model distinguishing 5 levels of observation, which was
extended recently by taking the microstructure at the bitumen level into account (Eberhard-
steiner et al., 2015b). As this model is based on closed formulas, it allows for software-based
and, hence, fast and economic mix design.
While the mentioned HMA multiscale model is described briefly in the following section,
details on identification and validation experiments are given in Section 3. Before conclud-
ing the work, the result of recently performed, extensive validation including over 100 data
points is presented.
The framework of continuum micromechanics (Hill, 1965, Zaoui, 2002, Suquet, 1997) pro-
vides solutions for the description of the mechanical behavior of a material. Regarding bulk
material as micro-heterogenous body filling a macro-homogeneous representative volume
element (RVE) allows for the consideration of features like material properties, morphol-
ogy or volume content of constituents at different length scales. Thereby, the microstructure
within such an RVE is described by reasonably chosen, quasi-homogenous subdomains, so-
called material phases, with known physical properties (volume fractions, elastic/viscoelastic
properties,...). The characteristic length of an RVE l has to fulfill the law of separation of
scales, d << l << L, where d represents the characteristic length of the inhomogeneities within
the RVE, and L denotes the characteristic lengths of geometry or loading of a structure built
up by the material defined on the RVE. If a single phase exhibits a heterogeneous micro-
structure itself, RVEs can be introduced within this phase in order to estimate its mechanical
behavior. These RVEs have dimensions, which fulfill l2 ≤ d, and imply again smaller inhomo-
geneities with characteristic lengths d2 << l2, and so on.
By solving matrix-inclusion problems of the Eshelby-Laws type the mechanical behavior
of the bulk material can be estimated. The physical properties of the material phases and
their interactions are used to correlate homogenous deformations acting on the boundary
of the RVE and resulting (average) stresses. As described in (Lackner et al., 2006, Eberhard-
steiner et al., 2015a) this leads to a homogenized stiffness tensor Chom of the general form
−1
⎧ −1 ⎫ ⎧ −1 ⎫
Ch ⎨∑ c : [ I P : (c
(c c )] ⎬ : ∑ fs [ I P : (c
(c c )] ⎬ (1)
⎩ r ⎭ ⎩ s ⎭
with the volume fraction fi, the stiffness tensor ci and the Hill tensor Pi0 accounting for the
morphology of phase i. I denotes the forth-order unity tensor. For matrix-inclusions prob-
lems, c0 is chosen as stiffness tensor of the matrix (Mori-Tanaka scheme (Mori and Tanka,
1973, Wakashima and Tsukamoto, 1991)).
For HMA, 5 scales of observation can be distinguished (see Figure 1):
• the bitumen-scale,
• the mastic-scale consisting of a bitumen matrix with embedded spherical filler particles (d
≤ 125 μm),
• the mortar-scale composed of matrix of mastic material with spherical sand inclusions
(d ≤ 2 mm),
• the asphalt-scale consisting of a mortar matrix with embedded spherical stone inclusions
(d > 2 mm) and spherical air voids, as well as
• the macroscale.
To estimate the homogenized stiffness behavior (in terms of Chom) on the next larger scale,
Equation (1) is applied to the mastic-, mortar—and asphalt-level taking the respective mate-
150
Figure 1. Micromechanical representation of hot mix asphalt by means of a three-step homogeniza-
tion scheme (Lackner et al., 2006).
rial phases (constituents) and their properties into account. Finally, the homogenized viscoe-
lastic HMA stiffness behavior can be derived.
While the homogenization according to Equation (1) is only valid for elastic problems,
bitumen—and hence bituminous materials—exhibit viscoelasticity expressed through the
relation between stresses σ and strains ε
dε
σ (t ) ∫ (t τ ) : dt (τ ) dτ [r ∗ ε ].
t
(2)
−∞
Thereby, r(t-τ) denotes the fourth-order tensorial relaxation function, τ represents the inte-
gration variable related to the time instant when d ε(t)/dt was imposed, and * is the Stieltjes
convolution operator. While r(t = τ) refers to instantaneous elasticity, r(t > τ) refers to vis-
coelastic deformations. To provide a formally elastic and, thus, very convenient description
of the viscoelastic problem, Equation (2) is represented in the Laplace Carson (LC) domain
defined through frequency variable p, with the LC transformation according to (Donolato,
2002), reading
∞
f * ( p) LC{ f (t )} pfˆ ( p ) p∫ f (t )e pt
dt (3)
0
Where f*(p) is the LC transform of the time-dependent function f(t) and fˆ (t ) is the Laplace
transform of f(t).
Applying Equation (3) to Equation (2) yields the formally linear elastic constitutive equa-
tion (Gurtin and Sternberg, 2002)
σ * ( p) *
( p) ε * ( p) (4)
defined in the LC domain. Accordingly, the relaxation tensors of the isotropic material bitu-
men reads as
where Ivol is the volumetric part of the fourth-order unity tensor I, with components
Ivol,ijkl = 1/3δijδkl and Kronecker delta δij (δij = 1 if i = j, and δij = 0 if i≠j), while Idev represents
the deviatoric part of the fourth-order unity tensor I (Idev = I-Ivol). The components of I are
defined as Iijkl = 1/2(δikδjl+δilδjk).
In this sense, the viscoelastic behavior of HMA follows
151
R*asph ( p ) {( fssto air )r* mor ( p )
+ fstoc sstoto : [I P* h mor ( p ) : (c sto r* mor ( ))] }
−1
with the volume fraction of stones fsto and air voids fair, the LC-transformed relaxation tensor
of mortar r*mor(p), the elastic stiffness tensor of stone csto and the LC-transformed Hill’s ten-
sor P*,sphmor(p) accounting for spherical inclusions in an isotropic matrix of mortar.
To give the predicted p-related relaxation tensor of HMA in a physically relevant format,
R*asph(p) has to be back-transformed into the time domain. Hence, the Gaver-Stehfest algo-
rithm (Stehfest, 1970) for a point wise back-calculation is employed. According to (Stehfest,
1970), f(t = t1) for a time instant t1 can be determined from the LC transform f*(p) by
N N
ln2 ln2 f * ( p)
f (t t1 ) i ( ) Vi (7)
t1 i 1 t1 i 1 p
3 EXPERIMENTS
A reliable prediction of HMA stiffness behavior strongly depends on the proper determina-
tion of input parameters. While volume fractions can be obtained from mix design (aggregate
gradation, binder and air void content), the mechanical properties of the constituents have to
be identified. Aggregate exhibits elastic behavior, which is usually declared by the manufac-
turer (in terms of Young’s modulus and Poisson ratio). The viscoelastic behavior of bitumen
can be characterized through Dynamic Shear Rheometer (DSR) tests (see Section 3.1).
To show the significance of predictions obtained from the model described in Section 2,
validation experiments using four point bending tests (see Section 3.2) were conducted.
a convenient frequency-time domain conversion, which was verified extensively using the
correlation between the results of static and dynamic shear tests. This allows a characteri-
zation of the viscoelastic response by a power-law type compliance function according to
(Füssl et al., 2014), reading
k
⎛t⎞
J (t ) = J 0 Ja ⋅ ⎜ ⎟ (10)
⎝τ ⎠
152
Figure 2. Schematic illustration of four point bending test for HMA with a maximum aggregate size
D > 11 mm.
with J0 as the elastic shear compliance, Ja and k as viscous parameters, and τ as a time vari-
able, which are identified by minimizing the error—using nonlinear least square fitting—
between experimentally obtained creep compliances from DSR identification tests and
predicted creep compliances.
The multistep homogenization scheme described in Section 2 was used to predict the Young’s
modulus of the investigated HMA mixtures. Thereby, each of the samples was investigated
individually by taking the viscoelastic behavior of the bitumen actually used (identified in
DSR tests), the elastic properties of the aggregates (as declared by the manufacturer) and
the specific volumetric composition (derived from aggregate gradation, binder and air void
content) into account.
Figure 3 shows the correlation between experimental results and model predictions. While
the Young’s modulus is overestimated for HMA mixtures containing paving grade bitumen,
the model predictions of the stiffness of samples with polymer-modified bitumen fit exper-
imental results well, especially at higher frequencies. This is expressed by a coefficient of
determination between R2 = 0.64 and R2 = 0.96.
Obviously, this approach delivers reliable stiffness predictions and, hence, its application
provides major advantages related to the design of bituminous pavements:
• The multiscale approach allows for a straight forward mix-design in terms of volumet-
ric composition through direct consideration of morphology and volume content of the
HMA constituents. This helps to reduce experimental efforts in mix design, and, hence,
makes mix design more economic.
• Taking the "real" (viscoelastic) behavior of bitumen, directly determined from DSR tests,
into account enables the prediction of viscoelastic HMA properties and their considera-
tion in the design process. So, benefits of innovative (e.g. modified) binder systems related
to the reduction of e.g. thermal stresses due to relaxation can be considered in the design
process.
• Implementation of the presented homogenization scheme (see Figure 1) into an analyti-
cal design method also enables the description of fatigue behavior from a microstructural
153
Figure 3. Validation of HMA multiscale model in terms of |E*| for different HMA mixtures with
polymer-modified bitumen PmB 45/80–65 and paving grade bitumen 70/100.
point of view (Füssl et al, 2014). Additionally, the effects of aging can be studied at the
micro scale and their consequences on the design result (macroscale) can be predicted
(Eberhardsteiner et al, 2015b, Eberhardsteiner et al, 2015c).
5 CONCLUSIONS
As a parameter for the resistance against deformation, the stiffness behavior of hot mix
asphalt (HMA) strongly influences the results of pavement design. Hence, a reliable predic-
tion of the mechanical behavior is crucial. Continuum micromechanics provide a framework
to estimate the viscoelastic behavior of HMA in terms of a multiscale model. Thereby, the
volumetric composition of the mixture, the mechanical properties of the constituents and
their morphology are considered on five levels of observation.
To validate these model assumptions, the results of four point bending stiffness tests were
compared to model predictions for 4 mixtures containing paving grade bitumen 70/100 or
polymer-modified bitumen PmB 45/80–65. While the elastic behavior of the aggregates was
defined by the manufacturer declaration, the viscoelastic response of the binder was identi-
fied in dynamic shear rheometer tests and the volumetric composition was derived from
gradation as well as binder and air void content.
This validation confirms that the proposed model is able to predict HMA stiffness
well, showing a coefficient of determination between R2 = 0.64 and R2 = 0.96. Hence, this
approach not only allows for proper pavement design but also has the potential to reduce the
experimental effort in mix design significantly. In contrast to already existing methods, the
proposed approach is based on a mechanistic description of the behavior of the described
material. Hence, predictions are valid for all mixtures, while empirical approaches are limited
to materials they are fitted to.
REFERENCES
Christensen, D. W., Pellinen, T. & Bonaquist, R. F. 2003. Hirsch model for estimating the modulus of
asphalt concrete. Journal of the Association of Asphalt Paving Technologists, 72, 97–121.
Donolato, C. 2002. Analytical and numerical inversion of the Laplace-Carson transform by a differen-
tial method. Computer Physics Communications, 145, 298–309.
Eberhardsteiner, L., Füssl, J. & Blab, R. 2015a. Prediction of Hot-Mix Asphalt Stiffness—A Multiscale
Approach. TRB 94th Annual Meeting Compendium of Papers.
154
Eberhardsteiner, L., Füssl, J., Hofko, B., Blab, R., Grothe, H., Handle, F. & Hospodka, M. 2015b.
Influence of asphaltene content on mechanical bitumen behavior—Experimental invesitgation and
micromechanical modeling. Materials and Structures, 48, 3099–3112.
EBerhardsteiner, L., Füssl, J., Hofko, B., Blab, R., Grothe, H., Handle, F. & Hospodka, M. 2015c
Towards a microstructural model of bitumen aging behavior. International Journal of Pavement Engi-
neering, 16 (10), 939–949.
Fritsch A., Hellmich Ch. & Dormieux L. 2009. Ductile sliding between mineral crystals followed by
rupture of collagen crosslinks: Experimentally supported micromechanical explanation of bone
strength. Journal of Theoretical Biology, 260(2), 230–252.
Füssl, J., Lackner, R. & Eberhardsteiner, J. 2014. Creep Response of Bituminous Mixtures—Rheologi-
cal Model and Microstructural Interpretation. Meccanica, 49, 2687–2698.
Gurtin, M. & E. Sternberg. 2002. On the linear theory of viscoelasticity. Archive for Rational Mechanics
and Analysis 11, 291–356.
Hill, R. 1965. Continuum micro-mechanics of elastoplastic polycrystals. Journal of Mechanics and Phys-
ics of Solids, 13, 89–101.
Hofstetter K., Hellmich, & Eberhardsteiner, J. 2005. Development and experimen—tal validation of a
continuum micromechanics model for the elasticity of wood. European Journal of Mechanics—A/
Solids, 24, 1030–1053.
Lackner, R., Blab, R., Eberhardsteiner, J. & Mang, H. A. 2006. Characterization and multiscale mode-
ling of asphalt—Recent developments in upscaling of viscous and strength properties. Computational
Mechanics: Solids, Structures and Coupled Problems, 6, 599–622.
Mori T. & Tanaka K. 1973. Average stress in matrix and average elastic energy of materials with misfit-
ting inclusions. Acta Metallurgica, 21(5), 571–574.
ONI 2004. ÖNORM EN 12697–26: Bituminous mixtures—Test methods for hot mix asphalt—Part 26:
Stiffness.
Scheiner, S. & Hellmich, C. 2009. Continuum microviscoelasticity model for aging basic creep of early-
age concrete. Journal of Engineering Mechanics, 135, 307–323.
Suquet, P. 1997. Continuum micromechanics, Springer, Wien—New York.
STEHFEST. 1970. H. Algorithm 368: Numerical inversion of Laplace transforms. Communications of
the ACM, 13, 47–49.
Van der poel, C. 1954. A general system describing the visco-elastic properties of bitumen and its rela-
tion to routine test data. Journal of Applied Chemistry, 4, 221–236.
Wakashima K. & Tsukamoto H. 1991. Mean-field micromechanics model and its ap—plication to the
analysis of thermomechanical behaviour of composite materials. Materials Science and Engineering
A, 146(1–2), 291–316.
Zaoui, A. 2002. Continuum micromechanics: Survey. Journal of Engineering Mechanics (ASCE), 128,
808–816.
155
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Changfa Ai
Southwest Jiaotong University, Chengdu, China
Chuan Xiao
Sichuan College of Architecture Technology, Deyang, China
ABSTRACT: The tensile strain at the bottom of asphalt concrete is well accepted as one
of key factors influencing pavement performance, as it is directly related to fatigue cracking.
This paper presents a procedure to estimate tensile strain at the bottom of asphalt layers
using the measured dynamic tensile strain and the deflection basin response from Falling
Weight Deflectometer (FWD) tests. Dynamic tensile strain sensors were installed during the
construction and strain measurements were made in conjunction with FWD loadings. The
experimental tests were performed in three separate test sections with semi-rigid base asphalt
concrete (AC) pavement, inverted asphalt pavement structure, and compound asphalt pave-
ment structure, respectively, in a recent constructed motorway in Sichuan Province, China.
Firstly, the influences of the pavement types, temperature conditions, and load levels on the
measured dynamic tensile strain at the bottom of asphalt layer were identified. Then, this
study analyzed the correlation of the measured dynamic tensile strain and the corresponding
deflection basin responses by using the grey relation entropy method. The results suggested a
highly significant correlation. Finally, a group of regression models to estimate the dynamic
tensile strain for various types of asphalt concrete pavement was proposed by using the meas-
ured deflection basin responses from FWD tests.
1 INTRODUCTION
The Falling Weight Deflectometer (FWD) is a non-destructive testing device that has been
widely used in pavement engineering to evaluate pavement structural condition (Zha et al.
2002). The FWD is a tool used to achieve rapid and repeatable in-situ characterization of
the pavement layer stiffness. The FWD plays a crucial role in selecting optimum pavement
maintenance and rehabilitation strategies (Wang et al. 2002; Qian et al. 2010).
Uddin et al. (1997) simulated the influences of FWD load pulse on the asphalt pavement
surface deflections response by using a three-dimensional finite-element pavement-subgrade
model. Hao and Wang (2002) developed a layered pavement finite element model based on
structural dynamic to evaluate the dynamic response of asphalt pavement under FWD load.
Ren et al. (2001) derived an analytical solution for the FWD dynamic response of the multi-
layered visco-elastic systems assuming the individual layers are in a fully contact condition.
Hou et al. (2002) adopted Newmark integration scheme to study the dynamic response of
layered pavement system under FWD load. Qiu et al (2012) recently studied the distribution
pattern of the dynamic deflection basin of semi-rigid base asphalt pavement with different
pavement damage conditions. However, there are several limitations in above-mentioned ana-
lytical and numerical simulation methods used to obtain the dynamic response of the com-
157
plex multilayered pavement systems under the FWD load, such as the improbably developing
the pavement structural model, to some extent arbitrary in selecting the material proper-
ties, and inadequately considering the boundary and the interface conditions. Therefore, the
actual response of pavement structure is difficult to simulate realistically. By contrast, in-situ
dynamic response through FWD test can accurately reflect the actual dynamic response of
pavement structures. Several researchers have developed relationship between deflection basin
parameters and pavement strain response, then been used for pavement condition assessment
(Thompson 1989; Feng et al. 1996; Lee et al. 1998; Xu 2000). This paper present the dynamic
strain measurements for three type of asphalt pavement structure under a moving load and
the falling weight deflectometer (FWD) load. According to the measurement results, the
correlation of the measured dynamic tensile strain and the corresponding deflection basin
responses is then analyzed by using the grey relation entropy method. Finally, a procedure to
evaluate the asphalt pavement structural condition is proposed using the measured deflection
values and the corresponding derived basin deflection indexes through a FWD test, which
provides a rapid and reliable method to assess pavement condition.
2.2 Instrumentation
During the construction phase of the test sections, the dynamic strain sensors were installed
both the longitudinal and transverse directions in the bottom of asphalt layers to capture the
pavement response under loading. A type of dynamic asphalt strain gages (ASGs), manufac-
tured by Applied Geomechanics Inc., was used in this study. This ASG consists of four active
elements of a Wheatstone bridge circuit, and they can withstand the high temperature and
vibratory rolled compaction required for asphalt placement. The complete dynamic response
under FWD load and BZZ-100 load were collected by using a high-frequency data acquisi-
tion system.
The strain gauge array was centered along the left side of the wheel path of the truck lane.
The wheel path was determined by wheel path measurement together with the rut location
measurement of another motorway having similar traffic conditions. In each test section, one
row of two longitudinal strain gauges and another row of two transverse strain gauges were
placed 0.15 m offset the center of the left wheel path. Each row was placed 1.0 m apart along
the traffic direction, as shown in Figure 2.
158
Figure 2. Layout of the instrumented asphalt strain gauges.
Figure 3. Asphalt strain gauges installation steps during asphalt pavement construction.
159
Figure 4. Falling weight deflectometer test and the layout of the deflection measuring sensors.
Figure 5. A typical measured strain response in longitudinal direction under increasing FWD load
levels.
Figure 6. Longitudinal and transverse strain at the bottom asphalt layer under various FWD load
levels.
be clearly distinguished as which are corresponding to six obvious changes in the measured
strain over time. Findings are concluded as follows:
1. The ASG response signal is similar in shape under different load levels. Figure 5(b) display
the response to only one of the six drops. As shown in Figure 5(b), there is a compression
first of the recorded strain under a falling weight, then obvious maximum tension and
subsequently gradually decreased bounce response. The initial compression of the meas-
ured strain is attributed to the temporarily released gravity of the heavy hammer before
it encounters the bearing plate. Then, the asphalt strain reaches to the first peak tension
when the harmer met and impact the bearing plate. The bearing plate then vibrated which
cause gradually decreased strain vibrations.
2. Relaxation is observed because of the viscoelastic nature of the bituminous materials
(Zhang 2011). As shown in Figure 5(b), there is an asymmetry on the measured asphalt
strain within each pulse cycle. Consequently, the asphalt surface has a viscosity effect
on the loading process, then strain retardation and strain relaxation occurs after FWD
impact loading.
160
3.2 Effect of pavement structure on the measured AC strain
Figure 6 presents the calculated longitudinal and transverse peak strain at the bottom of the
top asphalt layer under various FWD load levels for three pavement structures. It shows that
both longitudinal and transverse strain increases with increasing the FWD load level. Under
the same FWD load level, the measured longitudinal strain at the bottom of asphalt layer is
always higher than the AC strain in transverse direction, indicating the anisotropic and the
heterogeneous properties of the pavement response under the traffic loading. It indicates that
a transverse fatigue crack would first appear due to the larger generated strain in longitudinal
direction. Therefore, only the longitudinal strain was studied in the subsequent analysis of
the dynamic strain at the bottom of the top asphalt layer.
ε = e aT + b (1)
Where, ε is the longitudinal AC strain at the bottom of the top asphalt layer, [10−6]; T is the
pavement surface temperature, [°C]; a and b are regression coefficients.
A temperature correction factor KT is defined for evaluating the influence of pavement
temperature on the strain in the asphalt layer, as follows:
161
Table 1. Regression coefficients of temperature correction factor.
Regression coefficients
ε
KT e a (Tr −Tm ) (2)
εm
Where, ε r is the longitudinal AC strain at the reference temperature on the pavement sur-
face Tr , [10−6];
ε m is the longitudinal AC strain at the measured field temperature Tm , [10−6]. In the present
study, the reference temperature Tr is chosen as 20°C. Table 2 lists the regression coefficient
a for three types asphalt pavement structures through the measured AC longitudinal strain
at the bottom of the top asphalt layer according to the above proposed Equation (2). Firstly,
a relative constant regression coefficient a is observed for each type of pavement structure
under various applied FWD load level, which indicates the influences of FWD load level
on regression coefficient a is insignificant. Moreover, it shows that the regression coefficient
a mainly depends on the type of pavement structure. According to the definition of the
temperature correction factor KT , it describes the temperature sensitivity of the pavement
structure in terms of the AC strain at the bottom of the asphalt layer. A larger regressed coef-
ficient a, the more temperature sensitive of this pavement type on the AC strain at the bottom
of asphalt layer. In other words, the larger regressed coefficient a shows a poor temperature
adaptability of this type pavement structure.
The FWD applies impact load to pavement surface, simulating the magnitude and duration
of a moving wheel load. The velocity of moving load affects the load duration and therefore
the induced strain in asphalt, which relates to the visco-elastic characteristics of asphalt layer
and the elasto-plastic response of the subgrade. To determine the speed, which probably pro-
duce an identical strain response to the FWD impact load, the instrumented sections were
also tested under a standard moving wheel load, namely BZZ-100 that is defined as a single
axle, double tires, and with tire pressure of 0.7 MPa). Considering the temperature depend-
ency on the asphalt response, the standard moving wheel load and various FWD load level
were applied to the test sections at three temperature conditions, 12 °C (cold), 21 °C (mod-
erate), and 43 °C (warm), respectively. Table 2 lists the measured AC longitudinal strain at
the bottom of the top asphalt layer under various FWD and BZZ-100 load conditions.
162
Table 2. Measured strain response at the bottom of asphalt surface layer under various load
conditions.
Equivalent BZZ-100
FWD load/kN speed (km/h)
Pavement
type Temperature 30 40 50 60 70 20 40 60 80
S1 cold 9.1 11.5 18.0 21.2 24.5 25.5 19.1 15.8 11.6
moderate 15.4 19.1 24.7 30.9 33.6 32.3 25.7 22.7 19.5
warm 34.3 40.6 55.6 68.6 75.1 76.6 53.6 47.0 47.4
S2 cold 6.9 10.0 14.1 17.5 18.5 15.2 12.6 8.1 7.0
moderate 14.7 19.9 25.6 30.1 32.3 28.3 20.8 18.2 15.9
warm 53.6 67.8 88.3 100.3 109.8 102.4 76.2 66.1 61.0
S3 cold 10.6 13.3 17.4 19.6 21.2 18.3 17.2 14.5 12.3
moderate 20.9 24.9 29.0 31.9 33.6 30.3 29.9 27.9 23.2
warm 35.1 45.9 62.7 71.8 78.6 80.4 57.0 51.3 46.0
Figure 8. Determination of the corresponding relationship between FWD load and BZZ-100 load (in
the case of S1structure in the moderate temperature condition).
A procedure to determine the equivalent FWD load level in terms of the standard moving
wheel load of BZZ-100 at a given speed is presented as follows:
1. Calculate the strain ration ε FW i
WD / ε v based on the measured strain response for each FWD
j
response is found for the pavement structure under either a FWD test under a given load
level or a moving load at the specific speed. As shown in Figure 8, the intersections of the
linear regression equations at various FWD load level for strain ratio ε FW i
WD / ε v and the line
j
of ε FW
i
WD / ε v
j
= 1 are the equivalent speed of the BZZ-100 load that produces the same strain
response for the pavement structural under a corresponding FWD load level.
Based on the above-mentioned methods, Table 3 lists the equivalent relationship between
FWD load level and the speed of BZZ-100 load. It shows that the type of the pavement
163
Table 3. Equivalent relationship between FWD load and BZZ-100 load.
Figure 9. Regressed linear equations describing the equivalent relationship between FWD load level
and the speed of BZZ-100 load.
structure dominates the speed of BZZ-100 equivalent to which of the FWD load level, while
the pavement temperature condition has relatively less influence. More specifically, the speed
of the BZZ-100 load decreases with the increasing of the equivalent FWD load level. As for
the influence of the pavement type, a larger equivalent speed of BZZ-100 load is found for
the S1structure, and a lowest value is observed for the S2 structure.
Figure 9 displays the regressed linear equation of the equivalent relationship between
FWD load level and the speed of BZZ-100 load, which defines the required FWD load level
that would produce the same strain response of the asphalt pavement under the practical
moving load with a predetermined speed.
164
Table 4. Selected basin deflection indexes and the regressed prediction models f
AC strain.
DRRi = di / d0, (i = 1∼8); (2) curvature index CII i = di di +1 , (i = 0∼7); (3) slope index
Si di − di ri + 2 − ri ), (i = 0∼6); (4) shape factor Fi di − − di di , (i = 1∼7); (5) area index,
A [ (d2 / d0 d 4 / d0 ) d5 / d0 ], AUPP U =( d d d − d ) / 2. Where, di . is the
measured deflection at each sensor of FWD, and ri is the distance of each sensor from the center
of the loading plate. Previous researches have shown that the deflection basin parameters have
a significant influence in the prediction of critical pavement response in flexible pavements.
The relationship between dynamic strain and dynamic deflection basin parameters was
studied by using gray relation entropy analysis method (Xiao et al. 2014). Then, three indi-
cators with the highest correlation were chosen as independent variables based on results
of the correlation analysis between the deflection basin parameters and the dynamic strain
response of asphalt pavement. Finally, the prediction models of longitudinal and lateral
strains for asphalt pavement surface layer were established based on the dynamic deflection
basin parameters through multiple linear regression analysis (Xiao et al. 2014). The regres-
sion equation and fitting degree situation are summarized in Table 4.
A very high value of coefficient of determination, close to 1, is found for each individual
AC strain estimation equation, which shows a stronger relationship among the AC strain and
the selected basin deflection indexes. Similar tendency is also verified by the results of F test
of goodness of fit test for these prediction models. Lastly, the absolute difference between
the estimated and the measured AC strain for three test sections varied from 1.65% to 5.88%,
which satisfies the practical engineering requirement very well.
6 CONCLUSIONS
165
Figure 10. Flowchart of the procedure to evaluated asphalt pavement condition based on FWD tests.
1. The dynamic asphalt strain response under FWD load can be captured by the instrumented
asphalt strain gauges. Field measurements have shown that the AC strain at the bottom of
the top asphalt layer increases with the increasing of the FWD load level. Besides, the
measured longitudinal strain at the bottom of asphalt layer is always higher than the strain
in transverse direction, indicating the anisotropic and the heterogeneous properties of the
pavement response under the traffic loading. It might indicate that a transverse fatigue
would first appear due to the larger generated strain in longitudinal direction.
2. A clear nonlinear temperature dependency are observed for the measured AC strain for
all these three surveyed pavement structures. The AC strain increase significantly as the
increasing of the pavement surface temperature, and is independent on the applied FWD
load level. A temperature correction factor is defined to quantify the sensitivity of the
temperature dependency of the AC strain based on the regression analysis of the meas-
ured AC strain and the corresponding pavement surface temperature. The analysis of
the temperature corrections factors shows that the inverted asphalt pavement structure is
more sensitive to the temperature for AC strain under FWD load, and the semi-rigid base
asphalt concrete pavement has a lower temperature dependency for strain response under
FWD load.
3. A procedure to determine the equivalent FWD load level in terms of the standard mov-
ing wheel load of BZZ-100 is proposed according to the principle of identical induced
strain response. Under the same FWD load level, the speed of the equivalent BZZ-100
load is largest for the pavement structure S1 and the lowest for pavement structure S2. In
order to better simulate the pavement response under the practical moving load, a sires of
regression equations are proposed to describe the relationship of the FWD load and the
standard moving load for each pavement structure.
4. The correlation between different pavement deflection basin parameters and surface layer
dynamic strain values of asphalt pavement is evaluated through gray relation entropy anal-
ysis. Then, the prediction models of longitudinal and lateral strains at the bottom of the
top asphalt concrete layer were established based on the dynamic deflection basin param-
eters through multiple linear regression analysis, which fit the measured strain response
166
very well. Finally, a general procedure to evaluate the asphalt pavement structural condi-
tion is proposed using the measured deflection values and the corresponding derived basin
deflection indexes through a FWD test.
ACKNOWLEDGEMENTS
This work was supported by the National Natural Science Foundation of China under Grant
51378438; the National 973 Program under Grant 2013CB036204; and the specific research
funding from Ministry of Education under Grant SWJTU12ZT05.
REFERENCES
Feng Xiao, Tang Bo-ming, Li Jin-hua. 1996. Nondestructive evaluation with FWD on structural bearing
capacity of asphalt pavements [J]. Journal of Tongji University, 1996, 24(5): 520–524. (in Chinese).
Hao Da-li, Wang Bing-gang. 2002. Dynamic response of pavement structure [J]. Journal of Chang’an
University: Natural Science Edition, 22(3): 9–12. (in Chinese).
Hou Yun, Guo Zhong-yin, Tian Bo, et al. 2002. Dynamic deflection response of layered pavement
structure subjected to dynamic load [J]. China Journal of Highway and Transport, 15(3): 6–10. (in
Chinese).
Lee Y C, Kim Y R, Ranjithan S R. 1998. Dynamic analysis-based approach to determine flexible pave-
ment layer moduli using deflection basin parameters [J]. Journal of the Transportation Research
Board, 1639(2): 36–42.
Luo Qi-tian, Qiu Xin, Yang Qing. 2011. Evaluation study on subgrade modulus based on FWD dynamic
deflection basin parameters [J]. Highway Engineering, 2011, 36(2): 4–8. (in Chinese)
Qain Zhen-dong, Zhang Lei, Chen Lei-lei. 2010. Pavement Structural Dynamics [M]. Nanjing: South-
east University Press, (in Chinese).
Qiu Xin, You Qing-long, Yang Qing. 2013. Distribution characteristics of surface dynamic deflection
basin of CTB asphalt pavement with cracking [J]. Journal of Highway and Transportation Research
and Development, 30(3): 1–8. (in Chinese).
Ren Rui-bo, Tan Yi-qiu, Zhang Xiao-ning. 2001. Solution for solving asphalt pavement multilayered
viscoelastic body surface deflection in the FWD dynamic case [J].China Journal of Highway and
Transport, 14(2): 9–13. (in Chinese).
Thompson M R. 1989. Mechanistic-empirical evaluation of the MN road low volume road test section
[R]. Project IHR-535, University of Illinois at Urbana-Champaign, United States.
Uddin W, Pan Z, Noppakunwijai P, et al. 1997. Finite-element dynamic analysis of distressed asphalt
pavements [C]. The 8th International Conference on Structural Design of Asphalt Pavements. Seat-
tle: Federal Highway Administration, 725–739.
Wang Xu-dong, Sha Ai-min, Xu Zhi-hong. 2002. The dynamic characteristics and dynamic parameters
of asphalt pavement materials [M]. Beijing: China Communications Press. (in Chinese).
Xiao Chuan, Qiu Yanjun, Zeng Jie, et al. 2014. Study on measured dynamic response of asphalt pave-
ment under FWD load [J]. Journal of Highway and Transportation Research and Development,
31(2): 1–8. (in Chinese).
Xu Bing. 2000. Assessing pavement layer condition using FWD deflection data [D]. Raleigh: North
Carolina State of University.
Zha Xu-dong. 2002. Summary of back calculation methods of pavement layer moduli [J]. Journal of
Traffic and Transportation Engineering, 2002, 2(4): 1–6. (in Chinese).
Zhang Xiao-ning. 2011. The viscoelastic mechanics theory and application of asphalt and mixture [M].
Beijing: China Communications Press. (in Chinese).
167
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
C. Oliviero Rossi
University of Calabria, Rende, Italy
ABSTRACT: In this paper test results of hot fine-grained asphalt concrete samples on
creep by the direct tensile scheme at temperature of 20±2°C and stress from 0.04 to 0.24 MPa
are presented. It is shown that the asphalt concrete creep curve has three characteristic sites.
On the first site creep strain is described by the integral equation of Boltzmann-Volterra with
the Abel's creep kernel. The kernel parameters determining method is given. On the second
site the creep curve is described by the straight line equation and characterized by the strain
rate which is a power function of stress. The creep curve sites durations and the asphalt con-
crete long-term strength have been defined. The creep curve sites durations and the asphalt
concrete failure time depend on stress. Stress influence degree is such that stress change on
one order causes their change more than three orders.
1 INTRODUCTION
Asphalt concrete is one of the main materials for highway pavements. Mechanical proper-
ties of an asphalt concrete are highly depending on temperature, value, duration, and rate
of loading (MS-4 2008, Papagiannakis & Masad 2008, Yoder & Witczak 1975). In real road
conditions temperature in points of asphalt concrete layers of pavement structures due to
variations of ambient temperature, track wheels load values, their action duration and rate
varies within wide limits. Therefore, determination of mechanical behavior of an asphalt
concrete taking into account the variation of the above mentioned factors has important
practical value.
It is known that the basic methods for evaluation of mechanical behavior of viscoelastic
materials are tests on creep and relaxation (Cristensen 1971, Ferry 1980, Tschoegl 1989).
Technically, realization of creep test is easier. It is possible to construct creep curves and
long-term strength with using its results. Relaxation curves can be obtained from the creep
curves by using known methods (Tschoegl 1989, Hopkins & Hamming 1957). The long-
term strength curves enable to determine service life of a road asphalt concrete pavement.
In this paper test results of hot fine-grained asphalt concrete samples on creep are pre-
sented. Creep tests were carried out by the direct tensile scheme until complete fracture of
the asphalt concrete samples. Test temperature was 20±2°C. The applied stress was changed
from 0.04 to 0.24 MPa. Creep curves under different loads and long-term strength curves
of the asphalt concrete have been constructed. Three characteristic sites of creep curves—
the unstabilized, stabilized and accelerating creep sites are shown. Dependences of these
sites durations on stress are established. The first creep curve site of the asphalt concrete
is satisfactorily approximated by the Abel's kernel. The Abel's kernel parameters have been
determined by the proposed method. The second creep curve site is characterized by con-
stant creep rate. Creep rate dependence on stress has been established. Description of the
third creep curve site should be carried out on the basis of continuum damage mechanics
169
approach, it is not considered here and will be given in other works of authors. Depend-
ences of creep curve sites durations and long-term strength of the asphalt concrete have
been approximated.
2 MATERIALS
2.1 Bitumen
In this paper bitumen of grade 100–130 has been used which meets the requirements of the
Kazakhstan standard (ST RK 1373 2013).The bitumen grade on Superpave is PG 64–40
(Superpave series No.1 2003). Basic standard indicators of the bitumen are shown in Table 1.
Bitumen has been produced by Pavlodar processing plant from crude oil of Western Siberia
(Russia) by the direct oxidation method.
Measurement Requirements of
Indicator unit ST RK 1373 Value
Value
170
Table 3. Basic standard indicators of the asphalt concrete.
Measurement Requirements of
Indicator unit ST RK 1225 Value
3 TEST METHODS
3.2 Test
Tests of hot asphalt concrete samples in a form of rectangular prism on creep were carried
out according to the direct tensile scheme until a complete failure. The test temperature was
equal to 20±2°C, stress was changing from 0.04 to 0,24 MPas. The tests were carried out in
a special assembled installation, which allows applying a load to the asphalt concrete sample
within 1 second. The sample strain was measured by means of two clock typed indicators
while data was recorded in a video camera.
171
4 THEORY
1⎡ ⎤
t
ε (t ) ⎢σ (t ) ∫ K (t τ )σ (τ ) dτ ⎥ (1)
E⎣ 0 ⎦
where ε(t) = the strain at the time point t; σ(t),σ(τ) = the stresses at time points t and τ;
K(t−τ) = the creep kernel; E = the instant modulus of elasticity; t = the observation time; and
τ = the time preceding the observation time.
Taking into account that for creep process σ0=const from the Equation 1 we find:
⎡ t
⎤
ε (t ) ε 0 ⎢1 ∫ K (t τ ) dτ ⎥ (2)
⎣ 0 ⎦
⎡ δ 1−α ⎤
ε ( ) ε0 1 t ⎥ (4)
⎣ 1− α ⎦
Thus the creep equation with the Abel’s kernel contains three unknown parameters – ε0, α
and δ.
172
2
m
⎡ ⎛ δ 1−α ⎞ ⎤
S (ε 0 δ ) ∑ ⎣ε 0 1 ti ε eei ⎥ → min (5)
i =1
⎝ 1 − α ⎠ ⎦
where S(ε0,δ) = the sum of squares of deviations; and m = the number of creep strain values
determined experimentally.
From two equations composed on the expressions and we find expressions to determine
parameters ε0 and δ:
m m m m
∑ε ∑ i
2(11−α )
∑ ( ) ∑ ε t(
i
1 α
1i
1−α )
ε0 = i =1 i =1 i =1 i =1
2 (6)
m
⎡m ⎤
m∑ t i
2(1−α )
− ⎢ ∑ ti( − ) ⎥
i =1 ⎣ i =1 ⎦
m
⎛ εi ⎞
∑ ⎜⎝ ε
i =1
− 1⎟ ti( −α )
⎠
δ= 0
(7)
1 m 2( −α )
∑t
1 − α i =1 i
Setting values of the parameter α in the interval (0, 1) with a certain step from the expres-
sion (6) we will find values of the parameter ε 0 ε 0 (α ) . Substituting the found values of the
parameter ε 0 and corresponding values of the singularity parameter α in the expression (7)
values of the parameter δ δ (ε 0 , α ) have been determined.
Further sequentially substituting values of the singularity parameter α and calculated cor-
responding values of the parameters ε0 and δ in the expression (4) values of creep strain
ε (t i ) ε (ti , α , ε 0 , δ ) are calculated.
If we designate an average deviation of calculated values of creep strains from experimen-
tal values using Δε m ( λ , ε , δ ) as a selecting criterion of the best values of the parameters α, ε0
and δ the following condition can be accepted:
Δε mi (α i , ε i , δ i ) → min (8)
5 EXPERIMENT
173
Figure 3. The asphalt concrete creep curve at stress 0.24 MPa.
Figure 4. The site I of the asphalt concrete creep curve at stress 0.24 MPa.
174
ing two values of the singularity parameter α and two values of the discrepancy Δε we can
construct straight line graphics. On this straight line we find a value of the parameter α cor-
responding to the average discrepancy Δε = 0. From the expressions (6) and (7) taking into
account the found value of the parameter α we define the optimum values of the parameters
ε0 and δ.
The optimal values of the creep parameters defined at approximation of the site I of the
asphalt concrete creep curves at different stresses are given in Table 4.
Figure 7 presents the site II of the asphalt concrete creep curve at stress of 0.24 MPa where
experimental strains are designated by points and the solid line represent the approximating
line. We can see that the asphalt concrete strains on the site II of the creep curve is accurately
ai
described by the straight line equation. On this site the creep process is characterized by a
constant strain rate ε // . The rate values at different test stresses are given in Table 5. As can
be seen the site II of the asphalt concrete creep curve at all testing stresses is described by
a straight line with high accuracy. It was found that strain rate depends on stress and this
dependence is satisfactorily described by a power function (Fig. 8). With increasing of stress
the strain rate also increases.
Table 4. The Abel’s kernel parameter values in the site I of the creep curve.
Parameters Average
discrepancy
Stress σ, MPa ε0 α δ ∆ε,%
Figure 7. The site II of the asphalt concrete creep curve at stress of 0.24 MPa.
175
Figure 8. The strain rate dependence on the site II of the asphalt concrete creep curve on stress.
on long-term strength of the asphalt concrete is practically the same as on the creep curve
sites durations: the stress change with one order causing change of the failure time of three
orders.
6 CONCLUSION
Hot fine-grained asphalt concrete samples were tested on creep by direct tensile scheme at a
temperature of 20±2°C and stress from 0.04 to 0.24 MPa.
An asphalt concrete creep curve as the most viscoelastic materials at average temperatures
and stresses has three characteristic sites: the site of unstabilized creep with decreasing strain
rate, the site of stabilized creep with constant (minimum) strain rate and the site of accelerat-
ing creep with increasing rate which precedes failure.
Herein, it is proposed to describe a creep process of asphalt concrete within the first site
by the integral equation of Boltsmana-Volterra with the Abel’s creep kernel. The method of
determining the kernel parameters is given. It is shown that the proposed model describes
with high accuracy a creep process on the first site.
Moreover, creep strain of the asphalt concrete on the second site is accurately described
by a straight line. Creep process at this site is characterized by a strain rate. It was found that
177
the strain rate depends on stress and this dependence is satisfactorily described by a power
function.
Durations of all three creep curves sites with high accuracy are approximated by a power
function. With increase in stress the durations of the sites are decreased. Influence degree of
stress on durations of all sites of the creep curves is identical: change of stress by one order
causes change of the durations by more than three orders.
Finally, long-term strength of the asphalt concrete is also satisfactorily approximated by a
power function. As well as in the case of the creep curve sites durations, the stress influence
degree on long-term strength of the asphalt concrete such that stress change by one order
causes change of the failure time by three orders.
REFERENCES
Cristensen, R.M. 1971. Theory of viscoelasticity: An introduction. New York: Academic Press.
EN 12697–33. 2003. Bituminous Mixtures. Test Methods for Hot Mix Asphalt. Part 33: Specimen pre-
pared by roller compactor. Brussels: European Committee for Standardization.
Ferry, J.D. 1980. Viscoelastic Properties of Polymers. New York: John Wiley & Sons, Inc.
Hopkins, I.L. & Hamming R.W. 1957. On creep and relaxation. Journal of Applied Physics 28:
906–909.
Kachanov, L. 1986. Introduction to continuum damage mechanics. Dordrecht: Martinus Nijhoff
Publishers.
MS-4. 2008. The Asphalt Handbook. 7th Edition. Lexington: Asphalt Institute.
Papagiannakis, A.T. & Masad, E.A. 2008. Pavement Design and Materials. New Jersey: John Wiley &
Sons, Inc.
Rabotnov, Yu.N. 1987. Introduction to fracture mechanics. Moscow: Nauka.
ST RK 1225. 2013. Hot mix asphalt for roads and airfields. Technical specifications. Astana.
ST RK 1284. 2004. Crushed stone and gravel of dense rock for construction works. Technical specifica-
tions. Astana.
ST RK 1373. 2013. Bitumens and bitumen binders. Oil road viscous bitumens. Technical specifications.
Astana.
Superpave series No. 1. 2003. Performance graded asphalt binder specification and testing. Lexington:
Asphalt Institute.
Tschoegl, N.W. 1989. The Phenomenological Theory of Linear Viscoelastic Behavior. An Introduction.
Berlin: Springer-Verlag.
Yoder, E.J. & Witczak, M.W. 1975. Principles of Pavement Design. New Jersey: John Wiley & Sons,
Inc.
178
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Hongfu Liu
Key Laboratory of Road Structure and Material of Ministry of Transport (Changsha),
Changsha, P.R. China
College of Traffic and Transportation Engineering Changsha University of Science and Technology,
Changsha, Hunan, P.R. China
ABSTRACT: This paper evaluates the variation of the residual tensile strength of prismatic
specimens made of Hot Mixture Asphalt (HMA) due to tensile cyclic loading. The pre-set
number of cyclic loads is the fatigue life of the average of 20%, 50%, 65% and 80%. The
specimens were submitted to a direct tensile static load test until failure. The tests show a
gradual reduction in the residual tensile strength values, in accordance with the number of
cycles that are applied. A new definition of damage is introduced, as the relative variation
of the tensile strength with the number of cycles. The nonlinear fatigue damage evolution
equation based on the residual strength degradation is established. The greater the nominal
stress level is, the lower the curve of the damage evolution curve is, and the more moderate
fatigue damage evolution speed is, which varies with life radio.
1 INTRODUCTION
Fatigue properties are important aspects of asphalt pavement structure design theory and
methods of applied basic research. At present, test conditions, test methods, material compo-
sition, load characteristics, environmental impacts, and other factors of fatigue life of asphalt
mixtures have been the subject of numerous research. However, traditional classical fatigue
analysis on the essence of fatigue properties is limited. The S-N curves provide no informa-
tion on the variation of residual tensile strength with the number of cycles. Still, establishing
a continuum damage mechanics model can better describe the essence of fatigue damage
characteristics. Nonlinear fatigue damage evolution and accumulation in asphalt mixtures
have been a research hotpot.
Many researchers have studied the fatigue life of hot mixture asphalt. In almost all cases,
the specimens were subjected to pre-set maximum and minimum stress levels and were tested
until failure. The result of the test was the determination of the number of cycles that the
specimen is able to withstand. Principles were outlined by Schapery, Kim, and continuous
damage theories were used to develop a visco-elastic continuum fatigue damage model. Zhi
Suo et al. conducted indirect tensile fatigue tests using three types of asphalt concrete to
establish a fatigue damage model as well as several damage criteria. Zheng Jianlong put for-
ward a new method of determining fatigue test stress ratio using the true stress of dam-
age mechanics. The aforementioned author also proposed asphalt mixture nonlinear fatigue
damage theory and analysis method.
In all cases, it may be noted how the cyclic loads cause increasing damage to the concrete
up until its final collapse. However, the way in which its mechanical parameters evolve with
the number of cycles is not sufficiently well documented. In particular, there is a significant
179
knowledge gap in the precise definition of how the flexural strength of concrete progresses
with the number of cycles. This is because it is not possible to measure such a parameter dur-
ing the application of cyclic load (contrary to what happens, for example, with the modulus
of elasticity, or the residual strain), given that the parameter is exclusively obtained from a
failure test.
The research work presented in this document analyses the way in which the direct tension
strength of asphalt concrete AC-13C varies in specimens that have previously been subjected
to a direct cyclic tension test. The number of loading cycles has previously been established.
In all cases, the chosen stress level and the number of loading cycles caused the failure of
the specimen. Having concluded the cyclic test, the specimens are subjected to a direct static
tension tests, up until fracture, and the residual tensile strength values are measured. This
paper evaluates the variation of the residual tensile strength of prismatic specimens made of
asphalt concrete due to tensile cyclic loading. A new definition of damage is introduced, as
the relative variation of the tensile strength with the number of cycles. The decay law of the
residual strength under different stress levels is revealed. The nonlinear fatigue damage evolu-
tion equation based on the residual strength degradation is established.
2 EXPERIMENTAL INVESTIGATION
2.1 Materials
Take the asphalt mixture of fine grained AC-13C as the research object, the raw materials
are the east China sea oil SBS modified asphalt and basalt aggregate production in Zhuzhou
city, Hunan province. The result of test of asphalt is shown in Table 1, and the aggregate
density and mechanical test results are shown in Tables 2 and 3. All properties of the asphalt
and the aggregate met technical requirements of the Standard Specification for Construction
and Acceptance of Highway Asphalt Pavements (JTG F40–2004). Aggregation gradation of
asphalt mixture AC-13C is shown in Table 4.
Optimal asphalt content was determined using Marshall test. Results are presented in Table 5.
Block samples 300 mm × 300 mm × 50 mm were made by combining asphalt mixtures.
Small beam specimens were trimmed to 250 mm × 50 mm × 50 mm for testing.
180
Table 2. Test results of density of basalt aggregate.
Table 3. Test results of Impact value, Polishing value, Abrasion value of basalt aggregate.
Aggregate type Impact value (%) Polishing value (BPN) Abrasion value (%)
Sieve size (mm) 16 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075
Asphalt Bulk specific Volume of Voids filled Voids in mineral Marshall Flow
aggregate gravity air voids, with asphalt, aggregate, stability value
ratio (%) (g/cm3) VV (%) VFA (%) VMA (%) (kN) (0.1 mm)
strength. Tensile failure is the criterion for fatigue failure of the test specimen. According to
the requirements of conventional fatigue test, first, direct tensile strength test on 3 specimens
under quasi-static loading, at the loading rate of 5 mm/min. Test temperature is 20°C. The
strength of direct tensile (S), while beam is fractured, is calculated as follows.
Sj Fmax /A
A (1)
In formula, S j means the direct tensile strength of specimens under quasi-static loading
conditions, unit of MPa. Fmax means the load at destroying, unit of N. A means average cross-
sectional area value of 3 specimens, unit of mm2.
The test results of the direct tensile strength of specimens under quasi-static loading con-
ditions are shown in Table 6. The average value of the 3 parallel test results is 1.228 MPa.
Average value of direct tensile strength test result is 1.228 MPa.
Figure 1. Direct tensile test of small beams and Figure 2. Loading wave of direct ten-
stress diagram. sion fatigue within 0.5 MPa.
an MTS 661.20H-03 load cell with a range of 100 kN, for both traction and compression,
and an average error of below 0.5% in its range. Displacement of each stress cycle and stress
are automatically recorded by the data acquisition system, thus allowing stress and strain to
be calculated. Data acquisition interval can be set according to specific test conditions. The
speed of data capture is 200 Hz.
The test campaign was in three phases:
1 Cyclic load tests to failure. These tests consist in subjecting a total of twelve specimens to
a direct cyclic tension test to failure, to obtain the characteristic number of cycle failures
of the concrete. Tensile stress levels used in the tension fatigue test are 1.00, 0.50, and
0.25 MPa. Nominal stress ratios are 0.81, 0.41, and 0.20, respectively.
2 Cyclic load tests to a pre-set number of cycles. This phase consists in subjecting a total of
forty-two specimens to a direct cyclic tension test, up to a pre-set number of load cycles,
with the purpose of causing controlled fatigue damage to the specimen. The specimens
were not taken to failure, but the cyclic load tests concluded when a number of pre-set
cycles had been reached.
3 Static tests after cyclic load: The final phase consists in subjecting a total of fifty-three
specimens (forty-two of them previously subjected to cyclic loading and another eleven
that had not been subjected to cyclic load) to a direct static tension test. In this way, the
variation of the residual tensile strength with the number of cycles is determined.
Fatigue test method involves applying direct tension fatigue to asphalt mixture beam
specimens. Figure 1 shows an ongoing direct tension fatigue test. Scheme and main con-
trol standards of fatigue test are as follows. (1) Loading mode adopts stress-controlling
fatigue; specimen-molding method adopts a rutting plate, which is then cut into small beams.
(2) Loading waveform and frequency are set at 10 Hz continuous alternating sine waveform.
Load intermission is not considered. Loading waveform for two cycles under a stress level is
shown in Figure 2. (3) Temperature is set to 20 ± 1°C. Prior to the test, the specimen is kept
in the chamber for 12 hours.
182
Centering the axis of tension strength or fatigue test is crucial. The pull head of bonding
specimens is engraved with concentric circles composed of fine lines to minimize effects of
eccentric tension. On one hand, bonding specimens improve bond strength, increase friction,
and ensure better axis-centering. On the other hand, connect MTS with spherical valve at
the Both ends of the pull head, making sure that specimens are under axial tension. Small
beam and pull head were bonded with epoxy resin steel adhesive which was made of A, B
component at the ratio of 10:3, mixing evenly, with one-hour solidifying. After three days,
the tension strength of specimen meets the testing requirements.
lg σ = A − B lg N f (2)
where, σ is stress level, unit of MPa. N f is fatigue life, unit of times. A and B are regression
coefficients.
In double logarithmic coordinates, stimulating Figure 3 with formula (2), we can get the
following formula. Then, the traditional fatigue equation (3) can be achieved.
Table 7. Testing results of direct tensile fatigue test under different stress levels.
183
Figure 3. Fitting curve of the tension fatigue equation.
Table 8. The design of different fatigue cycles in the residual strength test.
Cycles
large, otherwise specimen would be failure before achieving the pre-set number of cycles and
reduce the likelihood of success considerably. The other principle is that obvious gaps are
essential for pre-set the number of cycles of residual strength. And the results can be fully
reflected the decay rule of the residual strength. Selection of standard values of fatigue life
of residual strength test is one of the key factors affecting test results. In this paper, the 20%,
50%, 65% and 80% of the average of the fatigue life were chosen as pre-set the number of
cycles. Dynamic load strength of specimen at the same loading speed with fatigue test was set
out three seconds later after unloading. The designed of different fatigue cycles under differ-
ent stress level are shown in Table 8.
N f is the average value of different specimens fatigue life.
First of all, the initial dynamic load strength of three different stress levels including
1 MPa, 0.5 MPa and 0.25 MPa were tested under the loading speed of 20 MPa/s, 10 MPa/s
and 5 MPa/s respectively. Then the residual strength of the specimens was tested after the
specimens undergo corresponding number of fatigue cycles at the same loading speed with
the determination of initial strength. The results of parallel test should be discarded selectiv-
ity, according to the discrete degree of the date. Results of fatigue residual strength test under
the stress of 1, 0.5, 0.25 MPa are shown in Tables 9, 10 and 11.
3.3 Analysis of tension fatigue damage characteristics based on the decay of strength
3.3.1 Definition of fatigue damage variable based on the decay of strength
Static tests after a pre-set number of cyclic load, residual tensile strength decay gradually
with the increase of number of fatigue. The performance degradation law is consistent with
the process of fatigue under repeated load. This macro physical and mechanical properties of
residual strength indicators describe the damage degree, with the change of residual strength
damage variables is appropriate. Characterization of asphalt mixture of macroscopic inter-
nal structure changes due to damage. Damage variable is defined based on the degradation
of the residual tensile strength, and is expressed as equation (4).
184
Table 9. Results of fatigue residual strength test under the stress of 1 MPa.
Table 10. Results of fatigue residual strength test under the stress of 0.5 MPa.
S0 Sr ( N ) S (N )
Ds ( N ) = = 1− r (4)
S0 S0
where, Ds ( N ) is the fatigue damage variable based on the decay of strength, and is the dam-
age value after loading N times. S0 is the average value of initial strength. Sr ( N ) is the residual
tensile strength after loading N times.
185
Table 11. Results of fatigue residual strength test under the stress of 0.25 MPa.
Table 12. The damage after a pre-set number of cyclic load under the stress of 1 MPa.
1 0 0.0 4.577 0 — —
2 344 20.0 4.497 0.061 0.001 0.077
3 5.311 −0.109
4 4.568 0.046
5 4.796 −0.002
6 860 50.0 4.573 0.045 0.048 0.008
7 4.517 0.057
8 4.584 0.043
9 1117 65.0 4.378 0.086 0.094 0.041
10 4.133 0.137
11 4.253 0.112
12 4.587 0.042
13 1375 80.0 2.956 0.383 0.242 0.127
14 3.429 0.284
15 4.406 0.080
16 3.726 0.222
17 1633 95.0 3.226 0.326 0.278 0.068
18 3.687 0.230
The specimen occurs fatigue fracture when the residual strength reduced to equal to the
stress level. According to the definition of fatigue damage variable, the damage which is cal-
culated after a pre-set number of cyclic load by the residual strength are shown in Table 12,
Table 13 and Table 14.
Sr ( N ) = (S
(S )( N / N f )u + σ max (5)
186
Table 13. The damage after a pre-set number of cyclic load under the stress of 0.5 MPa.
1 0 0.0 4.070 0 — —
2 283 20.0 3.948 0.036 0.025 0.064
3 3.757 0.083
4 4.275 −0.043
5 5707 50.0 3.867 0.056 0.085 0.053
6 3.974 0.030
7 3.487 0.149
8 3.666 0.105
9 7419 65.0 3.357 0.181 0.128 0.050
10 3.593 0.123
11 3.765 0.081
12 9131 80.0 2.969 0.275 0.200 0.063
13 3.513 0.143
14 3.469 0.153
15 3.162 0.228
Table 14. The damage after a pre-set number of cyclic load under the stress of 0.25 MPa.
1 0 0.0 3.476 0 — —
2 24426 20.0 3.398 0.022 0.019 0.078
3 3.145 0.095
4 3.689 −0.061
5 61066 50.0 2.813 0.191 0.082 0.112
6 3.593 −0.034
7 3.168 0.089
8 79385 65.0 3.133 0.099 0.143 0.088
9 3.180 0.085
10 2.626 0.245
11 97705 80.0 3.114 0.104 0.216 0.106
12 2.686 0.227
13 2.377 0.316
where, σ max is the stress amplitude applied by the fatigue test, N f is the fatigue life, and u is a
damage parameter of material.
According the definition of fatigue damage variable based on the decay of strength, the
nonlinear fatigue damage model is expressed as equation (6).
= ⎜1 − mmax ⎟ ⎢1 − ⎜1 − ⎥ = D ⎢ 1 − 1 −
⎟ ⎥
⎝ S0 ⎠ ⎢ ⎝ N f ⎟⎠ ⎥ cs ⎜
⎢⎣ ⎝ N f ⎠ ⎥⎦
⎣ ⎦
where, Dcs is the critical fatigue damage calculated by the residual tensile strength. Other
parameters are the same with equation (5).
187
3.3.3 Determining tension fatigue damage parameter
Based on equation (6), results of the tension fatigue damage test are fitted to determine
model parameters. Tension fatigue damage fitting results of the asphalt mixture under differ-
ent stress levels are shown in figures 4a, b, c. The solid line is a fitting curve. The black points
are dates tape formed by test results.
By substituting fitting results of the damage parameter into equation (6) of the nonlinear
fatigue damage model, tension fatigue damage evolution equation related to stress level is
obtained.
Tension fatigue damage evolution curve under different stress levels according to damage
evolution equation (6) is plotted in figure 5.
It can be seen from figure 5 that the fatigue damage evolution curve can be roughly divided
into two stages. The first stage is the stable growth of damage, where damage growth presents
an approximately uniform and slow increase. The second stage is damage failure, where dam-
age growth increases rapidly until the specimen fracture. In figure 5, from top to bottom, cor-
responding to the damage evolution curve is respectively stress level at 0.25, 0.50, and 1 MPa.
The greater the nominal stress level is, the lower the curve of the damage evolution curve is,
and the more moderate fatigue damage evolution speed is, which varies with life radio.
Figure 5. Tension fatigue damage evolution curve under different stress levels.
188
4 CONCLUSIONS
1. The tests carried out on specimens subjected to a pre-set value of cyclic loads, without
reaching failure, show a gradual reduction in the residual tensile strength values, in accord-
ance with the number of cycles that are applied.
2. A concept of damage is proposed, which is defined as the relative variation of the residual
tensile strength of the asphalt concrete with the number of cycles.
3. Tension fatigue damage evolution of asphalt mixtures exhibited roughly two stages as
stable growth and unstable failure. The greater the nominal stress level was, the lower the
curve of the damage evolution curve was, and the more moderate fatigue damage evolu-
tion speed was, which varied with life radio.
ACKNOWLEDGEMENTS
The project was supported by open fund of Key Laboratory of Road Structure and Material
of Ministry of Transport (Changsha University of Science & Technology) (kfj120204), and
Hunan Province Natural Science Youth Talent Cultivating Mutual Fund (14 JJ6018), and sci-
ence and technology project of Hunan province department of education (09c068), and key
engineering projects of science and technology of Guangdong Provincial Communications
Department (2013-01-002), and project of applied basic research of Ministry of Transport
(2013 319 223 020), and Guizhou province science and technology fund. The authors would
like to express gratitude to their financial support.
REFERENCES
Castro, M. & Sanche, J. 2007. Estimation of asphalt concrete fatigue curves-A damage theory approach
[J]. Construction and Building Materials 12: 1–7.
Di Benedetto, H. & de la Roche, C. 1998. State of the art on stiffness modulus and fatigue of bituminous
mixtures in bituminous binders and mixtures: state of the art and interlaboratory tests on mechanical
behavior and mix design [J]. E&FN Spon, Ed. L. Francken: 137–180.
Di Benedetto, H., C. de La Roche & H. Baaj. etc. 2004. Fatigue of bituminous mixture [J]. Journal of
Materials and Structures 37: 202–216.
Didier Bodin & Gilles Pijaudier-Cabot. & Chantal de La Roche. etc. 2002. A continuum damage
approach of asphalt concrete fatigue tests [J]. ASCE.
Dongre, R. & Myers, L. & D. Angelo, J. & Paugh, C. & Gudimettla, J. 2005. Field evaluation of witczak
and hirsch models for predicting dynamic modulus of hot-mix asphalt [J]. AAPT
Guang-song, Yang. 1995. Damage mechanics and composite material damage [M]. Beijing: National
Defence Industrial Press: 25–34. (in Chinese)
Hong-zhou, Zhu & Xiao-ming, Huang. 2005. Fatigue model of asphalt mixtures based on damage the-
ory [J]. Journal of Highway and Transportation Research and Development 22(2): 4–7. (in Chinese)
Hong-zhou, Zhu & Zhao-yi, He. & Xiao-ming, Huang. et al. 2007. Experimental research on fatigue
performance of asphalt treated base mixtures [J]. Journal of Highway and Transportation Research
and Development 24(4): 7–11. (in Chinese)
Jaeseung Kim, Ph.D & Randy, C. & West, Ph.D. P.E. 2010. Application of the viscoelastic continuum
damage model to the indirect tension test at a single temperature [J]. Journal of Engineering Mechan-
ics ASCE.
Jian-long, Zheng & Song-tao, Lv. 2009. Non-linear fatigue damage model for asphalt mixtures [J].
China Journal of Highway and Transport 22(5): 21–29. (in Chinese)
Kim, J., Sholar, G. & Kim, S. 2008. Determination of accurate creep compliance and relaxation modu-
lus at a single temperature for viscoelastic solids [J]. Journal of Materials in Civil Engineering, ASCE
20(2):147–156.
Kim, J. & West, R.C. 2010. Application of the visco-elastic continuum damage model to the indirect
tension test at a Single temperature [J]. Journal of Engineering Mechanics, ASCE15(3): 59–93.
Kuang-huai, Wu. & Xiao-ning, Zhang. 2007. Experimental research on uniform model for nonlinear
evolution equation of fatigue damage of asphalt mixture [J]. Highway 30 (5): 125–130. (in Chinese)
Lundstrom, R. & Isacsson, U. & Ekblad, J. 2003. Investigations of stiffness and fatigue properties of
asphalt mixtures [J]. Journal of Materials Science 38 (7): 4941–4949.
189
Seoy & Kim, R. 2008. Using acoustic emission to monitor fatigue damage and healing in asphalt
concrete [J]. Journal of Civil Engineering 12(4): 237–243.
Song-tao, Lv. & Jian-long, Zheng. 2011. Fatigue properties of asphalt mixtures under broad stress
ratio conditions: improved S-N model [J]. Journal of Highway and Transportation Research and
Development 28(8): 1–6. (in Chinese).
Zhe-sheng, Ge, Zhao-hui, Huang & Xiao-ming, Huang. 2002. Study on effect factors of asphalt mixes
fatigue properties [J]. Journal of Highway and Transportation Research and Development 19(6): 1–4.
(in Chinese).
Zhi-lin, Sun, Xiao-ming, Huang & Rui, Zhang et al. 2008. Study on effect of asphalt modifier on asphalt
mixtures fatigue performance [J]. Journal of Highway and Transportation Research and develop-
ment 25(4): 33–37. (in Chinese).
Zhi Suo & Wing Gunwong. 2009. Analysis of fatigue crack growth behavior in asphalt concrete material
in wearing course [J]. Construction and Building Materials 23(5): 462–468.
Zu-zhong, Li, Shuan-fa, Chen & Deng-liang, Zhang. et al. 2007. Fatigue test of material
tension-compression in stress tension and absorbing Layer [J]. Highway 32(10): 190–195.
(in Chinese).
190
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Ankit Gupta
Civil Engineering Department, Indian Institute of Technology (BHU) Varanasi, Uttar Pradesh, India
Sanjeev Adhikari
School of Engineering and Information Systems, Morehead State University, Morehead, KY, USA
1 INTRODUCTION
191
vehicular loadings, structural layers and environmental factors. Before M-E pavement, the
design procedures followed with more empirical in nature rather than mechanistic. A new
philosophy has been adopted in the recent MEPDG guidelines (NCHRP, 2004) to carry
out detailed distress development over the design life. It provides three levels of design. The
empirical Paris law of cracks propagation for bottom-up cracks length is further empiri-
cally correlated with the percentage of surface cracks area. However, it could not correlate
adequately (Rajbongshi, 2008; Rajbongshi and Das, 2009). In the similar line, for the first
time this guide has also introduced an approach for top-down fatigue cracking, where the
top-down cracking has assumed at longitudinal direction. The use of vertical compressive
strain at top of subgrade for traditional rutting evaluation has been abandoned, and instead,
it considers the prediction of accumulated plastic strain with repetitions in each layer.
Traditionally, the fatigue and rutting performances for a given pavement section are eval-
uated primarily based on the initial critical horizontal tensile strain (εt) at the bottom of
asphalt layer and the initial critical vertical compressive strain (εz) at the top of subgrade
layer respectively. To calculate horizontal tensile strain and vertical compressive strain in a
3-D multi-layered composite structure with distributed loads, it requires high technical skill,
expertise manpower and good computing facility as well.
Damage in pavement structures propagates with repetitive loads. This happens due to
material(s) degradation due to load repetitions, which ultimately leads to failure situation.
To represent the structural degradation, various parameters like strain levels, cracks length,
phase angle, energy dissipation, asphalt stiffness variation, permanent settlement, etc. can be
used. In case of rutting distress, pavement damage is the accumulation of plastic strain as
function of repetitions which directly reflects with the rutting evaluation criterion (i.e. total
rut depth). Therefore, for any given repetitions the rutting damage may be taken as the ratio
of rut depth to the allowable rut at failure (say, 20 mm). This is not case of fatigue perform-
ance evaluation. Fatigue is caused due to repetitive loads in the asphalt layer and thus, cracks
initiate at the bottom of asphalt layer and subsequently it propagates towards the surface.
However, the fatigue failure is measured in terms of percentage of surface cracks area in the
form of alligator or map cracking. Various techniques can be applied for fatigue evaluation.
Many researchers attempted to evaluate fatigue damage using fracture mechanics (Birgis-
son et al., 2007; Di Pasquale and Cakmak, 1989; Ghuzlan and Carpenter, 2000; Huang, 2004;
Krawinkler and Zohrei, 1983; Lundström, 2002; Jacobs et al., 1996; Majidzadeh et al., 1971;
Ramsamooj, 1991; Roufaiel and Meyer, 1987; Zhang et al., 2001; Zhang et al., 2013). Fracture
based cracks propagation in the asphalt mixtures is very complex and uncertain due to non-
homogeneity and visco-elastic properties. This approach assumes an initial cracks and it fails
to predict the damage prior to initiation of micro-cracks. Most of the traffic repetitions in an
in-service asphalt pavement, micro cracks appears prior to any visible cracks. Also many litera-
tures used dissipated energy approach to describe the asphalt fatigue behavior (Baburamani and
Porter, 1996; Chomton and Valayer, 1972; Daniel and Bisirri, 2005; Pronk and Hopman, 1991;
Shen and Carpenter, 2005; Shen et al., 2010; Van Dijk and Visser, 1977; Zhang et al., 2013). Dis-
sipated energy approach assumes that all dissipated energy produces damage. Moreover, this is
not true specially due to damping effect in visco-elastic materials like asphalt mixtures.
On account of structural degradation with load repetitions, the incremental fatigue damage
in each repetition would be different. The accumulated damage would vary nonlinearly with
repetitions. Consideration of nonlinearity becomes important when the pavement performance
is assessed at an intermediate condition of the structures either for maintenance purpose or for
reliability based pavement design. Considering different load levels, nonlinear damage models
is similar to double linear damage, damage curve, double damage curve, etc. Nonlinear dam-
age models were developed by various researchers (Manson et al., 1967; Manson and Halford,
1981; Richart and Newmark, 1948). However, these approaches do not account the effect of
material degradation with repetitions. Miner’s hypothesis (Huang, 2004) also postulates linear
damage with different loads or strain levels. Most of M-E design practices also (AI, 1999; Aus-
troads, 2004; French, 1997; IRC, 2012; Shell, 1978) adopt a linear damage principle.
In context of the fatigue damage evaluation in M-E approach, the asphalt stiffness reduc-
tion with load repetitions may adequately represent the material deterioration which ulti-
192
mately leads to fatigue failure. Traditionally, 50% reduction in asphalt stiffness is used as
fatigue failure criteria (Baburamani, 1999; Chakroborty and Das, 2003; Huang, 2004; Kim
et al., 2003; Lundström, 2002; Mackiewicz, 2013; Read and Collop, 1997; SHRP, 1994). Vari-
ous researchers (Collop and Cebon, 1996; Lundström, 2002; Oliveira et al., 2008; Under-
wood and Kim, 2013) studied the stiffness variation in different asphalt materials, vehicular
loadings, structural layers and environmental factors.
An asphalt pavement is idealized as a layered structure which contains normally two to four
layers made up of asphalt surfacing, base, sub-base and subgrade layers. The structure failures
of pavements happen due to repetitive vehicular and/or thermal loading (except shrinkage
cracking). A pavement design should be sustainable to the repetitive loadings for a specific
design period. Conceptually, Mechanistic-Empirical (M-E) principles of pavement design
intend to correlate the most appropriate mechanistic parameter of the structure with the
number of load repetitions for a specific failure or damage mechanism. Fatigue and rutting
are two primary mechanisms of pavement failures. Traditionally, the initial critical horizontal
tensile strain (εt) at the bottom of asphalt layer for fatigue case and initial critical vertical
compressive strain (εz) at the top of subgrade layer for rutting case are empirically correlated
with the fatigue life and rutting life of pavement section respectively.
For the first time, Kerkhoven and Dormon (1953) suggested to use of vertical compressive
strain at the top of subgrade for rutting evaluation. Saal and Pell (1960) was the first to rec-
ommend the use of horizontal tensile strain at the bottom of asphalt layer to minimize fatigue
cracking. Dormon and Metcalf (1965) also adopted these concepts for development of a
rational design method. The concept of using εt and εz in pavement design has been adopted
in various guidelines (AASHTO, 1993; AI, 1999; Austroads, 2004; French, 1997; IRC, 2012;
NCHRP, 2004; Shell, 1978; TRL, 1993). Recent MEPDG guideline (NCHRP, 2004) has
adopted a little different procedure than that of the traditional M-E design method.
Accurate estimation of εt and εz in the pavement structures are complex task Primary
reasons of such complexities are—multilayered and its interaction between two layers, non-
elastic behavior of asphalt material, structural boundary conditions, axle and wheel configu-
rations including moving loads, environmental factors, etc. To avoid these limitations, certain
simplifications and approximations are made in the pavement analysis and the solutions for
εt and εz are obtained through numerical analysis. Software’s such as ABAQUS, ANSYS,
KENPAVE, FPAV, etc. are used by various researchers (Chandra et al., 2008, Das and Pan-
dey, 1999; Gupta et al., 2015; Hadi and Bodhinayake, 2003; Kuo and Chou, 2004; Helwany
et al., 1998; Lacey et al., 2007; Rahman et al., 2011) for numerical analysis and determined
the various stress-strain parameters in pavement structures.
k2 k3
⎛ 1⎞ ⎛ 1⎞
Nf k1 × ⎜ ⎟ ×⎜ ⎟ (1)
⎝ εt ⎠ ⎝ E1 ⎠
where, Nf is fatigue life in terms of a specific load level (say, standard axle load); εt is initial
critical horizontal tensile strain at the bottom of asphalt layer; E1 is initial stiffness of asphalt
material and, k1, k2 and k3 are regression constants. Different literatures (Behiry, 2012; Huang,
2004; Lee et al., 2003; Lundstr, M. 2002; Matthews et al., 1993; Monismith, 2004; Shukla and
Das, 2007; Sun et al., 2003; Matthews et al., 1993; Oliveira et al., 2008) suggest different val-
ues for these regression coefficients. Some of them are listed in Table 1.
Some literatures (Huang, 2004; Majidzadeh et al., 1983; Powell et al., 1984; Thompson,
1987; Verstraeten et al., 1982) reported fatigue equations without considering E1 exclusively
in Equation 1. The air voids and effective binder content of the asphalt mix are also included
in some fatigue equations (AI, 1999; Hajj et al., 2005; Huang, 2004; NCHRP, 2004). Further,
NCHRP (2004) has included an additional factor as function of asphalt layer thickness in
the calibrated fatigue equation.
Initially, the fatigue equation is developed under certain laboratory conditions and subse-
quently, it is calibrated using field data. While developing the laboratory equation, a 2-point,
3-point or 4-point bending test on asphalt beam samples are performed either in control strain
or in control stress mode (Baburamani, 1999; Behiry, 2012; Huang, 2004; Lundström, 2002;
Martono et al., 2007; Oliveira et al., 2008; Shen et al., 2006). Cyclic loading is allowed nor-
mally at 8–10 Hz frequency till failure of the sample. Traditionally, 50% reduction of asphalt
stiffness is used as failure criterion (Baburamani, 1999; Chakroborty and Das, 2003; Huang,
2004; Lundström, 2002), and the corresponding numbers of load repetition is recorded as
laboratory fatigue life. However, under the field conditions a pavement structure is subjected
to neither in strain control nor in stress control. There is significant mismatched between the
laboratory and field conditions in terms of loading conditions, structural conditions, failure
194
Table 1. Parameters of different fatigue equations.
definition, multiple failure mechanisms, environmental aspects, etc. and therefore a robust
calibration is essential. Rajbongshi and Das (2009) had presented a systematic procedure for
calibration of laboratory fatigue equation.
To evaluate the fatigue damage progression in asphalt materials, fracture mechanics and dis-
sipated energy approaches are also used. Of course, they are involved with many approxima-
tions, complexities and empiricisms at different levels and their laboratory test conditions do not
simulate the field situations. Fracture based models basically assumed an initial crack and subse-
quently, the number of load repetitions is empirically correlated with the propagated crack length
under certain test conditions. Recent MEPDG guideline (NCHRP, 2004) uses the empirical Paris
law of crack growth, where the bottom-up surface crack (in percentage of area) is further empiri-
cally correlated with the cracks length. However, the cracks length prediction in asphalt mixes is
very complex and uncertain, due to non-homogeneity and non-elastic material behaviour. An in-
service pavement can sustain large number of repetitions before any visible surface cracks (Shen
and Carpenter, 2007; Shen et al., 2006). Detail studies on cracks propagation in asphalt mixtures
can be seen elsewhere (Birgisson et al., 2007; Ghuzlan and Carpenter, 2000; Jacobs et al., 1996;
Lundström, 2002; Majidzadeh et al., 1971; Ramsamooj, 1991; Zhang et al., 2001).
There have been many efforts to use dissipated energy to describe the asphalt fatigue behav-
ior (Baburamani and Porter, 1996; Chomton and Valayer, 1972; Daniel and Bisirri, 2005; Pronk
and Hopman, 1991; Shen and Carpenter, 2005; Shen et al, 2010; Van Dijk and Visser, 1977;
Zhang et al., 2013). The amount of energy dissipated varies with load repetitions due to mate-
rial degradation, and the number of load repetitions is empirically correlated with the energy
dissipated under the test conditions. This approach assumes that all dissipated energy produces
damage. However, in visco-elastic materials like asphalt mix tends to store energy and a part of
the dissipated energy can be converted into thermal energy through visco-elastic damping.
In view of the complex fatigue behaviour of asphalt materials, high computational efforts,
and complex field situation, the pavement design methods adopted in various guidelines are
still phenomenological and empirical in nature. Therefore, the most popular pavement design
method adopted in various countries is the M-E approach which requires comparatively less
computational effort and easy understanding.
195
Normally, a pavement is considered to be failed by rutting when it exhibits a rut depth of
20 mm (AI, 1999; IRC, 2012; Lekarp et al., 2000; NCHRP, 2004). Various rutting models are
developed under accelerated pavement testing facility in the laboratory (Al-Khateeb et al.,
2007; Barker et al., 1977; Bejarano and Harvey, 2002; Choubane et al., 2006; Claessen et al.,
1977; Das and Pandey, 1999; Gokhale et al., 2005; Pidwerbesky et al., 1997; Prowell et al.,
2007) and subsequently, it is calibrated using field data. Vertical compressive strain at top
of subgrade layer is correlated with the number of load repetitions at failure (Behiry, 2012;
Erlingsson, 2012; Gupta et al., 2014; Huang, 2004; IRC, 2012; Ramsamooj et al., 1998; Saride
et al., 2014; Selvi, 2015; Shell, 1978; Tarefder et al., 2010; TRRL, 1970; Qiao et al., 2015). The
generic form of rutting equation used in M-E pavement design can be expressed as:
c2
⎛ 1⎞
Nr c1 × ⎜ ⎟ (2)
⎝ εz ⎠
where, Nr is rutting life in terms of a specific load level (say standard axles load); εz is initial
critical vertical compressive strain at the top of subgrade layer and, c1 and c2 are regression
constants. A large pool of rutting equations can be seen in different literatures. Some of these
are listed in Table 2.
Various researchers developed the rutting prediction model for asphalt mixtures with dif-
ferent perspectives. For example, Fwa et al. (2004) developed different regression equations
using c—φ parameters, Collop et al. (1995) included the visco-elastic properties in rut predic-
tion, Zhou et al. (2004) considered three different stages of rutting and so on. Ali et al. (1998)
presented a rutting model based on data from the in-service pavements and Simpson et al.
(1995) tried to predict the rutting using neural network. Researchers had tried to evaluate the
rutting performance without utilizing the concept compressive strain (εz). The contribution
of individual layer on total rut depth can be estimated from the accumulated plastic strain
with repetitions (Archilla and Madanat, 2000; Archilla and Madanat, 2001; Chen et al. 2004;
NCHRP, 2004; Stuart et al., 1999; Uzan, 2004). Thus, the total rut depth as function of
number of load repetitions (n) may be expressed as:
where, ε ip ( n ) is accumulated plastic strain after n repetitions of the ith layer; and hi is thickness
of the ith layer. This approach predicts rut depth under certain laboratory conditions for dif-
ferent materials. Therefore, a robust field calibration is necessary based on field observations.
Recent MEPDG (NCHRP, 2004) has adopted this method to find rutting life using various
empirical factors and, could obtained the correlation coefficient (R2) of 0.399 between the
observed versus predicted rut depth. NCHRP (2006) suggested omitting this approach in the
design process due to certain inadequacy and complexity in evaluation process.
196
3 DISCUSSION AND FUTURE SCOPE
REFERENCES
Al-Khateeb, G., Gibson, N., and Qi., X. (2007). Mechanistic analysis of FHWA’s accelerated load-
ing facility pavements primary response, Jr. of the Transportation Research Board, Transportation
Research Record No. 1990, pp.150–161.
Ali, H.A., Tayabji, S.D., and Torre, F.L. (1998). Calibration of mechanistic-empirical rutting model for
in-service pavements, Jr. of the Transportation Research Board, Transportation Research Record
No. 1629, pp.159–168.
American Association of State Highway and Transportation Officials (AASHTO). (1993). Guide for
design of pavement structure, AASHTO, Washington, D.C.
197
Archilla, A.R., and Madanat, S. (2000). Development of permanent rutting model from experimental
data, Jr. of Transportation Engineering, ASCE, 126(4), pp.291–299.
Archilla, A.R., and Madanat, S. (2001). Estimation of rutting models by combining data from different
sources, Jr. of Transportation Engineering, ASCE, 127(5), pp.379–389.
Asphalt Institute (AI). (1999). Thickness design—asphalt pavements for highways and streets, Manual
Series No.1, 9th Edition, The Asphalt Institute, Lexington, Ky, USA.
Austroads. (2004). Pavement design, Austroads, Sydney, Australia.
Baburamani, P. (1999). Asphalt fatigue life prediction models: a literature review. Research Report ARR
344. ARRB Transport Research Ltd., Vermont South, Victoria.
Baburamani, P.S., and Porter, D.W. (1996). Dissipated energy approach to fatigue characterization of
asphalt mixes, Proc. of the combined 18th ARRB TR Conf. Transit, New Zealand Symposium,
pp.327–347.
Baek, C., Underwood, B.S., and Kim, Y.R. (2012). Effects of oxidative aging on asphalt mixture proper-
ties, Jr. of the Transportation Research Board, Transportation Research Record No.2296, pp.77–85.
Barker, W.R., Brabston, W.N., and Chou, Y.T. (1977). A general design for the structural design of
flexible pavements, Proc. of 4th Int. Conf. on Structural Design of Asphalt Pavements, Ann Arbor,
Vol.1, pp.209–248.
Behiry, A.E.A.M. (2012). Fatigue and rutting lives in flexible pavement, Ain Shams Engineering Jour-
nal, Ain Shams University, Vol.3, pp.367–374.
Bejarano, M.O., and Harvey, J.T. (2002). Accelerated pavement testing of drained and undrained pave-
ments under wet base conditions, Jr. of the Transportation Research Board, Transportation Research
Record No.1816, pp.137–147.
Birgisson, B., Wang, J., and Roque, R. (2007). Numerical implementation of a strain energy-based frac-
ture model for HMA materials, Jr. of Road Materials and Pavement Design, 8(1), pp.7–45.
Chakroborty, P., and Das, A. (2003). Principles of transportation engineering, Prentice Hall of India
Pvt. Ltd., New Delhi, India.
Chandra, S., Viladkar, M., and Nagrale, P. (2008). Mechanistic approach for fiber-reinforced flexible
pavements, Jr. of Transportation Engineering, ASCE, 134(1), pp.15–23.
Chen, J.S., Lin, C.H., Stein, E., and Hothan, J. (2004). Development of a mechanistic empirical model
to characterize rutting in flexible pavements, Jr. of Transportation Engineering, ASCE, 130(4),
pp.519–525.
Chen, Y., Tebaldi, G., Roque, R., Lopp, G., and Su, Y. (2012). Effects of interface condition charac-
teristics on open-graded friction course top-down cracking performance, Jr. of Road Materials and
Pavement Design, 13(Sup.1), pp.56–75.
Chen, Z., Hainian, W., You, Z., and Biao, M. (2015). Sensitivity analysis of longitudinal cracking on
asphalt pavement using MEPDG in permafrost region, Jr. of Traffic and Transportation Engineer-
ing, Elsevier, 2(1), pp.40–47.
Chomton, G., and Valayer, P.J. (1972). Applied rheology of asphalt mixes practical application, Proc. of
the 3rd Int. Conf. on the Structural Design of Asphalt Pavements, London, pp.214–225.
Choubane, B., Gokhale, S., Sholar, G., and Moseley, H. (2006). Evaluation of coarse and fine graded
superpave mixtures under accelerated pavement testing, Jr. of the Transportation Research Board,
Transportation Research Record No.1974, pp.120–127.
Claessen, A.I.M., Edwards, J.M., Sommer, P., and Uge, P. (1977). Asphalt pavement design - the shell
method”, Proc. of 4th Int. Conf. on Structural Design of Asphalt Pavements, Ann Arbor, Vol.1,
pp.39–74.
Collop, A.C., Cebon, D., and Hardy, S.A. (1995). Viscoelastic approach to rutting inflexible pavements,
Jr. of Transportation Engineering, ASCE, 121(1), pp.82–93.
Collop, A.C., and Cebon, D. (1996). Stiffness Reductions of Flexible Pavements due to Cumulative
Fatigue Damage, Jr. of Transportation Engineering, 122(2), pp.131–139.
Collop, A.C., and Roque, R. (2004). Report on the prediction of surface-initiated longitudinal
wheel path cracking in asphalt pavements, Jr. of Road Materials and Pavement Design, 5(4),
pp.409–434.
Daniel, J.S., and Bisirri, W.M. (2005). Characterizing fatigue in pavement materials using a dissipated
energy approach, Proc. of the Pavement Symposium at the Geo-Frontiers Conference (EM2005),
ASCE, pp.1–10.
Das, A., and Pandey, B.B. (1999). Mechanistic-empirical design of bituminous roads: an Indian per-
spective, Jr. of Transportation Engineering, ASCE, 125(5), pp.463–471.
Di Pasquale, E., and Cakmak, A.S. (1989). On the relation between local and global damage indices
and local stiffness degradation, Technical Report, NCEER-89–0034 National Center for Earthquake
Engineering Research, State University of New York, Buffalo, N.Y.
198
Dormon, G.M., and Metcalf, C.T. (1965). Design curves for flexible pavement base on layered system
theory, Jr. of Transportation Research Board, Highway Research Record No.71, pp.69–84.
Erlingsson, S. (2012). Rutting development in a flexible pavement structure, Jr. of Road Materials and
Pavement Design, 13(2), pp.218–234.
French. (1997). French design manual for pavements structures, Guide Technique, LCPC and SETRA,
Francaise.
Fwa, T.F., Tan, S.A., and Zhu, L.Y. (2004). Rutting prediction of asphalt pavement layer using c - φ
model, Jr. of Transportation Engineering, ASCE, 130(5), pp.675–683.
Ghuzlan, K.A., and Carpenter, S.H. (2000). Energy-derived, damage-based failure criterion of fatigue
testing, Jr. of the Transportation Research Board, Transportation Research Record No.1723,
pp.142–149.
Gokhale, S., Choubane, B., Byron, T., and Tia, M. (2005). Rut initiation mechanisms in asphalt mix-
tures as generated under accelerated pavement testing, Jr. of the Transportation Research Board,
Transportation Research Record No.1940, pp.136–145.
Gupta, A., Kumar, P., and Rastogi, R. (2014). Critical review of flexible pavement performance models,
Jr. of Civil Engineering, KSCE, 18(1), pp.142–148.
Gupta, A., Kumar, P., and Rastogi, R. (2015). Critical pavement response analysis of low volume pave-
ments considering non linear behavior of materials, Jr. of the Transportation Research Board, Trans-
portation Research Record No.2474, pp.3–11.
Gupta, A., Kumar, P., and Rastogi, R. (2015). Mechanistic-empirical approach for design of low vol-
ume pavements, Int. Jr. of Pavement Engineering, 16(9), pp.797–808.
Hadi, M.N.S., and Bodhinayake, B.C. (2003). Non-linear finite element analysis of flexible pavements,
Advances in Engineering Software, Elsevier, Vol.34, pp.657–662.
Hajj, E.Y., Sebaaly, P.E., and Weitzel, D. (2005). Fatigue characteristics of superpave and hveem mix-
tures, Jr. of Transportation Engineering, ASCE, 131(4), pp.302–310.
Helwany, S., Dyer, J., and Leidy, J. (1998). Finite-element analyses of flexible pavements, Jr. of Trans-
portation Engineering, ASCE, 124(5), pp.491–499.
Huang, Y.H. (2004). Pavement analysis and design, Second Edition, Pearson Prentice Hall, New
Jersey.
Indian Roads Congress (IRC). (2012). Guidelines for the design of flexible pavements, IRC: 37–2012,
3rd Revision, The Indian Roads Congress, New Delhi, India.
Jacobs, M.M.J., Hopman, P.C., and Molenaar, A.A.A. (1996). Application of fracture mechanics prin-
ciples to analyze cracking in asphalt concrete, Jr. of the Association of Asphalt Paving Technologists,
Vol. 65, pp.1–39.
Kerkhoven, R.E., and Dormon, G.M. (1953). Some consideration of california bearing ratio method
for the design of flexible pavements, Shell Bitumen Monograph No.1.
Kim, H.B., and Buch, N. (2003). Reliability-based pavement design model accounting for inherent vari-
ability of design parameters, Proc. of the 82nd Annual Meeting, Transportation Research Board,
National Research Council, Washington, DC.
Krawinkler, H., and Zohrei, M. (1983). Cumulative damage in steel structures subjected to earthquake
ground motions, Computers and Structures, 16(1–4), pp.531–541.
Kuo, C.M., and Chou, F.J. (2004). Development of 3-D finite element model for flexible pavements, Jr.
of the Chinese Institute of Engineers, 27(5), pp.707–717.
Lacey, G., Thenoux, G., and Rodríguez-Roa, F. (2007). Three-dimensional finite element model for
flexible pavement analysis based on field modulus measurements, The Arabian Jr. for Science and
Engineering, 33, No.1B, pp.65–76.
Lee, H.J., Kim, Y.R., and Lee, S.W. (2003). Prediction of asphalt mix fatigue life with viscoelastic mate-
rial properties, Jr. of the Transportation Research Board, Transportation Research Record No.1832,
pp.139–147.
Lekarp, F., Isacsson, U., and Dawson, A. (2000). State of the art. II: permanent strain response of
unbound aggregates, Jr. of Transportation Engineering, ASCE, 126(1), pp.76–83.
Lu, Y., Yang, S., and Wang, J. (2014). Research on pavement longitudinal crack propagation under non-
uniform vehicle loading, Engineering Failure Analysis, Elsevier, Vol.42, pp.22–31.
Lundström, R. (2002). Rheological and fatigue characterisation of asphalt concrete mixtures using
uniaxial testing, Licentiate Thesis, Research Report TRITA-VT FR 02:02, Kungl Tekniska Högsko-
lan. http://www.sbuf.se/sa/node.asp?node = 133.
Mackiewicz, P. (2013). Fatigue Tests of Bituminous Mixtures with Inclusion of Initial Cracks. Jr. of
Materials in Civil Engineering, ASCE, 25(1), pp.140–147.
Majidzadeh, K., Ilves, J., and May, R.W. (1983). Overlay design of flexible pavements: OAF program,
Jr. of the Transportation Research Board, Transportation Research Record No.930, pp.18–25.
199
Majidzadeh, K., Kuaffmann, E.M., and Ramsamooj D.V. (1971). Application of fracture mechanics in
the analysis of pavement fatigue, Proc. of the Association of Asphalt Paving Technologists, Vol.40,
pp.227–246.
Manson, S.S, and Halford, G. (1981). Practical implementation of the double linear damage rule and
damage curve approach for treating cumulative fatigue damage, International Journal of Fracture,
17(2), pp.169–192.
Manson, S.S., Freche, J.C., and Ensign, S.R. (1967). Application of a double linear damage rule to
cumulative fatigue in fatigue crack propagation, American Society for Testing and Materials, ASTM
STP 415, Philadelphia, PA, pp.384–412.
Matsuno, S., and Nishizawa, T. (1992). Mechanism of longitudinal surface cracking in asphalt pave-
ment, Proc. of the 7th Int. Conf. on Asphalt Pavements, The University of Nottingham, Vol.2,
pp.277–291.
Matthews, J.M., Monismith, C.L., and Craus, J. (1993). Investigation of laboratory fatigue testing proce-
dures for asphalt aggregate mixtures, Jr. of Transportation Engineering, ASCE, 119(4), pp.634–654.
Mollenhauer, K., and Wistuba, M. (2012). Evaluation of hot-mix asphalt susceptibility to temperature-
induced top -down fatigue cracking by means of uniaxial cyclic tensile stress test, Jr. of Road Materi-
als and Pavement Design, 13(1), pp.171–190.
Monismith, C.L. (2004). Evolution of long-lasting asphalt pavement design methodology: a perspec-
tive, Int. Symposium on Design and Construction of Long-Lasting Asphalt Pavements, ISAP. http://
www.asphalt.org/Pubs/Pubset.html.
Myers, L.A., Roque, R., and Ruth, B.E. (1998). Mechanisms of surface initiated longitudinal wheel path
cracks in high-type bituminous pavements, Jr. of the Association of Asphalt Paving Technologists,
Vol.67, pp.401–432.
National Cooperative Highway Research Program (NCHRP). (2004). Mechanistic-empirical design of
new and rehabilitated pavement structures, NCHRP Project 1–37 A, Transportation Research Board,
Washington, D.C.
National Cooperative Highway Research Program (NCHRP). (2006). Independent review of the mech-
anistic-empirical pavement design guide and software, Research Results Digest 307, NCHRP, Wash-
ington, D.C. http://onlinepubs.trb.org/ onlinepubs/ nchrp/nchrp-rrd−307.pdf
Oliveira, J.R.M., Thom, N.H., and Zoorob, S.E. (2008). Design of pavements incorporating grouted
macadams, Jr. of the Transportation Engineering, ASCE, 134(1), pp.7–14.
Pidwerbesky, B.D., Steven, B.D., and Arnold, G. (1997). Subgrade strain criterion for limiting rutting in
asphalt pavements, Proc. of 8th Int. Conf. on Asphalt Pavements, Seattle, Vol.2, pp.1529–1544.
Powell, W.D., Potter, J.F., Mayhew, H.C., and Numm, M.E. (1984). The structural design of bituminous
pavements, TRRL Laboratory Report 1132, UK.
Priest, A.l. (2005). Calibration of fatigue transfer functions for mechanistic-empirical flexible pavement
design. M.Sc. thesis, Civil Engineering Dept., Auburn University, USA.
Pronk, A.C., and Hopman, P.C. (1991). Energy dissipation: The leading factor of fatigue, highway
research: sharing the benefits, Proc. Conf. of the United States Strategic Highway Research Program,
London, Institution of Civil Engineers, pp.255–267.
Prowell, B.D., Hurley, G.C., and Crews, E. (2007). Field performance of warm-mix asphalt at national
centre for asphalt technology test track, Jr. of the Transportation Research Board, Transportation
Research Record No.1998, pp.96–102.
Qiao, Y., Andrew, D., Anders H., and La-Tanttu, L.K. (2015). Calculating rutting of some thin
flexible pavements from repeated load triaxial test data, Int. Jr. of Pavement Engineering, 16(6),
pp.467–476.
Rahman, M.T., Mahmud, K., Ahsan S. (2011). Stress-Strain characteristics of flexible pavement using
finite element analysis, Int. Jr. of Civil and Structural Engineering, 2(1), pp.233–240.
Rajbongshi, P. (2008). A Comprehensive Design Approach for Asphalt Pavements using Mechanistic-
Empirical Framework. Ph.D. Thesis, Civil Engineering Department, IIT Kanpur, India.
Rajbongshi, P., and Das, A. (2009). A systematic approach of field calibration of fatigue equation for
asphalt pavements, Jr. of Road Materials and Pavement Design, 10(1), pp.109–124.
Ramsamooj, D.V. (1991). Prediction of fatigue life of asphalt concrete beams from fracture tests, Jr. of
Testing and Evaluation, 19(3), pp.231–239.
Ramsamooj, D.V., Ramadan, J., and Lin, G.S. (1998). Model prediction of rutting in asphalt concrete,
Jr. of Transportation Engineering, ASCE, 124(5), pp.448–456.
Read, M.J., and Collop, A.C. (1997). Practical Fatigue Characterization of Bituminous Paving Mix-
tures. Jr. of Association of Asphalt Paving Technologists, Vol. 66, pp.74–108.
Richart, J.F.E., and Newmark, N.M. (1948). An hypothesis for the determination of cumulative damage
in fatigue, Proc. American Society for Testing and Materials, ASCE, pp.279–312.
200
Roque, R., Zhang, Z., and Sankar, B. (1999). Determination of crack growth rate parameters of asphalt
mixtures using the superpave IDT, Jr. of the Association of Asphalt Paving Technologists, Vol.68,
pp.404–433.
Roufaiel, M.S.L., and Meyer, C. (1987). Analytical modeling of hysteretic behavior of R/C frames, Jr.
of Structure Division, ASCE, 113(3), pp.429–444.
Saal, R.N.J., and Pell, P.S. (1960). Fatigue of bituminous road mixes, Kolloid-Zeitschrift, 171(1),
pp.61–71.
Saride, S., and Rayabharapu, V., and Vedpathak, S. (2014). Evaluation of rutting behaviour of geocell
reinforced sand subgrades under repeated loading, Indian Geotechnical Journal, pp.1–11.
Selvi, P. (2015). Fatigue and rutting strain analysis on lime stabilized subgrades to develop a pavement
design chart, Transportation Geotechnics, Vol.2, pp.86–98.
Shell. (1978). Shell pavement design manual – asphalt pavement and overlays for road traffic, Shell
International Petroleum Company Limited, London.
Shen, L., and Carpenter, S.H. (2005). Application of dissipated energy concept in fatigue endurance
limit testing, Jr. of the Transportation Research Board, Vol.1929, No.1, pp. 165–173.
Shen, S., Airey, G.D., and Carpenter, S.H. (2006). A dissipated energy approach to fatigue evaluation,
Jr. of Road Materials and Pavement Design, 7(1), pp.47–69.
Shen, S., Chiu, H., and Huang, H. (2010). Characterization of fatigue and healing in asphalt binders, Jr.
of Materials in Civil Engineering, ASCE, 22(9), pp.846–852.
Shen, S., and Carpenter, S.H. (2007). Development of an asphalt fatigue model based on energy prin-
ciples (with discussion and closure), Jr. of the Association of Asphalt Paving Technologists, Vol.76,
pp.525–573.
Shukla, P.K., and Das, A. (2007). A re-visit to the development of fatigue and rutting equations used for
asphalt pavement design, The Int. Jr. of Pavement Engineering, 9(5), pp.335–364.
Simpson, A.L., Daleiden, J.F., and Hadley, W.O. (1995). Rutting analysis from a different perspective, Jr.
of the Transportation Research Board, Transportation Research Record No.1473, pp.9–16.
Strategic Highway Research Program (SHRP). (1994). Fatigue Response of Asphalt-Aggregate Mixes,
Report SHRP-A-404, Washington D.C.
Stuart, K.D., Mogawer, W.S., and Romero, P. (1999). Validation of the superpave asphalt binder and
mixture tests that measure rutting susceptibility using an accelerated loading facility, FHWA-RD-
99–204, Federal Highway Administration, McLean, Va.
Sun, L., Hudson, W.R., and Zhang, Z. (2003). Empirical-mechanistic method based stochastic mod-
eling of fatigue damage to predict flexible pavement cracking for transportation infrastructure man-
agement, Jr. of Engineering Mechanics, ASCE, 129(2), pp.109–117.
Svasdisant, T., Achorsch, M., Bala di, G.Y. and Pinyosunun, S. (2002). Mechanistic analysis of top-
down cracking in asphalt pavements, Jr. of Transportation Research Board, Transportation Research
Record No. 1809, pp.126–136.
Tarefder, R., Saha, N., and Stormont, J. (2010). Evaluation of subgrade strength and pavement designs
for reliability, Jr. of Transportation Engineering, ASCE, 136(4), pp.379–391.
Theyse, H.L., Beer, M., and Rust, F.C. (1996). Overview of the South African mechanistic pavement
design analysis method, Jr. of the Transportation Research Board, Transportation Research Record
No.1539, pp.6–17.
Thompson, M.R. (1987). ILLI-PAVE based full-depth asphalt concrete pavement design procedure, 6th Int.
Conf. of Structure Design of Asphalt Pavements, Vol.1, University of Michigan, Ann Arbor, pp.13–24.
Transport Research Laboratory (TRL). (1993). A guide to the structural design of bitumen-surfaced
roads in tropical and sub-tropical countries, Overseas Road Note 31, 4th Edition, Overseas Center,
TRL, London.
Transport and Road Research Laboratory (TRRL). (1970). A guide to the structural design of pave-
ments for new roads, 3rd Ed., Road Note 29, Transport and Road Research Laboratory (TRRL),
HMSO, London.
Underwood, B.S., and Kim, Y.R. (2013). Nonlinear viscoelastic behavior of asphalt concrete and its
implication for fatigue modeling. Jr. of the Transportation Research Board Transportation Research
Record, No. 2373, pp.100–108.
Uzan, J. (2004). Permanent deformation in flexible pavements, Jr. of Transportation Engineering,
ASCE, 130(1), pp.6–13.
Van Dijk, W., and Visser, W. (1977). The energy approach to fatigue for pavement design, Jr. of the
Association of Asphalt Paving Technologists, Vol.46, pp.1–40.
Verstraeten, J., Vevereka, V., and Francken, L. (1982). Rational and practical design of asphalt pave-
ments to avoid cracking and rutting, Proc. of 5th Int. Conf. on Structural Design of Asphalt Pave-
ments, Delft University of Technology, Netherlands.
201
Wang, L.B., Myers, L.A., Mohammad, L.N., and Fu, Y.R. (2003). Micromechanics study on top-
down cracking, Jr. of Transportation Research Board, Transportation Research Record No.1853,
pp.121–133.
Washington State Department of Transportation (WSDOT). (1995). WSDOT pavement guide - for
design, evaluation and rehabilitation, Pavement Note, Vol.2: Washington State Department of Trans-
portation (WSDOT), Olympia, Washington.
Zhang, J., Sabouri, M., Guddati, M.N., and Kim, Y.R. (2013). Development of a failure criterion for
asphalt mixtures under fatigue loading, Jr. of Road Materials and Pavement Design, 14 (sup. 2),
pp.1–15.
Zhang, Z., Roque, R., Birgisson, B., and Sangpetngam, B. (2001). Identification and verification of a suit-
able crack growth law, Jr. of the Association of Asphalt Paving Technologists, Vol.70, pp.206–241.
Zhou, F., Scullion, T., and Sun, L. (2004). Verification and modeling of three stage permanent deforma-
tion behavior of asphalt mixes, Jr. of Transportation Engineering, ASCE, 130(4), pp.486–494.
Zou, J., Roque, R., and Byron, T. (2012). Effect of HMA ageing and potential healing on top down
cracking using HVS, Jr. of Road Materials and Pavement Design, 13(3), pp.518–533.
202
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Three point bending test is an important laboratory experiment for asphalt
mixture. In this paper, a Finite Element Model (FEM) simulation of this test is done. Bilinear
Cohesive Zone Model (CZM) elements are settled in the mid-span of the beam to simulate
the damage and fracture of the beam. The result of the simulation is in agreement with the
laboratory experiment. Finally, the sensibilities of CZM parameters are studied and the frac-
ture process of the beam is discussed. The result shows that the critical value of traction and
the fracture energy of CZM have a great effect on the FEM simulation, while the effect of
the initial stiffness of the CZM is not obvious. And it is found that, during the process of the
test, the decrease of the bearing capacity of the beam happens after the damage has occurred
at the bottom of mid-span.
1 INTRODUCTION
Cracking is a common phenomenon for asphalt pavement, which would greatly reduce the
pavement performance and cause a great deal of further damages. The simulation of the
fracture of asphalt concrete can improve the understanding of fracture process of the asphalt
pavement. But it has been pointed out that traditional numerical methods have difficulties in
simulating the damage and cracking as a continuous process (Lee et al. 2000, Kim 2011).
Cohesive Zone Model (CZM) is an interface model, which assumes the separation hap-
pens at that interface. The constitutive model of CZM is defined by the relationship between
the stresses acting on the interface and the separation of the interface. CZM can be used in
simulate the gradual cracking process in which the separation of the crack is constrained
by cohesive stress. CZM was put forward by the pioneering papers by Dugdale (1960) and
Barenblatt (1962), after that the finite element implementation of CZM was presented by
Hilleborg et al. (1976).
Recently, CZM has been used in the fracture analysis of the asphalt concrete. Espinosa &
Zavattieri (2003) adopted bilinear cohesive zone models to reduce the compliance by provid-
ing an adjustable initial slope in the cohesive law. Song et al. (2006) investigated the fracture
behavior of asphalt concrete based on bilinear CZM and the viscoelasticity of asphalt was
also considered. Kim et al. (2008) combined CZM with discrete element method and simu-
lated the fracture of asphalt concrete. Jin et al. (2013) integrated a fatigue damage evolution
model with the CZM and an indirect tensile fatigue test was simulated.
Three point bending test is an essential test to get the mechanical properties of asphalt
mixture, such as bending tensile strength and modulus. However, the results are inaccurate
if they are calculated using elastic theory or linear elastic fracture mechanics. So the damage
and fracture process of the three point bending test should be analyzed.
The object of this study is to simulate the damage and fracture process of asphalt con-
crete beam in three-point bending test based on bilinear CZM. Firstly, the result of FEM
is compared with the experimental result. Then, the influence of the parameters in CZM
is studied. Finally, the influence of fracture process to the normal stress in mid-span is
discussed.
203
2 BILINEAR COHESIVE ZONE MODEL
Espinosa & Zavattieri (2003) formulated a bilinear model to reduce CZM compliance by
providing an adjustable initial slope in the cohesive law. Figure 1 shows the bilinear cohesive
law in terms of opening traction and opening displacement. The pre-peak region represents
the elastic part, whereas the softening portion after the peak represents the damage occurring
in the fracture process.
⎛ tn ⎞ ⎡ K nn 0 ⎤⎛ εn ⎞ 1 ⎡ K nn 0 ⎤ ⎛δ n ⎞
t=⎜ ⎟ =⎢ = (1)
⎝ ts ⎠ ⎣ 0 K ss ⎦ ⎝ ε s ⎠ H 0 ⎢⎣ 0
⎥ ⎜ ⎟ K ss ⎥⎦ ⎜⎝ δ s ⎟⎠
Where the traction stress vector, t, consists of the following components: tn and ts which are
normal and shear tractions, respectively. δ n and δ s denote a normal displacement opening
and shear sliding. K nn and K ss are initial stiffness components of CZM. H 0 is the thickness
of the cohesive zone.
⎧ t t ⎫
max ⎨ nc , sc ⎬ = 1 (2)
⎩ tn ts ⎭
Where, tnc and tsc represent the critical values of the normal and shear traction. The sym-
bol ◊ represents the Macaulay bracket, which is used to signify that a purely compressive
stress state does not initiate damage.
204
damage level. It initially has the value 0. The value of D evolves from 0 to 1 after the damage
initiation. The normal and shear stress components are affected by D, which is showed in
Equation 3.
⎛ tn ⎞ ⎡ K nn 0 ⎤⎛ εn ⎞
t = ⎜ ⎟ = ( − D) ⎢ (3)
⎝ ts ⎠ ⎣ 0 K ss ⎥⎦ ⎜⎝ ε s ⎟⎠
To describe the evolution of damage under a combination of normal and shear separa-
tions across the interface, an effective displacement is defined as Equation 4.
δm ( δ n > +δ s ) (4)
For bilinear CZM, the damage variable for the linear softening can be represented as
Equation 5 (Dávila et al. 2001).
δ mf ( − c
)
D= m m
(5)
δm ( f
m − c
m )
Where, δ mf represents the effective displacement at failure, δ mc represents the effective dis-
placement at the initiation of the damage.
For bilinear CZM, the fracture energy of mode 1 is simply the area under the curve in
Figure 1, so it can be calculated by Equation 6.
1 f c
GC δ n tn (6)
2
Figure 2. Three point bending test simulation: (a) geometry, mesh and boundary condition (Units: N,
mm); (b) Mises stress contour plot after the beam was fractured.
205
the bending and fracture. The asphalt mixture used here can be categorized as Shell AH-70.
The aggregate gradation is listed in Table 1.
The bending and fracture process of 5 beams were measured. The results are listed in the
Table 2. The experimental data about the relation between mid-span vertical displacement
and the reaction force can also be seen in Figure 3.
Assuming the normal stress and strain at the bottom of mid-span get the peak value when
reaction force equals to Fmax , the maximum normal stress and strain and elastic modulus,
σ max , ε max E f can be calculated based on material mechanic as Equation 7–9 (Gere &
Goodno 2009).
3Fmax L
σ max = = 9.4541MPa (7)
2bd 2
6 Dd
ε max = = 2.092 × 10 −3 (8)
L2
σ max L3F
Ef = = = 4519.411MPa (9)
ε max 4bd 3D
Sieve size/mm 16 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075
Case 1 2 3 4 5 Average
* The maximum reaction Load. ** The deflection of the center of the beam.
206
In these formulas the following parameters are used.
− σmax = The maximum stress at the bottom of mid-span, (MPa)
− εmax = Strain corresponding to σmax at the bottom of mid-span, (mm/mm)
− Εf = Flexural Modulus of elasticity, (MPa)
− Fmax = The maximum reaction Load, (N)
− L = Support span, (mm)
− b = Width of the beam, (mm)
− d = Depth of the beam, (mm)
− D = Deflection of the center of the beam, (mm)
The value of σmax and Εf will be used in CZM for FEM simulation, in Section 4.2.1.
4 DISCUSSION
207
The initial value of tnc can be chosen as the maximum normal stress σ max = 9.4541MPa
calculated in the Equation 7 The initial value of Gc can be chosen as G 0c . J / m2
according to the research of Zhang et al. (2010).
The cohesive element is designed to simulate the fracture progress of an interface. So the
initial thickness of cohesive element is often chosen as zero. Theoretically, the stiffness K nn
should be infinitive large, so that the cohesive elements will not change the overall stiffness
of the structure, otherwise the overall stiffness of the structure will be reduced by cohesive
elements. However, K nn cannot be too large, otherwise it will cause the convergence problem
in FEM computation. According to the research of Turon et al. (2007), take the K nn to be
around 50 times of the elastic modulus of the material in the rest area will be large enough.
So the initial value of K nn can be chosen as K nn0 3 105 MMPa .
In concussion, the initial value of the parameters are tn σ max 9.4541M MPa ,
Gc Gc0 = 206.4 J / m 2 , K nn K nn0 = 3 × 105 MPa . Based on these initial values of CZM
parameters, the sensibility and the optimum value of these parameters are discussed in the
following section.
208
Figure 4. The influence of CZM parameters to FEM simulation: (a) the influence of tnc to force-
deflection curve; (b) the influence of Gc to force-deflection curve; (c) the influence of K nn to force-
deflection curve.
209
Figure 5. Comparison of changing process of normal tensile stress and reaction force.
5 CONCLUSION
REFERENCES
Barenblatt, G.I. 1962. The mathematical theory of equilibrium cracks in brittle fracture. Advances in
Applied Mechanics 7:55–129.
210
Dávila, C. G. 2001. Mixed-Mode Decohesion Elements for Analyses of Progressive Delamination. Pro-
ceedings of the 42nd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials
Conference
Dugdale, D.S. 1960. Yielding of steel sheets containing slits. Journal of the Mechanics and Physics of
Solids 8:100–104.
Espinosa, H. D. & Zavattieri, P. D. 2003. Agrain level model for the study of failure initiation and evolu-
tion in polycrystalline brittle materials. Part I: Theory and numerical implementation. Mechanics of
Materials 35: 333–364
Gere J.M. & Goodno B.J. 2009. Mechanics of materials. Stamford: Cengage Learning.
Hilleborg, A., Modéer, M. & Petersson, P.E. 1976. Analysis of crack formation and crack growth in con-
crete by means of fracture mechanics and finite elements. Cement and Concrete Research 6:773–782.
Jin, G., Huang, X., Zhang, S. & Liang, Y. 2013. Numerical study of fatigue damage of asphalt concrete
using cohesive zone model. JOURNAL OF SOUTHEAST UNIVERSITY(English Edition) 29(4):
431–435.
Kim, H., Wagoner, M.P. & Buttlar, W.G. 2008. Simulation of Fracture Behavior in Asphalt Concrete
Using a Heterogeneous Cohesive Zone Discrete Element Model. Journal of Materials in Civil Engi-
neering 20(8): 552–563.
Kim, Y.-R. 2011. Cohesive zone model to predict fracture in bituminous materials and asphaltic pave-
ments: state-of-the-art review. International Journal of Pavement Engineering 12(4): 343–356.
Lee, B.H., Danile, J.S. & Kim, Y.R. 2000. CONTINUUM DAMAGE MECHANICS-BASED
FATIGUE MODEL OF ASPHALT CONCRETE. JOURNAL OF MATERIALS IN CIVIL ENGI-
NEERING 12(2): 105–112.
Song, S. H., Paulino, G. H., & Buttlar, W. G. 2006. A bilinear cohesive zone model tailored for fracture
of asphalt concrete considering viscoelastic bulk material. Engineering Fracture Mechanics 73(18):
2829–2848.
Turon, A., Dávila, C. G., Camanho, P. P. & Costa, J. 2007. An engineering solution for mesh size
effects in the simulation of delamination using cohesive zone models. Engineering Fracture Mechan-
ics 74(10): 1665–1682.
Zhang, D., Huang, X. & Zhao, Y. 2010. Simulation of indirect tension test of asphalt mixtures based
on cohesive zone model. JOURNAL OF SOUTHEAST UNIVERSITY(Natural Science Edition)
40(6): 1276–1281.
211
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
J. Voskuilen
Rijkswaterstaat, Utrecht, The Netherlands
B. Gaarkeuken
BAM Infra Asfalt, Utrecht, The Netherlands
D. van Vliet
TNO, Delft, The Netherlands
M. Poot
Delft University of Technology, Delft, The Netherlands
ABSTRACT: The KGO mixing method was developed 40 years ago. Instead of adding
the bitumen at the end of the mixing cycle, the bitumen was poured down first together with
the larger aggregates. After mixing the bitumen with the larger aggregates, the reamining
aggregates were mixed down. As benefits of the KGO method were mentioned improving
the workability and shortening the needed compaction time. In this study it was investigated
if the KGO method and a longer mixing time could improve the mechanical properties of
asphalt mixtures. To investigate the influence on workability and compactability the indirect
tensile strength was determined of gyratory compacted specimens produced with the KGO
method and different mixing times. Form the results it can be concluded that a longer mixing
time didn’t have an effect om both. A modified mixing sequence has a positive influence on
the workability and the indirect tensile strength.
1 INTRODUCTION
1.1 General
In the early days several mix design methods have been developed starting from trial and error.
The developments were based on some principle thoughts on materials and volumes which
are at present not that obvious anymore. It is already long time known that the mechanical
properties can be influenced by the mix composition (grading and bitumen content) and the
choice of raw materials, but also the mixing and compaction process are important. A well
designed asphalt mixture with excellent properties, can perform poor due to failures during
the mixing or laying process. It is important that the mixing and compaction of asphalt
mixtures during the mix design mimics the mixing and compaction process in the field.
If they don’t, unexpected premature failures can be the result of this. If you look at the mixing
process, a shorter or longer mixing time, but also a higher or lower mixing temperature have an
influence on the mechanical asphalt properties. With Marshall lab tests it was determined that
the longer the mixing time, the better the Marshall stability and lower the air voids content. It
is amazing that asphalt mix plants can mix 4000 kg asphalt within about 35 seconds.
And too low mixing temperature results in poor coating of the aggregate particles, because
the aimed equiviscous temperature (EVT) temperature is not reached. At the EVT temperature
the bitumen has the best viscosity (170 mm2/s) to get the optimal mixing of bitumen with min-
eral aggregate. A too high mixing temperature will cause early ageing and dripping of bitumen.
213
But also the mixing sequence of the raw materials can influence the mechanical properties.
In this paper a limited study will be discussed about how compactability and indirect tensile
strength can be influenced by the mixing sequence and/or mixing time of raw materials.
60 seconds X – –
120 seconds X (= reference) X X
180 seconds X X X
The difference between the reference and the KGO-mixing methods is that with the KGO
methods first the coarse aggregate is coated with bitumen, and then the filler and later the fine
aggregate is added. The KGO II and III mixing method are specifically designed for asphalt
mix plants. In a laboratory mixer, there is a great risk of bitumen drain down if the total
amount of bitumen is mixed with the coarse aggregate in one time. To avoid this problem a
preliminary investigation was carried out in this study to determine how much bitumen can be
added to coat the aggregate without drainage of the bitumen. The result was that in the first
step 20 mass% of the total bitumen could be added without dripping problems. The remaining
80% of the bitumen is added after the filler addition. So, when is this study KGO II or III is
discussed, the authors actually mean a modified KGO II or KGO III method, because in both
cases first 20 mass% of the bitumen is added and later on the remaining 80 mass%.
Also the mixing time was varied in this limited study. As a reference the BAM mixing
procedure is chosen with a mixing time of 120 seconds. In Table 1 an overview is given of the
research program.
2 EXPERIMENTAL
2.1 Materials
For this study an AC was chosen that was used for binder and base layers, which is a standard
mixture of BAM. In this mixture no Reclaimed Asphalt (RA) is used. The mix composi-
tion of this AC 11 bin/base 40/60 mixture is given in Table 2. Bestone is a sandstone quarry
material from Norway. Wigro 40K is a fabric filler containing 10 to 35% Ca(CO)3, 50 to 85%
secondary materials and 5 to 15% Ca(OH)2. The binder is a pen grade 40/60 bitumen.
215
Table 2. Mix composition of AC 11 bin/base 40/60.
3 RESULTS
3.1 Compactability
To examine the influence of the mixing sequence and mixing time on the compactability the
relation between the number gyrations and density was investigated, see Figure 1.
The relationship of gyrations versus specimen density is very well accessible by the follow-
ing formula:
b
ρo .w . = a −
gyr + c
ρo.w. : Density of sawn gyratory specimens determined by weighing under and above
water [kg/m3]
gyr : Number of gyrations [−]
a, b, c : Regression constants
The constant a is a horizontal asymptote and is the specimen density which is achieved
after an infinite number of gyrations.
216
Table 3. Regression constants, densities and gyrations.
Average
Average number Average Average Degree of
Mixing density1 gyrations2 density3 number compaction
method A b c [kg/m3] [−] [kg/m3] gyrations4 [%]
Table 4. Average density and indirect tensile strength per mixing method.
According to the Dutch national standard (CROW 2015) the target density is achieved
when over the last 10 gyrations a maximum of 0.2% decrease in height ( = increase in density)
is recorded. The target density can be calculated from the formula. In Table 3 the results are
presented in another way.
1. Average realized density of the sawn gyratory specimens (goal reference target density of
2385kg/m3)
2. Average number of gyrations to achieve target density
3. Calculated target density in conformity with (CROW 2015)
4. Average number of gyrations to achieve calculated target density in conformity with [3]
4 DISCUSSION RESULTS
4.1 Compactability
From the results in Table 3 the following should be noted:
1. The mixing time has a small influence on the needed compaction energy to achieve the tar-
get density. Longer mixing times reduce the compaction energy to a small extent (except
for KGO III)
2. The mixing methods KGO II-120, KGO II 180 and KGOIII 180 result in a higher target
density. For KGO II the target density is reached after the same amount of gyrations as
217
Figure 2. Relation between density and indirect tensile strength.
the reference (BAM 120) method with the same gyrations. The mixing method KGO III
180 results in a higher target density, a little bit more compaction energy is necessary.
Generally it can be seen that compaction of AC11 bin/base 40/60 mixtures, of which the
coarse aggregate is pre-coated with bitumen, results in higher target densities with the same
compaction energy. The relationship between degree of compaction and compaction energy
(number of gyrations) hardly changes with the mixing method.
4 CONCLUSIONS
On the basis of this study with an AC mixture and determination of the effect of the mixing
time and mixing sequence with the indirect tensile strength test the following conclusions can
be drawn:
1. All the test specimens have been prepared with the same target density. The difference in
density between the test specimens with various methods of mixing and mixing times are
limited.
2. The average ITS-values for the different methods of mixing and mixing times do not differ
significantly and are comparable. The ITS-values are found to be dependent on the den-
sity of the specimens. If results of specimens with the same density are compared, mixing
method and mixing time don’t have any effect.
3. Using the KGOII method shows that the specimen density becomes higher, when the same
compaction energy is applied. Compaction energy for KGOIII is comparable with the
BAM mixing method.
4. Mixing methods KGOII and KGOIII result in a more compactable asphalt concrete with
higher target densities than the BAM method. The course of the degree of compaction
versus gyrations is for all mixing methods and mixing times comparable. The KGOII mix-
ing method results with the same number of gyrations in a higher density in comparison
with the KGOIII and the standard BAM mixing method.
5. As a consequence that both KGO mixing methods result is higher target densities, also the
ITS results are better than when the standard mixing method is used.
6. A longer mixing time with both KGO mixing methods has a positive effect on the ITS-
values.
219
7. By applying a KGOII-mixing method, the ITS-value for the same compaction energy is about
10% higher than when the reference BAM mixing method is used. These higher ITS-values
are also reachable for KGOIII but with slightly higher compaction energy than KGOII.
Pre-coating of coarse aggregate with bitumen takes care of a smoother blend which results
in an easier compaction. As a result of this a higher density is achieved with the same com-
pacting energy. With the same compaction energy level pre-coating of coarse aggregate with
bitumen leads to a higher indirect tensile strength.
5 RECOMMENDATIONS
The conclusions are based on a limited research work on only one AC mixture. It is recom-
mended to carry out a more extensive research work on different AC mixtures to investigate
if the shown conclusions are general for AC mixtures.
It is recommended to investigate if the same conclusions can be drawn if other functional
tests are carried out instead of only the indirect tensile test.
It is recommended to carry out similar research work on gap-graded mixtures as Porous
Asphalt and Stone Mastic Asphalt. Find out what will give the most benefits:
− first mix the coarse aggregate with 20 mass% of the bitumen and add the fine aggregate
and filler later on and mix at last the remaining 80 mass% of the bitumen
− because in gap-graded mixtures actually the mortar is the real binder of the stone skeleton,
it can be considered to mix first the mortar completely followed by mixing the mortar with
the coarse aggregate.
Because the first step of mixing due to relatively high amount of bitumen against the spe-
cific surface area of the coarse aggregate, there is a risk of the bitumen draining down. To
avoid this problem in this research only 20 mass% of the total bitumen was added in the first
step and the remaining 80% mass% of the bitumen was added after the filler addition. Maybe
the same happens during production of AC mixtures in an asphalt mix plant with the KGO
methods. It is recommended to determine if bitumen partly drains down if you add the total
amount of bitumen in one time If so, consider to add at first sufficient bitumen to coat the
coarse aggregate and add the remaining bitumen in the last phase.
It is recommended to find out if it is possible to reduce the production temperature if the
mixing sequence is changed. Although the total specific surface area of the total aggregate
for both, the conventional and the KGO mixing method, is the same, less adhesive forces are
needed to overcome during the mixing when addition of final fractions take place at a later
stage. Due to this maybe a lower production temperature is possible with the same mixing
energy. If this is true, the alternative mixing sequence could help du reduce energy and CO2
without changing the mixture composition or adding additives.
REFERENCES
Andersson P. 2008. Technical Description of Flow-Mixed Asphalt – KGO III, the National Road
Administration, Sweden.
CROW. 2015. Dutch National Standard, RAW Bepalingen.
Hesami E, Birgisson B and Kringos N. 2015. Effect of Mixing Sequence on the Workability and Per-
formance of Asphalt Mixtures, AAPT 2015, Portland USA.
Oveson L. 2012. http://www.kgo.se/press/
Voskuilen J, Ven van de M. 2006. De invloed van proefstukvervaardiging op mechanische gedrag van
asfalt (ruimtelijke benadering), CROW Wegbouwkundige Werkdagen 2006.
220
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The present investigation is centered around flexible pavement design taking
into consideration fatigue cracking as the performance criterion according to the Mechanis-
tic-Empirical Pavement Design Guide (MEPDG). Bottom up fatigue cracking is defined as
the maximum area of alligator cracking that is permitted to occur over the design period.
HMA dynamic modulus (E*) is used to predict pavement response parameters and it is a key
input for the prediction of fatigue. On these grounds, the impact of the E* of HMA mixes,
that differentiate with respect to the asphalt binder, on the predicted fatigue cracking is inves-
tigated. For this purpose three different asphalt binders were examined, one pure and two
modified. A flexible pavement cross section was considered with varying asphalt mix layer
thickness in order for the pavement design process to be carried out. Results are presented
and discussed in the paper.
In flexible pavement design process the two strains that are believed to be critical for design
purposes are the horizontal tensile strain at the bottom of the asphalt base layer and the ver-
tical compressive strain at the top of the subgrade layer. In case of excessive vertical strain,
permanent deformation occurs on the surface of the pavement structure from subgrade over-
loading and the pavement is distressed due to rutting. Excessive horizontal tensile strain
will result in the occurrence of cracking on the surface layer and pavement distress due to
fatigue.
The fatigue resistance of asphalt mixes is generally defined as their ability to respond to
repeated traffic loading under the prevailing environmental conditions without the devel-
opment of significant cracking or premature failure. One of the main objectives of the
mechanistic approach to the design of asphalt pavements is, therefore, to limit the maximum
horizontal tensile strain and hence fatigue cracking in the asphalt layers.
A number of fatigue performance models have been developed to predict the number of
repetitions allowed before failure occurs. The development of a fatigue model is generally
defined by considering strain level, temperature, loading frequency and modulus of the hot
mix asphalt (HMA) mixture (Carpenter 2006).
A major aspect of the new mechanistic-empirical pavement design (MEPDG) (AASHTO
2008) is that beyond the usage of a recommended fatigue model to predict the number of
load repetitions for fatigue cracking, a performance criterion is also developed and consid-
ered within the pavement design process. Bottom up fatigue cracking is considered as a per-
formance criterion and defined as the maximum area of alligator cracking expressed as a
percentage of the total lane area that is permitted to occur over the design period.
The key input for the prediction of fatigue and rutting damage in MEPDG is the dynamic
modulus (E*) (Plati et al. 2013). The dynamic modulus is used to predict pavement response
parameters that determine the strains and displacements of layered pavement structures
under different temperatures and loading conditions (Birgisson 2005, Flintsch et al. 2008).
221
E* is a fundamental property capable of elucidating the viscoelastic property of the
asphalt mix under consideration (Witczak 2001, Witczak 2002, Bari and Witczak 2006). The
MEPDG calls for preparing an E* master curve of the HMA based on the E* reported
at different temperatures and frequencies (Biligiri and Way 2014). Mathematically, dynamic
modulus is defined as the absolute value of the complex modulus. The complex modulus
consists of two parts: the real part which represents the elastic stiffness and the imaginary
part representing the internal damping of materials (Robbins 2009). The dynamic modulus
is determined from the maximum applied stress and peak recoverable axial strain. The phase
angle describes the lag in the induced axial strain relative to the applied compressive stress
and illustrates the time dependency of hot mix asphalt (HMA). E* can be determined upon
laboratory testing, but it can also be estimated through prediction algorithms.In the frame of
the present research study the E* determined in the lab is considered.
In light of the above, the objective of the present research study is to investigate the impact
of the E* of various asphalts (one pure and two modified) on the long term asphalt mix
behavior, with respect to the estimated percentage of fatigue cracking. For this purpose flex-
ible pavement cross sections with varying asphalt layer thickness and three different types of
asphalt are considered. In the framework of the pavement design process the percentage of
the total lane area with alligator cracking is estimated. Results on the sensitivity analysis are
presented and discussed in the following.
2 EXPERIMENTAL STUDY
To meet the objectives of the present study, a three layer pavement structure (asphalt layer,
base layer of unbound material and subgrade) typically implemented on motorways of the
national transportation network was selected (Figure 1).
The subgrade and the crushed gravel base modulus based on in situ testing modulus are con-
sidered to be equal to 150 MPa and 350 MPa, respectively. The crushed gravel base layer thick-
ness is equal to 30 cm. For the asphalt layer three mixtures are considered in the design process
with different binders, one pure asphalt 30–45 PEN and two asphalts 50–70PEN modified with
a 8% and 12% selenizza additive. Selenizza is an additive used among others, for manufacturing
high performance bituminous materials. It’s physical properties are given in Table 1.
The purpose of producing the modified binders was twofold. On the one hand, an effort
was made to improve the characteristics and the performance of the pure 50–70 PEN asphalt
with the addition of a modifier. On the other hand, the amount of the selenizza additive was
investigated in order to achieve the same mechanical characteristics and performance with
the pure 30–45 PEN asphalt.
222
The thickness of the Hot Mix Asphalt (HMA) layer ranges from 12 to 19 cm. Table 2
presents the gradation of the HMA mixtures considered, while Table 3 shows the binder
characteristics (penetration index-PEN and softening point) and volumetric properties (air
void percentage-Va and effective binder content by volume- Vbeff) of the HMA mixtures
investigated.
Four specimens from each mixture were compacted with the gyratory compactor accord-
ing to EN standards (EN 12697–31 2007). Initial dimensions of the specimens were 170 mm
high and 150 mm diameter in accordance with related standards. The specimens were then
internally cored and cut in order to achieve 150 mm highand 100 mm diameter according to
related standards.
E* testing was performed at the NTUA Laboratory of Pavement Engineering according
to AASHTO standards (T 342–11). A controlled sinusoidal (harvesine) compressive loading
was applied to each specimen for a range of loading frequencies and temperatures. Testing
was conducted at six temperatures 4, 15, 20, 25, 37°C and six frequencies, 25, 10, 5, 1, 0.5 and
0.1 Hz. The mean value of the E* of the four replicates is considered to be the representative
modulus for each mixture.
The viscoelastic analysis of the investigated pavement sections was completed utilizing the
3-D Move software program. The finite-layer approach three-dimensional moving load anal-
ysis (3-D Move) was developed by the University of Nevada, Reno (UNR) (Siddhartan et al.
2000) and treats each pavement layer as a continuum and uses the Fourier transform tech-
nique. The finite layer method is much more computationally efficient than the moving load
models based on the finite element method (Huhtala and Pohlajamaki 1992). In addition,
rate-dependent material properties (viscoelastic) can be accommodated by the approach.
Passing (%)
19 95 93 95
9.5 66 65 70
4.75 40 38 46
2.36 25 24 26
0.3 8.1 8.98.4
0.075 4.7 5 4.5
PEN 32 36 32
Softening point 56 54 56
Vα (%) 3.3 3.6 3.1
Vbeff (%) 11.6 11.3 11.5
223
The applied load considered was a standard axle load (18-kip dual wheel). All the simula-
tions were performed using the typical 18-kip equivalent single-axle load (ESAL) approach.
The number of traffic repetitions used in the analysis was calculated at 2.5 × 107 ESALs for
20 years. The dynamic analysis was performed considering a moving vehicle with 60 km/h.
For the fatigue analysis, the approach was to calculate the allowable number of axle load
applications needed for the incremental damage index according to NCHRP 1–37 A and it is
presented in 3D MOVE as shown in Eq. 1 (Baus and Stires 2010).
where Nf-HMA: allowable number of axle-load applications for a flexible pavement, εt: tensile
strain at critical locations calculated by the structural response model, inches/inches, E*HMA:
dynamic modulus of the HMA, psi, kf1, kf2, kf3: global field calibration parameters (from
the NHCRP 1–37 A calibration; kf1 = 1, kf2 = −3.9492 and kf3 = −1.281), βf1, βf2, βf3: local or
mixture specific field calibration constants; for the global calibration effort, these constants
were set to 1.0, C = 10M, where
⎛ Vbeff ⎞
M = 4.84 ⎜ − 0.69⎟ (2)
⎝ a
V V beff
f ⎠
where Vbeff: effective asphalt content by volume,%, Vα: air void percent in the HMA mixture,
CH: thickness correction term
1
CH = (3)
0.003602
0.000398 +
1+ exp(11.02 − 3.49 H HM
MA )
⎛ n ⎞
DI = ∑( DI ) = ∑ ⎜ ⎟ (4)
⎝ f HMA ⎠
N
Once the cumulative damage index was predicted, transfer functions were used to convert
data into alligator cracking using Eq.5.
⎛ C3 ⎞
Cbottom = 0.18 ⎜
FC (5)
⎝ 1 + e p (C C + C C DI × ) ⎟⎠
Where FCboFigttom: area of alligator cracking that initiates at the bottom of the HMA lay-
ers,% of total lane area, DI: cumulative damage index at the bottom of the HMA layer,
C1, 2, 3: transfer function regression constants; C3 = 6,000; C1 = 1.00; C2 = 1.00, C1* = −2C2*,
C2* = −2.40874–39.748(1+(39.370HHMA )−2.856, where HHMA: total HMA thickness, in.
Constant values of the fatigue law and alligator cracking equations are considered equal to
those of the global field calibration, since the aim of the present paper is to run a sensitivity
analysis of the various mixtures on the performance indicator (alligator cracking) and not for
pavement design purposes.
For the purpose of the present research a reliability level (R) of 95% was considered
according to the recommendations by (AASHTO 2008) for primary roads.
224
3 DATA ANALYSIS AND RESULTS
Figure 2 presents the E* master curves for the three mixtures at a reference temperature of
25°C, which is the reference temperature for the pavement design. From a first glance, E*
values for the mixture with 12% selenizza do not seem to differ significantly from those of
the 30–45 PEN mixture, while the HMA mixture with 8% selenizza exhibit lower values of
E* when compared to the above mentioned mixtures.
In order to quantify the differences observed, the percentage difference of the E* was
calculated considering E* values of HMA mixes with 30–44 PEN asphalt as reference. Fig-
ure 3 shows the percent error in the form of box plots. The box itself contains the middle 50%
of the data. The upper edge (hinge) of the box indicates the 75th percentile of the data set
and the lower hinge indicates the 25th percentile. The range of the middle two quartiles is the
inter-quartile range. The line in the box indicates the median value of the data. If the median
line within the box is not equidistant from the hinges, then the data is skewed. The ends of the
vertical lines or whiskers indicate the minimum and maximum data values, unless outliers are
present in which case the whiskers extend to a maximum of 1.5 times the inter-quartile range.
The points outside the ends of the whiskers are outliers or suspected outliers.
The maximum difference between E* values corresponding to HMA mix with 8% selenizza
and those of HMA mix with 30–45 PEN asphalt is around 32% while the minimum difference
is around 10%. The value of the median is 20%. Therefore, E* values of HMA mixes with 8%
selenizza are lower by 20% with respect to those of HMA mixes with 30–45 PEN asphalt.
On the other hand, the percent difference between E* values of HMA mixes with 12%
selenizza and 30–45 PEN asphalt ranges from around −21 to 10%. Median value is −4%,
indicating that in general E* values of HMA mix with 12% selenizza are higher but close
enough to the ones of HMA mixes with 30–45 PEN asphalt. Figure 4 shows the developed
strains at the bottom of the asphalt layer and at the centre of the load for each HMA layer
thickness and asphalt binder.
As expected, strains reduce with the increase of the HMA layer thickness in every case.
Strains induced in case of the HMA mix with 30–45 PEN are lower while in case of the
HMA with 8% selenizza are higher. Figure 5 shows the absolute percentage differences of the
strains for the various HMA mixes with respect to the one with 30–45 PEN asphalt.
It is observed that the percentage difference of the strains reduces with the increase in
HMA layer thickness in every case. Strains developed at the bottom of the HMA layer with
12% selenizza are closer to those developed at the bottom of the HMA layer with 30–45 PEN
asphalt, comparatively with the HMA layer with 8% selenizza. This is agreement with the
trend observed at the E* values.
225
Figure 3. E* percent difference.
Figure 4. Strains at the bottom of the HMA layer and the centre of the load.
226
Figure 5. Percentage difference of strains at the bottom of the HMA layer.
Figure 6 shows the estimated fatigue cracking percent (FC) for each mixture versus the
various HMA layer thicknesses considered. Regression analysis was performed and it was
found that the relation between FC and HMA thickness is best described by logarithmic
based equations shown in Figure 4. Values of regression coefficients in case of 30–45 PEN
and 12% selenizza HMA mixtures are similar, as expected, due to the small differences
observed in the estimated FC.
The Root Mean Square Error (RMSE) was calculated in order to determine the differences
of the estimated FC resulting from the different HMA mixtures. The RMSE value (expressed
in percentage) of FC between 30–45 PEN and 8% selenizza HMA mixtures is equal to 60.8%,
while between 30–45 PEN and 12% selenizza HMA mixtures 5.6%.
Considering the above it can be concluded that HMA mixture with 12% selenizza per-
forms, in terms of fatigue cracking, similar to the 30–45 PEN HMA mixture.
4 CONCLUSIONS
The aim of the present research study was to investigate the impact of the E* of various
asphalts on the long term asphalt mix behavior, with respect to the estimated percentage of
fatigue cracking at the end of the design life. The binders considered were pure asphalt 30–45
PEN and two modified with 8% and 12% selenizza.
227
Analysis results showed that HMA mixtures with 30–45 PEN asphalt and those with 12%
selenizza appear to have small differences as far as dynamic modulus is concerned. Moreover, the
above mentioned mixtures exhibit similar performance with respect to the estimated percentage
of fatigue cracking at the end of the design life. On the other hand, E* values of mixtures with
8% modified asphalt with selenizza are lower and percentage of fatigue cracking is higher.
Strain analysis showed that with the increase in the HMA layer thickness, the percent dif-
ferences with respect to the HMA mix with 30–45 PEN decrease. However, it is always the
case that stains developed at the bottom of the HMA layer with 12% selenizza are closer to
those developed at the bottom of the HMA layer with asphalt 30–45 PEN. This can be attrib-
uted to the fact that binder properties of 30–45 PEN and 12% selenizza are similar.
Overall, sensitivity analysis showed that mixes with different asphalt binders can achieve
the same E* values and pavements are expected to have similar performance. On the other
hand, differences in the E* of HMA mixes account for an unalike behavior in terms of stain
analysis and fatigue cracking prediction. Aggregate gradation of the mixtures did not vary
significantly and therefore mixture performance is related to the binder type.
The above are considered to be a supporting tool in the frame of pavement design and rehabili-
tation strategies which can be used as part of a cost effective analysis process. It is also noted that
the performance model of fatigue cracking considered in the present study is not calibrated to local
conditions and as such it is not advised to be used for design purposes. In the frame of this research
study it is used for comparison purposes in order for the sensitivity analysis to be performed.
The present research is ongoing considering the E* prediction algorithms as well.
REFERENCES
228
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
A road pavement ability to maintain the functional characteristics during the expected use-
ful lives represents one of the most delicate design aspects. In particular, the continuous
increasing of heavy traffic determines a growing in superstructure stress state and, conse-
quently, a higher performance request (Agostinacchio et al. 2014). However, it is important
to also consider the possibility of exceptional stresses such as a seismic event (Coni et al.
2000).
The possible local amplification conditions derived from seismic excitation must be
accounted for in order to identify and correctly design more suitable solutions for its mitiga-
tion. Generally, the maximum effects in terms of local amplification are due to resonance
phenomena due to the different soil nature located above the bedrock. It is possible to iden-
tify the areas potentially affected from amplified seismic phenomena mainly with monitoring
techniques in situ and historical documentation and testimonies regarding past events.
An European research project entitled: “REinforcement of FLEXible road structures with
steel fabrics to prolong service life” (REFLEX 2002) has been developed focussing on the
behaviour characterization of road pavements reinforced with welded steel mesh. Specialized
European laboratories have been also involved in order to reproduce this behaviour through
modelling techniques with a finite element approach. The achieved results have confirmed
229
the benefits of the presence in road pavements in order to counteract the different patholo-
gies typical of road superstructures.
The REFLEX project also provided the optimization of the flexible road pavement response
as a function of various pathologies by determining the optimal position of the steel mesh
corresponding to different depths from road surface. However, the study of seismic influence
on road behaviour represented a limit of the REFLEX project. This paper will provide a
contribution on the behaviour of this type of reinforced pavements especially focussing on
the correlation between soil nature-seismic action-steel mesh reinforcing interaction. In this
paper the ANSYS® FE software will be used in order to evaluate the stress/strain analysis of
road pavements reinforced with steel mesh and to minimize the negative effects both induced
by soil nature and the normal vehicular traffic.
In critical conditions, the reinforcement of flexible asphalt concrete for road superstructures
requires the use of different steel mesh configuration (Fig. 1). The steel yield strength must be
necessarily higher in applications where thermal cracks are due by the frost and in preventing
the reflected cracks (La Agostinacchio et al. 2008).
In such applications, it is also important to use a smooth steel mesh because the interaction
between asphalt concrete and steel is guaranteed by mesh and then from the welding points.
In order to counteract the permanent deformation (rutting) a steel mesh made with ribbed
wire can be usually introduced in the first layer of wearing asphalt concrete. In this case the
greatest possible adhesion between steel and asphalt concrete is realized (Wiman 2001).
It is also important that steel elongations must be accounted for in order to avoid exces-
sively work-hardened materials with a low adaptation capacity to the surface in step of lay-
ing. The meshes of a welded steel mesh must be rather dense (typically 100 mm side) while the
diameters should not be very high. This reinforcement is slightly used with higher diameters
in the case of the thermal cracking resistance and in particular in the longitudinal direction
of the roadway where the maximum benefit is achieved. In order to also prevent rutting phe-
nomena, various configurations and shapes of steel mesh have been studied. In particular, in
order to develop a tri-dimensional effect a steel mesh with a diagonal mesh has been studied
(REFLEX 2002) and a steel passivation state must be also accounted for.
Before the steel mesh installation in situ, an excessive rust layer can affect the complete
adhesion between steel and asphalt concrete and, in exercising, the steel corrosion can reduce
the section of the steel bars reducing the contribution of the reinforcing.
The steel mesh to be utilized in the rehabilitation of road asphalt concrete pavements gen-
erally has an hex mesh reinforced through transversely torsional flat wires arranged at regular
intervals which present highest inertia if compared to the smooth wires of the same section
and further increasing the global stiffness (Vuong et al. 2009). The particular geometry (hex-
agonal mesh instead of rectangle) and the junction technology (torsional wires rather than
welded) promotes the adhesion with the asphalt concrete. The steel mesh can be used with a
230
specially optimized slurry seal which, in addition to performing a waterproof function and
for the sealing of cracks, ensures the perfect adhesion between the asphalt concrete layers and
the steel mesh itself.
For this particular application, the slurry seal is suitable made of sand (about 90%), cement
(1–1.5%) and modified bitumen emulsion. The cement acts as a catalyst in the process of
breaking down of the modified emulsion with polymer ensuring the elasticity over the time.
For these applications, the steel mesh has an higher axial rigidity reducing the stress state.
It is also fundamental to stretch a steel mesh reinforcing secured to the existing pavement
layer through a proper nailing. This operation presents more time consuming if compared
with a geosynthetic reinforcing material but more benefits have been shown. During steel
mesh removal, instead, milling operations are necessary because during this operation the
milled material could contains metal elements (not reusable for the production) of hot or cold
recycled asphalt concrete.
Then, it is milled until the sharing with the steel mesh is realized and then it is mechanically
divided from the surface. In this way, it is possible to completely separate the steel mesh from
asphalt concrete that can be reused as milled material.
3 THE FE MODEL
The pavement studied in this paper relates to a secondary-ordinary suburban road (C1 cat-
egory) according to the Italian Highway Code (MIT 1992). The thicknesses of the pavement
layers were derived from the Italian Pavements Catalog (CNR 1995) that reports the dimen-
sional solutions, for each type of road, as a function of the number passages of commercial
vehicles and the resilient modulus value of the subgrade.
In this work, the road pavement studied is characterized by a number of commercial
vehicles passages equal to 4,000,000 and a resilient modulus value of the subgrade equal to
90 N/mm2.
The pavement is composed by a wearing course layer (in asphalt concrete) with a thickness
of 5 cm; a binder layer (in asphalt concrete) of 6 cm; a base layer (in asphalt concrete) of
13 cm and a foundation layer (in granular aggregate) of 15 cm. The described pavement rests
on a subgrade with resilient modulus of 90 N/mm2.
The soil properties of the constituent materials of the implemented model are implemented
as individual layers. These values, summarized in Table 1, have been defined in terms of elas-
ticity modulus E, Poisson coefficient ν and density γ.
In order to make the continuous model adhering to reality and get more reliable results,
the subgrade soil is modelled considering an embankment section of 7 m in height from the
embankment support plane and assuming the presence of a bedrock layer placed below. In
addition, for the embankment support plane, different soil types are considered, respectively
(A, B, C, D category) in accordance with the Italian code (MIT 2008).
Four different models have been implemented in ANSYS® software, one for each type of
considered soil (A, B, C and D) characterized from different values of dynamic elasticity
modulus Ed as shown in Table 2.
E γ
Layer N/mm2 ν kg/m3
231
Table 2. Ed values for the different soil type.
E
Soil type N/mm2
A 480
B 97
C 25
D 7.5
Figure 2. 2D and 3D ANSYS® model of the road configuration studied with meshing elements.
Figure 4. Vertical displacements (a) and vertical stresses (b) for 2D model considering different soil
types for configuration without reinforcing.
233
Figure 5. Vertical displacements (a) and vertical stresses (b) for 2D model considering different rein-
forcing depth for 2D model.
Table 4. Period value for 2D and 3D models, with and without adding steel mesh reinforcing at
11 cm from the rolling plane.
T2D (without mesh) T2D (mesh at 5 cm) T3D (mesh at 6 cm) T3D (mesh at 6 cm)
Id mode s s s s
Figure 7. Numerical outcomes for the harmonic analysis in terms of stress/strain state for 3D model.
comes performed show the 2D limit on the modelling of the real road pavement if compared
with 3D model.
235
Table 5. Differences in terms of Vertical Displacements (VD) obtained through
static analysis between 2D and 3D model with steel mesh at 11 cm.
Table 6. Differences in terms of Vertical Stresses (VS) obtained through static anal-
ysis between 2D and 3D model with steel mesh at 11 cm.
Table 7. Differences in terms of Shear Stresses (SS) obtained through static analy-
sis between 2D and 3D model with steel mesh at 11 cm.
of separation between the asphalt concrete layers, must be less than the sliding resistance
between the layers or: τmax ≤ 0,4 ÷ 0.5 N/mm2. The results of the static analysis carried out on
the 3D model highlight that all of the structural checks are satisfied.
In this paragraph the results of the static analysis, related to the response of the only continu-
ous 3D with steel mesh to a depth of 11 cm from the rolling plane in presence of fractures
will be shown. In order to obtain the stress-strain response within the pavement layers, the
modelling of a “longitudinal” fracture inside the pavement base layer with an amplitude
equal to 1 mm was implemented within ANSYS®. The results and the structural checks are
summarized in Tables 8–10.
It is important to note that the presence of the fracture in the base layer negatively affects
the stress-strain response of all the upper layers (wearing course and binder). Corresponding
to the foundation layer underlying the presence of the fracture is almost negligible. It is also
important to highlight that the presence of the steel mesh reduces (15–25%) the stress-strain
level of the entire pavement even with fracture.
236
Table 8. Vertical Displacements (VD) comparisons between with vs without adding steel
mesh.
Table 9. Vertical Stresses (VS) comparisons between with vs without adding steel
mesh.
Table 10. Shear Stresses (SS) comparisons between with vs without adding steel
mesh.
6 CONCLUSIONS
This paper has addressed the flexible pavements reinforcement theme through the inser-
tion of welded steel reinforcements at different depths. In particular, it has investigated the
dynamic response of superstructure reinforced and not through implementation in ANSYS®
finite element software of modal, harmonics and static analysis.
In order to obtain the maximum benefits in terms of stress-strain states an optimal depth
of the steel mesh reinforcing has been identified corresponding to 11 cm from the rolling
plane.
The study has also allowed to evaluate, in seismic conditions, the nature and type influence
of soil. From results of modal analysis with the worst soil (type D) it is clear that the inclu-
sion of the steel mesh makes no change, in both cases 2D and 3D, in global response in terms
of frequencies and vibration modes.
With reference to the stress-strain amplification effects, the beneficial effects offered by the
steel mesh inserted at the optimum depth of 11 cm have been also highlighted. As obtained
from static and harmonic analysis, the three-dimensional modelling of the domain has been
found to be more responsive to real case and is to be preferred if compared to the two-
dimensional approach.
The study has also shown that the inclusion of a welded steel mesh in the construction of
new flexible road superstructures has the mainly effect to realize the global stress-strain state
237
lower if compared with the case without reinforcing. The inclusion of the reinforcement at
the optimum depth of 11 cm from the rolling plane manages to contain the pavement crisis
strain conditions even in the presence of a marked fracturing assumed present inside the
pavement base layer.
REFERENCES
Agostinacchio, M. Ciampa, D. Olita, S. (2014). The vibrations induced by surface irregularities in road
pavements—a Matlab® approach. European Transport Research Review, 6 (3), pp. 267–275. DOI:
10.1007/s12544-013-0127-8.
CNR Consiglio Nazionale delle Ricerche (1995). Catalogo delle pavimentazioni stradali. Bollettino
Ufficiale n.178 del 15/09/1995-Parte IV.
Coni, M. & Bianco, P.M. (2000). Steel reinforcement influence on the dynamic behaviour of bituminous
pavement, Reflective cracking in pavements—Research in practice. Proceedings of the 4th interna-
tional Rilem Conference, 26–30 march 2000, Ottawa, Canada, ISBN: 2-912143-14-4.
La Agostinacchio M., Ciampa D. Olita, S. (2008). Cracking response and service life prediction of
flexible and semi-rigid road pavements implementing M-E PDG 2002 code. Pavement Cracking:
Mechanisms, Modeling, Detection, Testing and Case Histories, pp. 201–210.
MIT Ministero delle Infrastrutture e dei Trasporti (1992). Nuovo codice della strada. D.Lgs 30 aprile
1992, n.285 (GU n.114 del 18/05/1992—SO n.74).
MIT Ministero delle infrastrutture e dei Trasporti (2008). D.M. 14 gennaio 2008 pubblicato sulla
Gazzetta Ufficiale n. 29 del 4 febbraio 2008 (S.O. n. 30).
REFLEX (2002). Reinforcement of flexible road structures with steel fabrics to prolong service life.
Project reference: BRPR980816. European Commission & CORDIS-Community Research and
Development Information Service. http://cordis.europa.eu/project/rcn/46802en.html.
Vuong, Binh, Choi, Xavier, Hoque, Zahid (2009). Australian Asphalt Pavement Association. Assess-
ment of reinforced asphalt products for road applications using finite element modelling.
Wiman, L.G. (2001). Accelerated load testing of pavements HVS-nordic tests in Sweden 1999. In Swed-
ish National Road and Transport Research Institute-VTI (eds), Swedish National Road and Trans-
port Research Institute (VTI). VTI rapport 477A. ISSN: 0347–6030.
238
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: One of the main distress criteria affecting the performance of bituminous
pavement is fatigue cracking. Such cracks appear in the wheel path and are due to myriad of
factors including aging, stiff binders, traffic repetitions and large magnitude of loads. Taking
into account the increased emphasis on the use of Warm Mix Asphalt (WMA) technologies,
it is necessary to investigate the fatigue life of WMA mixtures since they are less susceptible
to aging. Different post-processing methods exist to analyze the fatigue experiments and
amongst them, the quantification of energy dissipation during fatigue loading is promising.
In this investigation, two types of WMA technologies namely, chemical additive based (EVO-
THERM) and organic additive based (SASOBIT) were benchmarked with hot mix asphalt.
Four point bending beam test was conducted to determine the fatigue life and the data was
analyzed using the evolution of energy dissipation.
1 INTRODUCTION
Warm-mix asphalt is classified as group of technologies that are used to produce asphalt mix-
tures at temperatures 30ºC lower than the traditional HMA (IRC: SP101, 2014). The most
commonly used WMA technologies are Chemical additives and Wax-based additives. These
additives impart reduced viscosities to the binder during mixing and compaction. Since these
additives allow the bituminous mixtures to be produced at lower temperatures, a considerable
reduction in aging of the binders is expected.
The understanding developed in the last few decades about fatigue damage has unequivo-
cally concluded that stiff binders, especially binders which are highly susceptible to aging lead
to early initiation of fatigue failure. Hence, it will be ideal if the binder is less aged during
production process so that one could offset the initiation of fatigue damage by few additional
years. Due to the reduced aging of the WMA binders, the tendency to exhibit increased
fatigue resistance when compared to HMA exists for some technologies and this needs to be
quantified.
Chemical additives improve the workability of an asphalt mix by reducing the friction
(due to reduction of surface tension of binder in the Newtonian regime) between the asphalt
binder and aggregates (Bower, 2011). Wax based additive contains long chain hydrocarbons
that offer lower viscosity at mixing and compaction temperature when compared to conven-
tional binder and at lower temperatures, it forms solid crystalline structure in the binder that
makes the binder to exhibit increased stiffness (Jamshidi et al., 2013). Such stiffness due to
the addition of wax-based additive may not contribute to the longer fatigue life. Hence, It
will be interesting to understand how the chemical additives and wax-based additives alter
the fatigue life of asphalt mixtures.
The important concern related to the determination of fatigue life of asphalt mixtures
is different post-processing methods exist to analyze the experimental data collected when
fatigue experiment on bituminous mixtures are carried out. Such methods include number
of cycles to 50% stiffness (AASHTO T321-07, 2007), normalized modulus (ASTM D7460,
2010), slope of dissipated energy curve, energy ratio (Rowe and Bouldin, 2000), evolution
239
of phase lag and Lissajous plots to name a few. Since each of these criteria follows different
approach to determine the fatigue failure, they may not show similar fatigue lives for same
material. It is important to understand how the fatigue life of WMA mixtures vary with
the post-processing technique used. Out of the various post-processing methods available,
approaches that appeal to energy dissipation are promising.
This investigation reports the quantification of fatigue life using the evolution of energy
dissipated for two types of WMA technologies and they are bench marked with HMA. For
the sake of comparison, the fatigue life as per AASHTO T321-07 (2007) is also reported.
2 EXPERIMENTAL INVESTIGATIONS
2.1 Materials
In this investigation, two types of WMA technologies were used. One is a chemical additive
(Evotherm) and the other is wax based additive (Sasobit). VG30 binder as per IS 73-2013,
(2013) was used to prepare HMA and WMA samples. Three types of samples namely, HMA,
WMA-Evotherm and WMA-Sasobit were tested. A binder content of 5% by mass of the
total mix was chosen for this study and this satisfied the mix design requirement as per the
guidelines given by MoRTH (2013). As recommended by the manufacturers, a dosage of
0.4% by weight of the binder was adopted for WMA-Evotherm. For WMA-Sasobit, 1.5% by
weight of binder was added. These additives were blended into the binder at a temperature
of 145ºC.
2.3 Mixing
For preparing WMA samples, WMA additive treated VG30 binder and the oven dried aggre-
gates were heated to 145ºC and mixed at 135ºC. In order to account the aging that occurs
in field during mixing and compaction, WMA mixtures were short-term aged at 117ºC for
4 hours + 5 minutes similar to HMA (AASHTO R30-2, 2006) and directly compacted at this
temperature. For preparing HMA samples, the oven dried aggregates and VG30 binder were
heated to 175ºC and the mixing was carried out at 165ºC. HMA mixtures were short-term
aged at 135ºC for 4 hours + 5 minutes and then placed at a compaction temperature of 150ºC
for 30 minutes (AASHTO R30-2, 2006).
2.5 Testing
The fatigue test was carried out using IPC EN standards testing equipment with 4 point
bending jig. These tests were carried out at five different strain levels 200, 400, 500, 600 and
800 με for both HMA and WMA beams. These strain levels were selected as per AASHTO
T321-07(2007) and ASTM D7460-10(2010) to generate the S-N plots. The selected frequency
was 10 Hz with a sinusoidal loading waveform and the testing temperature was 20ºC. The
fatigue beams were kept for conditioning in an environmental chamber at test temperature
240
Table 1. Number of cycles for termination of test for HMA and WMA.
for 2 hours prior to testing. After conditioning, the sample was placed into the loading frame
and clamped. The LVDT was placed on top of the sample at the mid-span and the initial
value was set to zero. The initial stiffness was calculated at the 50th load cycle and the test was
terminated either at 80% reduction in stiffness or at 1 × 106 cycle, whichever occurred earlier.
Table 1 shows the total number of loading cycles applied for the samples tested.
Figure 1 shows the variation of stiffness modulus with cycles as a function of strain for
HMA, WMA-Evotherm and WMA-Sasobit samples. It was seen that the termination stiff-
ness reached within the test duration at 400, 500, 600 and 800 με for both HMA and WMA
samples. It was interesting to note that the WMA-Evotherm samples showed more number
of cycle to reach its 80% reduction in stiffness than HMA, followed by WMA-Sasobit at all
strain levels (Table 1).
To understand the collected experimental data and interpret the results, two different post-
processing methods were used and they are discussed in the following.
S ae bx (1)
where, S = stiffness (Pa); x = number of cycles; a and b are regression constants. The initial
stiffness was calculated at x = 50 and the fatigue life for 50% reduction in stiffness was deter-
mined using exponential curve fitting.
For low strain level 200με, 50% reduction in stiffness was not seen for all the three materi-
als. Hence an exponential fit was carried out for such data (Figure 2).The number of cycle
corresponding to 50% reduction in stiffness was quantified for all the strain levels and sum-
marised in Table 3. It was observed that the WMA-Evotherm showed better fatigue life when
compared to HMA samples, followed by WMA-Sasobit samples at all tested strain levels.
for any given temperature and frequency. Initially when the loading was applied, a slight
distortion was seen at 10th cycle. With repeated fatigue loading (strain controlled), the stress-
strain response changes and the progressive distortions in the stress-strain hysteresis loops
are shown in Figure 3. Dissipated energy for each loading cycle can be determined by meas-
uring the area within the stress-strain loop and the equation to measure the dissipated energy
is given below (EN 12697-24, 2004).
Wi = πσ i ε i δi (2)
242
Figure 3. Progressive distortions of WMA-Evotherm at 800 με.
where Wi is the dissipated energy in cycle i, σi is the maximum tensile stress (kPa), εi maxi-
mum tensile strain, δi is the phase angle (degrees) for cycle i. The amount of energy dissipated
per loading cycle changes through the test period. In a strain controlled test, the dissipated
energy decreases continuously and is consistent with the reduction in flexural stiffness of the
material.
Figure 4 shows the dissipated energy curves for all the three materials at selected strain
levels. It is observed that the rate of reduction of dissipated energy was greater for WMA-
Sasobit samples compared to HMA and WMA-Evotherm samples at higher strain levels.
243
Figure 5. Dissipated energy curve for WMA-Evotherm at 400 με.
At lower strain level 200 με, it was found that the all the three mixtures exhibited identical
trends.
As part of RILEM initiatives, Benedetto et al., (2004) postulated the three stage damage
evolution for the fatigue of bituminous mixtures. These three stages were called as adaptation
phase (Phase 1), quasi-stationary phase (Phase 2) and failure phase (Phase 3). According to
Benedetto et al., (2004), in the first phase the decrease of stiffness modulus (and hence dissi-
pation) can be ascribed to ‘healing’ and ‘thixotropy’. It is expected that when the experiments
are conducted with rest periods during the first phase, substantial ‘healing’ of material can
occur. The precise fatigue damage accumulation occurs during the second phase wherein
every cycle adds a quantifiable amount of damage. Phase 3 relates more to the onset of
macro-cracks. With this background, it is worthwhile to quantitatively assess the evolution
of dissipated energy for all the experiments carried out here and find out the WMA-Strain
combination wherein the second stage manifests. Figure 5 shows a sample graph in which
phase 1, phase 2 and phase 3 are demarked.
To systematically determine the three stages, the following empirical equation was chosen.
This expression is in a sense similar to the equation used for modelling the three stage creep
curve (Francken 1977).
Wi = aN −bb − c(e d
dN
− ), (3)
d 2Wi
= ab(b )N − ( b )
− cd e dN . (4)
dN 2
The cycle number at which the tertiary phase is reached is given by the point where the rate
of change of slope changes sign. This indicates the inflection point in the curve of dissipated
energy versus number of cycles. Figure 6 shows the model fitting of dissipated energy data
for all the three materials at 600 με. The model parameters were estimated for all the three
materials and is given in Table 2.
244
Figure 6. Model fitting for dissipated energy curves- 600 με.
Onset of
Material a b c d R2 3rd stage
For low strain level (200 με), third phase was not seen within the test duration, for which
the parameters 'c’ and ‘d’ are 0 and hence it was not considered to determine the fatigue life
using this method.
The fatigue lives obtained from exponential fitting method and energy dissipation method
are tabulated in Table 3. It is seen that the fatigue lives are not identical using both the
245
Table 3. Fatigue life values using different post processing methods.
HMA
AASHTO 2.63E+06 2.54E+05 8.04E+04 1.64E+04 2.62E+03
Dissipated energy – 8.36E+04 4.35E+04 3.69E+03 6.90E+02
WMA-Sasobit
AASHTO 4.12E+06 1.17E+05 4.06E+04 6.78E+03 1.35E+03
Dissipated energy – 2.92E+04 1.00E+04 2.48E+03 1.27E+03
WMA-Evotherm
AASHTO 5.23E+06 4.48E+05 1.23E+05 2.29E+04 3.23E+03
Dissipated energy – 1.44E+05 4.18E+04 2.97E+04 1.38E+03
methods. From exponential fitting method, it was observed that the fatigue life of WMA-
Evotherm was higher than HMA, followed by WMA-Sasobit for all the higher strain levels.
While such order is preserved when dissipated energy is used for finding the onset of 3rd
stage, the number of cycles to reach such criteria is comparatively less than the AASHTO
criterion.
4 CONCLUSIONS
In this investigation, fatigue life was determined using two post-processing methods and
they are exponential fitting method (AASHTO T321- 07, 2007) and dissipated energy curve
method. While the exponential fitting method assumes the fatigue failure point as the number
of cycles corresponding to the 50% reduction of initial stiffness, in the dissipated energy
approach, the failure point was identified based on the slope variation of dissipated energy
curve. WMA-EVOTHERM exhibited better fatigue life when compared to WMA-SASOBIT
and HMA. It is expected that the improved fatigue life of Evotherm treated mixture can be
due to the less aging of the material and the presence of wax in the Sasobit might be one
reason for the reduced fatigue life at lower temperatures.
REFERENCES
AASHTO R30-2. 2006. Standard practice for mixture conditioning of hot mix asphalt (HMA). Ameri-
can Association of State Highways and Transportation Official, Washington, D.C: USA.
AASHTO T321-07. 2007. Determining the fatigue life of compacted hot mix asphalt subjected to
repeated flexural bending, American Association of State Highways and Transportation Officials,
Washington, D.C: USA.
ASTM D7460-10. 2010. Standard test method for determining fatigue failure of compacted asphalt
concrete subjected to repeated flexural bending. ASTM International, West Conshohocken, Penn-
sylvania: USA.
ASTM D7981-15.2015. Standard Practice for Compaction of Prismatic Asphalt Specimens by Means
of the Shear Box Compactor. ASTM International, West Conshohocken, Pennsylvania: USA.
Bower, N. 2011. Laboratory Evaluation of Performance of Warm mix asphalt in Washington state.
Master of Science thesis, Washington university.
EN 12697-24 (2004). Four point beam test on prismatic shaped specimen (resistance to fatigue)- Annex-
ure D. Technical report, European Committee for standardization.
Francken, L., Pavement deformation law of bituminous road mixes in repeated load triaxial compres-
sion. Fourth international conference on the structural design of asphalt pavements, University of
Michigan, Ann Arbor, 1977, USA, pp. 483–496.
IRC SP 101. 2014.Interim Guidelines for Warm Mix Asphalt, Indian Roads Congress, New Delhi:
India.
IS 73-2013. 2013. Paving bitumen - specification, Bureau of Indian Standards. New Delhi: India.
Jamshidi, A. Hamzah, M.O. & Zhanping, Y. 2013. Performance of Warm Mix Asphalt containing Saso-
bit: State-of-the-art. Journal of Materials in Civil Engineering: 530–553.
MoRTH. 2013. Specification for road & bridge works, fifth revision. IRC publications, Ministry of Road
Transport & Highways, New Delhi: India.
Rowe, G. & Bouldin, M. 2000. Improved techniques to evaluate the fatigue resistance of asphalt mix-
tures, 2nd Eurasphalt and Eurobitume Congress. Barcelona.
247
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
India has the second largest road network in the world. The National Highways have a length
of 96,260 kms, which comprise of 1.7% of the total road length and these highways carry
more than 40% of the total traffic (Indian road network, 2016). Most of the national high-
ways are two lane roads and the Government of India had setup a program called National
Highway Development Project (NHDP) to expand and upgrade the highway network. As
part of the NHDP, 46,902 kms (NHDP, 2016) are under construction and most of the high-
ways are bituminous pavements.
The national highways in India are designed as per the guidelines of Indian Roads Con-
gress (IRC: 37–2012, 2012).The main inputs for the design are Annual Average Daily Truck
Traffic (AADTT), axle load data and subgrade properties. Using the axle load data, Vehicle
Damage Factor (VDF) is calculated, which is used to calculate the total traffic in terms of
Equivalent Standard Axle Load (ESAL). Depending on the subgrade properties and ESAL
value, the design thickness are considered. The different layers of a new pavement cross-
section provided by IRC: 37–2012 (2012) is shown in Figure 1.
The design guidelines provided in IRC: 37–2012 are in the developmental stages. Sepa-
rate critical pavement temperatures are not taken into account at this point of time. An
annual average pavement temperature (AAPT) is used with the understanding that the use
of materials with the appropriate “modulus” value at such AAPT can, not only handle rut-
ting but fatigue cracking also. Also, the material database related to bituminous mixtures
are not available and a resilient modulus (MR) is used similar to the approach used in the
earlier AASHTO design code during the 1990’s. Such modulus values are measured at one
selected temperature for a specific frequency of 10 Hz. Over and above to this, due to the
substantial economic development taking place in the country, the truck traffic has increased
exponentially bringing within the attendant problems of overloading. With this background,
it will be interesting to see whether the pavement design sections given in the guidelines of
IRC: 37–2012 can actually handle the expected distresses within the provided design life.
Towards this end, the current approach outlined in Mechanistic-Empirical Pavement Design
Guide (M-EPDG) is used.
249
Figure 1. Different layers of pavement cross-section (IRC: 37–2012,2012).
As a part of the NCHRP project 1–37A, pavement analysis system called AASHTOWare was
released. This system implements the latest version of the Mechanistic-Empirical Pavement Design
Guide (M-E PDG) as envisioned by AASHTO (M-E PDG, 2008). It is quite comprehensive in
the manner in which pavement design can be proof-checked. It does not provide any design solu-
tion, instead it provides the framework for proof-checking the given design cross-section taking
into account all possible variations in traffic, climate and material properties. M-E PDG requires
total axle load spectrum, axle load distribution, and monthly variation of traffic, material proper-
ties and hourly climate data of the location to carry out the simulation. The requirement of data
to analyze a pavement is substantial, however, it also helps the designer to choose an optimal
solution. In this study, simulations were carried out for various design scenarios. The pavement
cross-section designed as per IRC: 37–2012 was proof-checked and sensitivity analysis of various
parameters was conducted on two National highways using AASHTOWare.
In this study, two National highways (NH 207 and NH 69) from two different locations pos-
sessing different climatic features were chosen.
Description NH 207 NH 69
which is not calibrated for India. Hence, to calculate AAPT for Bangalore and Bhopal, the
regression equation developed by Nivitha and Krishnan (2014) shown in Equation 1 was
used. The AAAT value was calculated from the daily maximum and minimum temperature
data for consecutive three years obtained from Indian Meteorological Department (IMD),
Pune. Table 2 shows the climatic properties for Bangalore and Bhopal.
where, α, β and γ are regression constants with values -0.7147, 1.3023 and 0.1103
respectively.
35 1700
40 1250
T (°C) BC DBM
Frequency (Hz)
5 8985 16526 22365 24516 27522 28003 9190 16087 21483 23545 26301 26696
15 2577 7266 12244 14438 17906 18462 2309 7136 12472 14835 18559 19175
25 531 2462 6149 8041 11230 11735 446 1922 5267 7165 10415 10946
35 172 659 2115 3189 5361 5684 127 506 1809 2855 5036 5323
45 124 262 780 1277 2552 2756 80 190 631 1077 2299 2547
55 76 111 236 360 757 906 65 90 206 346 775 960
252
Figure 4. Design cross-sections of NH 207 and NH 69.
Table 6. Fatigue and rutting life of designed cross-sections as per IRC. 37–2012.
NH 69 54 1055
NH 207 26 599
As can be seen, figures 5 to 7 clearly show that all the distresses reached their threshold val-
ues much before the design life. In order to understand the influence of speed, AADTT and
thickness of bituminous layers on the performance of the pavement, sensitivity analysis was
carried out and the various parameters chosen for sensitivity analysis are shown in Table 8.
Sensitivity analysis was carried out by varying one particular parameter by keeping all other
parameters constant (Kim et al., 2005).
4 SENSITIVITY ANALYSIS
Using the above parameters, simulations were carried out in AASHTOWare and the sen-
sitivity analysis was conducted using ‘one at a time’ approach (NCHRP 1–47, 2011). The
results from the simulations are plotted separately for all the three distresses such as Asphalt
Concrete (AC) rutting, bottom up fatigue cracking and total rutting. Figures 8 to 13 show
the variation of AC rutting and Figures 14 to 19 show the variation of bottom up fatigue
cracking. Figures 20 to 25 show the variation of total rutting with the parameters which are
mentioned earlier.
253
Figure 5. Variation of AC rutting for Figure 6. Variation of bottom up fatigue
the chosen sections. cracking for the chosen sections.
Parameters NH 207 NH 69
4.1 AC rutting
From the above graphs, it can be clearly identified that the sensitiveness of speed and
AADTT towards AC rutting was comparatively higher than thickness. The sensitiveness of
speed and AADTT were almost identical
Figure 10. Variation of AC rutting with speed. Figure 11. Variation of AC rutting with speed.
Figure 12. Variation of AC rutting with Figure 13. Variation of AC rutting with bitu-
bituminous layer thickness. minous layer thickness.
255
Figure 14. Variation of bottom up fatigue Figure 15. Variation of bottom up fatigue
cracking with AADTT. cracking with AADTT.
Figure 16. Variation of bottom up fatigue Figure 17. Variation of bottom up fatigue
cracking with speed. cracking with speed.
Figure 18. Variation of bottom up fatigue Figure 19. Variation of bottom up fatigue
cracking with bituminous layer thickness. cracking with bituminous layer thickness.
256
Figure 20. Variation of total rutting with Figure 21. Variation of total rutting with
AADTT. AADTT.
Figure 24. Variation of total rutting with bitu- Figure 25. Variation of total rutting with bitu-
minous layer thickness. minous layer thickness.
257
lower speeds (lower frequency) the modulus value will be lower and this leads to more rut-
ting. The parameters such as speed exhibited more sensitivity to total rutting and AC rutting
while thickness exhibited greater influence towards bottom up fatigue cracking and total rut-
ting. AADTT showed sensitiveness for all the distresses
5 SUMMARY
Two national highway sections (NH 207 and NH 69) with different climatic regions were consid-
ered and sections were designed as per IRC. 37–2012 guidelines. The designed cross-sections were
analysed using AASHTOWare and the performance was predicted. To improve the performance
and to find out an adequate design to achieve good performance throughout the design life,
sensitivity analysis was conducted. The results of the sensitivity analysis showed the effect of
each parameter on the performance. This sensitivity analysis was conducted using ‘one at a time’
approach. Using the results collated, it is clearly seen that for a country like India with heavy
axle loads and low speeds, substantial thickness in the bituminous layers are required. This is the
first ever exercise conducted for Indian Highways and substantial work in terms of creation of
material database and appropriate distress parameters properly validated are required.
ACKNOWLEDGEMENT
The authors thank Department of Science and Technology for funding this investigation.
The grant number is DST/TSG/STS/2011/46. The authors thank M/s V.R. Techniche, Delhi
and L&T IDPL, Chennai for sharing the traffic and axle load data.
REFERENCES
ASTM D 3381. 2013. Standard Specification for Viscosity-Graded Asphalt Cement for use in Pavement
Construction. American Society for Testing and Materials Standards, West Conshohocken, Pennsyl-
vania, USA.
Indian road netwok (2016). http://www.nhai.org/roadnetwork.htm. Retrieved or Accessed on 2nd Janu-
ary, 2016
IRC: 37–2012 (2012). Guidelines for the design of flexible pavements. Indian Roads Congress, New
Delhi, India.
IS: 73–2013 (2013). Paving Bitumen – Specifications. Bureau of Indian Standards, New Delhi, India.
Kim, S., Ceylan, H., Heitzman, M. (2005). Sensitivity Study of Design Input Parameters for Two Flex-
ible Pavement System Using the Mechanistic–Empirical Pavement Design Guide. Proceedings of the
2005 Mid-Continent Transportation Research Symposium, Iowa, August 2005.
MEPDG. (2008). Mechanistic Empirical Pavement Design Guide: A Manual Practice. American Asso-
ciation of State Highway and Transportation Officials, Washington, D.C., USA.
NCHRP 9–29: PT 01 (2008). Proposed Standard Test Method for Determining the Dynamic Modulus
and Flow Number for Hot Mix Asphalt (HMA) Using the Simple Performance Test System. Trans-
portation Research Board, Washington, D.C.
NCHRP 1–47, (2011). Sensitivity Evaluation of MEPDG Performance Prediction, Transportation
Research Board, Washington, D.C.
NHDP (2016). http://www.nhai.org/WHATITIS.asp. Retrieved or Accessed on 2nd January, 2016
Nivitha, M.R., and Krishnan J.M. (2014). Development of Pavement Temperature Contours for India,
Journal of Institution of Engineers (India: series A), 95 (2): 83–90.
258
Pavement & Bitumen
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Ming Wu
Highway Administration Bureau of Zhejiang Province, Hangzhou, China
1 INTRODUCTION
The recycling of asphalt pavements has become a routine procedure normally by an average
rate of 10 to 20% in a given mix design and large part of the Reclaimed Asphalt Pavement
(RAP) is stockpiled or degraded for low value applications (Zaumanis & Mallick 2015,
Swiertzet al. 2011, Mogawer et al. 2012). A survey conducted by Federal Highway Administra-
tion (FHWA) showed in US that although many state transportation agencies specifications
allow up to 30% RAP in the surface layers, a majority of these states are still only using RAP
percentages of 10 to 20% (Copeland 2011).
Generally, the major limitations for a high percentage RAP content in the mix design come
from three aspects (Newcomb et al. 2007, Howard et al. 2009, Zaumanis & Mallick 2015):
the properties of aged binder in RAP that can cause pavement cracking failures, the degree
of blending and diffusion that occurs between the virgin and RAP binder, and the RAP
aggregate properties, especially the fines content. There are several methods that can help to
apply high percentage of RAP and the principal are as follows (Karlsson & Isacsson 2006,
Al-Qadi et al.2007): use of softer binder, addition of softening additives or rejuvenators and
use of warm mix asphalt (WMA). Al-Qadi et al. (2012) performed an investigation on the
effect of softer binder and concluded that reducing one or two grades of virgin binder at a
50% RAP rate ensured equal or even better performance in laboratory compared with virgin
reference mixture. Similarly, virgin binder grade should be reduced to lower stiffness of 40%
RAP mixtures according to McDaniel et al. (2012). Recycling agents can be used to meet the
target PG of the aged RAP binder, resulting in improved cracking resistance without failing
the rutting resistance requirements (Zaumanis et al. 2013). Previous studies have shown that
inclusion of WMA additives has allowed the production of workable mixtures containing 50
to 100% RAP (Vargas-Nordcbeck & Timm 2012, D’Angelo et al. 2007, Mallick et al. 2008,
Ma & Mallick 2009).
Despite the efforts above, the blending process of RAP with virgin materials significantly
influences the properties of resultant mixtures and the relevant knowledge is limited. Mogawer
261
et al. (2013) evaluated the effect of rejuvenators on the degree of blending and performances
of high RAP, reclaimed asphalt shingle (RAS), and RAP/RAS mixtures. The authors found
rejuvenators can mitigate the stiffness of aged binder and improve the cracking characteris-
tics but adversely impact the rutting and moisture susceptibility of the mixture.
This study carried out the investigation of blending process and its impacts on the mixture
properties with two objectives: 1) determination of preheating temperature of aggregate sub-
ject to RAP content; 2) investigation of the impact of blending process, especially the feeding
sequence of materials, on mixture properties to recommend optimum blending procedure.
Aged asphalt mixture samples were collected from an in-service highway in China, which
has provided service more than one decade and experienced significant rutting distress and a
medium level of transverse and longitudinal cracks. The Rotavapor method (ASTM D5404
2012) was used for recovery of asphalt cement from solvent. Extracted binders were then
rejuvenated by one commercial rejuvenator Evoflex and SBR latex, respectively. The binder
properties are listed in Table 1.
Binder properties are improved with the addition of rejuvenator and SBR latex and the
modification degree increases with the additive content. Addition of 2%-4% rejuvenator or
4–6% SBR latex can largely restore the binder properties except for the ductility indicator.
Moreover, 6% SBR latex significantly increases the softening point which may cause diffi-
culty in the blending process. Therefore, at the consideration of binder property, workability
and economic factors, 2% rejuvenator and 4% SBR latex of aged binder mass were used in
the subsequent mixture preparation.
262
Table 2. Temperature data of different mixtures.
60 175 100 94 90
60 190 108 107 105
60 205 115 111 108
60 220 125 116 114
80 175 120 118 115
80 190 129 124 118
80 205 138 131 125
80 220 144 137 130
100 175 134 127 123
100 190 141 136 129
100 205 144 141 136
100 220 149 149 143
120 175 143 138 131
120 190 147 142 138
120 205 153 147 143
120 220 158 154 150
Temperature (°C)
Added (WMA) Density (g/cm3) 2.46 2.46 2.46 2.44 2.41 2.39
Air void (%) 3.9 3.8 4.3 4.7 5.7 6.5
VMA (%) 13.3 13.5 13.4 14.1 15 15.7
VFA (%) 70.7 71.8 67.9 66.7 62.0 58.7
Not added (HMA) Density (g/cm3) 2.44 2.41 2.38 2.39 2.36 2.31
Air void (%) 4.6 5.8 6.3 6.8 7.9 9.7
VMA (%) 14.0 15.1 15.5 16.0 17.0 18.6
VFA (%) 67.2 61.6 59.5 57.5 53.5 47.9
2008), the temperature of HMA shall be between 160°C and 170°C, which cannot be achieved
if the RAP content exceeds 30% at the given conditions. To elevate the mixture temperature,
it needs to either increase the RAP or aggregate temperatures. The former may give rise to
over aging concern and the latter is not economic to adopt a very high aggregate heating
temperature.
WMA is another solution. Warm agent at 0.5% of binder mass was used. The mixtures
(with 30% RAP) with and without warm agent were prepared at 100°C, 110°C, 120°C, 130°C,
140°C, 150°C and 160°C, respectively. Table 3 shows the volumetric information of the mix-
tures at different temperatures.
For the density indictor, WMA reaches and maintains the maximum value at temperatures
greater than 140°C while HMA increases with temperature. Air voids for both mixtures gen-
erally decrease with the increasing mixture temperature. For WMA, 130°C is inferred to be
threshold, below which the air void increases quickly. VMA indicator can satisfy the require-
ment (greater than 13%) at the given test temperature range for both mixtures. VFA increases
with the mixture temperature. However, VFA of WMA can satisfy the requirement (65–70%)
within the temperature range while VFA of HMA can only meet the requirement at 160°C.
263
Based on the above analysis, the acceptable temperature for WMA is 140°C and 160°C for
HMA. The determined temperature for WMA at 30% RAP was used for WMA with 40%
and 50% RAP contents in the following sections.
RAP temperature(°C) 60 80
RAP content 30 40 50 30 40 50
Aggregate temperature (°C) NA1 NA1 NA1 210 NA1 NA1
RAP content 30 40 50 30 40 50
Aggregate temperature (°C) 190 205 215 —2 183 195
Note: 1the WMA temperature cannot reach 140°C for aggregate temperature ranging from 175–220°C;
2
the WMA temperature can reach 140°C at any aggregate temperature ranging from 175–220°C.
264
Figure 2. RAP at 100 and 120°C.
of WMA, which is not economic. On the other way, over-heating of RAP (≥120°C) leads to
further aging and melting of RAP, as shown in Fig. 2, which may stick to RAP preheating
facilities in the mix plant.
Therefore, at the requirement of WMA temperature of 140°C, this study recommended
the RAP and aggregate preheating temperatures in Table 5.
Aside from the temperature factors, other blending process parameters, e.g. the feeding
sequence of materials, blending duration, also influence the mixture properties, which are
investigated in this section.
1 Add to RAP 120 Add to asphalt 2.442 4.5 90.3 4800 1895
2 120 Add to mixture 2.457 4.2 88.3 5300 1964
3 180 Add to asphalt 2.455 5.0 90.5 5200 2083
4 180 Add to mixture 2.462 4.4 91.2 6500 2210
5 Add to asphalt 120 Add to asphalt 2.435 3.8 86.3 3300 1939
6 120 Add to mixture 2.448 4.3 87.1 4100 1978
7 180 Add to asphalt 2.441 4.7 89.4 3700 2105
8 180 Add to mixture 2.459 3.4 88.6 4360 2383
Note: 1TSR = splitting tensile strength after freeze-thaw/ splitting tensile strength before freeze-thaw;
2
FN is the flow number for the dynamic creep test; 3 strain represents the tensile strain at damage for
the bending beam test.
Plan ID 1 2 3 4 5 6 7 8
Gray absolute correlation degree1 0.797 0.847 0.799 0.987 0.702 0.747 0.718 0.728
Note: 1the higher the gray absolute correlation degree, the better the plan is.
5 CONCLUSION
266
2. The preheating temperature of RAP used 100°C; the preheating temperatures of aggre-
gate were recommended to be 190°C, 205°C and 215°C at RAP content of 30%, 40% and
50%, respectively.
3. The feeding sequence of additives is important to mixture properties and this study rec-
ommended to add rejuvenator to RAP first and then SBR latex to mixture with a blending
duration of 180 s.
REFERENCES
Al-Qadi, I.L., Elseifi, M.A., & Carpenter, S.H. 2007. Reclaimed asphalt pavement: a literature review.
Springfield: Illinois Center for Transportation, Report Number, Report Number FHWA-ICT-
07−001.
Al-Qadi, I.L., Aurangzeb, Q, Carpenter, S.H., Pine, P.J., & Trepanier, J. 2012. Impact of high RAP con-
tents on structural and performance properties of asphalt mixtures. Springfield: Illinois Center for
Transportation, Report Number FHWA-ICT-12-002.
ASTM D5404. 2012. Standard Practice for Recovery of Asphalt from Solution Using the Rotary Evap-
orator, ASTM International, West Conshohocken, PA
Copeland, A. 2011. Reclaimed asphalt pavement in asphalt mixtures: state of the practice. Publication
FHWA-HRT-11-021, Turner-Fairbank Highway Research Center, Federal Highway Administration,
Department of Transportation, China. 2008. Technical specification of highway asphalt pavement
recycling. China Communications Press, Beijing.
D’Angelo, J., Harm, E. & Bartoszek, J., et al. 2007. Warm-mix asphalt: European practice. International
Technology Scanning Program, Federal Highway Administration, December, McLean, VA.
Gao, Y.F., Wang, J. & He, Z.D. 2013. Gray correlation analysis on influencing factors of postgraduates’
innovative capacity. Journal of Theoretical and Applied Information Technology 49(1):419–425.
Howard, I.L., Cooley Jr., AL. & Doyle, J.D. 2009. Laboratory testing and economic analysis of high rap
warm mixed asphalt. Jackson: Mississippi Department of Transportation, Report Number FWHA/
MS-DOT-RD-09-200.
Karlsson, R. & Isacsson, U. 2006. Material-related aspects of asphalt recycling-state-of-the-art. Journal
of Materials in Civil Engineering 18 (1), 81–92.
McDaniel, R.S., Shah, A.& Huber, G. 2012. Investigation of low- and high-temperature properties of
plant-produced RAP mixtures. McLean, VA: Turner-Fairbank Highway Research Center, Report
Number FHWA-HRT-11-058.
Ma, T.,& Mallick, R.B. 2009. Anevaluation of the effects of warm mix asphalt additives on workability
and mechanical properties of Reclaimed Asphalt Pavement (RAP) material. Paper No. 09–3503 Pre-
sented at the 88th Transportation Research Board Annual Meeting, Washington, DC.
Ma, T., Zhao, Y.L., Huang, X.M. & Zhang, Y. 2016. Using RAP Material in High Modulus Asphalt
Mixture. Journal of Testing and Evaluation. doi:10.1520/JTE20150194.
Mallick, R.B., Kandhal, P.S., & Bradbury, R.L. 2008. Using warm-mix asphalt technology to incor-
porate high percentage of reclaimed asphalt pavement material in asphalt mixtures. Journal of the
Transportation Research Board, 2051:71–79.
Mogawer, W.S., Bennert, T., Daniel, J., Bonaquist, R., Austerman, A.J., & Booshehrian, A. 2012. Per-
formance characteristics of plant produced high RAP mixtures. Journal of the Association of Asphalt
Paving Technologist (AAPT)81: 403–440.
Mogawer, W.S., Booshehrian, A., Vahidi, S. and Austerman, A.J. 2013. Evaluating the effect of rejuve-
nators on the degree of blending and performance of high RAP, RAS, and RAP/RAS mixtures. Road
Materials and Pavement Design 14(S2):193–213.
Molenaar, A., Scarpas, A., Liu, X., & Erkens, S. 2002. Semi-circular bending test; simple but useful?
Journal of Asphalt Paving Technologists (71):794–815.
Newcomb, D., Brown, E., and Epps, J., 2007. Designing HMA mixtures with high RAP content: a practi-
cal guide. Lanham, MD: National Asphalt Pavement Association.
Swiertz, D., Mahmoud, E., & Bahia, H. 2011. Estimating the effect of recycled asphalt pavements and
asphalt shingles on fresh binder, low-temperature properties without extraction and recovery. Journal
of the Transportation Research Board 2208: 48–55.
Vargas-Nordcbeck, A & Timm, D.H. 2012. Rutting characterization of warm mix asphalt and high
RAP mixtures, Road Materials and Pavement Design, 13(sup1):1–20.
Visintine, B., Khosla, N.P. & Tayebali, A. 2013. Effects of higher percentage of recycled asphalt pave-
ment on pavement performance. Road Materials and Pavement Design 14(2):432–437. West, R., et al.,
267
2012. Phase IV NCAT pavement test track findings. Auburn, AL: National Center for Asphalt Tech-
nology, Report Number NCAT 12–10.
Zaumanis, M., Mallick, R.B., and Frank, R. 2013. Use of rejuvenators for production of sustainable
high content RAP hot mix asphalt. In: XXVIII international Baltic road conference, 26–28 August
2013 Vilnius, Lithuania.
Zaumanis, M. & Mallick, R.B 2015. Review of very high-content reclaimed asphalt use in plant-pro-
duced pavements: state of the art, International Journal of Pavement Engineering 16(1):39–55.
268
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
70 74 46.6 39 0.02 6
269
Table 2. Technical indices of RPM.
Rubber
Relative density Moisture Carbon black hydrocarbon
Mesh g ⋅ cm−3 content% content% content%
60 1.24 0.45 29 57
Extracted RPM
RMA Trichloroethylene
Ageing time g g Mass, g Content, %
Ro 1.5511
Rd 1.5532
Ru 1.4815
RT+P0 1.3680
RT+P10 1.3312
RT+P20 1.3001
RT+P40 1.2381
RT+P60 1.2511
RT+P80 1.2232
271
Figure 2. Apparent morphologies of RPMs.
The original boundaries of RT+P0, RT+P10 and RT+P20 gradually vanish with the prolongation
of ageing time. The RPM granules become viscous and tend to aggregate, which shows that
desulfurization reaction occurs. This finding corresponds to that of a study (Wang, 2011).
The RPM aggregates are mutually adhered to form viscoelastic structures during long-
time ageing. Thus, it can be observed that RT+P40, RT+P60 and RT+P80 are in melting state.
Overall, change in RPM morphology reveals its degradation behavior.
3 CORRELATION ANALYSIS
The second part confirms that RPM degrades obviously with the ageing of RMA. Accord-
ingly, the behavior characteristic of RPM has correlation with the ageing mechanism of
RMA, and this correlation can be further proved by the IR spectrum analysis.
Figure 3 displays the IR spectra of RMA samples.
In Figure 3, the strong absorption peaks at 1460 cm−1 and 1375 cm−1 correspond to the
in-plane bending vibrations of C—H bonds. Absorption peaks also expand at 1700 cm-1
as a result of C=O stretching vibrations, and the adjacent peaks at 1600 cm-1 belong to the
deformation vibrations of C=C bonds. In addition, the stretching vibrations of S=O bonds
are observed at 1030 cm-1 (Liu, 2014; Siddiqui, 1999).
Many studies have proven that C=O and S=O are the major functional groups formed dur-
ing asphalt oxidative ageing. Therefore, the C=O and S=O indices are considered as indica-
tors to evaluate the oxidation extent (Chen, 2014; Pang, 2014). They can be calculated as the
following Formulas 1–2 (Jia, 2014).
272
Figure 3. IR spectra of RMA samples.
Figure 4. IR spectra of Ru, RT+P0, RT+P10 and RT+P20. Figure 5. IR spectra of RT+P40, RT+P60 and RT+P80.
RMA: the integral range of S=O is [985 cm−1, 1070 cm−1]; the integral range of C—H is [1400 cm−1,
1500 cm−1]. RPM: the integral range of S=O is [975 cm−1, 1140 cm−1]; the integral range of C—H is
[1380 cm−1, 1470 cm−1].
a b R2
multiples. Moreover, the formula model (Allometric1) is applied to both RMA and RPM,
which indicates that the oxidation behavior of RPM is consistent with that of RMA.
M a ⋅T b (3)
4 CONCLUSIONS
In this study, the ageing mechanism of RMA is explained from a new perspective. Since RPM
is an important component of RMA, the behavior characteristic of RPM should receive
much attention during the ageing process of RMA.
Test results confirm that the ageing of RMA is accompanied by the degradation of RPM,
which behaves as mass loss, decrease in crosslink density and change in apparent morphol-
ogy. Moreover, the oxidation behavior of RPM is consistent with that of RMA.
274
REFERENCES
Chen, M.Z. & Leng, B.B. 2014. Physical, chemical and rheological properties of waste edible vegetable
oil rejuvenated asphalt binders. Construction and Building Materials, 66: 286–298.
Jiang, G.M. 2010. Microbial desulfurization technology of crosslinked elastomer and research of its
structure and properties. Beijing University of Chemical Technology, Doctoral Dissertation.
Jia, X.Y. & Huang, B.S. 2014. Infrared spectra and rheological properties of asphalt cement containing
waste engine oil residues. Construction and Building Materials, 50: 683–691.
Liu, H.Y. & Chen, Z.J. 2014. Investigation of the rheological modification mechanism of crumb rubber
modified asphalt (CRMA) containing TOR additive. Construction and Building Materials, 67 (Part
B): 225–233.
Pang, L. & Liu, K.Y. 2014. Effect of LDHs on the aging resistance of crumb rubber modified asphalt.
Construction and Building Materials, 67: 239–243.
Siddiqui, M.N. & Ali, M.F. 1999. Studies on the aging behavior of the Arabian asphalts. Fuel, 78(9):
1005–1015.
Wang, X.F. & Cao, R.J. 2011. Rubber asphalt modification mechanism. Journal of Chang’an University
(Natural Science Edition), 31(2): 6–11.
Xiao, P. & Wang, Y.Q. 2014. Linear analysis and comprehensive evaluation of rubber asphalt ageing
characteristics. Highway, 2014(11): 170–174.
Zhang, G.Y. 2013. Crosslinking and crystallization of natural rubber studied by soild-state NMR spec-
troscopy. East China Normal University, Master’s Degree Dissertation.
275
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Tao Ma
School of Transportation, Southeast University, Nanjing, China
Xiaoming Huang
School of Transportation, Southeast University, Nanjing, China
ABSTRACT: Crumb rubber asphalt with good stability, thus named as stable crumb rub-
ber asphalt, was produced in plant based on wet process. Both conventional performance test
and Superpave performance test were conducted to evaluate the performance characteristics
of the stable crumb rubber asphalt. Based on Marshall mix design, SMA mixtures with stable
crumb rubber asphalt and SBS modified asphalt were designed for performance evaluation.
The results show that, SMA with stable crumb rubber asphalt has better high-temperature
performance and similar low-temperature performance and moisture stability as compared
to SMA with SBS modified asphalt. The SMA with stable crumb rubber asphalt shows better
fatigue life at low strain condition and similar fatigue life at high stain condition as compared
to SMA with SBS modified asphalt. The findings in this study indicate that the stable crumb
rubber asphalt provides a promising replacement for SBS modified asphalt used for SMA.
1 INTRODUCTION
The disposal of scrap tires has been a serious issue due to the lack of landfill space and envi-
ronmental concerns (Sienkiewicz et al. 2012). On the other hand, it was found that crumb
rubber from scrap tires can be used in asphalt mixture, producing asphalt rubber binder or
replacing part of the fine aggregates in asphalt mixtures. And the use of Crumb Rubber
Modified (CRM) binder has been of interest to the asphalt industry since 1980s (Wang et al.
2012). It has been reported that the asphalt industry can recycle up to 40% scrap tires if the
CRM binder is widely used (Júnior et al. 2012).
Currently, there are two different processes, the wet or dry process, to apply crumb rubber in
CRM asphalt mixtures. In the dry process, crumb rubber is added to the aggregate before the
asphalt binder is charged into the mixture. In the wet process, asphalt cement is pre-blended
with the rubber at high temperatures and specific blending conditions. However, the wet proc-
ess is considered as the more efficient way to improve the performance of asphalt binders and
mixtures (Zhang et al. 2014.). According to the American Society for Testing and Materials
(ASTM), crumb rubber modified (CRM) asphalt (also named as asphalt rubber) is defined
as modified asphalt composed of virgin asphalt and no less than 15% crumb rubber by the
weight of virgin asphalt. And researches proved that 15∼25% of crumb rubber by the weight
of virgin asphalt can produce asphalt rubber with satisfied properties (Wang et al. 2012, Nav-
arro et al. 2004, Tortum et al. 2005). In recent years, the use of CRM asphalt mixtures has
been increasing steadily. Previous studies have found that CRM asphalt binder could produce
asphalt pavements with lower traffic noise, improved pavement performance, and thus reduced
maintenance costs (Xiao et al. 2009, Wang et al. 2013, Xiao et al. 2007, Moreno et al. 2013).
Previous studies have found that although complicated chemical and physical reactions
occur during mixing of crumb rubber and asphalt, physical reaction shown as light compo-
277
nents of asphalt absorbed by crumb rubber and swelling of crumb rubber is predominant
(Gao et al. 2010, Frantzis et al. 2004, Zhang et al. 2015, Jong et al. 2001). Although thermal
dissociation of crumb rubber also happens, most of the crumb rubber particles still stay in
asphalt as solid particles forming a net structure in asphalt. Due to the composition change
of asphalt and the existence of crumb rubber particles, the performance of base asphalt
can be well improved. However, the microstructure of asphalt rubber is unstable at hot melt
condition leading to easy separation at storage of asphalt rubber. Therefore, it is difficult to
produce and store asphalt rubber like virgin asphalt or other polymer modified asphalt. Usu-
ally it needs special equipment to produce asphalt rubber right before production of asphalt
rubber mixture that results in difficulties for field construction and quality control. It is one
of the barriers preventing the wide application of asphalt rubber technology.
This study aims to characterize the performance of stable crumb rubber asphalt and Stone
Matrix Asphalt mixture (SMA) with stable crumb rubber asphalt.
2.1 Materials
The stable Crumb Rubber (CR) asphalt was produced in plant based on wet process and
the general producing process of stable crumb rubber asphalt was shown in Fig. 1. The base
asphalt with penetration grade of 70 was used to prepare the crumb rubber asphalt and
the rubber content of was 20% by the weight of base asphalt. Compare to the common
crumb rubber asphalt, the production of stable crumb rubber used 80∼100 mesh crumb rub-
ber instead of 40 mesh crumb rubber and the combined process of reaction kettle and colloid
mill instead of the high-speed stirring process. Besides, stabilizing agents were important to
produce the stable crumb rubber asphalt. To conduct comparative analysis, the SBS modified
asphalt with penetration grade of 50 was also used in this study. The separation test (JTG
E20, 2011) results for the stable crumb rubber asphalt, common crumb rubber asphalt and
SBS modified asphalt are shown in Table 1. It can be seen that, the storage stability of stable
crumb rubber asphalt is worse than the SBS modified asphalt but much better than the com-
mon crumb rubber asphalt.
278
Table 2. Volumetric parameters of different asphalt mixtures.
Both the stable crumb rubber asphalt and SBS modified asphalt were used to prepare
stone matrix asphalt mixtures with nominal maximum aggregate size of 13.2 mm (SMA13).
Marshall mix design method based on Chinese technical specification (JTG F40, 2004)
was used to determine the gradation and asphalt content of SMA13 with different asphalt.
Fig. 2 shows the design gradation for SMA13. Since the stable crumb rubber asphalt has
much higher viscosity than the SBS modified asphalt, 0.3% wood fiber by the weight of
aggregates were used during mix design for SMA13 with SBS modified asphalt while no
wood fiber was used during the mix design for SMA13 with stable crumb rubber asphalt. The
volumetric parameters of SMA13 with both SBS asphalt and stable crumb rubber asphalt
are shown in Table 2.
Figure 3. Graphs for different asphalt: (a) base asphalt, (b) SBS modified asphalt, (c) stable crumb
rubber asphalt.
280
3 RESULTS AND DISCUSSION
Figure 4. SEM graphs for (a) base asphalt and (b) SBS modified asphalt.
281
Figure 5. SEM graphs for (a) stable crumb rubber asphalt, (b) original crumb rubber, and (c) reclaimed
crumb rubber from asphalt.
282
Figure 6. Conventional performance testing results for different asphalt: (a) penetration at 25 °C,
(b) softening point, (c) ductility at 5 °C, (d) elastic recovery, (e) viscosity at 135 °C and (f) viscosity at
177 °C.
and low temperature performance as indicated by the higher G*/sinδ values and the lower
creep stiffness (S) and creep rate (m-value) at the same testing temperature. It indicated
that the performance grade for the base asphalt is PG64–22 while the performance grade
for the stable crumb rubber asphalt and the SBS modified asphalt are PG 88–28. Further
analysis showed that, although the performance grade of the stable crumb rubber asphalt
is the same with SBS modified asphalt, the stable crumb rubber asphalt has obviously
higher G*/sinδ value than the SBS modified asphalt at the same temperature, which means
the stable crumb rubber asphalt has better performance at high temperature than the SBS
modified asphalt.
283
Figure 7. Superpave testing results for different asphalt: (a) G*/sinδ of origninal asphalt, (b) G*/sinδ
after RTFO aging, (c) S after RTFO and PAV aging, and (d) m-value after RTFOT and PAV aging.
284
Figure 8. Fatigue test results for different asphalt mixtures.
modified asphalt which means that SMA13 with stable crumb rubber asphalt has better high
temperature stability than SMA13 with SBS modified asphalt. It confirms well with the previ-
ous analysis of asphalt performance. From Fig. 8, it can be seen that, compared to SMA13
with SBS modified asphalt, SMA13 with stable crumb rubber asphalt has similar anti-fatigue
performance at high test strain while better anti-fatigue performance at low test strain.
4 CONCLUSIONS
This study investigated the performance of the stable crumb rubber asphalt as compared to
the SBS modified asphalt. The following conclusions were concluded from the analysis:
1. The microstructure of stable crumb rubber asphalt is different with SBS modified asphalt.
Generally, the dissolution of crumb rubber into base asphalt and the interaction between
crumb rubber and base asphalt are weaker than SBS which leads to worse storage stability
of crumb rubber asphalt than SBS modified asphalt. However, with the crumb rubber dis-
solving into and interacting with base asphalt, the stable crumb rubber asphalt has better
storage stability than common crumb rubber asphalt.
2. Conventional performance tests show that stable crumb rubber asphalt has smaller ductil-
ity and much higher viscosity than SBS modified asphalt while Superpave performance
tests show that stable crumb rubber has the same performance grade with SBS modified
asphalt. Overall, compared to SBS modified asphalt, stable crumb rubber asphalt has
similar low-temperature performance but better high-temperature performance.
3. Mixture performance tests indicate that the low-temperature performance and moisture
stability of SMA13 with stable crumb rubber asphalt are similar to that of SMA13 with
SBS modified asphalt while the high-temperature performance and anti-fatigue perform-
ance of SMA13 with stable crumb rubber asphalt are better than that of SMA13 with SBS
modified asphalt.
4. The performance difference in the desulfurized rubber asphalt mixture and the crumb rub-
ber asphalt mixture is due to the combined effects of asphalt binder type and content and
aggregate gradation. However, compared to the crumb rubber asphalt mixture, the com-
paction temperature of desulfurized rubber asphalt mixture can be lowered by 10 to 20 °C
to achieve the same performance requirements and volumetric design parameters. Both
285
the desulfurized rubber asphalt mixture and the crumb rubber asphalt mixture require the
higher compaction temperature than the SBS asphalt mixture. Thus, the stable crumb rub-
ber asphalt provides a promising replacement for SBS modified asphalt used for SMA.
Further study will be conducted on the performance mechanism of stable crumb rubber
asphalt and its application in different types of asphalt mixtures.
ACKNOWLEDGEMENTS
The study is financially supported by the National Natural Science Foundation of China
(No. 51378006), National Science and Technology Support Program (2014BAG05B04),
Huoyingdong Foundation of the Ministry of Education of China (No. 141076), Excellent
Young Teacher Program of Southeast University (2242015R30027), and State Key Labora-
tory of High Performance Civil Engineering Materials (2014CEM008).
REFERENCES
ASTM D6648–08, Standard Test Method for Determining the Flexural Creep Stiffness of Asphalt Binder
Using the Bending Beam Rheometer (BBR), ASTM International, West Conshohocken, PA, 2008.
ASTM D7175–15 Standard test method for determining the rheological properties of asphalt binder
using a dynamic shear rheometer, ASTM International, West Conshohocken, 2015.
Frantzis, P., et al. 2004. Crumb rubber-bitumen interactions: diffusion of bitumen into rubber. Journal
of Materials in Civil Engineering 16: 387–390.
Gao, Y., et al. 2010. Interaction theory of asphalt and rubber. Journal of Wuhan University of Technol-
ogy-Materials Science Edition 25: 853–855.
JTG E20, Standard Test Methods of Bitumen and Bituminous Mixtures for Highway Engineering,
Ministry of Transport of the People’s Republic of China, Beijing, 2011.
JTG F40, Technical Specification for Construction of Highway Asphalt Pavement, Ministry of Trans-
port of the People’s Republic of China, Beijing, 2004.
Jong, R.K., et al. 2001. Characteristics of crumb rubber modified (CRM) asphalt concrete. KSCE Jour-
nal of Civil Engineering 5: 157–164.
Júnior, A.F., et al. 2012. Use of scrap tire rubber in place of SBS in modified asphalt as an environmen-
tally correct alternative for Brazil. Journal of Cleaner Production 33:236–238.
Moreno, F., et al. 2013. The effect of crumb rubber modifier on the resistance of asphalt mixes to plastic
deformation. Materials and Design 47: 274–280.
Navarro, F.J., et al. 2004. Thermo-rheological behaviour and storage stability of ground tire rubber-
modified bitumens. Fuel. 83: 2041–2049.
Sienkiewicz, M., et al. 2012. Progress in used tyres management in the European Union: a review. Waste
Management 32: 1742–1751.
Tortum, A., et al. 2005. Determination of the optimum conditions for tire rubber in Asphalt concrete.
Building and Environment 40: 1492–1504.
Wang, H.N., et al. 2012. Laboratory evaluation on high temperature viscosity and low temperature
stiffness of asphalt binder with high percent scrap tire rubber. Construction and Building Materials
26: 583–590.
Wang, H.N., et al. 2012. Laboratory evaluation on high temperature viscosity and low temperature
stiffness of asphalt binder with high percent scrap tire rubber. Construction and Building Materials
26: 583–590.
Wang, H.N., et al. 2013. Analysis on fatigue crack growth laws for crumb rubber modified (CRM)
asphalt mixture. Construction and Building Materials 47:1342–1349.
Xiao, F.P., et al. 2007. Rutting resistance of rubberized asphalt concrete pavements containing reclaimed
asphalt pavement mixtures. Journal of Materials in Civil Engineering 19: 475–483.
Xiao, F.P., et al. 2009. Fatigue behavior of rubberized asphalt concrete mixtures containing warm
asphalt additives. Construction and Building Materials 23: 3144–3151.
Zhang, D., et al. 2014. Rubberized asphalt mixture design using a theoretical model. Construction and
Building Materials 67: 1028–1034.
Zhang, F., et al. 2015. The research for structural characteristics and modification mechanism of crumb
rubber compound modified asphalts. Construction and Building Materials 76: 330–342.
286
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The objective of this study is to simulate the asphalt binder using the Molec-
ular Dynamics (MD) method, and the density and thermal conductivity of the asphalt model
are calculated and evaluated. The asphalt model consists of three components: asphaltenes,
aromatics, and saturates. The Dreiding Force Field and Amber Cornell Extension Force
Field (ACEFF) were assigned to the model. The methods of steepest descent and conjugated
gradient were used to optimize the energy of the MD system, and the Savitzky-Golay filter
was employed to smooth the data. The model was built through a relatively low density sys-
tem, and the NPT simulation was employed to equilibrate the MD model. The density of the
model was computed after stabilizing system. The MD results show that the density of the
asphalt model with different force fields differs and the density of the model with ACEFF is
relatively close to that of the real asphalt.
1 INTRODUCTION
Asphalt material is the byproduct, non-dissolution matters of the petroleum refinement. The
complicated composition of the asphalt determine the properties of asphalt, and the proper-
ties vary from different sources (Yao et al., 2015, Yao et al., 2013). Different numerical meth-
ods were used to simulate the asphalt, such as the Discrete Element Method (DEM) (Yao
et al., 2012) and Finite Element Method (FEM) (Dai and You, 2007), and most methods
focused on the macro scale for material simulation. The different constitutive models were
used to simulate the properties of asphalt and asphalt mixtures. However, this Molecular
Dynamics (MD) simulation is based on the atomic scale, and the timestep is a femtosecond.
Molecular Dynamics (MD) is a computer simulation to allow molecules or atoms of mate-
rials to move at a certain time frame based on the physical principles of atoms and molecules.
The atoms and molecules interact with each other during this time period. The trajectories
of atoms and molecules are recorded based on Newton’s law, and the energies of systems are
defined and calculated by the force fields. There are five kinds of simulations to control the
MD systems, such as Microcanonical ensemble (NVE ensemble), Canonical ensemble (NVT
ensemble), Isothermal-isobaric ensemble (NPT ensemble), Isoenthalpic-Isobaric ensemble
(NPH ensemble), and Generalized ensembles. The different ensembles control different con-
ditions in the MD systems to fulfill different requirements for simulations. In addition, dif-
ferent mathematical optimization methods can be used for energy optimization in the MD
system, and this is also another way to optimize the system.
The objective of this study is to model the asphalt material using the Molecular Dynamics
(MD) with different force fields. Different components of the asphalt models were used to
represent asphaltenes, aromatics, and saturates. Different optimization methods were also
287
Figure 1. The structures of components in the MD asphalt model: (a) structure of the asphaltene in
the asphalt model: C–C symbol connection shows the double bond of carbon atoms; No C–C symbol
connection represents a single bond of carbon atoms; (b) structure of 1, 7-dimethylnaphtalene in the
asphalt model, it represents the naphthene aromatic; (c) structure of docosane in the asphalt model,
represents saturate.
used for different asphalt models with force fields. After NPT simulation in the asphalt mod-
els, the energy, pressure and temperature of the MD systems were recorded, and the densities
of the asphalt models with force fields were compared and analyzed.
3 MODEL GENERATION
The three components, asphaltenes, aromatics, and saturates, were used in this MD asphalt
model based on the Corbett Method. The asphalt can be separated to four components
(asphaltenes, saturates, naphthene aromatics, and polar aromatic) by the Corbett Method
(Asphalt Institute, 2003). The 1,7-dimethylnaphtalene and docosane were adopted to rep-
resent aromatics and saturates, respectively (Zhang and Greenfield, 2007). The ratio of
asphaltenes, aromatics, and saturates in the asphalt model is 5:27:41. The chemical structures
of these components are shown in Figure 1.
4 DENSITY CALCULATIONS
The two asphalt models were built using different force fields, the Amber Cornell Extension
Force Field (ACEFF) and Dreiding Force Field (Figure 2a, visualized by Ovito (Alexander,
2010)). The asphalt model with ACEFF was optimized by the conjugate gradient method,
and the asphalt model with the Dreiding Force Field was optimized by the method of steep-
est descent. The two asphalt models were built with low density, and the NPT ensemble
(Isothermal-isobaric ensemble) was used to compress the models at room temperature and
1 atm pressure (Nose-Hoover thermostat and barostat). In this MD simulation, a 700 ps
simulation time was used to maintain a stable energy system, and the data was recorded every
100 fs. The results of the MD simulation are shown in Figure 2.
288
Figure 2. Energy, pressure (atm) and temperature results of the asphalt model during the NPT simula-
tion: (a) the asphalt model with low density (0.1 g/cm3) and six types of atoms in the asphalt model (CA,
CT, HC, HA, SS, and CC with different colors); (b) system energies during NPT (Dreiding Force Field);
(c) pressures in the asphalt model (Dreiding Force Field); (d) temperatures in the asphalt model (Drei-
ding Force Field). Atom types used in the asphalt model, CA type is sp2 aromatic carbon in 6-mem-
bered ring with 1 substituent; CC type is sp2 aromatic carbon in a 5-membered ring with 1 substituent
and next to a nitrogen; CT type is aliphatic carbon (tetrahedral); HC type is hydrogen bonding hydrogen
(charged group); HA type is aliphatic or aromatic hydrogen; and S type is sulfur in disulfide linkage or
methionine.
Figure 2a shows the initial state of the asphalt model with low density (0.1 g/cm3). During
this NPT simulation of the asphalt model, the energies (Figure 2b), pressures (Figure 2c),
and temperatures (Figure 2d) of the systems were recorded. From Figure 2b, the energies
were optimized and decreased significantly, reaching a stable level after 100 ps. A relatively
stable asphalt model was produced. Figure 2c and Figure 2d demonstrate 1 atm pressure and
a 298.15 K temperature in the systems after smoothing the data. The data fluctuated due to
small molecules (Yao et al., 2016), considering around 1024 molecules per mole in materials,
and the moving average of the data was optimized by the Savitzky-Golay filter with a span
of 10%. The densities of the asphalt models with different force fields were calculated and
are shown in Figure 3.
Figure 3 demonstrates the densities of the asphalt model with different force fields under
the NPT ensemble. From Figure 3, the densities in different asphalt models increase rapidly
and reach a stable range within 100 ps, and different densities of the asphalt models with
different force fields are observed. It indicates that 1) the MD calculation speed is fast with
289
Figure 3. Densities of asphalt model during the NPT simulation.
a few thousand atoms; 2) different energy calculations in the force field significantly affect
the density results, as well as other properties. It is important that the appropriate force field
and parameters are used in the MD systems, and it is also a determinant for calculating
the properties using different algorithms. The density of the asphalt model with the Amber
Cornell Extension Force Field is around 0.91 g/cm3, and the density of the asphalt model
with the Dreiding Force Field is around 0.81 g/cm3 at the conditions of 1 atm pressure and
room temperature. In the laboratory test, the densities of asphalt binders from different
sources is around 0.95 g/cm3 to 1.05 g/cm3 at the conditions of 1 atm pressure and room
temperature. There is still a minor difference between the density of the MD simulation and
the laboratory data. It is obvious that the density of the asphalt model with the Amber Cor-
nell Extension Force Field is better than that of the asphalt model with the Dreiding Force
Field. It can be deduced that different materials may be suitable for different force fields in
MD simulation.
5 CONCLUSIONS
In the MD study, the asphalt model was composed of three components: asphaltenes, aromat-
ics, and saturates. The 1, 7-dimethylnaphtalene and docosane (n-C22) were used to represent
aromatics and saturates, respectively. The asphalt models with different force fields, Amber
Cornell Extension Force Field and Dreiding Force Field, were optimized by the conjugate
gradient method and method of steepest descent, respectively. The asphalt models were gen-
erated with a low density, and the NPT ensemble was used to control the temperature and
pressure of the systems. The densities were calculated after the MD simulation. Different
densities were observed in the different asphalt models using different force fields. During
the NPT simulation, it is verified that the temperature and pressure were constant with the
desired conditions, and the variations were also observed. The system energies in different
asphalt models were maintained at a stable level after MD simulation. Therefore, the differ-
290
ent definitions and calculations of energies in force fields have significant impacts on the
properties of MD systems. In addition, the selection and optimization of interaction param-
eters between atoms is determinant for the predictions of the properties of the MD systems.
ACKNOWLEDGEMENTS
The authors appreciate Dr. Andreas Bick’s valuable suggestions and help on the generation
of the asphalt model and calculation of the properties. Any opinion, finding, and conclusion
expressed in this paper are those of the authors and do not necessarily represent the view of
any organization.
REFERENCES
Alexander, S. 2010. Visualization and analysis of atomistic simulation data with OVITO–the Open Visu-
alization Tool. Modelling and Simulation in Materials Science and Engineering, 18, 015012.
Asphalt Institute 2003. Superpave Performance Graded Asphalt Binder Specification and Testing.
Superpave Series No 1(SP-1), Lexington KY, USA.
Dai, Q. & You, Z. 2007. Prediction of Creep Stiffness of Asphalt Mixture with Micromechanical Finite-
Element and Discrete-Element Models. Journal of Engineering Mechanics, 133, 163–173.
Yao, H., Dai, Q. & You, Z. 2015. Chemo-physical analysis and Molecular Dynamics (MD) simulation
of moisture susceptibility of nano hydrated lime modified asphalt mixtures. Construction and Build-
ing Materials, 101, Part 1, 536–547.
Yao, H., Dai, Q. & You, Z. 2016. Molecular dynamics simulation of physicochemical properties of the
asphalt model. Fuel, 164, 83–93.
Yao, H., Liu, Y., You, Z., Li, L. & Goh, S.W. 2012. Discrete Element Simulation of Bending Beam
Rheometer Tests for Asphalt Binder. International Journal of Pavement Research and Technology, 5,
161–168.
Yao, H., You, Z., Li, L., Goh, S.W., Lee, C.H., Yap, Y.K. & Shi, X. 2013. Rheological properties and
chemical analysis of nanoclay and carbon microfiber modified asphalt with Fourier transform infra-
red spectroscopy. Construction and Building Materials, 38, 327–337.
Zhang, L. & Greenfield, M.L. 2007. Analyzing Properties of Model Asphalts Using Molecular Simula-
tion. Energy & Fuels, 21, 1712–1716.
291
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: In the present study four different binders are tested on Dynamic Shear
Rheometer (DSR) to evaluate the effect of spindle diameter and plate gap on the rheologi-
cal properties of asphalt binders. Frequency sweep tests are conducted on the binders from
10–70 ºC by varying the spindle diameter and gap between the plates. It was found that the
change in rheological properties with variation in spindle geometry and plate gap is more
pronounced at higher temperatures and lower frequencies. These deviations in the properties
with change in spindle attributes are found to be a function of stiffness of asphalt binder.
It is also established from the study that the 25 mm diameter spindle geometry gives better
and consistent results as compared to the 8 mm spindle geometry for the evaluation of the
rheological properties of different asphalt binders.
1 INTRODUCTION
Dynamic Shear Rheometer (DSR) is one of the most widely used instrument to study the
rheological properties of the asphalt binder. With the help of rheometer different rheological
properties such as complex modulus, phase angle, creep compliance etc. are studied. In DSR
two different spindle of diameters 8 mm and 25 mm are used. 8 mm diameter spindle is used
to test samples at lower temperatures i.e. 0–30 ºC whereas 25 mm diameter spindle is used
for higher temperatures i.e. 40–90 ºC. American Association of State Highway and Trans-
portation Officials (AASHTO) recommends 2 mm gap between parallel for 8 mm spindle
and 1 mm gap for 25 mm diameter spindle which is based on the extensive experiment and
research results. Strategic Highway Research Program (SHRP) suggests the following spindle
diameters for different conditions and binder stiffness:
From the Table 1 it is clear that the spindle diameter is based on the complex modulus
(G*) values of the binders which is related to the load (Shear) application capability of the
instrument.
DSR measures two basic parameters i.e. angular rotation and applied torque. All the oth-
ers parameters are calculated using different formulae based on these parameters. For the
calculation of stress and strain following equations are used:
2T
τ= (1)
π .r3
293
θ .r
γ = (2)
h
τ Gγ (3)
2 MATERIALS
Two virgin bitumen i.e. VG 10 and VG 30 and two polymer modified bitumen i.e. PMB (S)
and PMB (E) are used in the study. PMB (S) is Styrene-butadiene-styrene (SBS) modified
bitumen whereas PMB (E) is Ethylene-vinyl-acetate (EVA) modified. Both the modified bitu-
men were prepared using virgin VG 10 binder.
3 EXPERIMENTAL PROCEDURE
Frequency sweep test using DSR was performed on all four binders. All the tests for all the
binders were performed under linear viscoelastic (LVE) limits. Mode of loading was controlled
strain. Tests were performed between 10–70 ºC temperatures with the increment of 10 ºC. Fre-
quencies were varied from 0.1–100 radians/sec. In order to examine the effect of spindle geom-
etry and gap on the rheological properties measurements of bitumen different combinations of
spindle diameter and gap were used. The following combinations were adopted for the study:
• 25 mm spindle diameter and 1 mm gap
• 25 mm spindle diameter and 2 mm gap
• 25 mm spindle diameter and 3 mm gap
• 8 mm spindle diameter and 1 mm gap
• 8 mm spindle diameter and 2 mm gap
Based on the test results most appropriate combination was also evaluated.
The results of frequency sweep tests are analyzed in three categories, effect of spindle diameter,
effect of plate gap and effect of binder to assess the effect of these three different factors.
lower contact area of spindle with bitumen which at higher temperature tend to delaminate
from the spindle. This behavior was more pronounced in virgin binders which may be due to
lower viscosity of conventional binder as compared to modified binders. This change should
not be confused with the thermorheological complexity of the binders which also forms
‘wavy’ nature of curves. Hence, deviations in rheological results may also result from choice
of spindle geometry rather than the type and properties of binders.
Complex modulus curves show different behavior than phase angle curves. The erroneous
behavior shown by phase angle curves at higher temperatures and lower frequencies were
absent in complex modulus curves. At higher temperatures 8 mm spindle gave higher values
but this gap in values reduced with reduction in temperature. At intermediate temperatures
both the spindles gave similar results. At lower temperatures behavior reversed, 8 mm spindle
gave lower values as compared to 25 mm spindle. The results were consistent for all the bind-
ers. Figure 3 (a-f) and Figure 4 (a-f) shows the complex modulus curves for the VG-10 and
PMB (S) binder respectively.
296
Figure 4. Complex modulus curves for PMB (S).
Figure 5. Phase angle curves for VG 10 for 25 mm plate diameter. (Note: 25, 3; 25, 2; and 25, 1 repre-
sents spindle diameter of 25 mm with 1 mm, 2 mm and 3 mm plate gap).
297
Figure 6. Phase angle curves for PMB (S) for 25 mm plate diameter.
298
Figure 8. Complex modulus curves for PMB (S) for 25 mm plate diameter.
temperatures the difference in values of phase angles of were more significant. All the bind-
ers showed similar behavior.
In case of complex modulus at higher temperatures higher gap gave lesser values. The
difference in values was more at higher frequencies. At intermediate temperatures this gap
reduced and all the gap gave approximately similar values for all the binders. As the tempera-
ture further decreased i.e. at lower temperatures, higher gap gave higher values of complex
modulus. As in the case of phase angle curves, difference in the values of complex modulus
with respect to different gaps were more significant at higher frequencies and lower tempera-
tures.. In terms of viscoelastic nature of the asphalt binders it can be said that higher plate
gap gave higher elastic behavior at higher temperatures, while at lower temperatures smaller
gap width displayed stronger elastic nature. Similar behavior was seen in all the binders.
Figure 5 (a-f) to Figure 8 (a-f) presents the effect of plate gap on the measurements of phase
angle and complex modulus from 10–70 °C for VG-10 and PMB (S) binders.
5 CONCLUSIONS
The present study was undertaken to evaluate the effect of spindle diameter and plate gap on
the evaluation of rheological properties of asphalt binders. The following conclusions can be
drawn from the results and discussions presented:
• 8 mm spindle gives erroneous results of phase angle at higher temperatures. This may be
caused due to the lesser surface area of bitumen between the plates. Due to lesser surface
area bitumen tends to delaminate which causes irregularities in the results.
• At higher temperatures 8 mm spindle gives higher values whereas at lower temperatures
25 mm gave higher values of complex modulus. At intermediate temperatures both the
spindles gave similar results.
299
• Higher gap gives lower values of phase angle at higher temperatures while this behavior
changes at lower temperature where it gives lower values.
• In case of complex modulus higher gap gives lesser values at higher temperature and higher
values at lower temperatures.
• At intermediate temperatures spindle geometry and plate gap doesn’t affect the rheological
properties much.
• All the binders shown similar behavior so the effect of gap and spindle geometry is not
binder specific.
REFERENCES
AASHTO.1994. Standard Method of Test for Determining the Rheological Properties of Asphalt
Binder Using a Dynamic Shear Rheometer (DSR), TP–5.
D.A. Anderson, D. Christensen, H.U. Bahia, R. Dongre, M.G. Sharma, C.E. Antle, 1994. SHRP-A-
369: Binder Characterization and Evaluation-Physical Characterization, in: Strateg. Highw. Res.
Progr., Washington, DC.
G.D. Airey, Use of Black Diagrams to Identify Inconsistencies in Rheological Data, Road Mater. Pave-
ment Des. 3 (2002) 403–424. doi:10.1080/14680629.2002.9689933.
G.D. Airey, B. Rahimzadeh, A.C. Collop, Linear Rheological Behavior of Bituminous Paving Materials,
J. Mater. Civ. Eng. 16 (2004) 212–220. doi: 10.1061/(ASCE) 0899–1561 (2004) 16:3 (212).
N. Saboo, P. Kumar, Study of flow behavior for predicting mixing temperature of bitumen, Constr.
Build. Mater. 87 (2015) 38–44. doi:10.1016/j.conbuildmat.2015.04.001.
N.I.M. Yusoff, F.M. Jakarni, V.H. Nguyen, M.R. Hainin, G.D. Airey, Modelling the rheological proper-
ties of bituminous binders using mathematical equations, Constr. Build. Mater. 40 (2013) 174–188.
doi:10.1016/j.conbuildmat.2012.09.105.
300
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Xanthan gum were used to mix with base bitumen at 150°C. The influence of
xanthan gum on the aging properties of bitumen was investigated by the Thin Film Oven Test
(TFOT) and the Pressure Aging Vessel (PAV) test. 0.5wt.% xanthan gum modified bitumen
was selected to compare to the base bitumen with the Fourier Transform Infrared Spectros-
copy (FTIR) measurements. The results indicated that the xanthan gum modified bitumens
have a higher complex modulus and lower phase angle than base bitumen at a low content.
Meanwhile, the total strain and permanent strain of xanthan gum modified bitumens were
increased compared to base bitumen at 0°C. Besides, the results of the FTIR test showed that
the xanthan gum modified bitumen has a better aging resistance than base bitumen. How-
ever, the results also implied that the extra xanthan gum did not contribute to the improve-
ment of the bitumen’s rheological properties and aging resistance at the same time.
1 INTRODUCTION
Bitumen is manufactured from crude oil in refineries, which is the most suitable material as
a binder of mineral aggregates in paving applications (Garcia-Morales et al. 2006; Liu et al.
2011). Due to its visco-elasticity, impermeability and cost, it has been a preferential choice
in the asphalt pavement construction all over the world (Petersen 1984; Wu et al. 2012; Yu
et al. 2009).
Due to it is soft in the hot environment and brittle in cold weather, the base bitumen
does not have good engineering properties (Chen et al. 2002). Therefore, many different
types of materials have been attempted as additives to modify the bitumen with the purpose
to improve its rheological and mechanical properties. Thus, it is not a new phenomenon
that many polymers such as Styrene-Butadiene-Styrene (SBS), Styrene-Butadiene-Rubber
(SBR) and Ethylene Vinyl Acetate (EVA) were used as modified binders intended for road
application (Isacsson and Lu 1995; Palaniraj and Jayaraman 2011). However, these types of
modification can increase the construction cost of the asphalt pavement. By the way, it is
necessary to mention that the mechanisms of action for most of the polymer modifiers are
insufficiently understood.
Xanthan gum is an anionic polysaccharide and an important industrial biopolymer (Isacs-
son and Lu 1995; Milas et al. 1990), which is produced by the aerobic fermentation of the
bacterium Xanthomonas campestris (Garcıa-Ochoa et al. 2000; Psomas et al. 2007). Fig-
ure 1 shows the primary molecular structure of xanthan gum, which is a linear backbone of
(1→4) linked β-D-glucose with a trisaccharide side chain on every other glucose at the C(3)
position (Becker et al. 1998; Casas et al. 2000; Jansson et al. 1975).
As a type of biopolymer, the xanthan gum is low-cost and environmentally friendly. Due
to its special structure, xanthan gum displays excellent rheological properties, such as pseu-
doplastic properties, high viscosity at low shear forces and may also display a viscosity yield
value (Kalogiannis et al. 2003). What’s more, it can be relatively stable over a wide range of
301
Figure 1. The molecular structure diagram of Xanthan gum.
temperatures and PH values (Katzbauer 1998). These characteristics make it ideal to be used
as thickening and stabilizing agent for emulsions and suspensions, especially in the food,
cosmetics and textile industry (Rosalam and England 2006). The side chains of xanthan gum
can form a large number of hydrogen bonds in the helix structure, which result in a resistance
to flow in aqueous solution (Fukuda et al. 2006; Song et al. 2006). Therefore, it can also be
used in the building materials to reduce the sedimentation and bleeding of the cement mortar
and grout (Plank 2004).
The main objective of this article is to investigate the related properties of the xanthan
gum modified bitumen binder. In this study, the influence of xanthan gum with differ-
ent contents on the dynamic rheological properties and aging properties of bitumen was
evaluated.
2 EXPERIMENTAL
2.1 Materials
2.1.1 Bitumen
The base bitumen (AH-90) was used and provided by Panjin Northern Asphalt Co., Ltd. Its
physical properties are given in Table 1.
2.2 Methods
2.2.1 Preparation of Xanthan gum modified bitumens
All the modified bitumens were prepared by using a high shear mixer. First, 400 g of base
bitumen was heated until it becomes well melting fluid at around 150°C and poured into a
1000 ml cylindrical container. Then, four contents (0.2%, 0.5%, 1% and 3%) of xanthan gum
by the weight of base bitumen were chosen, and the gum powders were gradually added into
the bitumen and the mixture was blended at 4000r/min rotation for 30 minutes to ensure the
uniform dispersion of xanthan gum. The base bitumen was also processed under the same
conditions for comparison.
302
Table1. Physical properties of the base bitumen.
Figure 2. Simultaneous thermogravimetric (TG) and differential scanning calorimetry (DSC) curves
of xanthan gum, conducted by a thermal analyzer equipment NETZSCH STA 449F3 using an atmos-
pheric air stream of 10 ml/min, while heating from room temperature to 1000°C involved a linear tem-
perature increase of 10°C/min.
Figure 3. Configuration of the cone and plate device (a) and parallel plates device (b) in the rheometer
(Liu 2011).
304
Area of the carbonyl band around 1700cm −1
IC=0 =
Area of the spectra
s l bands between 2000 and 600cm −1 (1)
305
Figure 4. Curves of complex modulus and phase angle as a function of temperature for the bitumens
modified with different contents of xanthan gum.
Figure 5. Creep strain curve of bitumens with different xanthan gum concentrations at 0°C.
Table 4. Total strain and permanent strain for bitumens with different xanthan gum concentrations.
aging, especially after PAV aging. The increase of carbonyl and sulfoxide groups indicates an
increase of large molecules in the asphalt binder, resulting in higher stiffness and more solid-
like behavior (Wu et al. 2010). The reason for this can be that the oxidation, dehydrogenation
and crosslinking reactions occur at the same time during the process of aging, which leads to
the appearance of more carbonyl and sulphoxide groups. Meanwhile, the functional group
sulfoxide, formed in bitumen from the oxidation of sulfide groups, is already present in the
fresh bitumen, indicating that they are probably formed in the stage of preparation.
It also can be seen from figure 6 that the sulphoxide group is more sensitive to the aging,
and its band becomes broader as the ageing proceeds from TFOT to PAV (see right part of
figure 6). With respect to the carbonyl group, figure 6 shows no obvious absorbance peak can
be observed for bitumen without aging and it seems that there is only a slight increase for this
group, especially after TFOT aging. However, the carbonyl group is affected by PAV aging.
Therefore, the sulphoxide group shows a clearer evolution of aging than the carbonyl group.
In this paper, the bitumen modified by 0.5wt.% xanthan gum was selected to compare with
base bitumen to characterize the effect of xanthan gum on the aging properties of asphalt
binder. Figure 7 shows the FTIR spectrum of xanthan gum modified bitumens from fresh
to the PAV aged state. As observed in Figures 6 and 7, the similarity in terms of the location
of peaks and the symmetry of the FTIR results indicate that there was no signification vari-
306
Figure 6. Ageing evolution of base (AH-90) bitumen characterized by the FTIR from fresh to the
PAV aged state.
ation in the functions groups even in any chemical reaction or otherwise occurred when the
xanthan gum was mixed with the base bitumen.
To avoid the variation of binder film thickness on the KBr plate, the spectral normaliza-
tion was performed by bringing the same absorbent series at the same point. As an indicator
of the aging extent, the corresponding functional groups indexes were calculated from equa-
tions (1) and (2) and were summarized in Table 5. In addition, the peak area for 1700 cm−1
was calculated between 1673 and 1727 cm−1, while the peak area for 1030 cm−1 was calculated
between 983 and 1072 cm−1.
Comparing the data of base bitumen with xantan gum modified bitumen, it can be easily
found that the latter’s indexes of carbonyl and sulfoxide groups are smaller than those of
before, which means the modified of bitumen by xanthan gum can lead to the formation of
less carbonyl and sulfoxide groups, indicating that xanthan gum modified bitumens had a
better resistance to the formation of carbonyl and sulfoxide groups hence might getting more
stable. Therefore, the xanthan gum can improve the aging resistance of xanthan gum modi-
fied asphalt binder. This may be explained as follows: duo to the thickening property of xan-
than gum, the viscosity of the modified bitumens increases, which resulting in the decrease of
the mobility of the molecules in the bitumen as well as the diffusion of oxygen. Therefore, the
amount of the carbonyl and sulfoxide functional groups produced is higher for the bitumen
with lower viscosity after going through the aging process.
However, with respect to the index of functional sulfoxide group for the 0.5wt.% xan-
than gum modified bitumen without aging, it is bigger than that of the xanthan gum modi-
fied bitumen after TFOT aging, which maybe lead to a confused conclusion. The reason
for this may lie in the fact that the xanthan gum modified bitumen has been already aged
during the process of sampling.
G*after TFOT
AI1 =
G* virgin
(3)
G*after PAV
AI 2 =
G* virgin
(4)
307
Figure 7. Ageing evolution of xanthan gum modified bitumen characterized by the FTIR from fresh
to the PAV aged state.
Table 5. Group indexes of base and xanthan gum modified bitumens before and after aging.
where, the |G*virgin|, |G*after TFOT| and |G*after PAV| are the complex modulus of the virgin binder,
after TFOT aged bitumen and after PAV aged bitumen, respectively. The higher value of the
aging indexes (AI1, AI2) are, the deeper degree of bitumen aging is.
However, the ratio of complex modulus as an aging index maybe lead to a confused con-
clusion. As indicated in Figure 8 and Figure 9, the trend of AI1 and AI2 curves for all the
bitumens were similar, i.e. that the AI1 and AI2 curves of base bitumen were higher than the
xanthan gum modified bitumens at relatively low concentrations, which suggests that the
base bitumen has a relatively low aging resistance. While when the addition of xanthan gum
was higher than 1.0wt.%, an obvious increase in AI1 and AI2 values was observed, especially
when the concentration up to 3.0wt.%, the AI1 and AI2 curves of xanthan gum modified
bitumens almost higher than that of base bitumen, indicating that the xanthan gum had no
effect on the improvement of base bitumen aging resistance at a concentration higher than
1.0wt.%. This reveals that the xanthan modified bitumens exhibit a better TFOT and PAV
aging resistance at a low concentration, namely the xanthan gum can availably improve the
aging properties of bitumen to a certain degree. On the contrary, the excessive addition of
xanthan gum will be harmful to the aging resistance for bitumen.
4 CONCLUSIONS
The influence of xanthan gum biopolymer on the rheological properties and aging proper-
ties of bitumen was investigated. Based on the results presented in this paper, the following
conclusions were drawn:
1. The xanthan gum modified bitumens show a higher complex modulus and lower phase
angle than base bitumen. However, the extra xanthan gum did not enhance this effect on
complex modulus and phase angle of bitumen.
2. With the addition of xanthan gum, the total strain and permanent strain were increased,
which means the improvement of low temperature properties. On the other hand, too
much content of xanthan gum is not appropriate.
3. The carbonyl and sulfoxide groups were used as the aging indexes to evaluate the effect
of xanthan gum on the aging resistance of asphalt binder. It was found that less carbonyl
and sulfoxide groups appeared in the aged xanthan gum modified bitumen than in the
aged base bitumen, indicating that the xanthan gum can improve the aging resistance of
asphalt binder when xanthan gum was introduced. However, the analysis of the AI curves
308
Figure 8. AI1 curve with different xanthan gumconcentrations as a function of temperature.
shows that the excessive addition of xanthan gum will show a relatively worse aging resist-
ance for bitumen. Therefore, further work will need to be conducted to establish the extent
of the reaction and compatibility between xanthan gum and base bitumen.
ACKNOWLEDGEMENT
The authors are grateful for the financial support from National Key Scientific Apparatus Devel-
opment Program from the Ministry of Science and Technology of China (No. 2013YQ160501),
the National Basic Research Program of China (973 Program (No. 2014CB932014), the Natu-
ral Science Foundation of China (No. 51278392), the Science and Technology Plan Projects
of the Ministry of Transport of China (No.2013 318 811 250), and the International Science
& Technology Cooperation Program of China (No. 2013DFE83100).
REFERENCES
Becker, A., F. Katzen, A. Pühler & L. Ielpi, 1998. Xanthan gum biosynthesis and application: a bio-
chemical/genetic perspective. Applied Microbiology and Biotechnology 50(2):145–152.
Casas, J., V. Santos & F. Garcıa-Ochoa, 2000. Xanthan gum production under several operational
conditions: molecular structure and rheological properties . Enzyme and Microbial Technology
26(2):282–291.
Chamberlain, E. & M. Rao, 1999. Rheological properties of acid converted waxy maize starches in
water and 90% DMSO/10% water. Carbohydrate Polymers 40(4):251–260.
Chen, J.S., M.C. Liao & M.S. Shiah, 2002. Asphalt modified by styrene-butadiene-styrene triblock
copolymer: Morphology and model. Journal of materials in civil engineering 14(3):224–229.
309
Chen, M., B. Leng, S. Wu & Y. Sang, 2014. Physical, chemical and rheological properties of waste edible
vegetable oil rejuvenated asphalt binders. Construction and Building Materials 66:286–298.
Collop, A.C., G.D. Airey & S. Khanzada, 2002. Creep testing of bitumens using the dynamic shear
rheometer. International Journal of Pavement Engineering 3(2):107–116.
Fukuda, M., N.A. Peppas & J.W. McGinity, 2006. Properties of sustained release hot-melt extruded tab-
lets containing chitosan and xanthan gum. International journal of pharmaceutics 310(1):90–100.
Garcia-Morales, M., P. Partal, F. Navarro & C. Gallegos, 2006. Effect of waste polymer addition on the
rheology of modified bitumen. Fuel 85(7):936–943.
Garcıa-Ochoa, F., V. Santos, J. Casas & E. Gomez, 2000. Xanthan gum: production, recovery, and
properties. Biotechnology advances 18(7):549–579.
Isacsson, U. & X. Lu, 1995. Testing and appraisal of polymer modified road bitumens—state of the art.
Materials and Structures 28(3):139–159.
Jansson, P.-e., L. Kenne & B. Lindberg, 1975. Structure of the extracellular polysaccharide from Xan-
thomonas campestris. Carbohydrate Research 45(1):275–282.
Kalogiannis, S., G. Iakovidou, M. Liakopoulou-Kyriakides, D.A. Kyriakidis & G.N. Skaracis, 2003.
Optimization of xanthan gum production by Xanthomonas campestris grown in molasses. Process
Biochemistry 39(2):249–256.
Katzbauer, B., 1998. Properties and applications of xanthan gum. Polymer Degradation and Stability
59(1):81–84.
Liu, G., 2011. Characterization and Identification of Bituminous Materials Modified with Montmo-
rillonite Nanoclay. TU Delft, Delft University of Technology.
Liu, G., M. van de Ven, S. Wu, J. Yu & A. Molenaar, 2011. Influence of organo-montmorillonites on
fatigue properties of bitumen and mortar. International Journal of Fatigue 33(12):1574–1582.
Mastrofini, D. & M. Scarsella, 2000. The application of rheology to the evaluation of bitumen ageing.
Fuel 79(9):1005–1015.
Milas, M., M. Rinaudo, M. Knipper & J.L. Schuppiser, 1990. Flow and viscoelastic properties of xan-
than gum solutions. Macromolecules 23(9):2506–2511.
Ouyang, C., S. Wang, Y. Zhang & Y. Zhang, 2006a. Improving the aging resistance of asphalt by addi-
tion of Zinc dialkyldithiophosphate. Fuel 85(7):1060–1066.
Ouyang, C., S. Wang, Y. Zhang & Y. Zhang, 2006b. Improving the aging resistance of styrene–butadi-
ene–styrene tri-block copolymer modified asphalt by addition of antioxidants. Polymer Degradation
and Stability 91(4):795–804.
Palaniraj, A. & V. Jayaraman, 2011. Production, recovery and applications of xanthan gum by Xan-
thomonas campestris. Journal of Food Engineering 106(1):1–12.
Petersen, J.C., 1984. Chemical composition of asphalt as related to asphalt durability: state of the art.
Transportation research record(999).
Plank, J., 2004. Applications of biopolymers and other biotechnological products in building materials.
Applied microbiology and biotechnology 66(1):1–9.
Psomas, S., M. Liakopoulou-Kyriakides & D. Kyriakidis, 2007. Optimization study of xanthan gum
production using response surface methodology. Biochemical Engineering Journal 35(3):273–280.
Rosalam, S. & R. England, 2006. Review of xanthan gum production from unmodified starches by
Xanthomonas comprestris sp. Enzyme and Microbial Technology 39(2):197–207.
Siddiqui, M.N. & M.F. Ali, 1999. Studies on the aging behavior of the Arabian asphalts. Fuel
78(9):1005–1015.
Song, K.W., Y.S. Kim & G.S. Chang, 2006. Rheology of concentrated xanthan gum solutions: Steady
shear flow behavior. Fibers and Polymers 7(2):129–138.
Taherkhani, H. & M. Javanmard, 2011. STEADY-STATE DEFORMATION BEHAVIOUR OF
ASPHALTIC MIXTURES*. Iranian Journal of Science and Technology Transactions of Civil Engi-
neering 35(C1):81.
Wu, S., J. Han, L. Pang, M. Yu & T. Wang, 2012. Rheological properties for aged bitumen containing
ultraviolate light resistant materials. Construction and Building Materials 33:133–138.
Wu, S., L. Pang, G. Liu & J. Zhu, 2010. Laboratory study on ultraviolet radiation aging of bitumen.
Journal of Materials in Civil Engineering 22(8):767–772.
Wu, S., Q. Ye & N. Li, 2008. Investigation of rheological and fatigue properties of asphalt mixtures
containing polyester fibers. Construction and Building Materials 22(10):2111–2115.
Yu, J.-Y., P.-C. Feng, H.-L. Zhang & S.-P. Wu, 2009. Effect of organo-montmorillonite on aging proper-
ties of asphalt. Construction and Building Materials 23(7):2636–2640.
310
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Yansheng Yang, Hao Liu, Enguang Li, Yuming Dong & Liying Yang
Beijing Municipal Road and Bridge Building Material Group Co. Ltd., Beijing Engineering Research
Center of Green Building Material for Road and Bridge, Beijing, China
1 INTRODUCTION
One of the main sources of pollution stemming from industries related to transportation
infrastructures (i.e. road construction) resides in the manufacturing, spreading, and conser-
vation of asphalt mixes. A cleaner production of bituminous mixes requires lowering the
manufacturing temperature of these mixes without impoverishing or reducing their level of
mechanical performance.
Warm Mix Asphalt (WMA), produced at a temperature lower than that required for Hot
Mix Asphalt (HMA), involve the use of chemical surfactant, the addition of a wax addi-
tive, and water-based or water-containing foaming processes (Hurley, 2005, 2005 & 2006).
Compared to these WMA technologies, chemical surfactants do not depend on foaming or
viscosity reduction for lowering the mixing and compaction temperatures. Instead, they gen-
erally include a combination of emulsification agents, surfactants, and additives to improve
coating, mixture workability, and compaction, as well as adhesion promoters (antistripping
agents). While these chemical surfactants are more frequently used in the USA and European
countries (i.e., France and Norway), they have also been used in China (John et al. 2008;
Jones, et al. 2001).
A large number of laboratory and field studies have been performed to characterize the
properties and performances of WMA mixtures with chemical surfactant. A study conducted
by Grzegorz and Karol (2015) suggests that the use of surfactants in WMA can result in a
temperature reduction of at least 20°C for SMA. Wang et al. (2014) analyze the effect of
surfactants on the SMA mixture’s water stability. The research indicates that the water stabil-
ity of the SMA mixture with surfactant conforms to the required specifications. Joel et al.
(2012) and Marisa et al. (2015) evaluated the performance of the WMA mixture containing
high percentages of recycled asphalt pavement (RAP). The research results indicate that the
chemical surfactant used in WMA with RAP is expected to perform as well as HMA in terms
of moisture susceptibility, rut resistance performance, and fatigue resistance performance.
311
Although these surfactant products are very recent, the third generation products have
emerged, which can be directly mixed with bitumen in asphalt plants to produce warm mix
asphalt. This warm mix asphalt is mixed with aggregates to produce an asphalt mixture.
Furthermore, there is no need to alter the existing equipment. However, this effect of asphalt
on the behavior of asphalt binders remains unknown, and the effects of the variation of
warm-mix surfactant content in the asphalt needs to be explored. Thus, the aim of this work
is to investigate the effect of warm-mix surfactant content on the properties of unmodified
and SBS modified asphalts. The properties of unmodified and styrene–butadiene–styrene
triblock copolymer (SBS) modified asphalts with various warm-mix surfactant contents were
probed using several different conventional testing techniques (e.g. penetration, softening
point, ductility, and viscosity), zero shear viscosity, temperature sensitivity, and frequency
sensitivity. It is anticipated that the research results will benefit the engineering practices of
using warm-mix surfactant technology to produce WMA.
312
Table 2. Properties of unmodified asphalt with various warm-mix surfactant.
0% 63 48.5 67 0.468
0.5% 63 48.0 63 0.459
1.0% 63 48.0 65 0.455
1.5% 65 47.0 64 0.419
2.0% 67 47.0 64 0.409
0% 62 62.0 30 1.607
0.5% 64 61.0 32 1.493
1.0% 68 61.0 39 1.382
1.5% 72 60.5 43 1.321
2.0% 79 60.0 50 1.264
2.3.3 Temperature sensitivity of unmodified and modified asphalt with various warm-mix
surfactant
Temperature sensitivity is closely related to the performance of the asphalt binder. There
are several evaluation criteria for the temperature sensitivity of asphalt binders, including
the penetration index (PI), penetration-viscosity number (PVN), viscosity-temperature sus-
ceptibility (VTS), modulus temperature susceptibility (GTS), etc. In this study, the complex
modulus of the asphalt binders with various warm-mix surfactants at various temperatures
was assessed, and regression equations were developed to depict the relationships between
the logarithms of the complex moduli and temperatures. Slopes of the linear regression equa-
tions were used to assess the warm-mix asphalts’ temperature sensitivity.
2.3.4 Frequency sensitivity of unmodified and modified asphalt with various warm-mix
surfactant
In this study, the complex modulus of the asphalt binders with various warm-mix surfactants
at various frequencies were assessed, and regression equations were developed to depict the
relationships between the logarithms of the complex moduli and frequencies. The slopes
of the linear regression equations were used to assess the warm-mix asphalts’ frequencies
sensitivity.
313
surfactant content increases from 0 wt% to 2.0 wt%, the softening point and viscosity at
135°C of the unmodified asphalt is slightly reduced, whereas the ductility at 10°C does not
significantly change.
Table 3 shows that the penetration and the softening point of the modified asphalt increases
and decreases, respectively, as the warm-mix surfactant content increases from 0 wt% to
2.0 wt%. This indicates that the addition of warm-mix surfactant reduces the hardness of the
modified asphalt, which is not preferable for pavement deformation resistance. The ductil-
ity at 5°C increases as the warm-mix surfactant content increases, which indicates that the
warm-mix surfactant improves the cracking resistance of modified asphalt. The warm-mix
surfactant decreases the viscosity of the modified asphalt at 135°C, which is beneficial to the
modified asphalt’s flow ability.
Table 2 and Table 3 show that the addition of warm-mix surfactant had different effects on
the SBS modified and unmodified asphalts. The penetration and ductility of the SBS modi-
fied asphalt were quite sensitive to warm-mix surfactant content.
Figure 1. Steady state flow behavior for unmodified asphalts with various warm-mix surfactant con-
tents at 60°C.
Figure 2. Steady state flow behavior for modified asphalts with various warm-mix surfactant contents
at 60°C.
314
Table 4. ZSV of unmodified and modified asphalt with various warm-mix surfactant contents.
0 341 3068
0.5% 332 3046
1.0% 322 2458
1.5% 298 2192
2.0% 281 1813
Figure 3. The logarithm of the complex modulus of unmodified asphalt with various warm-mix
surfactants.
Figure 4. The logarithm of the complex modulus of modified asphalt with various warm-mix
surfactants.
Table 4 shows that as the warm-mix surfactant increases from 0 to 2.0%, the ZSV values of the
modified asphalt decrease, whereas the change in the ZSV values of the modified asphalt is not
obvious when the warm-mix surfactant content is within 1.0%. When the warm-mix surfactant
content is more than 1.0%, the ZSV values of the modified asphalt decrease, which indicates that
the warm-mix surfactant reduces the high-temperature performance of modified asphalt.
315
tions to quantitatively depict their relationships. The slopes of the linear regression equations
in the figures give information on the sensitivity of complex moduli to temperature changes.
A steeper slope indicates a higher sensitivity of the modulus to temperature. The absolute
values of the slopes for the two types of unmodified and modified asphalt binders at different
warm-mix surfactant contents are shown in Table 5.
As shown in Fig. 3, when the warm-mix surfactant content increases from 0 to 2.0%, the
complex modulus of unmodified asphalt decreases, and the decrement shrinks as the temper-
ature increases from 22°C to 82°C. However, the reduction is not as great as in the warm-mix
surfactant content within 1.0%. The warm-mix surfactant significantly reduces the complex
modulus of unmodified asphalt when its content is greater than 1.0%.
As observed in Fig. 4, when the warm-mix surfactant content increases from 0 to 2.0%, the
complex modulus of modified asphalt decreases, and the reduction shrinks as the tempera-
ture increases from 22°C to 82°C.
Table 5 shows that, for the unmodified and modified asphalt, the temperature sensitivity
of complex moduli decreases as the warm-mix surfactant content increases from 0 to 2.0%
by mass. However, the effect of the warm-mix surfactant on the temperature sensitivity of
the two asphalts is different. For the unmodified asphalt, the change of the temperature sen-
sitivity of the complex modulus is slight as the warm-mix surfactant content increases from
0 wt% to 1.0 wt% and the temperature sensitivity of the complex modulus reduces from 40.84
to 35.70, with the warm-mix surfactant content increasing from 1.0% to 2.0%. However, for
the modified asphalt, the temperature sensitivity of the complex modulus displays a slow
decline and a sharp decrease as the warm-mix surfactant content increases.
Table 5. Values of slope for logarithms of the change of complex moduli with temperature for unmodi-
fied and modified asphalt with various warm-mix surfactants.
0 41.38 37.15
0.5% 41.14 36.01
1.0% 40.84 34.308
1.5% 37.25 29.71
2.0% 35.70 26.99
316
Table 6. Values of slope for logarithms of the change of complex moduli with frequency for unmodi-
fied and modified asphalt with various warm-mix surfactants.
0% 0.9348 0.7836
0.5% 0.9379 0.7897
1.0% 1.0058 0.7926
1.5% 1.0178 0.7961
2.0% 1.0433 0.7970
Figure 5. The logarithm of the complex modulus versus frequency for unmodified asphalt with vari-
ous warm-mix surfactants.
Figure 6. The logarithm of the complex modulus versus frequency for modified asphalt with various
warm-mix surfactants.
2.0% by mass. The complex modulus of unmodified asphalt is more sensitive to warm-mix
surfactants than modified asphalt.
4 CONCLUSION
The effects of various warm-mix surfactant contents on the properties of asphalt are inves-
tigated in this research. The rheological properties of unmodified and modified asphalt with
various warm-mix surfactant contents were tested and compared, based on zero shear viscos-
ity (ZSV) obtained by steady-state flow behaviours and complex moduli at different tempera-
317
tures and frequencies. Their conventional performances were also measured and compared.
Based on the results of this study, the following conclusions can be made.
The ZSV test indicates that the warm-mix surfactant changes the ZSV of unmodified
and SBS modified asphalt. For the unmodified asphalt, the ZSV values gradually decrease
when the warm-mix surfactant content is greater than 1.0%, while for the modified asphalt,
the ZSV values decrease when the warm-mix surfactant content is greater than 0.5%, and
the decrement of the ZSV for SBS modified asphalt is greater than that of the unmodified
asphalt.
In the temperature sweep tests, the complex moduli of unmodified and SBS modified
asphalt and their sensitivities to temperature decrease as the warm-mix surfactant content
increases.
In the frequency sweep tests, for the unmodified asphalt, the complex moduli decreases
in the range of lower frequencies and increases in the range of higher frequencies as the
warm-mix surfactant content increases. For the SBS modified asphalt, the complex moduli
decreases as the warm-mix surfactant content increases. In addition, the complex modulus of
unmodified asphalt is more sensitive to warm-mix surfactant content than modified asphalt.
However, the sensitivity of the complex moduli for unmodified and SBS modified asphalt to
frequency increases as the warm-mix surfactant content increases.
In the conventional tests, as the warm-mix surfactant content increases, for unmodified
asphalt, there is only slight change in the softening point and ductility, and viscosity decreases.
For the modified asphalt, the penetration and viscosity decrease, ductility increases, and the
change of softening point is slight.
REFERENCES
Grzegorz, M. and Karol N. 2015. The evaluation of SMA mixture properties with the surface-active
agent in WMA technology. 7th Scientific-Technical Conference Material Problems in Civil Engineer-
ing, Procedia Engineering, 108: 22–29.
Hurley, G. and Prowell, B. 2005. Evaluation of Aspha-Min for use in warm mix asphalt. NCAT report
05–04, Auburn.
Hurley, G. and Prowell, B. 2005. Evaluation of Sasobit for use in warm mix asphalt. NCAT report
05–06, Auburn.
Hurley, G. and Prowell, B. 2006. Evaluation of Evotherm for use in warm mix asphalt. NCAT report
06–02, Auburn.
Joel, R.M, Hugo, M.R.D, Liliana, P.F. and Juan, A. 2012. The role of a surfactant based additive on the
production of recycled warm mix asphalts – Less is more. Constr Build Mater, 35:693–700.
John, D., Eric, H. and John, B. 2008. Warm-Mix Asphalt: European Practice. US: Federal Highway
Administration’s International Technology Scanning Program.
Jones, W. 2001. Warm Mix Asphalt A-state-of-the-art Review. Australia: Australian Asphalt Pavement
Association.
Marisa, D.A and Marcia, L.A. 2015. Warm Mix Recycled Asphalt - a sustainable solution. Journal of
Cleaner Production, 107: 310–316.
Wang, Y., Chen, J.Y. and Wang, K. 2014. Effect of Evotherm on SMA mixture water stability. Journal
of Dalian Jiaotong University, 35(5): 62–64, 84 [In Chinese].
318
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Dissipation is a powerful tool for characterizing fatigue behaviour and has
been employed for various materials. But the use of dissipation in characterizing fatigue
of bituminous concrete is complicated by the presence of other modes of dissipation than
fatigue, particularly dissipation due to viscous friction. The challenge is therefore to separate
fatigue dissipation due to repeated loading from viscous dissipation. Some methods have
already been suggested the same including the use of the Dissipated Pseudo-Strain Energy
(DPSE) and the Ratio of Dissipated Energy Change (RDEC).
In this study, the two primary methods available for separating fatigue dissipation, DPSE
and RDEC, were analysed using principles of thermodynamics and experimental observa-
tions in the literature. It was found that while they are useful in certain specific cases, these
methods cannot be universally used in all laboratory and field tests.
Accurate estimation of fatigue and viscous dissipation requires accurate characterization
of the viscoelastic behaviour of damaged bituminous concrete. However, fatigue of bitumi-
nous concrete occurs at higher stress levels. It is difficult to conduct experiments to elicit the
pure viscoelastic behaviour at such stress levels as experiments at high stress levels tend to
damage the specimen. It is therefore necessary to make some constitutive assumptions about
the viscoelastic behaviour of damaged bituminous concrete. One such constitutive assump-
tion was used in this study to develop an alternative method to estimate fatigue dissipation.
Fatigue dissipation was estimated using this method for bituminous concrete prepared with
one unmodified binders and two binders modified by a plastomer and an elastomer.
1 INTRODUCTION
For the past two decades, the main focus of research with respect to fatigue of bituminous
concrete has been to develop a model describing the fatigue behaviour of bituminous mix-
tures that is consistent with the experimental results, particularly a model that can describe
the fatigue behaviour in both load-controlled and displacement-controlled fatigue tests.
While many a framework has been proposed to realise these objectives, the approach used by
many, possibly because of its simplicity, is the energy dissipation approach. Fatigue damage
occurs mainly through formation of new surfaces by cracking, which is in turn caused by loss
of cohesion within the bituminous mastic, and the loss of adhesion between aggregates and
the mastic. These mechanisms of fatigue-related changes to the internal structure of bitumi-
nous mixtures involve dissipation of energy. The dissipation through these mechanisms of
damage is often referred to as the “fatigue dissipation”. In repeated loading tests in uniaxial
condition, the total dissipation over a cycle when the strain in reasonably small is given by
W ∫ σ (t)ε(t) dt (1)
319
where σ and ε are the axial stress and strain, respectively. By measuring the accumulation of
dissipation in repeated loading tests, the extent of damage in the material can be estimated.
This forms the basis of the dissipation approach to characterise fatigue and has been used
to successfully characterize the fatigue behaviour of various materials (Inglis 1927, Coffin
Jr 1957, Feltner & Morrow 1961). Since mechanical dissipation can be easily measured in
repeated loading tests, this approach promises a simple and easy method to characterise
fatigue behaviour of bituminous mixtures in such tests.
The use of dissipation framework in characterizing fatigue behaviour of bituminous mix-
tures, however, is complicated to an extent by the presence of other modes of dissipation such
as that due to viscous friction, in addition to that due to fatigue damage. Fatigue dissipation
needs to be separated from other modes of dissipation before it can be used for character-
izing fatigue of bituminous mixtures. The methods proposed in the literature for this purpose
can be broadly classified into two groups: use of a dissipated pseudo-strain energy (DPSE)
for characterization of fatigue and the use of a parameter known as the ratio of dissipated
energy change (RDEC) to characterize fatigue. These two approaches are analyzed in this
study the details of which are presented in the next few sections. A new approach for estimat-
ing fatigue dissipation is then proposed as an alternative. This is followed by an evaluation of
the proposed approach by applying it to certain experimental results.
ε (t ) ε 0 si ωt (2)
is applied on a linear viscoelastic material, the corresponding sinusoidal stress response when
there is no damage is given by
where Gve* and δ ve are the dynamic modulus and the phase angle of the viscoelastic material
and the subscript ve is used to denote that the stress response is purely viscoelastic devoid of
any damage. The pseudo strain in this case is defined in terms of σ ve as
σ ve (t ) Gv*
ε R (t ) = = ε 0 ve i (ωt + δ ve ) (4)
GR GR
Now, the pseudo-strain energy dissipation, Wps , over a cycle is defined in terms of pseudo-
strain as follows:
Wps ∫ σ (t ) ε (t) dt
R (5)
where σ (t ) is the actual stress in the material. It can be seen that the stress σ experienced by
the material is expected to be equal to σ ve only when there is no damage to the material, and
320
is expected to be different otherwise. Consequently, the pseudo-strain energy dissipation in a
cycle is zero when there is no damage and is positive otherwise. In view of this, Kim & Little
(1989) regarded the pseudo-strain energy dissipation as equal to the fatigue dissipation of
the material and used it to characterise fatigue. Following their work, Lee et al. (2000), Kim
et al. (2003), Lee et al. (2003), Masad et al. (2008) and Castelo Branco et al. (2008) further
developed the approach and used pseudo-strain energy dissipation in crack-growth models,
continuum-damage models, etc., to characterise fatigue.
The pseudo-strain dissipation energy in a displacement-controlled fatigue test is typically
calculated in the following manner, as illustrated by Bhasin et al. (2009). In a displacement
controlled fatigue test involving sinusoidal loading as shown in (2), the stress response will
not be sinusoidal but would possess some other waveform. However, conventionally in the lit-
erature, it is approximated as a sinusoidal waveform, in which case the stress response would
be approximately of the form
where Gn and δ n are parameters similar to the modulus and phase angle, respectively,
that depend on the number of cycles of loading that has been applied on the material.
For the DPSE approach, the viscoelastic part of the stress response is taken to be of the
form
where δ ve is the phase angle measured for undamaged bituminous concrete. Accordingly, the
pseudo-strain in the nth cycle is
Gn
ε Rn (t ) = ε 0 sin (ω δ ve ) (8)
GR
Now, the total dissipation Wtotaln ½, the viscoelastic dissipation Wven and the pseudo-strain
energy dissipation Wpsn in the nth cycle are given by
(11)
Bhasin et al. (2009) and others consider the pseudo-strain energy dissipation Wpsn to be
equivalent to the dissipation due to fatigue damage in the cycle. If viscous dissipation and
fatigue dissipation are the only modes of dissipation of bituminous concrete, this assumption
implies that the sum of viscous dissipation and pseudo strain energy dissipation should be
equal to the total dissipation in the cycle. However,
Gn2 2
Wpsn Wven = π Gn δ vve + π ε 0 si (δ n − δ vve )
GR (12)
⎛ G ⎞
=π n δ ve + π (δ n − δ vvee )⎟
⎝ GR ⎠
(13)
⎛ ⎡G ⎤ ⎡ G ⎤ ⎞
=π n 0 ⎜ ⎢G δ ve ⎥ sinδ n + ⎢ − δ n ⎥ sinδ ve ⎟
⎝⎣ R ⎦ ⎣ GR ⎦ ⎠
(14)
321
which can never be equal to Wtotaln for all possible values of δ n for any choice of GR , unless δ ve
is zero or δ n is equal to δ ve . Moreover, when GR is assumed to be equal to Gn and δ n is taken
to be always greater than or equal to δ ve , as it is usually done (Bhasin et al. 2009),
Wtotaln π Gn 0 δn
(15)
=π n
⎡⎣ δ ve cos ( − ) + sin ( − ) cosδ ve ⎤⎦ . (16)
Since, both δ ve and δ n δ ve are both expected to be such that
π
0 ≤ δ ve ,δ n − δ vvee ≤ , (17)
2
Wtotaln π Gn 0
⎡⎣ δ vve + sin ( n − vve )⎤⎦ , (18)
Wtotaln Wvveen + Wpsn ,
(19)
with equality occurring only when δ ve = 0 or δ n δ ve . Thus, the standard assumptions for
reference modulus and the phase angle parameter results in a condition where the sum of
viscous dissipation and pseudo strain energy dissipation is greater than the total dissipation.
For any other assumptions for the reference modulus, total dissipation being equal to the sum
of viscous dissipation and pseudo-strain energy dissipation is not guaranteed for all δ n . Since
the fatigue dissipation W fddn would be such that
in the absence of other modes of dissipation, it can be inferred that the pseudo-strain energy
dissipation is not the same as fatigue dissipation.
E n +1 − DE n
DE
Cn =
RDEC , (21)
DE En
where DE E n is the total dissipation in the nth cycle. In a typical sinusoidal fatigue test, Ghu-
zlan & Carpenter (2000) found this RDEC parameter to initially start with a relatively high
value and decrease with number of repetitions at a steep rate until it reaches a minimum. This
is followed by a period in which RDEC remains constant with number of repetitions which
Ghuzlan & Carpenter (2000) identified as the plateau period. Failure in both load-controlled
and displacement-controlled tests is marked by a rapid increase in RDEC with the number
of repetitions. For most bituminous concrete, the plateau period is the predominant part of
322
Figure 1. A sample displacement-controlled fatigue test result with total dissipation in a cycle decreas-
ing with each cycle (Source: Kim et. al. (2003)).
the RDEC vs. number of repetitions curve. Ghuzlan & Carpenter (2000) used the RDEC in
the plateau called the plateau value (PV) for characterisation of fatigue of bituminous con-
crete. Typically, the value of RDEC in the plateau period is not a constant but varies over a
certain range and the value of RDEC at the point where the stiffness of bituminous concrete
decreases to 50% of the initial stiffness is taken as the plateau value (PV).
The key assumption made in the rate of dissipation approach is that any change in total
dissipation between successive cycles is only due to fatigue damage. However, in many dis-
placement-controlled fatigue tests, it has been observed that the total dissipation in a cycle
may decrease with successive repetitions (see Figure 1) and yet the material may suffer fatigue
damage as evidenced by the decrease in the dynamic modulus with the progress of the fatigue
test (Kim et. al. 2003, Yoo & Al-Qadi, 2010). RDEC measured in such tests would be nega-
tive over a large portion of the test. The change in dissipation between cycles will be due to
both change in the viscous dissipation characteristics and fatigue damage. Thus, the assump-
tion that damage occurs only when the dissipation increases with load repetitions may not be
valid for all fatigue tests.
In this study, a new approach for estimating fatigue dissipation for bituminous concrete is
proposed. The approach involves making a suitable constitutive assumption on the viscoe-
lastic behaviour of damaged bituminous concrete, essentially an assumption relating the vis-
coelastic behaviour on the extent of damage. In any fatigue test, the constitutive assumptions
can be used to determine the viscous dissipation in each cycle as the test progresses. Since
the total dissipation can be calculated from the stress and strain measurements, the fatigue
dissipation can also be determined.
The constitutive assumption made in this study is that the total strain in bituminous con-
crete is the sum of a viscoelastic strain and another strain part, and the viscoelastic strain
response of damaged bituminous concrete is the same as the purely viscoelastic strain
response of undamaged bituminous concrete when subjected to the same stress loading. This
assumption implies that in a load-controlled fatigue test with the stress of the form
σ (t ) σ 0 sinωt, (22)
323
σ0
ε (t ) = sin (ω δ n ). (23)
Gn
σ0
ε ven (t ) = sin (ω δ ve ). (24)
Gvve*
σ 2
Wven ∫ σ (t ) ε (t) dt = π G
ven
0
*
ve
sinδ vve (25)
In a displacement controlled fatigue test with the strain of the form shown in (2), the stress
response is as shown in (6).The viscoelastic strain in this case is given by
σn
ε ve (t ) = sin(ω δ ve ), (26)
Gvve*
where σ n is the amplitude of the stress in the nth cycle. Using the expression for σ (t ) from
(6), the above equation can be rewritten as
Gn
ε ven (t ) = ε 0 sin(ω δ n − δ ve ), (27)
Gvve*
In both cases, the fatigue dissipation can be determined by subtracting viscous dissipation
from the total dissipation:
The constitutive assumption is essentially equivalent to assuming that fatigue damage does
not change the viscous dissipation characteristics of bituminous concrete.
4 EXPERIMENTAL PROCEDURE
Three bitumen were used in the study—one unmodified bitumen conforming to VG 30 grade
as per the Indian standard IS 73, a bitumen modified by a plastomer and another modified
by elastomer, both conforming to PMB 40 grade as per IS 15462. Locally available granite
aggregates with a gradation adhering to BC-Grade 2 gradation specified by MORT&H 2013
specifications (Ministry Of Road Transport and Highways 2013) was used for the mix. 5%
binder content was used for all the bitumen. Mixing and conditioning were conducted as per
AASHTO R30 and compaction was carried out using a PReSBOX shear compactor manu-
factured by IPC, Australia. Beams were cast with a target air voids of 4 ± 0.5%. From com-
pacted beams of dimensions 450 × 150 × 145–185 mm, three cylindrical specimen of diameter
93 ± 0.5 mm and height 150 ± 0.5 mm were cored out. Cylindrical specimens not meeting the
air void requirement of 4 ± 0.5% were discarded.
324
Fatigue tests were conducted in uniaxial tension using an Asphalt Mixture Performance
Tester (AMPT) manufactured by IPC, Australia. The specimen was glued to two steel plates
with a suitable epoxy adhesive to apply tension. Three linear variable differential transducers
(LVDT’s) of gauge length 70 mm and a range of ± 0.5 mm were fixed onto the specimen, all
oriented along the axial direction and separated at 120° angle from each other. The deforma-
tion measured by the LVDT’s were averaged and divided by the gauge length to obtain the
axial strain of the specimen. All the fatigue tests were conducted in load-controlled mode
with a haversine loading of 10 Hz frequency. All the tests were conducted at a temperature
of 20°C.
Tests were conducted in the following manner. Initially small amplitude haversine stress load-
ing was applied on the specimen to measure the dynamic modulus Gve* and the phase angle
δ ve of the bituminous concrete at 10 Hz. Followed by this, a larger amplitude haversine load-
ing at 10 Hz frequency was applied in addition to a seating load of 0.1 kN until complete
failure of the sample. For the mixtures with the two modified bitumen a stress amplitude of
550 kPa was applied, and for the mixture with unmodified bitumen a stress amplitude of
515 kPa was applied during the fatigue test. The dynamic modulus, phase angle of the bitu-
minous concrete specimens at 10 Hz frequency, the amplitude of the applied stress for the
fatigue tests, the corresponding viscous dissipation in a cycle and the number of repetitions
until complete failure are presented in Table 1.
Using the measured strain along with the applied stress, the total dissipation in each cycle
was calculated by numerically integrating equation (25). The LVDT’s recorded the axial
deformation only till the deformation exceeded their range. The total dissipation therefore
could be calculated only until that point. The cumulative total dissipation thus estimated for
each bituminous concrete is plotted against the number of load repetitions in Figure 2. The
total dissipation vs. number of repetitions curve appear to be nearly a linear function for all
the bitminous concrete. The total dissipation was highest for the bituminous concrete with
the elastomer modified binder and least for the bituminous concrete with the unmodified
binder.
According to the constitutive assumptions made in this study, the viscous dissipation does
not change with the cycles of load repetition in a stress-controlled test. The viscous dissipa-
tion in a cycle was calculated using equation (25). The fatigue dissipation in each cycle was
then determined by subtracting the viscous dissipation from the total dissipation. The cumu-
lative fatigue dissipation was determined by summing over the fatigue dissipation estimated
for each cycle. The cumulative fatigue dissipation vs. the number of cycles of repetition for all
the bituminous concrete are presented in Figure 3. At about 50,000 cycles, the total cumula-
tive dissipation for the bituminous concrete with elastomer modified binder is 4.35 × 105 J/m3,
while the cumulative fatigue dissipation is 2.9 × 104 J/m3, only 6.67% of the total dissipation.
The value for bituminous mixtures with plastomer modified binder and unmodified binder
are 22% and 11%, respectively. Thus, a significant part of total dissipation is due to viscous
friction rather than fatigue dissipation.
325
Figure 2. Variation of cumulative total dissipation with the number of cycles of repetition for the
bituminous mixtures.
Figure 3. Variation of cumulative fatigue dissipation with the number of cycles of repetition for the
bituminous mixtures.
The cumulative fatigue dissipation versus number of load repetitions was fitted by a second
order polynomial of the form aN + bN2, where N is the number of cycles of load repetition.
This quadratic form was found to completely fit the cumulative fatigue dissipation curve, with
326
Figure 4. Quadratic polynomial t to the cumulative fatigue dissipation curves and extended up to the
point of failure.
the R2 value of the fits being higher than 0.99 for all the tests. Such a quadratic form of the vari-
ation of cumulative fatigue dissipation with the number of cycles of load repetition implies that
the fatigue dissipation in a cycle increases linearly with the number of cycles of load repetition.
Fatigue dissipation in load-controlled tests can therefore be characterized by two constants.
Assuming the quadratic variation would continue until failure of the specimen, the cumu-
lative fatigue dissipation was predicted until the end of the test for the three bituminous mix-
tures and are presented in Figure 4. The results show that the bituminous concrete with the
plastomer modified bitumen dissipated the maximum energy through fatigue damage until
complete failure. This was followed by the bituminous concrete with the elastomer modified
bitumen. The bituminous concrete with the unmodified bitumen dissipated the least energy
through fatigue damage before failure. This order could be indicative of the resistance to
fatigue failure of the bitumen tested in this study.
6 CONCLUSION
Dissipation in repeated loading tests of bituminous concrete can be due to both viscous friction
and fatigue damage. Approaches existing in the literature to separate fatigue dissipation from
viscous dissipation, particularly the pseudo-strain energy dissipation approach and the rate of
dissipation approach, were analyzed in the light of experimental observations in fatigue tests.
Some issues and shortcomings were found in both the approaches. While they could be useful
in certain circumstances, both methods cannot be universally applied for all fatigue tests.
A new approach to estimate fatigue dissipation in fatigue tests was proposed. The base of
this approach is a constitutive assumption on the viscoelastic behaviour of damaged bitu-
minous concrete. Once such a constitutive assumption is in place, the viscous dissipation in
a cycle can be determined and can be used in the estimation of fatigue dissipation. In this
study, it was assumed that the viscoelastic response of damaged bituminous concrete would
be such that the viscous dissipation characteristics do not change with the extent of damage.
327
Fatigue dissipation was estimated in this manner for three different bituminous concrete from
load-controlled fatigue tests. It was found that a considerable portion of the total dissipation
in fatigue tests is due to viscous friction rather than fatigue damage.
The next step in the development of the new approach would be to validate the constitu-
tive assumption. Validation would require the measurement of viscoelastic characteristics of
damaged bituminous concrete. Measurements should be taken at different levels of damage
to obtain the evolution of viscoelastic behaviour with fatigue damage.
ACKNOWLEDGEMENT
The authors thank the Department of Science and Technology (India) for funding this inves-
tigation. The grant number is DST/TSG/STS/2011/46.
REFERENCES
Bhasin, A., Castelo Branco, V.T., Masad, E. and Little, D.N., 2009. Quantitative comparison of energy
methods to characterize fatigue in asphalt materials. Journal of Materials in Civil Engineering, 21(2),
pp.83–92.
Carpenter, S., Ghuzlan, K. and Shen, S., 2003. Fatigue endurance limit for highway and airport pave-
ments. Transportation Research Record, (1832), pp.131–138.
Carpenter, S. and Shen, S., 2006. Dissipated energy approach to study hot-mix asphalt healing in fatigue.
Transportation Research Record, (1970), pp.178–185.
Castelo Branco, V., Masad, E., Bhasin, A. and Little, D., 2008. Fatigue analysis of asphalt mixtures
independent of mode of loading. Transportation Research Record, (2057), pp.149–156.
Coffin Jr, L.F., 1957. Strain cycling and thermal stress fatigue. Proc. Sagamore Conference
Feltner, C., Morrow, J., 1961. Microplastic strain hysteresis energy as a criterion for fatigue fracture.
Journal of Basic Engineering, 83(1):15–22.
Ghuzlan, K. and Carpenter, S., 2000. Energy-derived, damage-based failure criterion for fatigue testing.
Transportation Research Record, (1723), pp.141–149.
Inglis, N.P., 1927. Hysteresis and fatigue of Wohler rotating cantilever specimen. The Metallurgist, 1(1),
pp.23–27.
IS 73. 2006. Paving Bitumen – Specification. Bureau of Indian Standards, New Delhi, India.
IS 15462. 2004. Polymer and Rubber Modified Bitumen – Specification. Bureau of Indian Standards,
New Delhi, India.
Kim, Y.R., Little, D.N. and Lytton, R.L., 2003. Fatigue and healing characterization of asphalt mix-
tures. Journal of Materials in Civil Engineering, 15(1), pp.75–83.
Kim, Y.R. and Little, D.N., 1989. Evaluation of healing in asphalt concrete by means of the theory of
nonlinear viscoelasticity. Transportation Research Record, (1228), pp. 198–210.
Lee, H.J., Daniel, J.S. and Kim, Y.R., 2000. Continuum damage mechanics-based fatigue model of
asphalt concrete. Journal of Materials in Civil Engineering, 12(2), pp.105–112.
Lee, H., Kim, Y. and Lee, S., 2003. Prediction of asphalt mix fatigue life with viscoelastic material prop-
erties. Transportation Research Record, (1832), pp.139–147.
Li, N., Pronk, A., Molenaar, A., van de Ven, M. and Wu, S., 2013. Comparison of Uniaxial and Four-
Point Bending Fatigue Tests for Asphalt Mixtures. Transportation Research Record, (2373), pp.44–53.
Masad, E., Castelo Branco, V.T.F., Little, D.N. and Lytton, R., 2008. A unified method for the analysis
of controlled-strain and controlled-stress fatigue testing. International Journal of Pavement Engineer-
ing, 9(4), pp.233–246.
Ministry of Road Transport and Highways (MoRT&H). 2001. Specification for roads and bridge work.
Government of India, Indian Roads Congress, 4th Revision, New Delhi, India.
Pronk, A.C., 2001. Partial healing in fatigue tests on asphalt specimens. Road Materials and Pavement
Design, 2(4), pp.433–445.
Shen, S. and Carpenter, S., 2005. Application of the dissipated energy concept in fatigue endurance limit
testing. Transportation Research Record, (1929), pp.165–173.
Yoo, P.J. and Al-Qadi, I.L., 2010. A strain-controlled hot-mix asphalt fatigue model considering low and
high cycles. International Journal of Pavement Engineering, 11(6), pp.565–574.
328
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Yongjie Ding
The University of Tennessee, Knoxville, TN, USA
China University of Petroleum, Qingdao, P.R. China
Baoshan Huang
Tongji University, Shanghai, China
The University of Tennessee, Knoxville, TN, USA
Xiang Shu
The University of Tennessee, Knoxville, TN, USA
ABSTRACT: While virgin and aged binders are blended together, the old molecular associa-
tions of binders may be broken and new ones rebuilt due to intermolecular interactions. In this
study, the intermolecular interactions between virgin and aged binders were investigated using Gel
Permeation Chromatography (GPC). Results indicate that intermolecular interactions occurred
between virgin and aged binders when virgin and aged binders were blended together in sol-
vent. The factors affecting the degree of interaction included difference in chemical composition
between the two binders, aged binder content, and molecular size. The degree of intermolecular
interaction increased with the increase in the difference in chemical structure between virgin and
aged binders. The intermolecular interactions were more likely to happen between large molecules
than other molecules. For large and medium molecules, it is found that there was a good liner rela-
tionship between the degree of their intermolecular interaction and the aged binder content.
1 INTRODUCTION
Recycled Asphalt Pavement (RAP) and Recycled Asphalt Shingles (RAS) have been increas-
ingly used by asphalt paving industry due to economic and environmental benefits. The
incorporation of RAP and RAS into asphalt paving mixtures has raised many concerns over
the overall performance of recycled asphalt mixtures (Daniel and Lachance, 2005, Huang
et al., 2005, Shu et al., 2008, Zhao et al., 2012, Huang et al., 2006) as well as the blending of
aged binder into virgin binder (Bowers et al., 2014, Zhao et al., 2013).
Asphalt binder is a colloidal material and the molecular associations between asphaltic
molecules have an important influence on the properties of binders (Gray et al., 2011, Dickie
and Yen, 1967). For many years, the dominant conceptual models for the intermolecular
structure of asphalt have been based on clusters of aromatic components in a surrounding
medium of less aromatic components (Gray et al., 2011). Yen and co-workers pointed out
that the formation of clusters of molecules by π-π stacking of aromatic rings was the main
reason of asphaltene molecular aggregation (Dickie and Yen, 1967). Some researchers have
challenged this paradigm and proposed new models. For example, Gray et al. (2011) pro-
posed a supramolecular model that included molecular recognition and host-guest interac-
tion in a three-dimensional porous network. In this model, the π-π stacking of aromatic rings
is a contributing factor rather than a dominant motif.
Although the dominant form of the molecular association among asphaltic molecules has
had no final conclusion yet, previous studies suggest that the molecular association among
asphaltic molecules has a significant influence on the properties of binders (Gray et al., 2011,
329
Figure 1. Interrelationship between molecular association and GPC curves.
Dickie and Yen, 1967). Consequently, an understanding of the factors that affect the molecu-
lar association will be helpful for interpreting the performance of binders.
The molecular structures of binders change during the ageing process, which include the
dissociation, isomerization, and fragmentation of asphaltene molecules, the dissociation of
aromatics molecules, and the association, cyclization, and then dissociation of saturate mol-
ecules (Siddiqui and Ali, 1999). The changes of molecular structures may lead to the varia-
tion of molecular associations. Therefore, when virgin and aged binders are blended together
and become homogeneous on molecular level, the molecular association of binders may be
changed due to the intermolecular interaction between virgin and aged binders. In fact, unlike
either virgin or aged binder, the molecular association of the mixed binder should exhibit
new characteristics because the molecules which lie beside each other in mixed binders have
changed. Researches have shown that the interaction between different binders can affect
the properties of mixed binders (Davison et al., 1994). However, researches with a focus on
the intermolecular interactions between virgin and aged binders are still very limited. In this
study, the changes in molecular association of mixed binders caused by the blending proc-
ess of virgin and aged binders were analyzed using gel permeation chromatography (GPC).
Figure 1 illustrates the interrelationship between molecular association and GPC curves.
2 GPC
GPC has been used by many researchers to characterize the molecular size distribution
(MSD) of asphalt binders as well as to establish relationships between MSD and the physical
and aging characteristics and the pavement performance of asphalt binders (Wahhab et al.,
1999, Kim and Burati Jr, 1993). Usually, GPC curves are divided into three parts, including
large molecular size (LMS), medium molecular size (MMS) and small molecular size (SMS)
(Terrenzio et al., 1997). In order to correlate GPC data to physical properties with more accu-
racy, GPC curves are divided into 8, 10, 13 and even 30 slices (Kim and Burati Jr, 1993).
Researches show that GPC test results varied with the concentration and the solvent of
the samples (Haley, 1971) because GPC curves are the results of molecular association of
asphaltic molecules instead of single molecules (Branthaver, 1993, Herrington and Ball,
1996). Therefore, the changes in molecular association caused by intermolecular interaction
of binders can be analyzed by GPC. In addition, intermolecular interactions can be helpful
for understanding the chemical structures of binders.
The intermolecular structure of asphalt binder has been a research hot topic for many years.
Among all the components, asphaltene has received the most attention for its complexity and
diversity. Asphaltene can agglomerate in hydrocarbon solvents because it contains many aro-
matic rings. Brandt et al. (Brandt et al., 1995) used thermodynamics modeling to simulate the
330
asphaltene stack solvents and found that the maximum agglomerated number for asphaltene
molecules was 5. Pacheco- Sánchez et al. (Pacheco-Sánchez et al., 2003) used molecular
dynamics simulation to analyze 96 asphaltene molecules and found that asphaltene agglom-
eration number decreased with the increase in temperature and agglomerate orientation is
mostly depended on crude oil characteristics. Aguilera-Mercado et al. (Aguilera-Mercado
et al., 2006) simulated the agglomeration of asphaltene and resin molecules and found that
asphaltene does not conform to a linear agglomeration model, but exhibits a more complex
multimodal agglomeration pattern.
Hydrogen-bonding and π-π stacking were found to be two main factors for the association
between asphaltene molecules (da Costa et al., 2012, Liu et al., 2010). da Costa et al. (2012)
concluded that hydrogen bonding is as important as π-π interactions. Liqun Zhang (Zhang,
2007) and De Leeuw et al. (De Leeuw et al., 1990) found that for a system having asphaltene,
resin and saturate, a delicate balance among the different intermolecular forces, including
polarity of aromatic compounds in the system, polarity of saturate controlled the size and
lifetime of molecular agglomerates.
The structure of asphaltene agglomeration is the basis of the intermolecular structure and
has been particularly analyzed. Yen et al. (Yen et al., 1961) found that interlayer distance
between agglomerated asphaltene molecules is 0.35 to 0.38 nm. Zhang & Greenfield (Zhang
and Greenfield, 2007) investigated the intermolecular orientation of asphaltene using molec-
ular dynamics simulation. They concluded that asphaltene molecules without long aliphatic
chains prefer to stay almost parallel to each other at low temperature and form agglomerates
in T-shaped orientation at high temperature. On the opposite, asphaltene molecules with
short aliphatic chains prefer to stay parallel at high temperature and pack almost T-shaped
at low temperature.
Tang et al. (Tang et al., 2014) conducted a study to investigate the agglomeration variation
pattern of asphaltic molecules using molecular dynamics simulation. Their results showed
that layer agglomeration structure of asphaltene molecules form in binders due to the stack-
ing interactions of π-π aromatic. But the aliphatic chain of asphaltene molecules can enlarge
torsion angle between aromatic rings and block the agglomeration. Asphaltene and resin mol-
ecules can agglomerate under the effect of the stacking interactions of π-π aromatic and be dis-
persed into maltene. As temperature or the length of molecular aliphatic chain of asphaltene
molecules increases, the agglomerated structure of asphaltene and resin is destroyed.
Based on the researches mentioned above, asphaltic molecules can form many association
structures. The objective of this paper was to investigate the influence of the intermolecular
interaction between virgin and aged binders on the molecular associations of mix binders.
Two parameters from GPC test were used to characterize the changes in molecular associa-
tions. The effects of three factors were investigated, including the ratio of aged binders, the dif-
ferences in chemical compositions between virgin and aged binders, and the molecular size.
As mentioned above, GPC is an analytical chemistry technique that yields the molecular size
distribution of a given medium in solution. The y axis of a GPC curve is representative of
the refractive index (RI) output in mV. An increase in refractive index indicates an increase
amount of molecules of a given size (Bowers, 2013). The degree of refraction and the magni-
tude of the signal generated by an RI detector are a function of solution concentration and
the sample being analyzed. Since the plot of RI against concentration is linear, the concen-
tration can be derived by measuring the RI and comparing this value with that of the pure
solvent (Yunus and Rahman, 1988).
331
In this study, the virgin and aged binders were proportionally mixed together. Therefore,
the GPC curves of the mixed binder should be equal to the sum of the GPC curves of virgin
and aged binders alone if there is no intermolecular interaction and molecular association
changes happening during the mixing process.
In order to evaluate whether intermolecular interaction occurred between virgin and aged
binders, the following procedure was run. At first, virgin and aged binders were proportion-
ally mixed together and the mixed binder was tested by GPC to obtain a tested curve. Then
the GPC curves of virgin and aged binders were proportionally added together to form a
new GPC curve, which was called the calculated curve. The calculated curve is expressed in
Eq. (1):
where V represents the values of GPC curves, and R means the ratio of aged/virgin binder
in the mixed binder.
The difference between the tested and calculated curves is called GPC curve difference
(which is the shaded area shown in Fig. (2)) and can be used to measure the changes in
molecular associations during mixing process. The computing method is shown in Eq. (2).
The GPC curves in Figure 2 is from one RAS-virgin mixed binder. It can be seen that the
tested curve is obviously higher than the calculated curve on the left side till the peak, which
suggests that there were obvious differences in molecular associations within this molecular
size range. To the right side of the peak, the tested curve is almost identical with the calcu-
lated one, which means that no obvious molecular associations changed within this region.
In order to evaluate the influence of molecular size on the molecular association, the dif-
ferences in MSD between tested and calculated curves were also investigated. The MSD
of GPC curve was separated into LMS, MMS and SMS. The ranges of molecular size for
LMS, MMS, and SMS were defined as >3000, 1000–3000, and 200–1000 Dalton, respectively
(shown in Fig. 2). The ratio of slice area to the whole area was considered to be the molecular
content. The elution time in the chromatogram was converted to molecular size using the
calibration curve obtained from the standard polystyrene GPC testing (Daly et al., 2013).
The MSD differences was calculated as follows:
332
In order to separate the MSD differences caused by intermolecular interaction and the
potential experimental error, one point was tested twice and their difference was considered
to be caused by experimental error. The differences in GPC curves and MSD should be sig-
nificantly higher than the experimental error so that it can be considered to be caused by the
changes in molecular association.
6 LABORATORY EXPERIMENT
In the laboratory experiment, a typical PG 64–22 virgin asphalt binder and two aged binders
including one RAP and one RAS were selected. All three materials were locally available in
Tennessee, USA. The GPC curves of binders are shown in Figure 3. It can be seen that the
difference in GPC curve between RAS and virgin binders was larger than that between RAP
and virgin binders, which means that the difference in chemical structure between RAS and
virgin binders was significantly larger than that between RAP and virgin binders. Therefore,
the comparison of the results from RAS and RAP can be used to investigate the influence of
difference in chemical structure on the intermolecular interaction
In the experiment, the RAP/RAS binder content varied from 0 to 100% at 10% intervals.
In order to enable the GPC curves of virgin and aged binders to be reasonably added up, the
concentration of the binders was set to be the same. The virgin binder and RAP/RAS binder
were first dissolved into tetrahydrofuran (THF) to make the solutions at target concentration
(1 mg/mL). Then 10 mL solution of the mixed binders was produced by adding the solutions
of different binders in the right proportions. The solution was shaken in a solution shaker at
high speed for 1 minute for complete dissolution, and then injected through a 0.2 μm filter
to filter out the undissolved impurities. An auto-sampler was used to collect the prepared
sample and then placed in the sample holder in EcoSEC GPC.
7 RESULTS
333
Table1. Experimental error of GPC curves differences.
Asphalt 1 2 3 4 5 6 7 8 9 10 11 means
Experimental 2.67 0.69 0.86 0.50 4.52 2.36 0.29 0.40 3.57 0.62 7.92 2.22%
error (%)
intermolecular interaction between virgin and RAP binders was not significant. The dif-
ference in GPC curve between RAS and virgin binder was significantly larger than that
between RAP and virgin binder, and all the value was outside of the experimental error
interval, which indicates that obvious intermolecular interaction happened between RAS
and virgin binders.
The phenomenon that RAS showed a clear trend in the difference between its GPC and
virgin binder’s indicates that one dominant intermolecular interaction may exist between
virgin and RAS binders. Compared to RAS, RAP did not show a clear trend, suggesting that
there existed mixed intermolecular interactions between virgin and RAP binders, none of
which played a dominant role.
Aged asphalt ratio 0 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Means
Experimental 1.45 2.31 1.83 0.90 1.18 1.92 1.44 1.35 1.43 1.64 0.38 1.44
Error (%)
Figure 7. Correlation coefficient from linear regression between MSD and aged asphalt content.
8 DISCUSSION
The GPC curves differences and MSD difference between RAS and virgin binders were both
significantly larger than those between RAP and virgin binders, suggesting that the aging degree
of old binders had an significant influence on the intermolecular interactions between virgin
and aged binders. The more the chemical composition of aged binder differs from that of virgin
binder, the stronger the intermolecular interactions between virgin and aged binders generated.
335
The GPC curves can reflect the intermolecular associations of asphaltic molecules. In this
study, due to the existence of solvent, only part of intermolecular association could be shown on
the GPC curves. While the dissolved virgin and aged binders were mixed together and shaken to
achieve a homogeneous blend, some intermolecular associations were disconnected and recon-
nected. According to Gray’s molecular model of asphalt, the intermolecular interactions include
molecular recognition and host-guest interaction in three-dimensional porous network (Gray
et al., 2011). Because the components of virgin and aged asphalt binders were different, some
molecules might find more appropriate positions than before and reconnected with other mol-
ecules to form a new intermolecular association, leading to the changes in the GPC curves.
Because MSD is the mass concentration of the binders, it can be used to investigate the
direction of some intermolecular interactions. It can be seen from Figure 6 that, for RAP
binder, the tested curves were higher than the calculated curves in the MMS region, whereas
the tested curves were lower than the calculated curves in the LMS region. This result implies
that some of large molecules were converted into medium molecules during the blending proc-
ess of virgin and RAP binders. It may be attributed to the fact that some intermolecular asso-
ciations of the large molecules were broken and the molecules could not find suitable places
to reconnect. In case of RAS binders, the tested curves were much higher than the calculated
curves in the LMS region, whereas the tested curves were lower than the calculated curves in
both MMS and SMS regions. This indicates that small and medium molecules agglomerated
together and formed large molecules during the mixing process. The reason for this may be
that the chemical composition of RAS differed significantly from that of virgin binder. Some
small and medium molecules which did not connect before found right positions to connect
with other molecules during blending process and formed large molecules. This may be due
to the polarity of RAS binder molecules, which had an impact on the virgin binder molecules
and improved the degree of intermolecular association. However, the results from MSD could
only reflect the trend of the intermolecular interaction. More future studies are needed to
learn the details of the intermolecular interactions between virgin and aged binders.
9 CONCLUSIONS
In this study, the intermolecular interactions between virgin and aged binder were investi-
gated using GPC. Virgin and aged binder were blended together in a chemical solvent and
tested using GPC to examine the changes in molecular association caused by intermolecular
interactions. Based on the results, the following conclusions are drawn:
1. The intermolecular interactions between RAS and virgin binders were found to be more sig-
nificant than that between RAP and virgin binders, indicating that the difference in chemi-
cal composition between aged and virgin binders affected the intermolecular interaction.
2. The molecular size had an influence on the intermolecular interactions. Intermolecular
interactions were more likely to happen between molecules with significantly different
chemical structures.
3. There existed a linear relationship between difference in MSD and the content of aged
asphalt binder in the LMS and MMS regions for both RAS and RAP binders, indicating
that the content of aged binder also affected the intermolecular interactions between aged
and virgin binders.
REFERENCES
Aguilera-Mercado, B., Herdes, C., Murgich, J. & M Ller, E. 2006. Mesoscopic simulation of aggrega-
tion of asphaltene and resin molecules in crude oils. Energy & Fuels, 20, 327–338.
Bowers, B.F. 2013. Investigation of Asphalt Pavement Mixture Blending Utilizing Analytical Chemistry
Techniques.
Bowers, B.F., Huang, B., Shu, X. & Miller, B.C. 2014. Investigation of reclaimed asphalt pavement
blending efficiency through GPC and FTIR. Construction and Building Materials, 50, 517–523.
336
Brandt, H., Hendriks, E., Michels, M. & Visser, F. 1995. Thermodynamic modeling of asphaltene stack-
ing. The Journal of Physical Chemistry, 99, 10430–10432.
Branthaver, J. 1993. Size exclusion chromatography studies of aged SHRP asphalts. Fuel science & tech-
nology international, 11, 123–139.
Da Costa, L.M., Stoyanov, S.R., Gusarov, S., Tan, X., Gray, M.R., Stryker, J.M., Tykwinski, R., De
M. Carneiro, J.W., Seidl, P.R. & Kovalenko, A. 2012. Density functional theory investigation of
the contributions of π–π stacking and hydrogen-bonding interactions to the aggregation of model
asphaltene compounds. Energy & Fuels, 26, 2727–2735.
Daly, W.H., Negulescu, I. & Balamurugan, S.S. 2013. Implementation of GPC Characterization of
Asphalt Binders at Louisiana Materials Laboratory.
Daniel, J. & Lachance, A. 2005. Mechanistic and volumetric properties of asphalt mixtures with recycled
asphalt pavement. Transportation Research Record: Journal of the Transportation Research Board, 28–36.
Davison, R.R., Bullin, J., Glover, C., Chaffin, J., Peterson, G., Lunsford, K., Lin, M., Liu, M. & Ferry, M. 1994.
Verification of an asphalt aging test and development of superior recycling agents and asphalts. Final Report.
De Leeuw, S., Smit, B. & Williams, C. 1990. Molecular dynamics studies of polar/nonpolar fluid mixtures.
I. Mixtures of Lennard-Jones and Stockmayer fluids. The Journal of Chemical Physics, 93, 2704–2714.
Dickie, J.P. & Yen, T.F. 1967. Macrostructures of the asphaltic fractions by various instrumental meth-
ods. Analytical chemistry, 39, 1847–1852.
Gray, M.R., Tykwinski, R.R., Stryker, J.M. & Tan, X. 2011. Supramolecular assembly model for aggre-
gation of petroleum asphaltenes. Energy & Fuels, 25, 3125–3134.
Haley, G.A. 1971. Molecular and unit sheet weights of asphalt fractions separated by gel permeation
chromatography. Analytical Chemistry, 43, 371–375.
Herrington, P.R. & Ball, G.F. 1996. Temperature dependence of asphalt oxidation mechanism. Fuel,
75, 1129–1131.
Huang, B., Li, G., Vukosavljevic, D., Shu, X. & Egan, B. 2005. Laboratory investigation of mixing
hot-mix asphalt with reclaimed asphalt pavement. Transportation Research Record: Journal of the
Transportation Research Board, 37–45.
Huang, B., Shu, X. & Burdette, E. 2006. Mechanical properties of concrete containing recycled asphalt
pavements. Magazine of Concrete Research, 58, 313–320.
Kim, K.W. & Burati Jr, J.L. 1993. Use of GPC chromatograms to characterize aged asphalt cements.
Journal of Materials in Civil Engineering, 5, 41–52.
Liu, D., Kong, X., Li, M. & Wang, Z. 2010. Study on the aggregation of residue-derived asphaltene
molecules. Energy & Fuels, 24, 3624–3627.
Pacheco-S Nchez, J., Zaragoza, I. & Mart Nez-Magad N, J. 2003. Asphaltene aggregation under vac-
uum at different temperatures by molecular dynamics. Energy & fuels, 17, 1346–1355.
Shu, X., Huang, B. & Vukosavljevic, D. 2008. Laboratory evaluation of fatigue characteristics of recy-
cled asphalt mixture. Construction and Building Materials, 22, 1323–1330.
Siddiqui, M.N. & Ali, M.F. 1999. Investigation of chemical transformations by NMR and GPC during
the laboratory aging of Arabian asphalt. Fuel, 78, 1407–1416.
Tang, B., Ding, Y., Cao, X. & Zhu, H. Molecular Dynamics Simulation to Investigate Variation Pattern
of Asphalt Molecules Agglomeration. Transportation Research Board 93rd Annual Meeting, 2014.
Terrenzio, L.A., Harrison, J.W., Nester, D.A. & Shiao, M. Natural v. artificial aging: use of diffusion
theory to model asphalt and fiberglass-reinforced shingle performance. Proceedings of the Fourth
International Symposium on Roofing Technology, 1997.
Wahhab, H.A.-A., Asi, I., Ali, F. & Al-Dubabi, I. 1999. Prediction of asphalt rheological properties
using HP-GPC. Journal of materials in civil engineering, 11, 6–14.
Yen, T.F., Erdman, J.G. & Pollack, S.S. 1961. Investigation of the structure of petroleum asphaltenes by
X-ray diffraction. Analytical Chemistry, 33, 1587–1594.
Yunus, W. & Rahman, A.B.A. 1988. Refractive index of solutions at high concentrations. Applied optics,
27, 3341–3343.
Zhang, L. 2007. Physical and mechanical properties of model asphalt systems calculated using molecu-
lar simulation, ProQuest.
Zhang, L. & Greenfield, M.L. 2007. Analyzing properties of model asphalts using molecular simula-
tion. Energy & fuels, 21, 1712–1716.
Zhao, S., Bowers, B., Huang, B. & Shu, X. 2013. Characterizing rheological properties of binder and
blending efficiency of asphalt paving mixtures containing RAS through GPC. Journal of Materials
in Civil Engineering, 26, 941–946.
Zhao, S., Huang, B., Shu, X., Jia, X. & Woods, M. 2012. Laboratory performance evaluation of warm-
mix asphalt containing high percentages of reclaimed asphalt pavement. Transportation Research
Record: Journal of the Transportation Research Board, 98–105.
337
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Long life bituminous pavement is widely pursed. The change of bitumen
properties after long term field aging, however, often receives little attention. Particularly
missing are oxidative aging mechanisms and oxidation products generated in sufficient
long term field aging conditions. Bitumen binders studied in this paper were extracted from
Hong Kong field pavement which has served for 36 years. In addition, four strategic highway
research program (SHRP) bitumen binders were artificially aged to simulate long term oxida-
tive aging. The bitumen binders extracted from field as well as their asphaltene components
were examined by using Fourier Transform Infrared spectroscopy (FTIR) and X-ray Photo-
electron Spectroscopy (XPS). The artificially aged bitumen binders were subject to analysis
by using FTIR. Main oxidation products in different bitumen components were analyzed.
The relationships between ketone, alcohol and the corresponding dynamic viscosity were
investigated. Alcohol was proven to be a significant long term aging product but not seri-
ously affect the dynamic viscosity of field-aged bitumen.
1 INTRODUCTION
Bitumen aging causes the embrittlement and cracking of bituminous pavement. Bitumen
aging mechanism and aging products have been a focus of study for many years. According
to Peterson & Glaser (1998) and Domke et al. (2000), bitumen aging follows a dual oxidation
mechanism based on different aging kinetics: (1) a rapid aging period followed by (2) a long
period with constant oxidation rate. Ketones and sulfoxides are recognized to be the main
aging products, and their formation was found to be dependent on temperature and oxygen
pressure (Liu et al.1998, Siddiqui & Ali 1999). Recently, alcohols were also noticed as a major
product of bitumen aging. Peterson & Glaser (2011) explained the generation of alcohols by
a modified dual oxidation mechanism. Lins et al. (2008) found the generation of ketones and
alcohols during photodegradation process of hot-mix bitumen. Aldehyde and esters were
also identified in aged bitumen (Yut & Zofka 2011). These findings suggest that a variety of
single bond carbon-oxygen products are formed during the aging of bitumen.
It is commonly agreed that ketone concentration is highly correlated to the physical prop-
erties of aged bitumen (Wang et al. 2014, Morian et al. 2011). Pan et al. (2012), however,
suggests that both sulfoxides and ketones control the physical properties of bitumen by using
quantum chemistry simulation. It is concluded by Peterson & Glaser (2011) that alcohols
generate “a similar effect on viscosity increase as do the ketones.” Therefore, the relationships
between the oxidized products and bitumen physical properties remain inconclusive.
Most aging simulation was conducted in laboratory with the help of accelerated aging
apparatuses, e.g. Rolling Thin Film Oven (RTFO) and Pressure Aging Vessel (PAV). Although
recently researchers investigated the physical and chemical aging behaviors of bitumen after
aging in natural environment (Hesp et al. 2007, Wu et al. 2008, de Sá et al. 2013), the aging
extent of those bitumen samples are far less than aging severity level of a long-life pavement
339
with more than 40 years (Nunn et al. 1997 & 2001). Little literature reports the change of
physical and chemical bitumen properties under natural aging with a sufficient “long term”.
Field-aged bitumen samples used in this study were extracted and recovered from a
36-year-old expressway (Tuen Mun Road) in Hong Kong. A total of 23 samples were used
to ensure that they can reflect the average field situation. The bitumen used for the pavement
was reported in a well-recognized paper by Mcleod (1989) published in 1980s. Four bitumen
types from the Strategic Highway Research Program (SHRP) were used to perform artificial
aging by using a pressure oxidation autoclave. The four SHRP bitumen types are: AAD
(Ca coast), AAG (Ca Valley), AAK (Boscan), and AAM (W Tx Inter). Fourier Transform
Infrared Spectroscopy (FTIR) was used to characterize aging products. The focus of the
investigation was placed on ketones and alcohols. X-ray photoelectron spectroscopy (XPS), a
surface-sensitive quantitative spectroscopic technique, was used to measure the presence and
relative quantities of chemical elements, chemical state and electronic state of the field-aged
bitumen samples.
2 EXPERIMENTAL SECTION
340
2.4 XPS examination
A Physical Electronics XPS equipment (PHI Quantum 2000) was used to study the field aged
bitumen and extracted asphaltene. Mild sputtering was first performed on the sample surface
to remove the surface contamination and moisture. Five regions of total four bitumen and
asphaltene samples were scanned, and each region is 100 μm in diameter. The average result
of the measurements from each specimen was used to represent the information of whole
sample.
Figure 1. Typical FTIR spectra of long term field-aged bitumen binder samples.
341
Figure 2. Comparison between long term field aged bitumen and long term artificial aged AAD
sample.
Figure 3. Comparison between long term field aged bitumen and long term artificial aged AAG
sample.
Figure 4. Comparison between long term field aged bitumen and long term artificial aged AAK
sample.
Figure 5. Comparison between long term field aged bitumen and long term artificial aged AAM
sample.
342
bitumen samples. The carbonyl group C = O at about 1700 cm−1, S = O at about 1030 cm−1 are
clearly present in both the field-aged and artificially aged specimens.
There are also small distinct peaks of the functional group C-O at the range between
3200 cm−1 to 3300 cm−1 in the FTIR spectra of the artificially aged SHRP bitumen binders.
The most obvious difference between the spectra of field-aged and artificially aged bitumen
samples occurs at the peaks at the wavenumber of 1260 and 1300 cm−1, which are circled out
in Figure 2 to Figure 5. The field-aged bitumen shows a strong evidence of these two peaks
while artificially aged bitumen does not (at least not obvious). The two peaks are the charac-
teristic peaks of carbon-oxygen stretch vibration of phenol and alcohol, especially for terti-
ary alcohol. The four artificially aged bitumen samples show the same oxidation products of
ketone and sulfoxide, but long term field-aged bitumen show more distinct types of alcohol
products or phenols.
In artificial aging, the temperature and oxygen pressure are usually both set to a fixed high
value. According to Arrhenius equation, the aging kinetic will be a constant:
k Ae − Ea /( RT ) (1)
where k is the constant rate of aging reaction at the absolute temperature T, A is the pre-
exponential factor and Ea is the activation energy.
The oxidation rate for bitumen is explicit in fixed aging temperature and oxygen pres-
sure through the parameters A and Ea. However, in field situation, oxygen diffuses through
maltene to polar aromatic and asphaltene aggregates and the temperature varies greatly.
Under such conditions, it is likely to have more decomposition of the molecular chains and
more side reactions. Meanwhile, the radical oxidation reactions always happen simultane-
ously with the transfer of bitumen components (maltene to asphaltene). Thus the aging
kinetics are more sensitive to the variations in temperature, oxygen pressure and bitumen
compositions. Overall, during long term aging in a natural environment, bitumen may have
more complex oxidation reactions and tends to produce more types of alcohol products. The
oxidized products are richer and more diverse in field conditions.
Figure 6 is the FTIR spectra of asphaltenes compared with the spectra of bitumen of
long term field aging. The two spectra show nearly same characteristic peak wavelength
locations. However, asphaltene showed different characteristic peaks for hydroxyl: There
are little signs of C-O bonding at 1260 cm−1 but the hydroxyl peak at 3260 cm−1 is very
intense. This indicates that the C-O bonding is formed mostly in the associated hydroxyl
form of asphaltene. The other proportions of hydroxyl such as phenol and tertiary alcohol,
reflected at the wavenumber of 1260 cm−1 and 1300 cm−1 respectively, may mainly exist in
other bitumen components. For the ketone characteristic peak at 1700 cm−1, asphaltene also
shows more intense peak than that of bitumen, indicating that ketone content is higher in
asphaltene.
Figure 6. FTIR spectra of asphaltene and bitumen with long term field aging.
343
3.2 XPS results of field-aged bitumen and asphaltene samples
The principle of XPS is to irradiate the sample using X-ray, so that the inner electrons or
valence electrons of atoms are stimulated. The stimulated electrons are called photoelectrons.
Binding energy of photoelectrons can be measured to obtain information on the valence and
chemical environment of the atom elements since every element has its characteristic electron
binding energy. In XPS test, survey scan is first conducted to identify the whole elemental
composition, and then narrow scan of each identified element photoelectron line is con-
ducted. The photoelectron peak in XPS spectra is marked by the photoemission element and
its electron orbit such like C1s.
Analysis of bonding state is one of the most important application of XPS especially for
measurement and analysis of oxidation valence state, which is by measuring the chemical
shift of the peak position of the binding energy for specific element in X-ray photoelectron
spectroscopy. Multiple peak spectra may exist for elements with several different bonding
situations.
The survey scan results of a typical field-aged bitumen sample is shown in Figure 7. It can
be seen from Figure 7 that the XPS spectra show only several sharp photoelectron peaks within
narrow ranges. Four field-aged bitumen samples show almost identical peak locations at spe-
cific binding energies. The C KLL and O KLL peak at high binding energy of about 1240 eV
and 960 eV are the Auger lines instead of photoelectron lines, which are produced by the resid-
ual atom energy during the relaxation process and should be excluded. The strong photoelec-
tron spectral lines of O1s and C1s at binding energy about 523.5 eV and 284.8 eV indicate that
oxygen (O) and carbon (C) are the main elements in the bitumen samples (besides hydrogen
which cannot be shown) and the weak spectral lines of S2s, S2p, Si2s, Si2p at binding energy at
about 230 eV to 80 eV indicate that there exist sulfur (S), and silicon (Si) heteroatoms.
The content of each element in the bitumen sample is shown in Table 2. Carbon has the
highest content of 87.2% and oxygen accounts for 9.2%, silicon and sulfur have the lower
content of 2.8% and 0.8% respectively. According to the peak information of binding energy
Figure 7. Typical XPS spectra of four field- Figure 8. Carbon spectra of bitumen
aged bitumen samples based on survey scan based on narrow scan.
344
at 102.5 ev, the bonding state of Si is identified as SiO2 (Trejo et al. 2009), which may be from
the very fine aggregates remained after binder extraction. The recalculated carbon, oxygen
and sulfur content after removing SiO2 is shown in Table 3.
Figure 8 shows the narrow scan results of C element in a typical field-aged bitumen sample. As
shown in Figure 8a, the peak shape of carbon element C1s is roughly symmetric within a small
range between 283 eV and 286 eV. This indicates that the basic element C has a similar bonding
form on the sample surface. The peak energy is at about 284.8 eV, which is close to a benzene
ring carbon system (Camacho-Bragado et al. 2002). There is a small bulge that can be identified
at high binding energy (286 eV), implying the presence of oxygen-containing functional groups
as induced by the chemical energy shift of carbon-oxygen bonding group shown in Table 1.
Figure 9 and Table 4 show the survey scan result and element content of a typical long
term field-aged asphaltene sample. Four spectra of asphaltene samples showed almost the
same peak locations at specific binding energies. Only C, O, S elements are found on these
sample surfaces, indicating that SiO2 is removed in asphaltene extraction. Then average car-
bon content is 93.9%, and oxygen content is about 4.0%, close to the content of bitumen
sample. However, the sulfur concentration is increased more than twice to 2.1%.
Figure 10 is the narrow scan results of C element in field-aged asphaltene samples. As
shown in Figure 10, different with bitumen sample, the C1s peak of asphaltene ranges from
282 ev to 290 ev. This implies that the oxidation degree of carbon in asphaltene is higher so
that the C1s peak has a large shift to high binding energy. To investigate the form and content
of detailed carbon bonding groups in asphaltene, the C1s peak is subjected to a sub-peak
analysis. It can be seen from Figure 10 that the asphaltene surface forms a C-C character-
istic photoelectron peak and two carbon-oxygen bonding peaks. One is the C-O bonding at
the binding energy of 286.5 ev which has a higher peak and larger area, and the other is the
C = O bonding at the binding energy 288.2 ev which is lower and smaller. The main C-C peak
of C1s is the same with bitumen sample at 284.7 ev, indicating that the C element mainly
forms a benzene ring system in asphaltenes too. Hence, we may speculate that the C-O bond-
ing and C = O bonding might be attached at the carbon ring system or its branch chains.
Table 2. Average element content in long term field-aged bitumen (excluding hydrogen).
Carbon
Element 284.8 eV Silicon Sulfur
Binding energy C, C-C, Graphite C Oxygen 102.5 eV 163.9 eV
Characteristic (Camacho-Bragado 523.5 eV SiO2 (Tery L. RSOR, Thiophene,
chemical bond et al. 2002) —* Barr 1978) (Trejo et al. 2009)
Element Sulfur
Binding energy Carbon Oxygen 163.9 eV
Characteristic 284.8 eV 523.5 eV S = O,
chemical bond C, C-C, Graphite C — Thiophene
345
Figure 9. Typical XPS spectra of four field aged Figure 10. Carbon spectra of asphaltene based
asphaltene samples based on survey scan. on narrow scan.
According to the average asphaltene concentration of total long term field aged sample
and the quantitative analysis of XPS, we can re-calculate the element distribution of carbon,
oxygen and sulfur in the maltene component, as shown in Table 5. As the table shown, oxygen
concentration is similar in asphaltene and maltene, while the sulfur content is much lower in
maltene than asphaltene.
Dependent on the colloidal system of bitumen, asphaltene is surrounded by resins, saturates
and aromatics (Yen 1992). Thus, the XPS results of a bitumen surface might reflect bonding
information of mainly maltene and a small proportion of asphaltene. In maltene component,
saturates are mainly mixtures of alkanes and cycloalkanes with few heteroatom, which are dif-
ficult to have oxidation reaction; therefore, the oxygen is mainly considered to react with resins
and aromatics on the bitumen surface. XPS result of long term aged bitumen shows that oxy-
gen existing in the samples but the bonding degree is low as no obvious chemical shift of carbon
element is found. The oxidation in aromatics and resins might be the crucial step for their con-
version to asphaltene. XPS results show that a great proportion of hydroxyl and sulfoxide prod-
ucts in long term aged asphaltene. The oxygen bonding degree is much higher in asphaltene as
it shows a large chemical shift and a significant characteristic peak of the carbonyl functional
group. Oxidization occurs in all the components of bitumen, but in long term oxidation it has
more effect on the carbon ring system in asphaltene than resins and aromatics.
346
Figure 11. Relationship between relative concentration of ketone (left) and alcohol (right) and log
dynamic viscosity.
FTIR, XPS and DSR tests of long term aged bitumen binders are conducted in the study.
The results help understands the aging products and their effects on the engineering proper-
ties of bitumen. Conclusions can be drawn as follows:
1. Ketone, alcohol and sulfoxide are the main oxidization products in long term aged bitu-
men under natural environment.
2. Artificially aged bitumen has the same main oxidation products of ketone and sulfoxide,
but the hydroxyl products and concentration of different aging products are not the same
as those shown in field-aged bitumen. Field-aged bitumen tends to produce more types of
hydroxyl compounds such as phenol and alcohols.
3. During the aging process, ketone and sulfoxide is mainly generated in asphaltene compo-
nent while alcohol or other hydroxyl compounds exist in all components.
347
4. Content of carbon, oxygen and sulfur elements in long term field aged bitumen and
asphalt are semi quantified.
5. Ketone content has an obviously positive correlation with the binder viscosity while alco-
hol has slightly negative correlation with the viscosity of long term field-aged bitumen.
REFERENCES
Barr, T.L. 1978. An ESCA study of the termination of the passivation of elemental metals. The Journal
of Physical Chemistry, 82(16), 1801−1810.
Camacho-Bragado, G.A., Santiago, P., Marin-Almazo, M., Espinosa, M., Romero, E.T., Murgich,
J.,... & Jose-Yacaman, M. 2002. Fullerenic structures derived from oil asphaltenes. Carbon, 40(15),
2761−2766.
de Sá, M.D.F.A., Lins, V.D.F.C., Pasa, V.M.D. & Leite, L.F.M. 2013. Weathering aging of modified
bitumen binders. Fuel Processing Technology,115, 19−25.
Domke, C.H., Davison, R.R. & Glover, C.J. 2000. Effect of oxygen pressure on bitumen oxidation
kinetics. Industrial & engineering chemistry research, 39(3), 592−598.
Hesp, S.A., Iliuta, S. & Shirokoff, J.W. 2007. Reversible aging in bitumen binders. Energy & fuels, 21(2),
1112−1121.
Liu, M., Ferry, M.A., Davison, R.R., Glover, C.J. & Bullin, J.A. 1998. Oxygen uptake as correlated to
carbonyl growth in aged bitumen and bitumen Corbett fractions. Industrial & engineering chemistry
research, 37(12), 4669−4674.
Lins, V.F.C., Araújo, M.F.A.S., Yoshida, M.I., Ferraz, V.P., Andrada, D.M. & Lameiras, F.S. 2008.
Photodegradation of hot-mix bitumen. Fuel, 87(15), 3254−3261.
Mcleod, N.W. 1989. Relationship of paving bitumen temperature susceptibility as measured by PVN to
paving bitumen specifications, bitumen paving mixture design and bitumen pavement performance.
J. of Association of Bitumen Paving Technologists, Vol. 58, 410−489.
Morian, N., Hajj, E., Glover, C. & Sebaaly, P. 2011. Oxidative aging of bitumen binders in hot-mix
bitumen mixtures. Transportation Research Record: Journal of the Transportation Research Board,
(2207), 107−116.
Nunn, M.E., Brown, A., Weston, D. & Nicholls, J.C. 1997. Design of long-life flexible pavements for
heavy traffic. TRL REPORT 250.
Nunn, M. & Ferne, B.W. 2001. Design and assessment of long-life flexible pavements. Transportation
Research Circular, 503(2001.12), 32−49.
Pan, T., Lu, Y. & Lloyd, S. 2012. Quantum-chemistry study of bitumen oxidative aging: an XPS-aided
analysis. Industrial & Engineering Chemistry Research, 51(23), 7957−7966.
Petersen, J. & Harnsberger, P. 1998. Bitumen aging: dual oxidation mechanism and its interrelationships
with bitumen composition and oxidative age hardening. Transportation Research Record: Journal of
the Transportation Research Board, (1638), 47−55.
Petersen, J.C. & Glaser, R. 2011. Bitumen oxidation mechanisms and the role of oxidation products on
age hardening revisited. Road Materials and Pavement Design, 12(4), 795−819.
Siddiqui, M.N. & Ali, M.F. 1999. Studies on the aging behavior of the Arabian bitumen. Fuel, 78(9),
1005−1015.
Siddiqui, M.N. & Ali M.F. 1999. Investigation of chemical transformations by NMR and GPC during
the laboratory aging of Arabian bitumen. Fuel, 78(12), 1407−1416.
Trejo, F., Ancheyta, J. & Rana, M.S. 2009. Structural characteristicization of asphaltenes obtained from
hydroprocessed crude oils by SEM and TEM. Energy & Fuels, 23(1), 429−439.
Valdes, H., Sanchez-Polo, M. & Zaror, C.A. 2003. Effect of ozonation on the activated carbon surface
chemical properties and on 2-mercaptobenzothiazole adsorption. Latin American applied research,
33(3), 219−223.
Wang, Y., Wen, Y., Zhao, K., Chong, D. & Wei, J. 2014. Connections between the Rheological and Chemical
Properties of Long-Term Aged Bitumen Binders. Journal of Materials in Civil Engineering, 04014248.
Wu, S., Pang, L., Mo, L., Qiu, J., Zhu, G. & Xiao, Y. 2008. UV and thermal aging of pure bitumen-
comparison between laboratory simulation and natural exposure aging. Road Materials and Pave-
ment Design, 9(sup1), 103−113.
Yen, T.F. 1992. The colloidal aspect of a macrostructure of petroleum bitumen. Fuel science & technol-
ogy international, 10(4−6), 723−733.
Yut, I. & Zofka, A. 2011. Attenuated total reflection (ATR) fourier transform infrared (FT-IR) spec-
troscopy of oxidized polymer-modified bitumen. Applied spectroscopy, 65(7), 765−770.
348
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Warm mix asphalt mixture, a certain technical measures to reduce the vis-
cosity of asphalt can make asphalt mix at relatively lower temperature than hot mix asphalt.
In order to improve the pavement performance, warm mix technology of rubber asphalt
mixture were discussed in this study. Then the influence of warm mix agent on the formation
of rubber asphalt mixture, ore heating temperature and road performance was analyzed by
the test. Results showed that the mix temperature and ore heating temperature were both
decreased by 10°C with the use of warm mix agent, Sasobit; at the same time, high tempera-
ture performance and fatigue property of the modified asphalt mixture were improved, while
the low temperature performance was decreased and the water stability was affected slightly.
349
tively in (160 ± 3)°C and (170 ± 3)°C. Both the temperatures were ascertained under the same
viscosity.
Table 1. SMA-13 mineral aggregate gradation of warm mix asphalt rubber mixture.
Screen opening/mm 16 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.75
Pass rate/% 100 92.5 62.3 32.1 27.3 21.5 18.7 15.9 14.3 11.1
350
Figure 3. Wet process.
Forming temperature/°C Stability (kN) Void fraction (%) Stability (kN) Void fraction (%)
125∼130 7.13 6.2 6.87 6.6
135∼140 8.35 5.2 7.33 5.3
145∼150 8.88 4.5 7.84 4.7
155∼160 7.53 4.2 8.19 4.4
160∼165 7.45 3.8 8.15 4.2
Code requirement ≥ 6.0 3–4.5 ≥ 6.0 3–4.5
Referring to the literature, the molding temperature of rubber modified asphalt mixture
was 150~160°C and the heating temperature of the ore was 185°C. Under the premise of
both the stability and the porosity meeting the porosity, Marshall test results of two kinds
of mixtures were compared at different rock heating temperatures. Obviously, the perform-
ance of warm mix rubber modified asphalt mixture was best at 175°C and the stability was
8.89 kN below this temperature. So the best heating temperature was ascertained at 175°C.
Comparing the Marshall test results of two kinds of mixtures at different temperatures,
it was not difficult to find that the performance index of warm mix rubber modified asphalt
mixture was the best in the temperature range of 145∼150°C and superior to rubber modi-
fied asphalt mixture. Therefore, the molding temperature was identified in 145∼150°C. In
comparison with warm mix rubber modified asphalt mixture and rubber modified asphalt
mixture, the former heating temperature in different stages was 10°C lower than the latter.
Hence, the conclusion could be drawn that the addition of warm mix agent reduced the con-
struction temperature of the modified asphalt mixture effectively (Hasanz, 2014).
351
3 THE COMPARISON TEST AND ANALYSIS OF WARM MIX RUBBER
MODIFIED ASPHALT MIXTURE PERFORMANCE THROUGH DRY
PROCESS AND WET PROCESS
Table 4. Rutting test results of warm mix rubber modified asphalt mixture.
Warm mix rubber modified Dry process 5478 6034 6265 5926 ≥ 3000
asphalt mixture Wet process 5940 5206 5624 5590
Rubber modified asphalt mixture 4772 4514 4826 4704
Table 5. Low temperature trabecular bend test results of warm mix rubber modified asphalt mixture.
352
Table 6. Immersion Marshall test results of warm mix rubber modified asphalt mixture.
Table 7. Freeze-thaw splitting test results of warm mix TOR rubber asphalt mixture.
Warm mix rubber modified Dry process 186 450 39 980 19 450
asphalt mixture Wet process 145 090 29 690 16 160
Rubber modified asphalt mixture 104 520 23 580 12 150
agent exists in the crystal structure of a web. That’s why the surface of the mix agent is
wrapped with a layer of rubber modified asphalt, the elasticity resuming performance is get-
ting stronger and the stiffness modulus becomes smaller.
According to the experimental results, the rutting resistance, low temperature crack resistance
and fatigue resistance performance of warm mix rubber modified asphalt mixture by dry proc-
ess are all better than that by wet process. But there is not obvious difference in water stability.
The addition of warm mix agent Sasobit has a negative impact in the low temperature
crack resistance of rubber modified asphalt mixture. And the Low temperature bending test
shows that the influence is slighter when the dry method is adopted.
The road performance of warm mix rubber modified asphalt mixture through dry way is
better than that through wet way. Furthermore, the construction technology of dry method
is simple and easy to operate, which can reduce the project costs. As a consequence, when
adding warm mix agent Sasobit to the warm mix rubber modified asphalt mixture, dry proc-
ess is a priority selection.
The low temperature performance and water stability performance of the mixture is mainly
depended on the performance of asphalt. So adding anti-stripping agent is advised to improve
the water stability performance of warm mix rubber modified asphalt. In addition, too much
warm mix agent will be harmful to the low temperature performance of the mixture.
REFERENCES
354
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
Cold In-place Recycling (CIR) is a method in which deteriorated pavement materials are
recycled in-situ with either foamed bitumen or emulsion to produce a new stabilized base
course for road rehabilitation. This technology has gained importance in the recent years
due to their environmental friendly nature in terms of energy savings and maximum use of
milled material. Reclaimed Asphalt Pavement (RAP) is obtained by cold milling or heating
and removal of pavement layer up to a depth where the distress has transferred or from full
depth removal of asphalt pavements. RAP materials treated with foamed bitumen are called
as Bitumen Stabilized Material (BSM) (Asphalt Academy, 2009). BSM mixture consists of
pulverized RAP material, fresh aggregates depending on the target gradation, foamed bitu-
men and active filler. Water is added to the pulverized material to ensure proper dispersion
of foamed bitumen. Active fillers are also added to provide the initial strength to the mix and
also to modify the fine fraction of aggregate gradation (Halles and Thenoux, 2009).
From the point of view of mechanical characterization in the mechanistic-empirical pave-
ment design framework, it should be understood that the material can show a behavior similar
to granular material or bituminous material. This study aims to explore the material behavior
exhibited by BSM. If we consider a pavement cross section which has to be rehabilitated as
shown in Figure 1, the BSM can be laid as base/binder course, after milling the pavement lay-
ers up to a depth to which the distress has permeated. The behavior of BSM mixes can be in
between that of a bituminous mix and granular material (Jenkins et al., 2007). Hence when a
pavement is rehabilitated with BSM layer, it is not very clear the kind of ‘modulus’ values to
be used to carry out the design and evaluate the performance.
Huan et al. (2010) treated BSM as a granular material and carried out resilient modulus
and unconfined compressive strength test. Khosravifar et al. (2013) reported that BSM is
a partially bound material based on their stiffness. Vorobeiff (2005) developed an interim
design method for pavements with BSM layers considering it as a ‘bound material’. Fatigue
355
Figure 1. Cross section of a pavement ( M-E PDG, 2004).
cracking, rutting of the surface and shrinkage cracking were the three modes of distress
identified in this study. Kim et al. (2009) measured the dynamic modulus of BSM at 4.4, 21.1
and 37.8°C and at six different loading frequencies (25, 10, 5, 1, 0.5 and 0.1 Hz). They found
out that the dynamic modulus of BSM was less sensitive to the loading frequencies compared
to Hot Mix Asphalt (HMA). Since most of the testing protocols presuppose a characteristic
material behavior, the mechanical response of the material will be dependent on the testing
protocols. For instance, the repeated load triaxial test methods (AASHTO: T 307–99, 2007)
are designed to capture the strain hardening behavior of granular material under repeated
loading in the presence of confinement pressure. Hence it is possible that one can measure
‘resilient modulus’ or ‘dynamic modulus’ and use it as an input for the design of pavement
structures, little realizing that such distinction does not really characterize a material.
BSM exhibit both pressure-dependent characteristics of granular materials and loading-
time dependent characteristics of bituminous materials. In this investigation resilient modu-
lus test (AASHTO: T 307–99, 2007) and dynamic modulus test (AASHTO: TP-79, 2010)
were carried out on BSM. At any given temperature (in this study, room temperature), the
data was analyzed to determine which of the characteristics, viz., granular or bituminous can
best describe the mechanical properties of the material.
2.1 Material
The properties of bitumen of grade VG10 (stipulated in IS: 73 (2013) of Indian Standards),
which was used to produce foamed bitumen, are shown in Table 1.
The RAP was obtained from milling the Bituminous Concrete (BC) layer of National
Highway 45 from Tambaram to Tindivanam in the southern part of India. The road was
constructed in 2005. The thickness of the BC layer (wearing course) which was milled was
50 mm. The nominal maximum aggregate size of the BC layer was 19 mm. The thickness of
the Dense Bituminous Macadam layer (binder course) was 160 mm.
Specification
Properties Value value
nominal maximum aggregate size of stone dust is 13.2 mm and 9.5 mm respectively. Indirect
Tensile Strength (ITS) test was carried out on dry and wet specimen after dry and wet cur-
ing as per ASTM D 6931 (2010). In the case of BSM mixture, for a design traffic more than
6 million standard axles, the specified values for ITSdry and ITSwet were 225 kPa and 100 kPa
respectively (Asphalt Academy, 2009). At 2% binder content, the ITSdry was 248 kPa and
ITSwet was found to be 216 kPa. The optimum foamed bitumen content was chosen as 2%.
3 TESTING
εr ( (i )) ε r ( ((ii − 1))
% Reduction in ε r = ∑
i = 4 ,7 ,10 ,18 ε r (Sequence(i ))
× 100 (1)
It can be seen from Table 3 that WMM showed substantial strain recovery during these
cycles (i.e. Sequence 4, 7, 10 and 13 of Table 2 when compared to BSM. Such variation in
strain response should reflect in the calculation of MR and this is shown in Figure 4. It is
358
Figure 3a. Strain response during 12th and Figure 3b. Strain response during 12th and 13th
13th sequence for WMM. sequence for BSM.
4 44 38 53
7 33 29 51
10 66 62 71
13 50 43 53
359
seen that the MR value of WMM was found to increase as each sequence progressed (Fig-
ure 4). Typically this is the expected behavior of a granular material in which the deformation
is reduced due to the increase in the confining pressure. However, the behavior of BSM is
different as the confinement pressure is increased. The percentage reduction in recoverable
deformation for BSM was less compared to WMM and hence there was a reduction in MR
values of BSM when the confinement pressure was increased at such sequences.
Influence of deviatoric stress
During repeated load triaxial testing, the deviator stress is increased (Table 2). The granular
material “strain hardens” due to the increase in the bulk stress. However as known, the devel-
opment of shear stresses during such loading will ensure that the MR did not increase signifi-
cantly (Figure 5a). However, BSM showed significant increase in MR when deviatoric stress was
increased as can be seen in Figure 5b. This can be due to the strain hardening behavior of BSM
and also due to less shear stresses developed in the material due to increase in deviatoric stress.
Use of resilient modulus models for WMM and BSM
In the literature, one comes across various models that relate MR to the stress conditions. It
will be interesting to see whether one can fit a resilient modulus model to BSM. Table 4 shows
the three resilient modulus models and the associated parameters for BSM and WMM. The
model parameters were estimated by regression analysis using the data of 15 sequences of
repeated load triaxial test. The first model is the k-θ model that was developed by Hicks
and Monismith (1971) which considers the influence of only bulk stress (θ) on MR. The
Model parameters
M-E PDG model BSM: 3.32 BSM: 0.106 BSM: 0.41 WMM: BSM: 0.86
k2 k3 WMM: 0.82 WMM: 0.62 -0.41 WMM: 0.97
⎛θ ⎞ ⎛τ ⎞
Mr = k1 ⎜ ⎟ ⎜ oct + 1⎟
⎝ ρ⎠ ⎝ ρ ⎠
360
second model is the model due to Uzan (1999) which takes into account the shear stresses
that develop in the material and it is applicable for both granular stress-hardening and cohe-
sive stress-softening materials. The model which is used in Mechanistic-Empirical Pavement
Design Guide (M-E PDG) considers bulk stress and shear stress (τoct) (M-E PDG, 2004). ρ
in the Uzan's model and M-E PDG model (Table 4) is atmospheric pressure and is taken
as 101.35 kPa. It can be seen in Table 4 that k1and k2 were positive in the case of BSM and
WMM indicating that as bulk stress increases, MR value increases. k3 is related to shear stress
and as shear stress increases, MR tends to reduce. k3 was found to be negative for WMM
which indicates that the shear stress developed soften the material due to which MR value
was reduced. Few researchers have found that k3 value to be positive in the granular material
and attributed it to the lower liquid limit of the soils (Hossain, 2010; George, 2004). k3 for
BSM was found to be positive and can be attributed to the high “cohesive” nature of BSM.
As can be seen, the use of material parameters such as resilient modulus and the associated
such models for BSM may not be appropriate as these models were developed to characterize
typical granular material response. It is clearly seen that while BSM might have considerable
amount of granular material content, due to the influence of binder and active filler in addi-
tion to the binders in RAP, the response of the material is more like a “cohesive” material
and it will be interesting to compare its response to the bituminous mixtures. Such analysis is
carried out in the next section.
361
during loading and unloading data of 100th cycle of 12th sequence (confining pressure of
103.4 kPa and axial stress of 206.8 kPa), is shown in Figure 8a. It was seen that during
unloading, the strain recovery of BSM and WMM were different. Thus, it can be inferred
that even though the BSM exhibits pressure dependent characteristics, the rate dependence
plays a more critical role. In an attempt to compare identical loading scenario for BSM and
DBM, the strain response during 10 Hz frequency of dynamic modulus testing was chosen
since the loading time is 0.1 seconds. It can be seen in Figure 8b that the recovery of BSM
during unloading was similar to that of the DBM material. The material shows behavior
similar to bituminous material and hence any test method and the associated post-processing
method which considers explicitly the rate dependent characteristics of the material can be
considered to characterize BSM.
Attribute Value
Figure 7a. Dynamic modulus as function of Figure 7b. Phase lag as function of frequency.
frequency.
362
Figure 8a. Normalized strain during MR test. Figure 8b. Normalized strain during dynamic
modulus test.
4 CONCLUSIONS
The main aim of this paper is to explore the mechanical response of BSM and compare with
a regular granular and HMA material. Repeated load triaxial test was carried out on BSM
samples at room temperature. The manner in which the MR values changed for WMM and
BSM was significantly different. The fact that the strains needed more time to recover for
BSM was clearly seen. Also, the history of loading played a critical role for BSM in the sense
that when the confinement pressure increased, the MR values actually decreased for the first
sequence. Since the MR values did not follow the expected trend, it is not logical to use any
MR models and this was demonstrated.
During dynamic modulus test, the viscoelastic characteristics of BSM, i.e. loading rate
dependency, was clearly observed. The normalized strain response during unloading was
found identical for DBM and BSM. While MR testing showed disparity between WMM
and BSM in terms of the MR value, no such disparity was seen when BSM and DBM were
compared. Hence one can conclude that the response of the BSM is predominantly similar
to that of a bituminous material and appropriate testing protocols and post-processing meth-
ods are necessary.
ACKNOWLEDGEMENT
The authors acknowledge M/s WIRTGEN, Germany for the donation of WLB-10S foam-
ing equipment and WLM-30 pug mill mixer to the pavement engineering laboratory at IIT
Madras. The authors acknowledge the technical assistance provided by M/s IPC Global,
Australia during the conduct of the experiments.
REFERENCES
AASHTO T-307–99. 2007. Standard Method of Test for Determining the Resilient Modulus of Soils
and Aggregate Materials. American Association of State Highway and Transportation Officials,
Washington D.C., USA.
AASHTO TP-79. 2010. Determining the dynamic modulus and flow number for Hot Mix Asphalt
(HMA) using the Asphalt Mixture Performance Tester (AMPT). American Association of State
Highway and Transportation Officials, Washington D.C., USA.
ASTM D 6931. 2010. Standard test method for Indirect Tensile (IDT) strength of bituminous
mixtures.
American Society for Testing and Materials Standards, West Conshohocken, Pennsylvania, USA.
Asphalt Academy. 2009. Technical guidelines: Bitumen stabilized materials, guidelines for the design
and construction of bitumen emulsion and foamed bitumen stabilized materials. Technical Report 2,
Pretoria, South Africa.
363
AASHTO TP-79. 2010. Determining the dynamic modulus and flow number for Hot Mix Asphalt
(HMA) using the Asphalt Mixture Performance Tester (AMPT). American Association of State
Highway and Transportation Officials, Washington D.C., USA.
ASTM D 6931. 2010. Standard test method for Indirect Tensile (IDT) strength of bituminous
mixtures.
American Society for Testing and Materials Standards, West Conshohocken, Pennsylvania, USA.
Asphalt Academy. 2009. Technical guidelines: Bitumen stabilized materials, guidelines for the design
and construction of bitumen emulsion and foamed bitumen stabilized materials. Technical Report 2,
Pretoria, South Africa.
George, K.P. 2004. Resilient modulus prediction employing soil index properties, FHWA/MS-DOT-
RD-04–172, Mississippi, USA.
Halles, F. and Thenoux, G. 2009. Degree of influence of active fillers on the properties of recycled
mixes with foamed asphalt. 88th TRB Annual Meeting (CD-ROM), Transportation Research Board,
Washington, D.C., USA.
Hicks, R.G. and Monismith, C.L. 1971. Factors influencing the resilient properties of granular materi-
als. Transportation research record, No. 345, 15–31.
Hossain, M.S. 2010. Characterization of unbound pavement materials from Virginia sources for use in
the new Mechanistic-Empirical Pavement Design procedure, FHWA/VTRC 11-R6,Virginia, USA.
Huan, Y., Siripun, K., Jitsangiam, P., and Nikraz, H. 2010. A preliminary study on Foamed bitumen
stabilization for Western Australian pavements. Scientific Research and Essays, 5 (23), 3687–3700.
IRC:109–1997. 1997. Guidelines for Wet Mix Macadam. Indian Roads Congress, New Delhi, India.
IS: 73–2013. 2013. Paving bitumen - specification. Bureau of Indian Standards, New Delhi, India.
Jenkins, K.J., Long, F., and Ebels, L.J. 2007. Foamed bitumen mixes = Shear performances?. Interna-
tional Journal of Pavement Engineering, 8(2), 85–98.
Khosravifar, S., Schwartz, C.W., and Goulias, D.G. 2013. Mechanistic structural properties of foamed
asphalt stabilized base materials. 92nd Annual Meeting (CD-ROM), Transportation Research Board,
Washington D.C., USA.
Kim, Y., Lee, H., and Heitzman, M. 2009. Dynamic modulus and repeated load tests of cold in-place
recycling mixtures using foamed asphalt. Journal of Materials in Civil Engineering, 21 (6), 279–285.
M-E PDG. 2004. NCHRP - Guide for mechanistic - empirical design of new and rehabilitated pave-
ment structures. National Cooperative Highway Research Program, Transportation Research Board,
National Research Council, Washington, D.C., USA.
MoRTH. 2001. Specification for roads and bridge work. Indian Roads Congress, 4th revision, New
Delhi, India.
Sparsha, N.S. 2015. Characterization of unbound granular material for bituminous pavements. M.S.
Thesis, Department of Civil Engineering, Indian Institute of Technology Madras, India.
Vorobeiff, G. 2005. Design of foamed bitumen layers for roads. Technical report No. 44, Australian
Stabilisation Industry Association (Auststab), Queensland, Australia.
Weston, C., Amsterdam, E., and Jenkins, K. 2002. A study into the mechanical properties of foamed
bitumen stabilized materials. 21st Annual South African Transport Conference, July pp. CD–ROM.
Uzan, J. 1999. Granular material characterization for mechanistic pavement design. Journal of trans-
portation engineering, 125 (2), 108–113.
364
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
M. Gong
Southeast University, Nanjing, Jiangsu, China
Purdue University, West Lafayette, Indiana, USA
H. Yao
Southeast University, Nanjing, Jiangsu, China
T. Pauli
Western Research Institute, Laramie, Wyoming, USA
ABSTRACT: Recent researches indicate that phase separation behavior in asphalt is wax
induced and the rippled phase (so-called bee structure) may consist of wax material. This study
aims to investigate the wax-induced phase separation behavior and its impact on asphalt's
rheological properties. Two asphalts were blended with three kinds of commercial petroleum-
derived microcrystalline waxes. In order to obtain the details of microcrystalline wax, nuclear
magnetic resonance was used to investigate molecular structure of microcrystalline waxes.
Phase separation behaviors of asphalts blended with microcrystalline waxes were scanned
using atomic force microscopy. Fourier transform infrared results of all samples were also
provided. Rotational viscosity and bending beam rheometer were employed to characterize
asphalts' rheological properties. Relationship between phase-separation behavior and macro
properties is investigated by correlating atomic force microscopy result and rheological
properties. The formation mechanism of bee structure is proposed.
1 INTRODUCTION
Existing research indicates that asphalt will separate into several phases under certain condi-
tions at micro scale (Masson et al. 2007; Pauli et al. 2011; Das et al. 2012, 2013; Nahar et al.
2014). Typically, its separation behavior can be influenced by several factors, e.g. temperature,
aging, modification and so on. Knowledge at such scale (10−4∼10−6 m) will provide a compre-
hensive understanding of asphalt’s property and thus may promote a better prediction of
asphalt pavement’s performance through both experimental and simulation methods.
In the past decade, research on asphalt’s separation behavior through atomic force micro-
scopy has drawn great attention all over the world. AFM is able to characterize both micro-
structure and micromechanical properties of asphalt without causing damage to scanned
sample, which shows promising application prospects in asphalt research field.
Recent emphasis of research pertaining to AFM studies of asphalt has tended to focus on
the morphological characteristics of asphalt surface, specifically the wrinkled “bee” phase.
Although great effort has been made to establish the relationship between the so-called
“bee structures” and macro properties (e.g. cracking, healing, thermal behavior and so on),
a precise interpretation of bee structure chemical composition remains somewhat elusive.
365
It has actually been widely discussed in the literature that regardless of whether bee struc-
tures mainly consist of asphaltenes or waxes, either fraction will have profound but distinct
influence on asphalt’s rheological properties. Since the nature of the bee structures remains
elusive, the microstructure to bulk property relationship is still not established. In order to
address these limitations, this paper attempts to investigate the chemical composition and
formation mechanism of bee structures. Asphalt with distinct phase-separation behaviors
induced by added microcrystalline are tested based on physicochemical composition and
rheological properties. The relationship between asphalt’s morphology features and rheologi-
cal performance is then discussed.
2 LITERATURE REVIEW
The application of Atomic Force Microscopy (AFM) has significantly extended researchers’
knowledge of asphalt binder surface microstructure. In 1996, with the aid of AFM, Loeber
observed rippled structures embedded in the surface of asphalt samples. Since these struc-
tures had similar features as bumble bees, Loeber named them “bee structures” and attrib-
uted the formation of these rippled structures to the aggregation of asphaltenes (Loeber
et al. 1996). Subsequently, several research groups began to investigate bee structure through
AFM. Some group concurred with Loeber’s opinion that asphaltenes could be responsible
for the structure of asphalt surface (Pauli et al. 2001). A few researches indicated that for-
mation of bee structure may also be related to asphalt’s metal content (Masson et al. 2006).
Researchers later argued that these structures more likely originate from wax (Schmets et al.
2010; Pauli et al. 2011, 2015). These controversies can be attributed to the complex composi-
tion of asphalt.
It is apparent that observation on asphalt’s microstructure/phase-separated behavior using
AFM is a phenomenon-based method to investigate asphalt material. And it was found that
AFM result was not in consistent with traditional colloid model. Thus, interpretation of
these microstructures would provide in-depth understanding on asphalt’s property at nanos-
cale and promote the establishment of new novel micro-model to characterize asphalt.
3 OBJECTIVES
By posing the hypothesis that bee structures consist of micro-crystalline wax, this paper
focuses on the interpretation of asphalts’ microstructure observed by AFM and establish-
ment of the relationship between microstructure and rheological properties. The main objec-
tives of this paper are as follows:
1. Following Pauli’s work in 2011 (Pauli et al. 2011), this study added different commercial
petroleum-based microcrystalline waxes into two asphalts at various dosages. Different
contents of microcrystalline waxes may contribute to distinct morphology features of
asphalt sample. Analyzing the molecular structure of these microcrystalline waxes may
help identify the substance which forms bee structure in asphalt.
2. Proton Nuclear Magnetic Resonance (H-NMR) spectroscopy was used to analyze the
carbon chain and branch structure of microcrystalline wax. Functional groups were stud-
ied by Fourier Transform Infrared (FTIR) spectroscopy. Of special interest, long chain
index which is related to wax crystal solid was calculated and compared with roughness
of asphalt sample surface measured by AFM. All these knowledge would enhance our
understanding on bee structures’ chemical composition as well as phase separation behav-
ior of asphalt.
3. Rotational Viscosity (RV) and Bending Beam Rheometry (BBR) tests were conducted
to characterize the rheological property of all asphalt samples. The relationship between
roughness and rheological performance was analyzed to reveal the impacts of micro struc-
ture on macro properties of asphalt.
366
4 MATERIALS AND TESTS
4.1 Materials
Pen30 (Penetration: 20–40(0.1 mm)) and Pen70 (Penetration: 60–70(0.1 mm)) asphalts were
used in this study. Both asphalts were blended (high shear mixer, 140°C, 10 min) with three
different microcrystalline waxes at dosages of 1%, 2% and 3%, respectively. These microcrys-
talline waxes are derived from petroleum and their basic properties are illustrated in Table 1.
4.2 Tests
4.2.1 AFM test
Samples were prepared using heat casting method. A Bruker AFM was used to scan all the
samples at rate of 1 Hz. Phase separation behavior is evaluated using the parameter of root
mean square roughness by NanoScope Analysis software.
4.2.2 H-NMR
DRX500 Bruker NMR spectrometer was used for all microcrystalline wax samples at a fre-
quency of 500MHz and chloroform was used to dissolve microcrystalline waxes. Distinction
of protons in H-NMR spectrum is listed in Table 2.
Branch and carbon chain information of microcrystalline wax were evaluated using Branch
Index (BI) and Number of Carbons (NC) (Yang et al. 1999). These two indexes can be cal-
culated as follows:
2 Aγ
BI =
(
3 A A )
3Aβ 2 Aγ
NC =
Aγ
where Aγ is the area percentage of Hγ, Aβ is the area percentage of Hβ, and Aα is the area
percentage of Hα.
4.2.3 FTIR
NEXUS 870FT FTIR was used to obtain asphalts’ spectra. The long chain alkane
index is defined as A724/(A1460 + A1376) which is related to solid wax phases in asphalt
Microcrystalline wax
367
(Lamontagne et al. 2001; Sastry et al. 2010). To note, A represents absorbance area and sub-
script represents the wave numbers.
Figure 1. (a) Pen30 asphalt; (b) 1% 70# microcrystalline wax blended Pen30 asphalt; (c) 2% 70#
microcrystalline wax blended Pen30 asphalt; (d) 3% 70# microcrystalline wax blended Pen30 asphalt;
(e) 1% 80# microcrystalline wax blended Pen30 asphalt; (f) 2% 80# microcrystalline wax blended Pen30
asphalt; (g) 3% 80# microcrystalline wax blended Pen30 asphalt; (h) 1% 90# microcrystalline wax
blended Pen30 asphalt; (i) 2% 90# microcrystalline wax blended Pen30 asphalt; (j) 3% 90# microcrystal-
line wax blended Pen30 asphalt.
368
them decreases. This phenomenon indicates that bee structure may consist of microcrystalline
wax and would coalesce into large size microstructure when wax concentration increases. For
80# microcrystalline wax blended asphalts, bee structures become larger once 1% wax was
incorporated. However, it is shown that addition of 90# microcrystalline wax may depress the
growth of microstructure. The surfaces of 90# microcrystalline wax blended asphalts become
much flatter than those of samples blended with 70# and 80# microcrystalline wax.
Figure 2 displays images of Pen70 asphalt and microcrystalline wax blended asphalts.
Unlike Pen30 asphalt, Pen70 asphalt has long and thin microstructures on its surface. It
is shown that 70# microcrystalline wax could increase the size of bee structures, while 80#
microcrystalline wax may increase the number of small-size microstructures at low dosage and
then promote the convergence of microstructures at relative high dosages (>1%). Figure 2 also
shows that addition of 90# microcrystalline wax would limit the growth of microstructure.
Surface roughness data are further listed in Table 3. Higher roughness value usually cor-
responds to more and/or large sized bee structures. It is proved again that addition of 70#
and 80# microcrystalline wax can promote the growth of bee structure. Asphalt samples with
higher content of 70# and 80# microcrystalline wax usually show higher roughness values.
Figure 2. (a) Pen70 asphalt; (b) 1% 70# microcrystalline wax blended Pen70 asphalt; (c) 2% 70#
microcrystalline wax blended Pen70 asphalt; (d) 3% 70# microcrystalline wax blended Pen70 asphalt;
(e) 1% 80# microcrystalline wax blended Pen70 asphalt; (f) 2% 80# microcrystalline wax blended Pen70
asphalt; (g) 3% 80# microcrystalline wax blended Pen70 asphalt; (h) 1% 90# microcrystalline wax
blended Pen70 asphalt; (i) 2% 90# microcrystalline wax blended Pen70 asphalt; (j) 3% 90# microcrystal-
line wax blended Pen70 asphalt.
369
Table 3. Roughness of different samples.
Roughness (nm)
It is also found that 2% 80# microcrystalline wax blended asphalt sample has lower rough-
ness than 1% 80# microcrystalline wax blended asphalt sample, even though the former one
shows larger sized bee structures.
Although additives are all microcrystalline waxes, results in this section show that bee
structures in asphalt are more likely to consist of substance which may have similar proper-
ties as 70# and 80# microcrystalline waxes. Moreover, even 70# and 80# microcrystalline
waxes have distinct influences on asphalt’s phase separation behaviors. All these differences
can be attributed to various molecule structures of these three microcrystalline waxes. Next
section will provide more details of microcrystalline waxes by H-NMR.
5.2 H-NMR
As illustrated in above section, 70# and 80# microcrystalline wax would promote the growth
of bee structure. In order to indentify the characteristics of molecules which may be com-
prised in bee structure, this section uses H-NMR to study the micro property of microcrystal-
line wax. H-NMR spectra are displayed in Figure 3.
It is obvious that 70# and 80# microcrystalline waxes have similar H-NMR spectra. There
is only slight difference in the peak of 7.283 ppm which can be attributed to the existence of
aromatics. 90# microcrystalline wax shows an obvious peak in the chemical shift of 1.566 ppm
which comes from the R3CH structure. Branch index and number of carbon are listed in Table 4.
It is evident that all microcrystalline waxes have no benzylic structures. 70# microcrystalline wax
has a low BI but a high NC. 90# microcrystalline wax has highest BI and lowest NC. H-NMR
results indicate bee structure may comprise of microstructures whose BI is lower than 0.104458
and NC is higher than 21.1464. Previous studies proposed that bee structure is wax crystal. In
general, wax molecules with longer carbon chains and shorter/fewer branches tend to crystallize
more easily. Result in this section seems to be in consistent with previous hypothesis.
5.3 FTIR
FTIR spectra of all asphalt samples and micro-crystlline waxes are shown in Figure 4. It can
be seen that all asphalt samples and micro-crystlline waxes have similar functional groups.
Long chain indexes are further demonstrated in Figure 5. It is shown that long chain index is
much higher in micro-crystlline wax (Figure 5a) than that in asphalt (Figure 5b). It has been
pointed out in literature that long chain index has a close relationship with liquid and solid
phase long chain methylene carbons in hydrocarbons. Hence, it can be used to indicate the
existence of wax crystal (Sastry et al. 2010). It is shown in Figure 5b that only 90# microcrys-
talline wax blended asphalt shows higher long chain index as the dosage increases. However,
AFM results show that 90# microcrystalline wax blended asphalt displays fewer and smaller
bee structures than other asphalt samples. Even though bee structure may comprise of wax,
it is concluded that formation mechanism of bee structure may be different from that of
traditional wax crystal in asphalt.
370
Figure 3. Spectra of three microcrystalline waxes: (a) 70#; (b) 80#; and (c) 90#.
Wax type Hγ Hβ Hα Ha BI NC
Figure 4. FTIR spectra of: (a) micro-crystlline waxes and (b) asphalt samples.
Figure 5. Long chain indexes of: (a) micro-crystlline waxes and (b) asphalt samples.
Figure 6. Rotational viscosities of all asphalt samples at different temperatures: (a) Pen 30 asphalt with
micro-crystlline wax and (b) Pen 70 asphalt with micro-crystlline with wax.
m S m S m S
372
Figure 7. Relationship between roughness and rheological results: (a) 70# microcrystalline wax
blended Pen70 asphalt’s S; (b) 80# microcrystalline wax blended Pen70 asphalt’s S; (c) 80# microcrystal-
line wax blended Pen70 asphalt’s S; and (d) all the samples’ viscosities.
induce compression stress on the elastic crystal. Then the wax crystal may undergo a diffu-
sion induced wrinkling process through relaxation of the curvature elasticity (Safran, 1999;
Vandeparre, 2008; Chung, 2009; Huang, 2013; Na et al. 2015).
The term of curvature elasticity here is not common in traditional asphalt research field.
It was first introduced in asphalt’s AFM study by Pauli in 2015 (Pauli et al. 2015). Typically,
if a film with finite thickness is constrained to lie in a plane, there will be compression stress
on the molecules and the energy of this system is not stable. Since the membrane can deform
in the normal direction, it would band itself in order to reach a stable energy state. This
out-of-plane bending deformation is known as curvature modes, and free energy associated
with this mode is named as curvature free energy. Supposing there is a cubic crystal, the com-
pression stress can be expressed as three in-plane components, or two principal components
and one principal angle. The wrinkling pattern depends on the ration between two principal
stresses as well as the principal direction within the isotropic film (Huang, 2013). Obviously,
this stress state is directly related to the material property of wax crystal and matrix. Hence,
once this curvature model is established, the bulk property of asphalt can be analyzed/back-
calculated through investigating the surface pattern by AFM. Then AFM result can provide
more useful information in research on aging and modification of asphalt material.
373
6 CONCLUSIONS
1. 70# and 80# microcrystalline wax would promote the formation of bee structure as well
as the phase separation behavior in asphalt. H-NMR result indicates that bee structure
may consist of molecules with long carbon chains and short branches.
2. It is found that there is no close relationship between bee structure and wax solid which is
characterized by long chain index using FTIR.
3. Formation of bee structures correlates well with rheological properties of 70# and 80#
microcrystalline wax blended asphalts. A model is proposed to explain the formation
mechanism of rippled structure. It is concluded that bee structure could reflect bulk prop-
erty of asphalt matrix. This result could be further used to explain the modification mech-
anism or aging mechanism in asphalt research field.
REFERENCES
Chung, J.Y., Nolte, A.J., Stafford, C.M. 2009. Diffusion-controlled, Self-organized Growth of Symmet-
ric Wrinkling Patterns. Advanced materials, 21, 1358–1362.
Das, P.K., Jelagin, D., Birgisson, B., & Kringos, N. (2012). Micro-Mechanical Investigation of Low
Temperature Fatigue Cracking Behaviour of Bitumen. 7th Rilem International Conference on Crack-
ing in Pavements.
Das, P.K., Kringos, N., Wallqvist, V., & Birgisson, B. (2013). Micromechanical investigation of phase
separation in bitumen by combining atomic force microscopy with differential scanning calorimetry
results. Road Materials and Pavement Design, 14, 25–37.
Huang, R. 2013. A kinetics approach to surface wrinkling of elastic thin films. In X. Chen (ed.) Mechan-
ical self-assembly science and application. Springer.
Lamontagne, J., Dumas, P., Mouillet, V., & Kister, J. (2001). Comparison by Fourier Transform Infrared
(FTIR) Spectroscopy of Different Ageing Techniques: Application to Road Bitumens. Fuel, 80, 483–488.
Loeber, L., Sutton, O., Morel, J., Valleton, J.-M., & Muller, G. (1996). New Direct Observations of
Asphalts and Asphalt Binders by Scanning Electron Microscopy and Atomic Force Microscopy.
Journal of Microscopy,182, 32–39.
Masson, J.F., Leblond, V., & Margeson, J. (2006). Bitumen Morphologies by Phase-detection Atomic
Force Microscopy. Journal of Microscopy. 221, 17–29.
Masson, J-F., Leblond, V., Margeson, J., & Bundalo-perc, S. (2007). Low-temperature Bitumen Stiff-
ness and Viscous Paraffinic Nano and Micro-domains by Cryogenic AFM and PDM. Journal of
Microscopy, 227, 191–202.
Nahar, S.N., Schmets, A.J.M., Scarpas, A., & Schitter, G. (2014). Microstructural Changes in Bitumen
at the Onset of Crack Formation. European Polymer Journal,56, 17–25.
Na, J.H., Kim, S.U., Sohn, Y. and Lee, S.D.. 2015. Self-organized wrinkling patterns of a liquid crystal-
line polymer in surface wetting confinement. Soft Matter. 11: 4788–4792.
Ocko, B.M., Wu, X.Z., Sirota, E.B., Sinha, S.K., Gang, O. and Deutsch, M. 1997. Surface freezing in
chain molecules: Normal alkanes. 55(3): 3164–3181.
Pauli, T., Grimes, W., Beiswenger, J., & Schmets, A. (2015). Surface Structuring of Wax in Complex
Media. Journal of Materials in Civil Engineering. 27(8), C4014001-1.
Pauli, A.T., Grimes, R.W., Beemer, A.G., Turner, T.F., & Branthaver, J.F. (2011). Morphology of
Asphalts, Asphalt Fractions and Model Wax-doped Asphalts Studied by Atomic Force Microscopy.
International Journal of Pavement Engineering, 12, 291–309.
Pauli, A.T., Branthaver, J.F., Robertson, R.E., Grimes, W., & Eggleston, C.M. (2001). Atomic Force Micro-
scopy Investigation of SHRP Asphalts. ACS Division of Fuel Chemistry Preprints, 46, 104–110.
Safran, S.A.. 1999. Curvature elasticity of thin films. Advances in physics. 48(4): 395–448.
Schmets, A., Kringos, N., Pauli, T., Redelius, P., & Scarpas, T. (2010). On the Existence of Wax-induced
Phase Separation in Bitumen. International Journal of Pavement Engineering, 11, 555–563.
Sastry, M.I.S., Raman, N.S., Jain, N., Mukherjee, S., Kagdiyal, V., Singhal, S.K., & Sarpal, A.S. (2010).
Determination of Wax Appearance Temperature and Wax Dissolution Temperature of Waxy Crude
Oils by Temperature Programmed IR Spectroscopy. Petrotech-2010. 20100930.
Vandeparre, H., & Damman, P. 2008. Wrinkling of Stimuloresponsive Surfaces: Mechanical Instability
Coupled to Diffusion. Physical review letter. 101, 124301.
Yang, Y., Liu, B., Cheng, J., Zhang, H., Sun, X. and Gu, Z. 1999. Characteristic of 1H-NMR spectra of
the heavy oils in refinery. Journal of petrochemical universities. 12(4).
374
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: In order to explore the effect of short-cut basalt fiber on asphalt mixture fatigue
performance, AC-13 and SMA-13 aggregate gradations were chosen. Through four-point bend-
ing fatigue life test, variation of the fatigue life and cumulative dissipated energy of the asphalt
mixtures adding basalt fiber were analyzed. In order to analyze the effect on the anti-fatigue
performance of mixture by adding basalt fiber, the formula which was admitted widely was used
to fit fatigue test results. The research indicated that the adding of basalt fiber could improve the
fatigue performance of asphalt mixture effectively. At the same time, the sensitivity of mixture to
strain was reduced and the durability of asphalt concrete pavement could be improved.
1 INTRODUCTION
Fiber asphalt mixture has been proved to be an effective mean of solving the problem of
asphalt pavement damage, and have an advantage over polymer modified asphalt in areas such
as thin overlay of pavement engineering (Ludwig, 2003; Serfass, 1996; Bueno, 2003; Fitzgerald,
2000; Peng, 2006). Following the organic fibers, asbestos fiber fibers, basalt fiber has caused
more attention in countries with excellent characteristics. basalt fiber is a new type of fiber,
it can made up of low strength of organic fiber and elastic modulus and high temperature
performance shortcomings, also can avoid the environment under threat and defects of the
human body caused by asbestos and other mineral fibers, and can be recycled (Sayed, 2010;
Huang, 1996; Zhang, 2006; Guo, 1991; Li, 2009; Guo, 2010). In the 1990s, the United States
paving the first asphalt concrete pavement mixed with basalt fiber in Georgia, and achieved
good performance. Since then, Basalt fiber asphalt mixture becomes a hot focus of research.
In this paper, AC-13 and SMA-13 gradation aere adopted, and through four-point bending
fatigue test, the role of basalt fibers on fatigue properties of asphalt mixtures was analyzed.
375
The basalt fiber was produced by Jiangsu Tianlong Continuous Fiber High-tech Limited
Company, and cellulose fiber, German Rettenmeyer Father & Son’s company. Cellulose fiber was
used only for SMA-13 gradation, basalt fiber was used both in the AC-13C and SMA-13 grada-
tion. The basalt fiber performance parameters was listed in Table 5, and cellulose fiber perform-
ance test results, Table 6. Two kinds of fiber were shown in Figure 1 and Figure 2.
Standard test
Test Standard result Test result method
Standard
Test Standard result Test result test method
Limestone Basalt
Aggregates 2# 3# 4# 1# 2# 3# 4#
376
The optimum asphalt content of fiber asphalt mixture is always larger than that with-
out fiber, owing to the oil absorbing ability of fiber (Yang, 2009). According to the related
experimental results, we could get the information that the bitumen aggregate ratio of general
AC-13C asphalt mixture was 5.1% and the basalt fiber asphalt mixture was 5.3%.
The amount of cellulose fiber was 3% of asphalt mixture’s mass when we combined with
practical engineering. According to “Technical Specification for Construction of Highway
Standard test
Test Standard result Test result method
length/mm 6 0.8
Ash content /% 18 ± 5 19.3
PH 7.5 ± 1.0 7.7
oil absorption /% 5 6.2
Water content /% 5 1.8
377
Table 7. Data of AC-13C aggregation.
Index
Index
Mixture Oil-stone Bulk specific Marshall Fort shellen Ken
type ratio /% gravity /g/cm3 air voids/% stability/kN leakage loss flying loss
Asphalt Pavement” (JTG F40—2004), the results of SMA modified asphalt mixture design
and Marshall Test was shown in Table 9–10.
According to the related experimental results, we could get the information that the bitu-
men aggregate ratio of general SMA-13 asphalt mixture was 6.1% and basalt fiber asphalt
mixture was 6.0%.
378
2.3 Test procedures
The method of asphalt mixture fatigue test in this paper is four-point bending fatigue life
test. The test was conducted on the Australian production of UTM–25. Test temperature
was15°C±0.5°C and the loading frequency was 10 Hz ± 0.1 Hz and continuous partial sine
loading mode controlled by the constant strain was adopted. The test would terminate when
the bending stiffness modulus of the specimens decreased to 50% of the initial modulus. The
size of the specimen was 380 mm × 50 mm × 63.5 mm, and the permissible error was 5 mm.
All the specimens was cut by the rolling and compacting asphalt mixture with the size of
400 mm × 300 mm × 75 mm. Specific test procedures are as follows:
1. Specimen preparation. First, the asphalt mixture was mixed indoors by using type SYD
– F02–20 Marshall automatic mixer produced by Shanghai ChangJi Geological Instru-
ment Co Ltd. Second, asphalt mixture plate specimen 400 mm × 300 mm × 75 mm size
was made by type QCZ-4 grinding tester produced by Beijing Valley and Arrow Com-
pany. Third, plate specimen would cut into the size of the test requirements by the cutting
machine produced by SAST Jiangyan Instrument Factory.
2. The selection of strain level. According to the rules, in order to get a complete fatigue
property, at least three stress levels should be chosen and the appropriate number of
fatigue life should cover the scope from thousands of times to millions of times. There-
fore, three strain control levels of 250 με, 450 με and 650 με were chosen in this paper for
the AC—13C mixture, SBS modified asphalt with the gradation AC-13C and SMA-13,
450 με, 650 με and 850 με were selected as strain control level.
3. The maintenance of specimens. Small beam specimens would be put into the box of plus
or minus 0.5°C environment and maintenance time should be not less than 4 hours.
4. Began to test. After installed correctly, the specimens should be preloaded 50 cycles under
the target stress level to determine the initial stiffness modulus. The test would be started
when the required target strain level was reached after preloading. Test processes were
shown from Figure 3 to Figure 8.
379
Figure 5. Grinding tester. Figure 6. Asphalt mixture plate specimen.
3.1 Data analysis and comparison based on fatigue life and cumulative energy dissipation
Results of asphalt mixture fatigue test were shown in Table 11 and Table 12.
By analyzing the test results, the discovery was that anti-fatigue performance of mixture
attenuated greatly with improvement of control strain level. But in general, the anti-fatigue
performance of mixture attenuated slower after adding basalt fiber. The anti-fatigue per-
formances of both AC-13C gradation and SMA-13 gradation improved greatly after using
basalt fiber. Especially in SMA-13 grading, the fatigue strength and cumulative energy dissi-
pation were nearly four times higher than that of using lignin fiber. The parameters improved
more than double in the SBS modified asphalt of AC gradation and around 70% in the
matrix asphalt of AC gradation. It means that basalt fiber could greatly improve the fatigue
performance of asphalt mixture.
double logarithmic was taken to fatigue life and strain level as lnNf=lnC-mlnε. And then
linear fitting was carried out on the fatigue life and the strain level under the logarithmic
coordinate, results was shown in Figure 9.
380
According to the results of linear fitting, the value C and m in formula would be deter-
mined, and the results were listed in Table 13.
In the formula, C and m represent material parameters. The larger the parameter C was,
the better fatigue resistance of material was. Parameter m represents the sensitive degree of
material on the strain level. And the smaller the value m was, the less sensitive the material
was to the strain level. As shown in table, after adding basalt fiber, the C values increased but
m values reduced both in AC grading and SMA gradation. We could get conclusion that after
adding fiber, the fatigue performance of asphalt mixture increased but its sensitive degree to
strain level decreased. Comparing the data, it could also be found that the sensitivity of the
matrix asphalt was smaller than SBS modified asphalt.
There were three reasons for the result. First, basalt fiber dispersed in the asphalt mixture
overlap each other and then forms a three-dimensional network. This structure limits the
development of fatigue cracks in some degree. Second, fiber can disperse the stress concen-
tration and lift off the accumulation of strain energy, the toughness of the asphalt mixture
will be improved. Finally, because the fiber has certain oil absorption ability, which makes
fatigue
fatigue life energy life /10 energy fatigue life energy
/10 thousands dissipation thousand dissipation /10 thousand dissipation
Mixture type times /J/m3 times /J/m3 times /J/m3
Based asphalt 71.131 473.472 1.128 19.609 0.172 4.605
AC-13
BF+Based 115.997 761.447 2.255 34.606 0.244 6.067
asphalt AC-13
381
Table 13. Results of Fatigue equation parameter
parameter
Mixture type C m
the optimum proportion of basalt fiber asphalt mixture higher than others, more asphalt can
also help improve the anti-fatigue performance of mixture.
4 CONCLUSIONS
The corresponding asphalt mixtures were prepared to do the research on fatigue perform-
ance of basalt fiber asphalt mixture. In this paper, main conclusions are as follows:
1. After adding basalt fiber, the fatigue life of mixture increased largely. It suggested that
basalt fiber could improve fatigue life of asphalt mixture greatly.
2. Comparing the mixture before and after adding basalt fiber, the changing trend of the
cumulative energy dissipation was less than the fatigue life. After adding basalt fiber, the
energy dissipations of the damage reduced and the fatigue life tended to be prolonged.
3. Through the fatigue equation fitting, it could be found that the addition of basalt fiber
could improve the fatigue performance of asphalt mixture effectively and reduce the sen-
sitivity to strain levels.
REFERENCES
Bo Peng, Ning feng Guan, et al. 2006.Experimental study on the fiber asphalt mixture properties [J].
Journal of China and Foreign Highway, 26(1):128–130.
Bueno, B.D., W.R. da Silva, D.C. de Lima, et al. 2003, Engineering properties of fiber reinforced cold
asphalt mixes [J]. Journal of Environmental Engineering, 129(10):952–955.
Fitzgerald, R.L. 2000.Novel application of carbon fiber for hot mix asphalt reinformortar and carbon-
carbon performs [D]. Michigan Technological University, Master Dissertation,
Huan Guo, Yan Ma, et al. 2010. Development and Application Prospect of Continuous Basalt Fiber [J].
China Fiber Inspection, (3):76–78.
Huang, H., White T.D. 1996. Dynamic Properties of Fiber-modified Overlay Mixture [J]. Trans Res
Rec., 1545:98–104.
Ludwig Dipl-Ing, Steve, B. 2003.The influence to the mechanism properties of bituminous mastics and
thus to open grain asphalt using synthetic fibers[C]//AAPA 12th International Flexible Pavements
Conference: 1–9.
Ming Yang, Xiao-ning Zhang. 2009. A Study on the Performance of Mineral Fiber Asphalt Mixture [J].
Journal of China and Foreign Highway.2 (6):254–257.
Nai Guo, Qiao Niu. 1991. A Brief View on the Harm and Control of Asbestos Fiber Dust [J]. Occupa-
tional Medicine, 18(5):318.
Quan-ming Li, Xuan Xiao, et al. 2009.Research Progress of Continuous Basalt Fiber and Its Composite
Materials [J]. HI-Tech Fiber and Application. 34(5):33–35.
Sayyed Mahdi Abtabi, 2010.Mohammad Sheikhzadeh. Fiber reinforced asphalt-concrete - A review [J].
Construction and Building Materials. 24:871–877.
Serfass, J.P., Samanos, J. 1996.Fiber-modified asphalt mixture characteristics applications and
behavior[J]. Asphalt Paving Technologier, 65:193–230.
Yan-jun Zhang. 2006. Performance Analysis of Highway Engineering Fiber Asphalt Concrete [D].
Shandong University, Master Dissertation, 20–21.
382
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
Recently, the high rate of urbanization has greatly accelerated economic and social develop-
ment of some countries, fast-growing cities are engines of financial growth and global cent-
ers of innovation, but urbanization has also caused several environmental concerns such as
increased air pollution, local climate alteration and increased energy demands (Cui, 2012).
Traffic congestion has been managed by either constructing new roadways or renovat-
ing the existing transport infrastructures, thus requiring large amounts of building materials
such as aggregates and bitumen. The growing interest in environmental sustainability, as
ascertained by the EU conference in Paris in 2015, has asked for a wiser use of available
resources to make them last longer and minimize their impact on the ecological system dur-
ing their life cycle.
The adoption of reclaimed asphalt pavement (RAP) material has been evaluated since
the 1970s but has gained momentum in the last decade due to increased number of studies
on performance assessment of RAP mixes by researchers and funded programs. Recycled
materials in road pavements have the potential to avoid disposal of waste in landfills and
reduce the energy consumption and the costs for supplying of virgin materials (Martinez et
al., 2015). However, Life-Cycle Analyses should be always performed to get a precise idea
of the benefits of using such materials during the pavement service life (Chiu et al., 2008)
(Giustozzi et al., 2012).
Including RAP in asphalt mixtures entails the introduction of aged and hardened binder,
which has already suffered a process of oxidation during his initial production and conse-
quent service life, thus being much stiffer than desired. One of the common practices to cope
383
with the aging effect of reclaimed binder from RAP consists in using rejuvenator agents, for
instance; the aim is to replace part of the maltenic phase lost during aging and provide ‘reju-
venated’ mechanical characteristics somewhat equivalent to those of virgin binders.
Increasing the amount of RAP binder within a binder blend induces an increased viscosity
at mixing and compaction temperatures and makes the blend stiffer (Colbert and You, 2012).
Even if the addition of rejuvenators could make technically possible the use of 100% RAP
mixes (Silva et al., 2012), road pavement designers are often wary about the effectiveness of
very high RAP content mixes to ensure adequate long time performance.
The effect of different type of rejuvenators on the rheological, physical, and chemical per-
formance of the asphalt binder has been subject of many studies (Zaumanis et al., 2014) (Asli
et al., 2012) (Zargar et al., 2012) (Romera et al., 2006) (Zaumanis et al. 2013) (Chen et al.,
2007). Conventional binder tests such as softening point, penetration and dynamic viscos-
ity, disclosed the rejuvenators’ ability to reduce stiffness and viscosity of RAP binder, thus
improving their cracking resistance at low temperatures and enhancing workability during
construction. Different type and percentage of rejuvenators can be adopted to achieve the
optimal correlation with the amount and type of RAP binder (Carrion A., 2015).
The adoption of Dynamic Shear Rheometer to investigate the trend of the complex shear
modulus of rejuvenated binders is common in pavement research (Ongel and Hugener, 2015)
(Shen et al., 2001) (Shen et al., 2007a) (Shen et al., 2007b) (Dony et al., 2013). When the test
is performed for different aging conditions—i.e., simulated through RTFOT (Rolling Thin
Film Oven Test) and PAV (Pressure Aging Vessel) – it is able to evaluate binder long-term
performance, including fatigue behavior.
Conventional DSR high-temperature tests can also be supplemented with Multiple Stress
Creep Recovery (MSCR) test, for instance, to get a comprehensive understanding of asphalt
binder’s high-temperature performance and correlate it to the HMA rutting behavior (Yang
and You, 2015) (Wasage et al., 2105) (D’Angelo, 2009) (Zhang et al., 2015).
The present research focused on the characterization of oxidized asphalt binders contain-
ing rejuvenator at different percentage; in particular, asphalt was rejuvenated without any
heating of materials. Rejuvenator was cold applied to 100% RAP mixes and rejuvenated
binder was cold-extracted for further testing. The experimental plan is further detailed in the
following sections.
2 EXPERIMENTAL PLAN
2.1 Materials
RAP material was gathered from surface course milling of existing road pavement; the sur-
face layer was initially constructed 12 years before according to the road agency in charge
of maintaining the road. It is indeed common practice, according to the records, to mill and
resurface after 10–12 years.
RAP was initially screened and the coarse part above the 12.5 mm sieve was eliminated
to avoid crusts resulting from milling. Gradation sieve size was analyzed before and after
the extraction of binder; as expected, before-extraction curve is almost free of fine particles
(Figure 1). Dotted lines in Figure 1 represent possible gradation upper and lower limits for
new wearing courses according to local technical specifications.
The binder content extracted from RAP was 5.24% (±0.11%) by weight of the aggregates
and Theoretical Maximum Density (TMD) of RAP was 2.439 g/cm3, according to the stand-
ard EN 12697-5.
The vegetal oil-based rejuvenator, to be applied directly on cold RAP, is dark-brown
colored and liquid at ambient temperature with specific gravity of 0.94 g/cm3; viscosity is
400–500 cP at 25°C and flaming point is 110°C. Brown color is given by a small part of fluxed
bituminous binder within the rejuvenator matrix.
Mixing time with RAP was 8–10 minutes; there is no need to dry the RAP although
a relatively low humidity of 2–4% is preferred for optimal results. Once rejuvenated, the
384
Figure 1. Sieve size gradation curve of RAP material (before and after binder extraction).
asphalt mix could be stored up to 72 hours and still retained workability; however, applica-
tion within 48 hours after mixing was suggested to avoid excessive strengthen of the mix
during compaction.
RAP was kept into a climate chamber before applying the rejuvenator to assess differences
in the rejuvenated binder performance due to initial RAP humidity.
3 ANALYSIS OF RESULTS
385
It can be noted that oxidized binder from RAP had, as expected, the smallest penetration
and the greatest softening point values; in addition, more than 2% of rejuvenator drasti-
cally increased penetration value and exhibited very low ring and ball temperature. How-
ever, standard bitumens for asphalt paving in temperate climate zones are usually within
the 70–100 dmm penetration range and 42–46 °C softening temperature. Less than 1.5% of
rejuvenator provided very small benefit according to physical properties testing.
Binders’ behavior after aging showed the variable effectiveness of rejuvenator contents; 1%
or less provided almost no benefit respect to oxidized original RAP binder while more than
1.5% still preserved greater penetration and lower softening point temperature.
These tests, although still greatly used as a first straight judgment on bitumen’s perform-
ance, could not be considered conclusive due to their very empirical nature; further rheologi-
cal tests had thus to be performed.
Brookfield viscosity test (EN 13302) was conducted on binder samples between 90 and
200°C when the binder is supposed to approximate Newtonian fluid behavior. Results are
shown in Figure 2; again, before-after PAV aging behavior is represented.
It can be noted that dynamic viscosity decreased whenever rejuvenator content was
increased for all temperatures. At 120°C, for instance, viscosity was as follows (before-PAV
condition):
− 11.027 Pa*s for oxidized RAP binder
− 1.948 Pa*s for 1% rejuv.
− 1.359 Pa*s for 1.5% rejuv.
− 1.220 Pa*s for 2% rejuv.
− 0.557 Pa*s for 3% rejuv.
PAV aging increased binder viscosity; however, rejuvenation of binders retained its effec-
tiveness after PAV since resulting viscosity was lower than oxidized RAP binder’s viscosity at
all temperatures. Smaller percentage of rejuvenator (i.e., 1%) resulted, as previously assessed,
in very similar performance to oxidized RAP binder.
Although viscosity test is more illustrative of real binder behavior respect to empirical
penetration and softening point tests, it still evaluates binder behavior in a very specific state
0 5.24 0.11
1 6.0 0.25
1.5 6.16 0.12
2 7.55 0.20
3 7.93 0.29
Soft. point
Rejuv. content Pen (before PAV) Pen (after PAV) Soft. point (before PAV) (after PAV)
0 5.4 - 71.8 -
1 14.8 −2.8 −60.9 69.7
1.5 24.2 15.6 57 60.4
2 34.8 18.2 51.7 60.2
3 95.4 50.6 44.7 52
386
Figure 2. Dynamic Brookfield viscosity of rejuvenated binders.
(Newtonian fluid, high temperature). Rheological test were therefore conducted over the in-
service pavement temperature domain.
387
Table 3. Low-temperature behavior of rejuvenated binders.
Test
temperature 0°C −6°C −12°C −18°C −24°C −30°C
Test
parameters S (MPa) m S (MPa) m S (MPa) m S (MPa) m S (MPa) m S (MPa) m
*bold values represent the failure temperature (whether S > 300 MPa or m 0.3)
Stiffness of binders decreased when the rejuvenator content was increased, this was even
more notable in the low-frequency domain (meaning also, high-temperature testing due to
time-temperature superposition principle). Oxidized and aged binder commonly exhibits an
asphaltene-reticulated structure where the asphaltene and resin components are strictly linked
to each other; the binder therefore behaves as in a sort of ‘gel’ state. Oil-based rejuvenators
tend to restore the maltenic part of the bitumen by dispersing reticulated asphaltene-resin
structure into oil. An excessive amount of rejuvenator leads to disproportionate the oil con-
tent and thus softens the binder, especially at high temperatures. However, a proper content
of rejuvenator (i.e., 2%) showed a similar rheological behavior of neat bitumen.
PAV aging increased stiffness, as expected; however, rejuvenated binders still preserved
a lower stiffness respect to oxidized RAP binder regardless of rejuvenator content. This is
significant at high frequencies (low temperatures testing) in order to improve resistance to
thermal cracking even after long-term aging.
DSR analysis in amplitude sweep mode was adopted to evaluate rejuvenated binder’s fatigue
behavior according to AASHTO TP101-14. Tests were conducted at 20°C, constant frequency
of 10 Hz and variable strain; the loading scheme consists of a continuous oscillatory strain
sweep. Loading is increased linearly from zero to 30% over the course of 3100 cycles. Binders’
complex modulus (G*) and phase angle (δ) were recorded every ten load cycles.
The damage accumulation in the specimen was computed according to the following equa-
tion (Kim et al., 2006):
D(t ) ≅ ∑ i=1[ * − C )]
N
(C * (t i − t i 1 )1+α (1)
where C(t) = G*(t)/G*(t0) which is |G*| at time t divided by the initial “undamaged” value
of |G*|; γ0 = applied strain for a given data point in percent; |G*| = complex shear modu-
lus in MPa; α = parameter depending on the storage modulus G′ value; t = testing time in
seconds.
388
Figure 3. Master curves of rejuvenated binders, before and after PAV aging.
The Damage factor D(t) progressively increases during the test and, consequently, the
complex shear modulus (G*) decreases even if temperature and frequency remain constant.
In particular, the loss modulus (G″ = |G*| sinδ) is usually represented against load cycles as a
fatigue-related parameter. Figure 4 depicts fatigue behavior of rejuvenated binders.
389
Table 4. PG of rejuvenated binders.
G*sinδ @25°C
Binder PG X-Y [kPa]
RAP binder had severely aged and grades as PG 88–10 but the rejuvenator addition lowered both the
PG temperatures. In particular, increasing rejuvenator content led to lower the high-PG (PG X) while
expand the low-PG (PG Y) value.
Although the oxidized RAP binder exhibited greater stiffness than rejuvenated binders, it
also had a steep reduction of G* after few load cycles. Due to oxidation, indeed, RAP binder
lost its elasticity and behaved like a brittle material with almost no capacity to plastically
accumulate damage before failure. Plot of D(t) emphasizes this behavior.
Rejuvenated binders provided a greater capacity of accumulate damage before failure and,
for specific contents, very similar to neat bitumen. Further testing on asphalt concrete sam-
ples should be conducted to confirm the fatigue behavior of binders due to rejuvenators.
Finally, Performance Grade (PG) of rejuvenated binders was evaluated according to
Superpave principles (Table 4) in AASHTO M320.
Particular rejuvenator contents (i.e., 1.5 and 2%) provided similar PG of neat binder, even
enhancing low-temperature performance and exhibiting the same resistance to rutting (G*/
sinδ criteria at high temperature). This showed that, despite general concerns, the proper dos-
age of rejuvenator could retain an acceptable rutting susceptibility; however, further testing is
necessary on asphalt concrete samples to confirm the rutting potential of rejuvenated binder.
Other phenomena should indeed be taken into account such as aggregate-binder adhesion,
proper blending of rejuvenator into the mix, etc.
The Superpave intermediate temperature fatigue parameter (G*sinδ) at 25°C showed a
significant reduction provided by the rejuvenator, respect to oxidized RAP binder which was
very close to the 5000 kPa maximum limit. Again, neat binder provided very similar value to
1.5 and 2% rejuvenator contents.
4 CONCLUSIONS
The experimental campaign investigated the rejuvenation of 100% RAP by means of veg-
etable oil-based rejuvenator. The latter was cold-applied to RAP material and no heat was
provided during the rejuvenation process. Based on the results obtained, the following con-
clusions can be drawn:
− 100% RAP mixes, or high-content RAP mixes, can be recycled and adopted as a new pave-
ment material under certain circumstances; technology now allows for recycling products
which are able to provide similar characteristics to virgin materials. However, an in-depth
knowledge of product to be recycled, recycling agent, and their interconnected properties,
is of primary importance.
− Increasing the amount of rejuvenator typically provides greater penetration, reduced ring-
and-ball temperature, lower dynamic viscosity, reduced stiffness, and decreased both PG
temperatures respect to oxidized RAP binder.
− After PAV aging, rejuvenated binders still preserved better performance (physical and rheo-
logical) than unaged RAP binder. However, it should be noted that PAV procedure is a labo-
ratory-controlled method which is intended to simulate between 5 and 10 years of in-service
aging while RAP material could have undergone greater aging during real conditions.
390
− Fatigue behavior of rejuvenated binders, evaluated through the linear amplitude sweep
test, showed a major ability to accumulate damage respect to oxidized RAP binder and
similar to neat bitumen.
− Specific rejuvenator dosage (1.5–2% by RAP weight) proved to perform equal or better
than virgin binder; on the other hand, less than 1.5% and more than 2% by RAP weight
was shown to be detrimental in terms of performance.
REFERENCES
Asli, H., Ahmadinia, E., Zargar, M., & Karim, M.R. (2012). Investigation on physical properties of
waste cooking oil–Rejuvenated bitumen binder. Construction and Building Materials, 37, 398–405.
Carrion, A.; Lo Presti, D.; Airey, G. (2015). Binder design of high RAP content hot and warm
asphalt mixture wearing courses. Road Materials and Pavement Design, 16 (1), 460–474. doi:
10.1080/14680629.2015.1029707.
Chen, J.S., Huang, C.C., Chu, P.Y., & Lin, K.Y. (2007). Engineering characterization of recycled asphalt
concrete and aged bitumen mixed recycling agent. Journal of Materials Science, 42(23), 9867–9876.
Chiu, C.T., Hsu, T.H., & Yang, W.F. (2008). Life cycle assessment on using recycled materials for reha-
bilitating asphalt pavements. Resources, Conservation and Recycling, 52(3), 545–556.
Colbert, B., & You, Z. (2012). The properties of asphalt binder blended with variable quantities of recy-
cled asphalt using short term and long term aging simulations. Construction and Building Materials,
26(1), 552–557.
Cui L., Shi J., Urbanization and its environmental effects in Shanghai, China, Urban Climate, Volume
2, December 2012, Pages 1–15, ISSN 2212-0955.
D’Angelo, J.A. (2009). The relationship of the MSCR test to rutting. Road Materials and Pavement
Design, 10(sup1), 61–80.
Dony, A., Colin, J., Bruneau, D., Drouadaine, I., & Navaro, J. (2013). Reclaimed asphalt concretes with
high recycling rates: changes in reclaimed binder properties according to rejuvenating agent. Con-
struction and Building Materials, 41, 175–181.
Giustozzi F., Crispino M., Flintsch G. (2012). Multi-attribute life cycle assessment of preventive main-
tenance treatments on road pavements for achieving environmental sustainability.
International Journal of Life Cycle Assessment, 17, 409–419.
Kim, Y., H.J. Lee, D.N. Little, and Y.R. Kim. A simple testing method to evaluate fatigue fracture and
damage performance of asphalt mixtures. Journal of Association of Asphalt Paving
Technologists, Vol. 75, 2006, pp. 755–788.
Martinez-Arguelles G., Giustozzi F., Crispino M., Flintsch G. (2015). Laboratory investigation on
mechanical performance of cold foamed bitumen mixes: Bitumen source, foaming additive, fiber-
reinforcement and cement effect. Construction and Building Material, 93, 241–248.
Medani, T.O. and Huurman, M. (2003). Constructing the Stiffness Master Curves for Asphaltic Mixes.
Report 7-10-127-3, ISSN 0169-9288. Faculty of Civil Engineering and Geosciences, Delft University
of Technology.
Ongel, A., & Hugener, M. (2015). Impact of rejuvenators on aging properties of bitumen. Construction
and Building Materials, 94, 467–474.
Romera, R., Santamaría, A., Peña, J.J., Muñoz, M.E., Barral, M., García, E., & Jañez, V. (2006). Rheo-
logical aspects of the rejuvenation of aged bitumen. Rheologica acta, 45(4), 474–478.
Shen, J., Konno, M., & Takahashi, M. (2001). Evaluation of recycled asphalt by SHRP binder specifica-
tions. J Pavement Eng Jpn Soc Civil Eng, 6, 54–60.
Shen, J., Amirkhanian, S., & Tang, B. (2007a). Effects of rejuvenator on performance-based properties
of rejuvenated asphalt binder and mixtures.Construction and Building Materials, 21(5), 958–964.
Shen, J., Amirkhanian, S., & Aune Miller, J. (2007b). Effects of rejuvenating agents on superpave mixtures
containing reclaimed asphalt pavement. Journal of Materials in Civil Engineering, 19(5), 376–384.
Silva, H.M., Oliveira, J.R., & Jesus, C.M. (2012). Are totally recycled hot mix asphalts a sustainable
alternative for road paving?. Resources, Conservation and Recycling, 60, 38–48.
Wasage, T.L.J., Stastna, J., & Zanzotto, L. (2011). Rheological analysis of multi-stress creep recovery
(MSCR) test. International Journal of Pavement Engineering, 12(6), 561–568.
Yang, X., & You, Z. (2015). High temperature performance evaluation of bio-oil modified asphalt bind-
ers using the DSR and MSCR tests. Construction and Building Materials, 76, 380–387.
Zargar, M., Ahmadinia, E., Asli, H., & Karim, M.R. (2012). Investigation of the possibility of using
waste cooking oil as a rejuvenating agent for aged bitumen. Journal of hazardous materials, 233,
254–258.
391
Zaumanis, M., Mallick, R., & Frank, R. (2013). Evaluation of Rejuvenator’s Effectiveness with Con-
ventional Mix Testing for 100% Reclaimed Asphalt Pavement Mixtures. Transportation Research
Record: Journal of the Transportation Research Board, (2370), 17–25.
Zaumanis, M., Mallick, R.B., Poulikakos, L., & Frank, R. (2014). Influence of six rejuvenators on the
performance properties of Reclaimed Asphalt Pavement (RAP) binder and 100% recycled asphalt
mixtures. Construction and Building Materials, 71, 538–550.
Zhang, J., Walubita, L.F., Faruk, A.N., Karki, P., & Simate, G.S. (2015). Use of the MSCR test to
characterize the asphalt binder properties relative to HMA rutting performance–A laboratory study.
Construction and Building Materials, 94, 218–227.
392
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
Bituminous mixes are composite materials that consist of a binder mixed with filler/
fines (together with bitumen called the mastic) and aggregates. The mixes of asphalt
pavements consist of bituminous binder that connect between the filler together and the
aggregates.
In India, MORTH Specifications provides different mix specifications for base courses,
binder courses and for wearing courses. The major properties in bituminous paving mix-
tures are stability, flow, durability, flexibility and skid resistance (in the case of wearing
surface).
The properties of bituminous mixes can be improved, making it more durable, cost effec-
tive and much more sustainable by adding polymers, plastics and various other additives.
Various forms of modifications of the physical properties of bitumen and stability of bitu-
minous mixes can be evolved using different materials like crumb rubber, waste plastics, rice
husk ash, mineral fillers, fly ash and different types of the polymers.
393
Table 1. Gradation of Dense Bituminous Macadam [MORTH, Clause 507.2.5].
Grading 1 2
Thickness 80–100 mm 50–75 mm
Nominal Aggregate Size 40 mm 25 mm
Sieve, mm
45 100
37.5 95–100 100
26.5 63–93 90–100
19 71–95
13.2 55–75 56–80
9.5 –
4.75 38–54 38–54
2.36 28–42 28–42
1.18 – –
0.6 – –
0.3 7–21 7–21
0.15 – –
0.075 2–8 2–8
Bitumen Content,% Min.4 Min. 4.5
MINIMUM FLOW(mm) 3
MAXIMUM FLOW(mm) 6
COMPACTION LEVEL 112 BLOWS ON EACH OF
(NUMBER OF BLOWS) TWO FACES OF THE SPECIMEN
PERCENT AIR VOIDS 3–6
PERCENT VOIDS FILLED 65–75
WITH BITUMEN (VFB)
The following steps were used to prepare the DBM samples as per ASTM D 5581 in the
lab:
a. Each mould of the DBM mix was prepared using approximately 4050 grams of ingre-
dients. The weight of the mix was taken in such a manner that compacted specimens of
DBM should have height between 88.9 mm to 101.6 mm.
b. Before placing in the mould, the mixture and a mould assembly (base plate, mould and
collar) was heated to approximately 143°C. The face of the compaction hammer was thor-
oughly cleaned and heated on a hot plate set at approximately143°C. The temperature of
the laboratory during compaction of the specimens was between 20 and 30°C.
c. Place a 152.4 mm diameter paper disc in the bottom of the mould before the mixture is
introduced. Place approximately one-half of the batching the mould with a heated spoon.
Spade the mixture vigorously, penetrating the entire mixture, with the heated spatula 15
times around perimeter and 10 times over the interior. Place the second half of the batch-
ing in the mould and repeat the spading procedure. Smooth the surface of the mix to a
slightly rounded shape.
d. If necessary, the mixture and mould shall be returned to an oven at the required tempera-
ture for the minimum time necessary to achieve the required compaction temperature.
394
Figure 1. Prepared specimen in a standard mould.
e. Place a 152.4 mm (six inch) paper disc on top of material, place the mould assembly on
the compaction pedestal in the mould holder, and apply the calibrated number of blows
(approximately 112) with the mechanical compaction hammer. Compaction was per-
formed at a minimum rate of 40 blows per minute. The compaction hammer shall apply
only one blow with each fall, that is, there shall not be a rebound impact. Remove the base
plate and collar, and reverse and reassemble the mould. Apply the calibrated number of
compaction blows to the face of the reversed specimen.
f. Remove collar, base plate, discs, and allow specimen to cool as shown in Figure 1. Cooling
may be accomplished at room temperature, in a25°C air bath, or if more rapid cooling is
desired the mould and specimen may be placed in front of a fan until cool.
g. Extrude the specimen from the mould as shown in Figure 2. Care must be taken in extrud-
ing the specimen from the mould, so as not to develop tensile stresses in the specimen or
tear the sides of the specimen. The extruded specimen is shown in Figure 3.
3 TESTING OF SPECIMENS
395
Figure 2. Extraction of sample using sample extractor.
of the three specimens shall not differ by more than 48 kg/m3. If this density requirement is not
met, the entire set of specimens shall be discarded and a new set of specimens prepared.
d. Determine the average specific gravity of the specimens and record to the nearest 0.001.
Calculate the average bulk density of the specimens, by multiplying the average specific
gravity by 998 kg/m3. Record the average bulk density to the nearest 1 kg/m3.
e. Bring the specimens to 60±1°C by immersing in the water bath 45to 60 minutes. Prior to
testing, it shall be assured that the inside of the test heads are clean, and that the guide
rods are clean and lubricated so that the upper test head slides freely over them.
f. The breaking head temperature was maintained between 21 to 38°C, using a water bath
when required. Remove the specimen from the water bath, quickly towel dry specimen and
place in the lower segment of the breaking head. Place the upper segment of the break-
ing head on the specimen, and place the complete assembly in position on the testing
machine.
396
Figure 4. Specimen testing.
g. Apply the load to the specimen with a constant rate of 50.8±2.5 mm per minute until the
maximum load is reached and the load decreases. The maximum load is defined as the last
point in the load/time curve before the load decreases. The elapsed time for the test from
removal of the test specimen from water bath to maximum load determination shall not
exceed 30 seconds. Figure 4 shows the specimen testing.
h. Record the stability of each specimen to the nearest 50 Newtons, and the flow to the near-
est0.1 mm.
i. Correct the stability obtained for each specimen, for the height of the specimen. Record
the corrected stability to the nearest 50 newtons.
j. Determine and record the average corrected stability to the nearest 50 Newtons, and the
average flow to the nearest 0.1 mm.
4 SIEVE ANALYSIS
Sieve analysis test was used to determine the aggregate sizes from a sample taken quarry.
Through this sieve test, the proportion of coarse aggregates, fines aggregate and filler was
determined and ensuring the aggregate were well blended within the gradation limit as speci-
fied in MORTH for DBM Grade 1. The sieve analysis test results are presented in Table 3.
The combined gradation of the sample taken is shown in Table 4.
This optimum binder content of 4.5% was used for evaluating the effect of different per-
centage of waste plastics, tyre tube, rice husk ash, crumb rubber and polymers on the Strength
and flow characteristics of mixes.
397
Table 3. Sieve analysis results.
Percentage
Required Percentage Percentage Percentage passing Combined Limit as per
sieve passing 40 mm passing 20 mm passing 10 mm stone dust gradation MORTH
45 16 16 29 39 100 100
37.5 15.296 16 29 39 99.296 95–100
26.5 9.28 16 29 39 91.28 63–93
13.2 0.3424 0.448 16.7823 39 56.5727 55–75
4.75 0 0 0.58 38.22 38.8 38–54
2.36 0 0 0.0406 34.632 34.6726 28–42
0.300 0 0 0.0029 7.9 7.9029 7–21
0.075 0 0 0 0 0 2–8
To determine the optimum binder content (OBC), Marshall Samples were prepared by vary-
ing percentage of 60/70 binder without any addition of any modifier. Stability-Flow analysis
and Volumetric analysis was carried out for the prepared Marshall Core samples with vary-
398
ing bitumen content from 3.5% to 5.5%. The test values obtained are plotted graphically in
Figure 5. From the graph, the optimum binder content was found to be 4.5 percent by wt. of
aggregates using stability flow values.
The waste plastic bottles were used in the bituminous mixes of DBM Grade 1 and stability flow
characteristics of the mix was carried out using Marshall Method of bituminous mix design.
The optimum binder content (OBC) of 4.5% was replaced with 8%, 12% and 16% of plastic
content to determine the Stability and Flow characteristics of the modified mix. The material
and mixing procedure are shown in Figures 6 and 7. The results are shown in Table 5.
Table 5. Test outputs for stability and flow by varying plastic content.
399
7 RESULTS OF DBM GRADE 1 MIX DESIGN WITH TYRE TUBE (POLY ISO
BUTYLENE)
Marshall method of bituminous mix design for DBM grade 1 was carried out by replacing
the optimum binder content with varying percentages of 6%, 10% and 14% tyre tube to
determine the Stability-Flow characteristics of the modified mix. The material and mixing
procedure are shown in Figures 8 and 9. The results are shown in Table 6.
Table 6. Test outputs of stability, flow with replacement of OBC by varying tyre tube content.
Replacement of OBC Height of sample Stability Correction Corrected
with tyre tubes (%) (mm) (kN) factor stability (kN) Flow (mm)
400
8 RESULTS OF DBM GRADE 1 MIX DESIGN WITH RICE HUSK ASH
Marshall method of bituminous mix design for DBM grade 1 was carried out by replacing
the optimum binder content with varying percentages of 6%, 10% and 14% rice husk ash to
determine the Stability-Flow characteristics of the modified mix. The outputs are as shown
in Table 7 and Figures 10 and 11.
Table 7. Test outputs of stability, flow with replacement of OBC by varying rice husk ash content.
Figure 10. Variation of stability to the variation in rice husk ash content.
Figure 11. Variation of flow to the variation in rice husk ash content.
401
9 CONCLUSIONS
The major conclusions drawn from the study carried out on stability flow analysis of DBM
(GRADE 1) by using different additives are as under:
1. The flow criteria for DBM grade 1 is satisfied only if the Bitumen is replaced by 8% plastic
waste, although the stability values lie within the specific range for all replacement levels.
2. The stability values for the DBM grade 1 mix increase with replacement of OBC by all the
considered percentages of discarded tyre tube waste.
3. The flow criteria for DBM grade 1 as per MS2 and MORTH specifications is satisfied on
replacing the optimum binder content of 4.5% with 10% discarded tyre tube waste only
indicating that this percentage is the only suited level of replacement.
4. Although the stability value increased for 6% replacement of OBC by rice husk ash and
reduced for 10% and 14% replacement levels, but the 6% replacement level only does not
satisfy the flow criteria. It indicates that 10 to 15% replacement level of rice husk ash is
suited for creating a stable and flow able DBM mix of grade 1. Higher limit of rice husk
ash replacement needs further investigations.
5. Of the three materials used, replacement of OBC by 10% discarded tyre tube has the high-
est stability value.
6. The optimum content of CRMB and PMB for use in DBM Grade 1 mix is 5%.
7. The bituminous mixes of DBM Grade 1 with 5% PMB having 40% stone dust shows the
maximum stability value and the bituminous mixes of DBM Grade 1 with 5% CRMB
having 44% stone dust shows the maximum stability value.
REFERENCES
ASTM D-5581 Standard Test Method for Resistance to Plastic Flow of Bituminous Mixtures Using
Marshall Apparatus (6 inch-Diameter Specimen). Designation: D 5581 – 96 (Reapproved 2001).
IRC SP-53 Guidelines on use of Polymer and Rubber Modified Bitumen in Road Construction. IRC,
New Delhi
IRC-94–1986 Specifications of dense bituminous macadam. IRC, New Delhi.
MORT&H Specifications for Road and Bridge Works (Fourth Revision).
402
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
W. Huang & B. Li
Key Laboratory of Road and Traffic Engineering, Ministry of Education, Tongji University, Shanghai,
China
P. Wang
Keda Transportation Design Institute, Shanghai Municipal Engineering Design Institute Limited Company,
Shanghai, China
1 INTRODUCTION
Asphalt pavement cracking induced by thermal stress is a problem in cold area and area
suffering large variations in daily temperature, thus the evaluation of low temperature per-
formance of mixture is fundamental for pavement design. Evaluation of low temperature
characteristics of mix is conducted from two sides: one is from binder such as BBR test in
Strategic Highway Research Program (SHRP), the other is from mixture such as bending
beam creep test, indirect tensile test [1] and TSRST [1–4]. The evaluation of mixture’s low
temperature performance were conducted mostly with conventional binders, some mixture
with polymer modified asphalt were tested but the Fracture Temperature (FT) is usually
higher than −34ºC [2, 5–7], the evaluation of binder suitable for −34ºC or lower temperature
is few, and the low temperature characteristics of this type of mixture is little studied.
The objective of this paper is to evaluate low temperature performance of the special mod-
ified binder mixture using TSRST test along with binder test of BBR and conventional test
with 7 types of modified binder, to character the mixture performance with binder indexes
and its influence on mixture.
Mixture using virgin asphalt with different penetration, SBS modified asphalt, PE modified
asphalt, fiber additive, Asphalt Rubber (AR), Chemical modified Asphalt Rubber (ARC) and
crumb rubber plus SBS modified binder were tested through TSRST.
403
2.1 Binders
The virgin asphalt used in this study is Esso70# (penetration between 60/80) according to Chines
specifications, and all other modified asphalt is made using the same virgin asphalt of Esso70#,
different penetration gradation of virgin asphalt (Esso70#, Esso110#, Esso160#) were prepared
via addition of light petrol oil to virgin asphalt with similar procedure in asphalt plant.
SBS modified asphalt uses linear SBS with index in Table 1. Light petrol oil was used to
get higher penetration and moderate viscosity, SBS modified asphalt with SBS content 6%wt,
7%wt, 8.5%wt and 12%wt were labeled with SBS1#, SBS2#, SBS3# and SBS4#, respectively.
Asphalt rubber was prepared under 180ºC for 1.5h after crumb rubber addition, and ARC
was made under more than 240ºC for 4h to make rubber desulfarize and melt in virgin asphalt, to
improve flexibility of asphalt, light petrol oil was added in the modifying processing, crumb rub-
ber content is 16.5%wt for AR1#, 15%wt for AR2#, 18%wt for ARC1# and 20%wt for ARC2#.
Crumb rubber plus SBS modified asphalt was made by adding SBS to AR and ARC respectively
with (AR+SBS) content (8.5+2.5)%wt and (16+4)%wt for ARS and ARCS respectively.
PE modified asphalt contains wet process and dry process, in wet process PE was added
to asphalt with stirring in high temperature while in dry process PE added to mixture when
mixture mixing. The content of PE are the same for wet process and dry process which is
5.7%wt of binder for wet process. Fiber was added to mixture while mixture mixing.
According to AASHTO specifications, low temperature of asphalt binder in BBR test is
obtained as the lowest temperature at which flexural stiffness S(t60 sec)≤300 MPa and slope
of log stiffness and log time curve m(t60 sec) ≥ 0.3. All modified asphalt characteristics were
illustrated in Table 2.
BBR test
Low
Binder Penetration Soften point temperature,
type (25ºC), 0.1 mm (T&B), °C °C S, MPa m
404
2.2 Aggregate and mixture
Mixture of AC-16 was used for Esso virgin asphalt, SBS modified asphalt, Fiber asphalt and
PE modified asphalt, the asphalt contend (AC) is 4.9%, and target Air Voids (AV) is 4%, fiber
with 0.3%wt of mixture was added to mixture while mixing. ARAC-16 was used for AR,
ARC and crumb rubber plus SBS modified asphalt, the AC is 5.7%, and target Air Voids
(AV) is 4%. Gradation of mixture is illustrated in Figure 1.
405
3 EXPERIMENTAL RESULTS AND DISCUSSION
Thermal induced stress in specimen increases with temperature decreasing because of the
fixed top and bottom mold, and there is an increase in the slope of thermal stress. The ther-
mal stress is similar to be linear after transition point and continue increasing to a peak point
and inducing cracking in specimen. Then the stress decrease sharply because of cracking
displacement or shrinking rate induced by temperature decreasing less than creeping rate. A
typical TSRST stress-temperature curve is illustrated in Figure 3.
It is recommended to use BBR low temperature to indicate asphalt low temperature char-
acteristics in SHRP, the mixture low temperature results of TSRST is compared with the
binder BBR results as illustrated in Figure 4.
It is clear that the BBR low temperature is close to TSRST FT, and is 0.3ºC to 6ºC higher
(except PE) compared to TSRST results, means that BBR is an effective method to evaluate
the mixture low temperature performance while only using the binder characteristic with the
result a few degree Celsius higher than the actual performance of mixture. Softer binder is
good for the low temperature performance of asphalt mixture, with the maximum differences
between TSRST results and BBR results within 6ºC. Fiber addition could improve low tem-
perature performance of mix with about 5ºC. The excellent low temperature of SBS modified
406
asphalt mixtures using high SBS content and large penetration is confirmed in this study,
with FT lower than −40ºC, also the AR has good performance in low temperature. PE has
negative influence on low temperature performance of mixture both in BBR low temperature
and TSRST FT, and the negative influence varies in modified process. Dry process is better
than wet process of PE in FT which may be caused by the distribution of PE particles in
asphalt binder as in wet process it is even but in dry process it is scatted, but both process have
lower FT than that of Esso70# mixture with a difference between 6–9ºC. PE modified mix-
tures show BBR low temperature lower than TSRST FT, which is contrary to other binders
in this study which illustrate a higher BBR low temperature than TSRST FT, implying that
BBR is not suitable for PE modified asphalt to evaluate low temperature performance.
The relationship between FT and TT is linear while considering all type of binders
(Figure 5), and the relationship is good with correlation coefficient being 0.82, the linear
relationship between FT and TT also was also found by Isacsson [2] and Carter [8]. The
asphalt used in this study is specialized with high penetration (such as 118.6, 154) or high
SBS content, and the relationship between FT and TT indicates this relationship is general
for virgin asphalt and modified asphalt.
Higher penetration asphalt is used in cold areas to prevent/retard thermal induced crack-
ing compared with warm areas, showing that hardness of asphalt affects the low temperature
performance of mixture. Scatter pot of FT and binder penetration is illustrated in Figure 6.
407
Figure 7. Scatter plot of fracture temperature and fracture stress.
For each type of asphalt, higher penetration corresponds to lower FT, validating the use of
softer binder in cold area. Virgin asphalt mixture with 160# asphalt have a FT with −34.1ºC,
it is amazing that low temperature of virgin asphalt is equal to that of AR and ARS. For
virgin asphalt, mixture with softer asphalt has lower strength in high temperature, indicating
that there is a risk of rutting while using softer asphalt to improve low temperature perform-
ance, so the penetration of virgin asphalt should be controlled in construction. There is no
definitive relationship between binder penetration and FT for all types of binder in this study,
while close FT corresponding to penetration from 34 to 188, close penetration correspond-
ing to FT with difference of 16ºC, it is clear that the relationship between penetration and
FT varies from binder to binder. In literature[6], it concluded that polymer modification of
binder did not improve the fracture temperature tested by TSRST of mixture, while consid-
ering the hardness of polymer modification of binder was about half of that of the virgin
asphalt, such as 40 compared to 84, it is generally true that softer binder with lower fracture
temperature, and the conclusion is also shown in literature [5].
Specimen fracture in TSRST test occurs while thermal induced stress is bigger than frac-
ture strength of specimen, making fracture strength is an indication of low temperature per-
formance of mixture[9], the relationship between fracture strength and FT in this study is
plotted in Figure 7. Fracture stress rise as FT reducing for the mixture, the trend is clear in
this study. SBS, AR and ARS mixture has the similar large fracture stress and low FT, means
they are all excellent materials for cold area. The fracture stress of fiber asphalt mixture is
biggest among the seven types of binders, but only correspond to FT around −27ºC which
is much less than the SBS or AR FT of −35ºC, means that the addition of fiber in mixture
is good for fracture strength but not enough positive influence for shrinkage coefficient and
modulus in low temperature of mixture that affect mixture fracture, resulting fiber asphalt
with bigger fracture stress but moderate FT.
4 CONCLUSIONS
With conducting TSRST of 7 different types of binder and comparison with their binder
characteristics, conclusions are drawn as fellows based on this study:
1. Good relationship is illustrated between TSRST and BBR test data, low temperature
performance of mixture is much related to binder characteristics; the difference between
TSRST and BBR varies for different binders, generally within 6ºC in this paper.
2. TSRST fracture temperature has good relationship with transition point temperature, and
the relationship is general for asphalt binder used in pavement.
408
3. Higher penetration binder is effective for improving mixture low temperature perform-
ance, but there is limitation while considering the rutting risk, and the penetration influ-
ence on FT is different from binder to binder.
4. Fiber has positive influence while PE negative influence on mixture fracture temperature,
wet-process is more negative than dry-process for PE modification mixture.
5. Virgin asphalt with high penetration could not reach a FT of −34ºC, while modified binder
with high polymer content and high penetration could reach a FT of lower than −34ºC.
REFERENCES
Carter, A., et al., Evaluation of cracking in overlays in Quebec: A case study. Advanced Testing and
Characterisation of Bituminous Materials, Vols. 1 and 2, 2009: p. 1091–1098.
Das, P.K., Y. Tasdemir, and B. Birgisson, Low temperature cracking performance of WMA with the use
of the Superpave indirect tensile test. Construction and building Materials, 2012, 30: p. 643–649.
Isacsson, U. and H. Zeng, Low-temperature cracking of polymer-modified asphalt. Materials and
Structures, 1998, 31(205): p. 58–63.
Jung, D. and T.S. Vinson, Low-temperature cracking: test selection. 1994.
Lu, X., U. Isacsson, and J. Ekblad, Influence of polymer modification on low temperature behaviour of
bituminous binders and mixtures. Materials and Structures, 2003, 36(10): p. 652–656.
Olard, F., et al., Properties of bituminous mixtures at low temperatures and relations with binder char-
acteristics. Materials and structures, 2005, 38(1): p. 121–126.
Raad, L., et al., Low Temperature Cracking of Modified AC Mixes in ALASKA, Executive Summary,
1997.
Raad, L., et al., Thermal cracking models for AC and modified AC mixes in Alaska. Design and Reha-
bilitation of Pavements 1998, 1998(1629): p. 117–126.
Sauzeat, C., et al., Low temperature behaviour of bituminous mixes: TSRS tests and acoustic emis-
sion. Advanced Characterisation of Pavement Soil Engineering Materials, Vols. 1 and 2, 2007:
p. 1263–1272.
409
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The binder course plays a very important role in strengthening the interface
between new and existing pavement layers, which is highly related to shear damage of the
interlayer and many other kinds of defect. Due to the inappropriate design of the binder
course, such defects caused by interlayer shear fracture frequently occurred in alpine regions
such as in Qinghai province in China. Taking into account the binder course shear strength as
the key evaluation parameter, the optimal aggregate covering ratio and asphalt content were
determined by tests, which simulated the working state of the binder course using composite
specimens shaped in the lab. Based on that, the influences of different aggregate size, lithology,
pretreatment and asphalt binder type were further researched and analyzed. The results of
this paper can be a supplement to the current design parameters for binder courses in the
Chinese specifications, and also can provide references for other similar research.
1 INTRODUCTION
The problem of early damage of asphalt pavements is more prominent in Qinghai province,
because this area is in a typical alpine region. Highway construction is less developed, while
the environmental and geological conditions are complicated. The key problem is vehicle
overloading which often happens in recent years. The earlier constructed highways gradu-
ally require massive major and medium repairs, and improvement of the asphalt pavement
maintenance will become the hot issue of the highway transportation industry in the next
few years.
In the past, for pavement maintenance road engineers often focused on the materials and
the pavement structure, such as thickening the asphalt layers or optimizing the mixture type to
improve the bearing capacity of the structure, and the driving comfort, while the mechanical
characteristics of the binder course between the old pavement and the overlay were ignored
unconsciously [1]–[2]. In addition, Qinghai has numerous mountains and different kinds of
411
rock which have been weathered seriously, so the region lacks high quality stone that can be
used in the binder course [3]–[5]. As discussed above, many unfavorable factors result in defects,
such as upheavals, of the asphalt pavement after maintenance resulting in a continuously
decreasing asphalt pavement condition [3]–[5].
The material design of the binder course mixture should be based on the mechanical per-
formance of this structural layer, considering the factors of aggregate lithology, size and
binder type. Unfortunately, the current specifications [5][6] do not give a clear guidance for
the material selection for the binder course design and construction technology. This paper
focuses on the aggregate and binder used for the binder course. Based on the evaluation of
the shear strength [7], this paper proposes an optimized design method for asphalt pavement
binder courses in alpine regions.
2 TESTING PROGRAM
In order to simulate the real working conditions of the binder course between the new and
old pavement [8], this study proposes a structural model of the old pavement layer, the binder
course and the overlay as shown in Figure 2. According to the model, the following prepara-
tion method for binder course composite specimens is proposed for laboratory tests.
The process of composite specimen preparation is as follows (Figure 3):
1. Drill four core samples from the old pavement with diameter of 150 mm[9].
2. Place the core samples in order in the composite rutting plate mold, then inject cement
mortar into the mold and level it, then cure at room temperature for 48 hours.
3. Clean the old asphalt pavement surface of the composite specimens with a steel brush,
and make sure that old pavement surface stones are exposed;
412
4. Spread the binder and aggregate on the surface of the old pavement in accordance with
the optimum dosage for the binder course;
5. Roll over the binder course paving material 2 to 3 times back and forth using a hammer
and level it;
6. Compact the asphalt mixture on top by the rutting molding apparatus to simulate the new
pavement;
7. Drill four composite specimens, with a diameter of 100 mm, and place them under dry
and ventilated conditions for 24 hours.
According to the experiences with the binder course of asphalt pavement design and con-
struction in alpine region [10], the raw materials of the binder course include 90# paving
Table 1. Relationship between aggregate covering ratio and shear strength of binder course composite
specimens with different asphalt content.
Note: Spraying volume is about 8.6 kg/m2 when the aggregate coverage was 100%.
413
Table 2. Relationship between asphalt content and shear strength of binder course composite
specimens.
Figure 6. Relationship between aggregate covering ratio and shear strength of binder course compos-
ite specimens with different asphalt content.
asphalt and limestone aggregate of particle size 5–10 mm. The asphalt contents are 1.2 kg/m2,
1.4 kg/m2 and 1.6 kg/m2, and the aggregate covering ratios are 50%, 60% and 70%. In the
laboratory, the shear tests [11] are carried out by using the UTM testing machine. The test
apparatus is shown in Figure 4 and Figure 5. Failure of the test specimens always occurred
at the interlayer.
414
Figure 7. Relationship between asphalt content and shear strength of binder course composite
specimens.
415
Table 3. The shear strength of specimens with different aggregate size at different temperatures.
Shear strength
Asphalt type and (MPa)
the optimum Aggregate
Aggregates Temperature spraying amount covering Test Average
lithology Particle size (°C) (kg/m2) ratio (%) results value
Figure 8. Shear strength of specimens with different aggregate size at different temperatures.
416
Table 4. Shear strength of specimens with different aggregate lithology at different temperatures.
Shear strength
Asphalt type and (MPa)
optimum spraying Aggregate
Particle size Aggregates Temperature amount covering Test Average
specification lithology (°C) (kg/m2) ratio (%) results value
Figure 9. Shear strength of specimens with different aggregate lithology at different temperatures.
Table 5. hear strength of specimens with different aggregate pretreatment methods at different
temperatures.
Shear strength
(MPa)
Aggregates Asphalt type and Aggregate
lithology and Pretreatment Temperature optimum spraying covering Test Average
particle size method (°C) amount (kg/m2) ratio (%) results value
417
Figure 10. Shear strength of specimens with different aggregate pretreatment methods at different
temperatures.
On the basis of the analysis of the characteristics of raw materials and design parameters of
asphalt pavements in alpine region, the following conclusions have been drawn.
1. Instead of the range given by the specifications or previous experience, in this paper it
has been determined that the optimum spraying amount of asphalt is 1.4 kg/m2 and the
optimum aggregate covering ratio is 60%.
2. There is a certain correlation between the shear strength of the binder course and the
aggregate particle size, the aggregate lithology and the pretreatment method of the
aggregate. In terms of particle size, the shear strength sequence is 4.75–9.5 mm > 9.5–
13.2 mm > 13.2–16 mm. In terms of aggregate lithology, the shear strength sequence
is limestone > dolomite > granite. In terms of pretreatment method, the shear strength
sequence is limewater pretreatment > washing > not washing.
3. In this paper, composite specimens were prepared to simulate the actual pavement structure,
and shear tests were conducted to evaluate the joint performance of the new and old pave-
ment. However, more test results are needed to improve the evaluation methods in the future.
REFERENCES
AASHTO. T 324-04 Standard Specifications for Transportation Materials and Method of Sampling
and Testing [S]. AASHT0, 2004.
Changshun Hu, Binggang Wang. Design principle and construction technology of composite pavement
[M]. Beijing; China communication press, 1999.
JTG D50-2006 Specifications for Design of Highway Asphalt Pavement [S].
JTG F40-2004 Technical specification for construction of highway asphalt pavement [S].
Jinan Shen. Analysis and preventive techniques of premature damage of asphalt pavement in express-
way [M]. Beijing; China communication press, 2004.
Jinan Shen. Technical approach to solve the premature damage of water disease on asphalt pavement
expressway [J]. Road, 2000, (5):71–76.
Kaiyu Huang. Research on interlayer bonding state of asphalt pavement for expressway [J]. Highway
engineering, 2010, (2):40–44.
Lei Zhu, Ning Liu, Wuju Wei. Analysis on factors influencing shear strength of binder course between
old and the new pavement [J]. Highway engineering, 2015, 40(1):121–125.
Qinglin Sha. Premature damage and its preservative measures of bituminous pavement on express-
way [M]. Beijing; China communication press, 2001.
Wuju Wei, Lei Zhu, Chi Ma. Research on the forming method of the composite specimen of the asphalt
pavement [J]. Journal of China & Foreign Highway, 2014,34(3):50–52.
Yaoting Zhu, Maojin Lei, Ning Shi. Influence of interface condition on structural performance and life
prediction of asphalt pavement [J]. Highway engineering, 2011,(5):23–26.
418
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
419
Figure 1. FLUKO FA25.
temperature 170±3°C. All performances tests, including softening point, 135°C viscosity and
rutting factor were conducted according to the standard test methods of bitumen and bitu-
minous mixtures for highway engineering (Sengoz, 2008; Kang, 2014). In order to observe
clear fluorescent images with high contrast between bitumen and polymer, cross linking
agents were not used for the binders with high polymer contents.
gotten from Figure 6 that when magnification was greater than 20 × , segmentation effect of
Ostu became worse. Meanwhile, 10 × cannot reflect more detailed information. So 20 × was
selected as the optimal magnification.
Storage format mainly influences the detailed information of images. Thus, what we need
is the format which can keep the most detailed information. And polymer particles of images
saved as different storage format were extracted for this purpose. Obviously, JP2 was selected.
421
Figure 5. Images captured under different magnification.
422
Figure 7. Relationship between polymer particle number and neighborhood level.
area of polymer particles A, area percentage P, ratio of major and minor axis R, and standard
deviation of R, these four representative parameters are extracted from fluorescent images. The
first three are average value of the whole image. The formulas of A, P and R are listed below.
Area of polymer
A= (1)
Number of polymer particles
Area off ppolymer
y
P= × 100%
Area of image (2)
Major axis
i of polymer
R= (3)
Minor axis of polymer
3 DATA ANALYSIS
423
Figure 8. Relationships between pavement performance and polymer content.
Different pavement performances are the results of comprehensive actions of polymer con-
tent and shearing time. At the micro level, these kinds of actions are presented as the changes
of morphology parameters.Fortunately, all these are included in the linear relations below.
Equation (4), (5) and (6) are further linear models. Due to direct proportion relation
between polymer content and area percentage, P has positive relations with pavement per-
formances, as well as polymer content.
Next work was to analyse the effects of parameters except P on pavement performances.
Softening
o point = 1.56 P 12.77 A + 64.09 R 2 = 0.896 (4)
Softening point is equal to the dripping temperature tested by ring and ball method. As
shown in Equation (4), this temperature correlated positively to area percentage,but negatively
to ratio of major and minor axis. 64.09°C is the basic temperature depending on raw materials.
The increase of P improve the basic temperature. But the greater the mean area, the lower the
temperature is. That’s because big polymer particles may form equivalent metal ball in high tem-
perature, which will increase the gravity of the dripping and weaken the internal resistance.
Vis
i cos ity
ity 0.004R 0 84SD
S 0.992 R 2 = 0.972 (5)
Viscosity is tested by brookfield viscometer. It is measured by the torque when the rotor
rotates in the PMB. Polymer showing as the ribbon has great resistance to the rotor. So R, a
parameter describing the degree of ribbon, is correlated positively to viscosity. However, if R
is not steady, that is to say, if SD of R is too large, the value of viscosity will decline.
424
Table 4. Results of stepwise regression analysis.
Parameters
Performances A / um2 P/% R SD of R
Softening point / °C √ √
Viscosity / Pa.s √ √
Rutting factor / Pa √ √
Parameters
Performances A P R SD of R constant
Rutting factor is the ratio of complex modulus G* and sine value of the phase angle sin δ,
tested by parallel plate of dynamic shear rheometer (DSR). This index reflected the average
performance more accurately. Diverse polymer shapes, quantitatively expressed by SD, can
form rich structure asphalt more easily. And this kind of structure asphalt is benefit to high
temperature stability of PMB.
The same conclusion can be drawn by data analysis of the other two PMBs. Differences
exist among estimated parameters of different PMBs, as shown in Table 5. Obviously,
S90-DG has greatest performance in raw materials. It proves the excellent modification effect
of star structure polymer.
4 CONCLUSIONS
Pavement performances are the results of comprehensive actions of polymer content and
shearing time. The former is corresponding with area percentage while the latter with the
other morphology parameters. Through parameters extraction and data analysis, simple but
meaningful linear models were obtained to describe the micro effect mechanism of polymer
to pavement performance.
REFERENCES
Chen, D.H. 2013. PC Stepwise Regression Forecasting Model of Climatic Field of Fujian Flood Period
Rainfall. Meteorology, 39(9): 1190–1196.
Chen, J. 2002. Asphalt Modified by SBS Tri-block copolymer: morphology and model. Journal of
Materials in Civil Engineering, 5(6): 224–229.
Gonzales, R.C. 2003. Digital image processing. Beijing: Electronic Industry Press.
Gurland. 1968. Distribution of discrete particles in quantitative microscope. New York: McGraw Hill.
Huang, W.D. & Sun, L.J. 2001. The phase structure of polymer modified asphalt. Highway traffic sci-
ence and technology, 10: 1–3.
425
Kang, A.H. & Kou, C.J. 2014. Evaluation methods of dispersed phase uniformity in SBS modified
asphalt based on digital image analysis technology. Journal of Sichuan University, 46(1): 172–176.
Kou, C.J. & Kang, A.H. 2015. Methods to prepare polymer modified bitumen samples for morphologi-
cal observation. Construction and Building Materials, 81: 93–100.
Larsena, O. José. Alessandrinib, L. 2009. Micro-structural and rheological characteristics of SBS-as-
phalt blends during their manufacturing[J]. Construction and Building Materials, 23(8): 2769–2774.
Liu, X. & Xiao,Y. 2005. Microscopic image acquisition and processing of modified asphalt. Highways
& Automotive Applications, 5: 97–99.
Pratt W.K. 1991. Digital image processing. New York: Wiley Inter Science.
Sengoz, B. & Isikyakar, G. 2008. Analysis of SBS polymer modified bitumen using fluorescent micros-
copy and conventional test methods. Journal of Hazardous Materials,150(2): 424–432.
Shacham, M. & Brauner, N. 2014. Application of stepwise regression for dynamic parameter estima-
tion. Computers & Chemical Engineering, 69: 26–38.
Sun, D.Q. 2003. Study on compatibility and engineering properties of SBS modified asphalt. Tongji
University, Shanghai.
426
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Ronald Corun
Axeon Specialty Products, Paulsboro, NJ, USA
Frank Fee
Frank Fee, LLC, Media, PA, USA
ABSTRACT: Re-refined Engine Oil Bottoms (REOB) are one of several products obtained
in the refining of recovered engine oil and have been used since the 1980’s in the asphalt
industry. Generally, REOB is used to help soften the base asphalt binder and is commonly
used from three to ten percent by weight in order to achieve desired low temperature asphalt
binder properties. Recently, poor cracking performance in a number of Canadian and north-
ern United States pavement sections have been blamed on the use of REOB to modify the
asphalt binder. This has prompted many state agencies in the northeast United States to ban
its use. This paper summarizes the laboratory performance of asphalt binders modified with
REOB. Two different sources of REOB were blended with different base grades at varying
dosage rates in the study. Performance grading, master stiffness curves, double edged notch
tension test, and Black Space analysis were conducted on the asphalt binders at different
levels of laboratory aging. The research study showed that while being able to achieve softer
asphalt binder grades, the addition of REOB accelerates the aging of the asphalt binder with
higher levels of age hardening occurring at higher REOB dosage rates. The study also indi-
cated that while the stiffness properties at low temperatures are not impacted by the REOB,
the relaxation properties, as measured using m-slope of the Bending Beam Rheometer (BBR),
are highly affected. Both the Black Space analysis, using the Glover-Rowe approach, and the
DENT test show promise at identifying the age hardening affects.
1 INTRODUCTION
The Performance Grading (PG) system was implemented to ensure that asphalt binder per-
formance met the climatic conditions required to mitigate asphalt pavement distress. To help
achieve the PG specifications, asphalt binder refineries commonly utilize various modifiers
that include, but not limited to; various polymers, acids, waxes, various aromatic and par-
affinic processing oils, and engine oil residues. According to the literature, one form of engine
oil residue, called Re-refined Engine Oil Bottoms (REOB), has been used in the asphalt
industry since the 1980’s (Gryzybowski et al., 2014; Planche et al., 2015). The addition of
REOB is typically used to soften the asphalt binder and help improve the intermediate and
low temperature properties of the asphalt binder. Although this may seem to be something
that is mainly a concern in colder climate regions, many state agencies in warmer climates
have expanded their use of Recycled Asphalt Pavement (RAP) and Recycled Asphalt Shin-
gles (RAS), which are commonly used in conjunction with softer asphalt binder grades to
help counter the stiff, recycled asphalt binder. Therefore, the potential usage of REOB has
expanded to all climate regions in North America.
427
Even though today’s asphalt binders meet the required PG grades, some agencies in the
United States and Canada are reporting premature durability failures. These premature
durability issues are being linked to the current asphalt binder production methods, as well as
the potential use of REOB when producing/modifying the asphalt binder (Hesp et al., 2009;
Johnson and Hesp, 2014; Ahern, 2015; Reinke, 2015). As of 2015, the following state agencies
have banned or limited the use of REOB for asphalt binder modification;
• Northeast United States—New Hampshire, Vermont, Connecticut, Massachusetts, Rhode
Island, New York;
• South/Southeast United States—South Carolina, Georgia, Texas (limits to ≤ 5%); and
• Midwest United States—Michigan, Illinois, Colorado
It has been hypothesized that the premature durability issues are due to accelerated physi-
cal hardening of REOB modified asphalt binders, possibly due to presence of metal catalysts
and oxidized engine components (Hesp and Shurvell, 2010; Planche et al., 2015). Meanwhile,
others have shown that REOB modification to asphalt binders do not result in durability
issues. Work by D’Angelo et al. (2012) noted that changes in age related performance was a
function of the base asphalt properties and not the REOB additives. Similar findings were
also found by D’Angelo et al. (2013) in a follow up study using a wider range of REOB dos-
age rates. Wielinski et al., (2014) showed that the addition of REOB at a “moderate level” of
9% did not compromise the mixture stiffness or resistance to moisture damage during labo-
ratory evaluation. The conflict in material performance has left agencies extremely confused
about how to specify and allow REOB modification of asphalt binders, resulting in many
agencies to either prematurely ban REOB, or take a “wait and observe” approach.
A laboratory testing program was developed that would utilize the REOB additives as a
means to modify a “stiffer” asphalt binder to produce a “softer” asphalt binder. This is a com-
mon application for REOB materials in the asphalt industry. Two PG grades were produced
from a PG70-22 asphalt binder. First, a PG64-22 was produced by blending 10% REOB by
weight with the PG70-22 asphalt binder. Second, a PG58-28 asphalt binder was produced
by blending 20% REOB by weight with the same PG70-22 asphalt binder. An additional
PG58-28 asphalt binder was produced by blending 6% REOB by weight with a PG64-22
asphalt binder. With two sources of REOB utilized in the study, this resulted in six different
REOB modified asphalt binders. The performance of the REOB modified asphalt binders
were compared to the straight run asphalt binders of the identical PG grades (i.e. – PG58-28
and PG64-22).
The eight asphalt binders in the study were evaluated for their respective asphalt binder
properties under a variety of aging conditions. For asphalt binder testing, Original, RTFO,
20 hour PAV and 40 hour PAV aging conditions were utilized.
identified this difference as a means of indexing the non-load associated cracking potential
of asphalt binders and defined it as follows:
where, ΔTcr = Difference in critical low temperature PG grade, Tcr = Critical low temperature
grade predicted using the BBR m-slope, Tcr = Critical low temperature grade predicted using
the BBR Stiffness (S)
In Equation (1), as the ΔTcr decreases and becomes negative, the asphalt binder is consid-
ered to be more prone to non-load associated cracking. Initially, Anderson et al., (2011) set
a limit of ΔTcr ≤ −2.5oC for when there is an identifiable risk of cracking and preventative
action should be considered. Rowe (2011) further advanced this methodology, eventually
developing a new asphalt binder fatigue property termed Glover-Rowe parameter, which will
be discussed later, and also recommended that at a ΔTcr ≤ −5oC immediate remediation should
be considered.
Figure 1 shows the ΔTcr results for the various asphalt binders evaluated at RTFO, 20 Hour
PAV and 40 Hour PAV aged conditions. The results indicate that as the dosage rate of the
REOB modification increases, the potential for cracking, as proposed by Anderson et al.,
(2011) and Rowe (2011), also increases. This would indicate that as the REOB modified
asphalt binder ages, it loses it relaxation properties, detrimentally affecting the critical low
temperature performance.
429
2.2 Rheological indices using Christensen-Anderson model (CAM)
The Christensen-Anderson Model (CAM) is a very useful tool because the master curve
parameters (ωo, R, and Td) have specific physical significance. The Crossover Frequency,
ωo, is the frequency where the phase angle is close to 45 degrees and is an indicator of the
hardness of the asphalt binder (Anderson and Bonaquist, 2012). Larger Crossover Frequen-
cies would indicate lower degrees of aging. The Rheological Index, R-value, is the difference
between the log of the glassy modulus and the log of the dynamic modulus at the Crossover
frequency (Anderson and Bonaquist, 2012). As R-value increases, the master curve becomes
flatter indicating a more gradual transition from elastic behavior to steady-state flow. Nor-
mally, R-value is higher for oxidized/aged asphalt.
Figure 2 shows the results of plotting the Crossover Frequency vs R-Value for the different
REOB sources and target PG grades. As the figure indicates, as the asphalt binders age, the results
move from the upper left part of the space (i.e. – high Crossover Frequency/Low R-Value) to the
lower right area of the space (i.e.—low Crossover Frequency/higher R-Value). At the lower dos-
age rate (6% REOB), similar location and movement in the ωo—R-Value Space occurs, indicat-
ing that minimal to no difference in age hardening characteristics occurs. However, at the higher
dosage rates of 10 and 20%, the REOB modified binders after 20 Hours PAV are very similar to
the non-REOB modified binders after 40 Hours PAV conditioning. This would indicate that as
aging/conditioning time increases, the REOB modified binders undergo an acceleration in age
hardening not witnessed in the asphalt binders not modified with REOB.
Figure 1. ΔTcr vs REOB dosage rate at varying laboratory aged conditions; (a) REOB source #1; (b)
REOB source #2.
430
testing results, making it much more practical than directly measuring ductility using traditional
methods. Rowe (2011) re-defined the Glover parameter based on analysis of a black space dia-
gram and recommended the use of Equation (2), called the Glover-Rowe (G-R) parameter.
| ( cos δ )
* 2
|
(2)
sinδ
Figure 2. Crossover Frequency vs R-Value for asphalt binder at different laboratory aged conditions;
a) source #1, b) source #2.
431
Figure 3. Binder fatigue resistance evaluation—Glover-Rowe parameter test results.
Figure 4. Binder fatigue resistance evaluation—Double Edged Notch Tension Test (DENT).
test, similar to the Direct Tension Test (DTT) used in the Superpave PG system with the
exception that notches are imposed on the specimen. The test can be conducted in a standard
force-ductility instrument, such as that used for the DTT test. The DENT test was devel-
oped by Queen’s University in Canada (Andriescu et al., 2004) and modified and adapted
for intermediate temperature testing by the FHWA (Gibson et al., 2011). The DENT test is
formalized in specifications in Ontario, Canada (Ontario Ministry of Transportation Test
Method LS-299).
The premise for the test is that energy required for fracturing ductile materials includes
essential energy to create new surfaces and nonessential energy away from the crack that
leads to yielding and plasticity. The test allows for calculating the Essential Work of Frac-
ture (EWF) and Critical Tip Opening Displacement (CTOD). To determine these quantities,
multiple specimens are tested with varying ligament lengths. For each ligament length, the
432
total work of fracture is determined using the area under the force—displacement curve
with normalization for specimen geometry. Then, the total work of fracture is plotted versus
ligament length to allow for separating essential and nonessential work components under
the assumption that EWF is proportional to ligament length whereas nonessential work of
fracture is proportional to volume of plastic zone surrounding the crack. CTOD is then
calculated by extrapolating trends to a zero ligament length. CTOD represents the ultimate
elongation for a zero ligament length, which represents the strain tolerance in the vicinity of
a crack. Binder CTOD was found to rank binders well with respect to mix push-pull fatigue
test results (Zhou et al., 2013) compared to other binder tests evaluated (including LAS,
MSCR, and elastic recovery). In addition, Gibson et al. (2011) demonstrated that DENT test
results were highly correlated to fatigue cracking severity observed in the FHWA ALF sec-
tions. The resultant CTOD results of the asphalt binders evaluated in this study are shown in
Figure 4. The asphalt binders were conditioned for 20 hours in the PAV and tested at a test
temperature of 15°C. The results indicate that the two 20% REOB modified asphalt binders
resulted in the lowest CTOD results for the PG58-28 asphalt binder group. Lower CTOD
results represents asphalt binders more prone to fatigue cracking. The 6% REOB modified
binders resulted in very similar CTOD results compared to the neat PG58-28 asphalt binder.
Meanwhile, the 10% REOB modified binders resulted in a very similar CTOD value to that
of the neat PG64-22 asphalt binder.
3 CONCLUSIONS
A laboratory research program was conducted to evaluate the impact of Re-refined Engine
Oil Bottoms (REOB) on asphalt binder and mixture performance. Two different REOB
sources were blended with different base asphalt binders to target a PG58-28 and PG64-22
asphalt binder, respectively. The performance of the REOB modified asphalt binders were
compared to neat asphalt binders of the same PG grade. The results of the study indicated;
• The magnitude of aging has a significant impact on the performance of REOB modified
asphalt binders and mixtures. When compared to neat asphalt binders of equivalent PG
grade, asphalt binders modified with REOB exhibit higher degrees of aging, with small
changes observed at RTFO aging and significant changes observed after 40 Hr PAV aging.
• As the dosage rate of REOB increased, the asphalt binders aged/hardened more than the
lower REOB dosage/neat asphalt binders. PG grading and master stiffness curve compari-
sons of the asphalt binders at RTFO, 20 Hr PAV and 40 Hr PAV aged conditions clearly
noted the higher levels of aging. The plotting of the rheological indices, Cross-over Fre-
quency vs R-value, indicated that after only 20 hours of PAV aging, the 20% REOB dosed
binders were similar in stiffness to the 40 hours PAV aged neat binders. This was found for
both REOB Source #1 and #2. Similar findings were also found after only 10% REOB
dosage rate for REOB Source #2.
• REOB source did appear to have a slight impact on the overall performance of the REOB
modified asphalt binders and mixtures. Differences were found in the asphalt binder and
mixture performance at the identical dosage rates for the two different REOB sources.
This would indicate that not all REOB materials would influence the asphalt binder and
mixture performance to the same degree and a greater effort to standardize REOB may
need to be considered.
• Asphalt binder testing using the BBR ΔTcritical, was found to be a sensitive parameter when
comparing the relative change in binder aging and hardening. The relaxation properties of
the REOB modified asphalt binders, as measured using the BBR m-slope, were affected
by laboratory oxidative aging, with minimal detrimental changes to the BBR stiffness. The
other asphalt binder properties measured (Rheological Indices, Glover-Rowe, and DENT
CTOD) were also found to be sensitive to REOB dosage rate and degree of aging and
could be utilized as an asphalt binder screening tool for asphalt binders susceptible to
aging, such as REOB modified asphalt binders.
433
REFERENCES
Ahern, W., 2015, Atypical Raveling in Vermont, 2015, Presented at the FHWA Expert Task Group Meet-
ing, Fall River, MA, April 9th, 2015.
Anderson, M., P. Kriz, G. King, and J.P. Planche, 2011, “Evaluation of the Relationship between
Asphalt Binder Properties and Non-Load Related Cracking,” Journal of the Association of Asphalt
Paving Technologists, 80, pp. 615–664.
Anderson, D.A. and R. Bonaquist, 2012, NCHRP Report 709, Investigation of Short-Term La bora-
tory Aging of Neat and Modified Asphalt Binders, National Cooperative Highway Research Program
(NCHRP), Transportation Research Board, Washington, D.C., 78 pp.
Andriescu, A., S. Hesp, and J.S. Youtcheff, 2004, “Essential and Plastic Works of Ductile Fracture in
Asphalt Binders,” Transportation Research Record, 1875, pp. 1–7.
D’Angelo, J., K. Grzybowski, and R. Walker, 2012, “Asphalt Binder Modification with Rerefined Heavy
Vacuum Distillation Oil (RHVDO)”, Proceedings of the Canadian Technical Asphalt Association,
Volume 57, pp. 257–275.
D’Angelo, J., K. Grzybowski, S. Lewis, and R. Walker, 2013, “Evaluation of the Performance of Asphalt
Mixes Produced with Re-refined Heavy Vacuum Distillate Bottoms”, Proceedings of the Canadian
Technical Asphalt Association, Volume 58, pp. 273–290.
Gibson, N., X. Qi, A. Shenoy, G. Al-Khateeb, M.E. Kutay, and A. Andriescu, 2011, “Full-scale acceler-
ated performance testing for Superpave and structural validation,” FHWA-RT-01946. Federal High-
way Administration, Washington, D.C.
Glover, C.R. Davison, C. Domke, Y. Ruan, P. Juristyarini, D. Knorr, and S. Jung (2005). “Develop-
ment of a New Method for Assessing Asphalt Binder Durability with Field Evaluation,” Report No.
FHWA/TX/05-1872-2. National Research Council, Washington, D.C.
Gryzybowski, K., A. Palmer, and S. Lewis, 2014, VTAE Oils in Asphalt, Presented at the Rocky Moun-
tain User/Producer Group, Coeur d’ Alene, Idaho, October 15th, 2014.
Hesp, S., S. Genin, D. Scafe, H. Shurvell, and S. Subramani, 2009, “Five Year Performance Review of a
Northern Ontario Pavement Trial: Validation of Ontario’s Double-Edge-Notched Tension (DENT)
and Extended Bending Beam Rheometer (BBR) Test Methods”, Presented and Published in the
Proceedings of the 54th Annual Conference of the Canadian Technical Asphalt Association (CTAA),
Moncton, New Brunswick, November 2009.
Hesp, S., and H.F. Shurvell, 2010, “X-ray Fluorescence Detection of Waste Engine Oil Residue and Its
Effect on Cracking in Service”, International Journal of Pavement Engineering, Volume 11 (6), pp.
541–553.
Johnson, K. and S. Hesp, 2014, “Effect of Waste Engine Oil Residue on the Quality and Durability of
SHRP MRL Binders”, Submitted for Presentation and Publication at the 93rd Annual Meeting of
the Transportation Research Board, Washington, D.C.
Planche, J-P, F. Turner, M. Farrar, R. Glaser, W. Grimes, R. Boysen, T. Pauli, 2015, Blended REOB
Binder Advanced Chemical and Physical Characterization, Presented at the FHWA Expert Task
Group Meeting, Fall River, MA, April 9th, 2015.
Reinke, G., 2015, Further Investigations into the Impact of REOB and Paraffinic Oils of the Perform-
ance of Bituminous Mixtures, Presented at the FHWA Expert Task Group Meeting, Fall River, MA,
April 9th, 2015.
Rowe, G.M., 2011, “Prepared Discussion for the AAPT paper by Anderson et al.: Evaluation of the
Relationship between Asphalt Binder Properties and Non-Load Related Cracking.” Journal of the
Association of Asphalt Paving Technologists, 80, pp. 649–662.
Rowe, G.M., G. King, and M. Anderson, 2014, “The Influence of Binder Rheology on the Cracking of
Asphalt Mixes on Airport and Highway Projects, ASTM Journal of Testing and Evaluation, 42(5).
Wielinski, J., A. Kriech, G. Huber, A. Horton, and L. Osborn, 2014, “The Impact of Asphalt Blended
with Re-refined Vacuum Tower Bottoms (RVTB) and Its Effect on HMA Mixture Performance”,
Proceedings of the Canadian Technical Asphalt Association, Volume 59, pp. 402–426.
Zhou, F., W. Mogawer, H. Li, A. Andriescu, and A. Copeland, 2013, “Evaluation of Fatigue Tests for
Characterizing Asphalt Binders,” Journal of Materials in Civil Engineering, 25, pp. 610–617.
434
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
B.A. Feyissa
Ingenieurbüro Abesha—Engineering Office, Freiburg, Baden-Württemberg, Germany
ABSTRACT: The physical and the rheological properties of four binder types had been
investigated. Both short—and long-term aged binders showed lower penetration value,
higher softening point and stiffer characteristics. A 70/100 binder showed higher penetra-
tion ratio and softening point increment for short—and long-term aging. A particular atten-
tion was given on the force-ductility behaviors of aged binders. Results indicated that the
long-term aged binders showed higher ductility-force of 6.8 N, 5.3 N, 2.8 N and 13.7 N for
30/45, 50/70, 70/100 and PmB binders, respectively. Among binder types, hard grade bitumen
(30/45) and polymer modified binder (PmB 45 A) showed relatively higher shear modulus
(G) and viscous parameter (η) due to their stiff properties they already inherited. The rate of
deformation and recovery of both binders were relatively lower than softer grades. The initial
shear modulus values of unaged binders were lower than that of the corresponding values
after long-term aging.
1 INTRODUCTION
1.1 Background
Binders are the most determinant components of the Asphalt Concrete (AC) pavement.
While exposed to combinations of irregular traffic loadings and cyclic environmental con-
ditions, binders undergo change in physical and rheological properties with time. Repeti-
tive traffic loading leads either crumbling of aggregate particles or causes dilation of the
pavement portion that ultimately causes fatigue and rutting failures. The simplest pavement
structural model asserts that each individual load inflicts a certain amount of unrecoverable
damage. However, trucks have a greater share and wheel loads of heavy vehicles are consid-
ered as primary contributors.
Binders are extremely sensitive to the surrounding environmental conditions. The physi-
cal and rheological properties of the binders change with time due to volatilization of light
weight components during high temperature and photo oxidation due to exposure to the
surrounding atmospheric air and solar radiation.
In order to address the real problem and pin point reliable solution in the construction
industry, several investigation techniques are adopted worldwide. Among which the binder
advanced testing techniques are getting more attention. The physical and rheological char-
acteristics were evaluated in this research using standard testing methods as well as with
advanced Dynamic Shear Rheometer (DSR) and Pressure Aging Vessel (PAV).
2 LABORATORY INVESTIGATIONS
2.1 Binders
A total of 4 binder grades, namely of 30/45, 50/70, 70/100 and polymer modified binders
(PmB 25/55–55 A) from the same source were used in this research. Aged and unaged sam-
ples were prepared and standard physical and rheological properties were evaluated. The
properties of all binders before and after short and long term aging were investigated. The
experiments were conducted according to the standard of European Norm as described
hereunder.
436
ing a jet of air. The oven was kept at 163 ± 1°C while the carriage was rotating at a rate of 15
rpm for 75 ± 1 min.
Simulations of accelerated long term aging tests were conducted to evaluate the suscep-
tibility of binders during the service period. The laboratory investigations were carried out
primarily on subjecting the binders with short term aging (with RTFOT). Then the samples
were placed in a pressure aging vessel (PAV) as per EN 14769 under a given pressure (usually
2.1 ± 0.1 MPa) and temperature in between 80–115 °C for specified period of time (20 hr).
437
Table 1. Standard binder test results.
Figure 1. Illustrations of the physical properties of unaged, short—and long-term aged binders for
penetration value and softening point.
The physical changes of binders due to aging were addressed in terms of the penetration
ratio and the change in the softening point. Penetration Ratio (PR) of the aged asphalt with
their unaged asphalt can be used to reflect the change of properties of asphalt binder during
aging and it s expressed as,
PAged
PR = × 100% (1)
PUnaged
438
Figure 2. Force ductility test results for four binder grades.
where SAged = softening point after aging; SUnaged = softening point before aging.
Long term aged binders showed relatively lower PR value, implying that these binder grades
had undergone harder during aging. The PR value of PmB binder was relatively higher dur-
ing the short term aging as shown in Table 2., indicating that the penetration properties of the
binder were less sensitive to aging when compared to other unmodified binders during the
initial phase, however, the change was slightly higher during long-term aging, meaning that
the physical properties of PmB binder were relative sensitive in the long term.
A similar trend showed that the changes in the softening point of both short and long term
aged binders were higher when compared with unaged binders. Among unmodified binders,
the change in softening point were comparable among each other, however, the difference was
higher when compared with PmB binder. The ΔS value for PmB was lower, however, the incre-
ment of the change in its softening point was rather higher with aging compared with other
binders, which has raised some questions regarding sensitivity of the PmB binder while aging
in long terms and it needs in-depth investigation (which was not part of this research work).
The force-ductility measurements were conducted to estimate the maximum force required
and the deformation characteristics while subjecting binders for long term and short term
aging.
As illustrated in the Figure 2, long term aged binders required the highest force to undergo
deformation due to the higher stiffness attained during the aging process. The force required to
deform almost all unmodified binders significantly dropped or vanished after 200 mm defor-
mation, however, both aged and unaged PmB relatively maintained a certain resistance beyond
400 mm deformation. A particular attention was given on the force-ductility behaviors of aged
binders. As can be seen from Figure 2, the long-term aged binders showed the highest ductility-
force of 6.8 N, 5.3 N, 2.8 N and 13.7 N for 30/45, 50/70, 70/100 and PmB binders, respectively.
439
3.2 Rheological test results
The viscoelastic properties of binders were evaluated with respect to creep-recovery and fre-
quency test results. The results showed the properties of all binders had significantly been
altered during aging. The various parameters investigated are discussed as follows.
γ (t ) 1
J (t ) = = (3)
τ (t ) G (t )
440
shear modulus, higher viscous parameter and lower rate of creep-recovery, the total defor-
mation decreased and recovery parts were relatively lower. On the other hand, binder grades
50/70 and 70/100 showed relatively lower shear modulus and viscous parameter, in particular
70/100 grade of binder showed the least, as expected. They underwent higher deformation due
to their softened properties (lower shear modulus and viscous parameter) and their higher rate
of deformation/recovery properties hardly influenced to lower the total deformation.
Figure 5. Illustration of the relationship between cyclic loading and rheological properties.
441
G* G iG , or, (4)
σ0
G* = = G ′ 2 + G ′′ 2 (5)
γ0
where G* = complex shear modulus; G′ = storage shear modulus; G″ = loss shear modulus;
σ0 = maximum shear stress; γ0 = maximum shear strain.
The phase angle is defined (in radian) as time lag between the maximum shear stress and
the maximum stain and expressed as,
G′
tanδ = (6)
G ′′
442
Figure 7. Phase angle at temperatures of 20 °C.
tures and frequencies. Typical results of all binders at 20 °C are demonstrated in Figure 7 below.
At lower temperatures, 30/45 and PmB unaged binders showed lower phase angle and 70/100
binder showed the higher phase angle. In general, decreased in phase angles were observed for
aged binders. In particular short—and long-term aged PmB binders showed the lowest phase
angle, due to the fact that aging had led an increased in storage modulus and a decreased in
corresponding loss modulus. For the three unmodified binders presented, the phase angles
generally exhibited higher values particularly at high temperatures, indicating a more viscous
behaviors and greater tendencies to flow under pressure. For the polymer modified binders,
however, the values of the phase angles were rather lower at higher temperature due to higher
storage moduli values which were also evidenced during the force-ductility experiments.
The physical and rheological properties of four binders have been investigated with accelerated
aging technique as per European Norms at TU-Darmstadt. Among the binders, 70/100 binder
grade showed relatively higher changes in the penetration ratio of 62.58% and softening point
increment of 5.85°C for short-term aging and 41.20% and 12.60 °C corresponding values for
long-term aging. For the PmB binders, corresponding values are 74.32% and 2.25 °C for short
term aging and 49.42% and 8.30% for long term aging, respectively. These figures shows that
the physical changes that PmB binder were relatively smaller compared to other binders for the
case of short term aging implying that PmB binders would perform better than other binders
during mixing and laying in practical situation where it is considered as the cause of the initial
(short-term) aging, however, during long-term aging, the penetration ratio was rather higher
than the remaining two binders (30/45 and 50/70), which gave the PmB binder an indisputable
stiff characteristics. Such characteristics had also been evidenced during force ductility test
with a maximum magnitude of 13.7 N for long term aged PmB binder.
443
The creep recovery result also showed that the shear modulus (G) and the viscous parameter
(η) increased with aging, particularly highest values were recorded for long term aged binders. At
20 °C, the initial shear modulus values for unaged binders of 30/45, 50/70, 70/100 and PmB were
4.77 × 105 N, 2.70 × 105 N, 1.83 × 104 N and 3.70 × 105 N and corresponding values after long-term
aging were 1.11 × 106 N, 9.46 × 105 N, 5.57 × 105 N and 1.42 × 106 N, respectively. With the same
temperature, the viscous parameter (η) for unaged binders were 8.52 × 105, 4.93 × 105, 2.62 × 105
and 8.79 x105 Pas and that of long-term aged binders of 3.48 × 106, 2.77 × 106, 2.04 × 106 and
6.12 × 106 Pas for binders of 30/45, 50/70, 70/100 and PmB, respectively.
Furthermore, the master curves representing the rheological properties of a binder at
selected temperature revealed that the complex shear moduli of all binders decreased and the
phaseangles increased with increased in temperature. At the same time, the moduli increased
and the phase angles decreased significantly due to aging. In most circumstances, for both
unaged and aged binders, the minimum moduli were recorded for 70/100 followed by 50/70,
30/45 and PmB binders except the 30/45 binder showed relatively higher moduli than the cor-
responding PmB at lower temperature ranges in both unaged and short-term aged cases.
The tests further indicated higher shear moduli and decreased stress relaxations and heal-
ing characteristics after aging. Among the samples, the PmB had shown comparatively better
results and performance characteristics. Further investigation is needed that can possibly
accommodate moisture, ultraviolet and other factors responsible for the reduction perform-
ance of binders. Besides, detailed investigations of the various components/composition of
the binders are necessary to come across with the option to enhance the binder performances
as well as to pin point important components which play an important role in the perform-
ance and properties of the binders, in which the research is under study at the institute.
ACKNOWLEDGEMENT
The authors would like to thank the Institute of Road and Pavement Engineering of TU Darm-
stadt for the provision of full laboratory facilities and materials necessary for this research.
REFERENCES
Bell C.A. 1989. Summary Report on Aging of Asphalt-Aggregate Systems, SR-OSU-A-003 A-89–2.
Bell, C.A. 1994. Aging-Binder Validation, Strategic Highway Research Program, SHRP-A-384, National
research Council.
Hagos, E.T. 2008. The Effect of Aging on Binder Properties of Porous Asphalt Concrete, PhD Thesis,
Delft University of Technology.
Jain, M.C. Negi, R.S. Krishna, A. & Tyagi, B.R. 2008. Multigrade Bitumen Verses conventional Bitu-
men, Journal of scientific and Industrial Research, Vol. 67, pp 307–313.
Kliewer, J.E. Zeng, H. & Vinson, T.S. 1996. Aging and Low-Temperature Cracking of Asphalt Concrete
Mixture, Journal of Cold Regions Engineering, ASCE, Vol. 10, No.3, pp 134–148.
Kumar, P. Mehndiratta, H.C. & Singh, K.L. 2009. Rheological Properties of Crumb Rubber Modified
Bitumen-A Lab Study, Journal of scientific and Industrial Research, Vol. 68, pp 812–816.
Malkin, A.Y. 1994. Rheology fundamentals-Fundamental topics in rheology, ISBN 1–895198–09–7,
ChemTec Publishing Canada.
Schmalz, M. Letsch, R. & Plannerer, M. 1990. Investigation on High and Low Temperature Behaviour
of Asphalt by Static and Dynamic Creep Tests, Mechanical Tests for Bituminous Mixes, Characteriza-
tion, Design and Quality Control, Proceedings of the Fourth International Symposium, pp 270–277.
Sybilski, D. 1996. Zero-Shear Viscosity of Bituminous Binder and Its Relation to Bituminous Mixture’s
Rutting Resistance. In Transportation Research Record: Journal of the Transportation Research Board,
No. 1535, Washington, D.C., pp. 15–21.
Wu, S. Pang, L. Liu, G. & Zhu, J. 2010. Laboratory Study on Ultraviolet Radiation Aging of Bitumen,
Journal of Materials in Civil Engineering, ASCE, Vol. 22, No. 8, pp 767–772.
Ye, Q. & Wu, S. 2009. Rheological Properties of Fiber Reinforced Asphalt Binders, Indian Journal of
Engineering and Material Sciences, Vol. 16, pp 93–99.
444
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Yanzhu Wang
China University of Petroleum, Qingdao, Shandong, China
Xudong Wang
Research Institute of Highway, Ministry of Transport, Beijing, China
ABSTRACT: The adhesion between asphalt and aggregate is the process that asphalt wet
aggregate. According to the wetting equation derived from the theory of surface free energy,
the adhesion properties between asphalt and aggregate can be evaluated quantitatively. This
paper made a research on the adhesion work between different asphalt and aggregates based
on the theory of surface free energy, which evaluated the adhesion properties of asphalt and
aggregate at a high temperature. The results show that the relationship between the adhe-
sion and temperature is in accordance with quadratic equation. And there exists a maximum
point, which indicates an optimum wetting state between asphalt and aggregate. The various
adhesion properties between different asphalt and different aggregate occur in corresponding
temperatures, which has guide meaning to determine the mixing and molding temperature
of asphalt mixtures.
1 INTRODUCTION
Moisture damage can greatly reduce the asphalt pavement performance and service life,
which have attracted more and more attention in recent years. The adhesion between asphalt
and aggregate is a direct factor of asphalt mixture of moisture susceptibility. The adhesion
between asphalt and aggregate is the process that asphalt wet and spread aggregate, affected
by the nature of the two mixtures and mixing temperature (Masad et al. 2006; Wasiuddin,
N.M. 2007). In order to make asphalt wet aggregate adequately, the appropriate mixing tem-
perature is very important. At present, mixing temperature is reported to be determined by
the asphalt viscosity-temperature property at home and abroad; but, this does not apply to
the modified asphalt. If setting the temperature of bitumen viscosity of 0.17 ± 0.02 Pa•s
as the mixing temperature, according to "Highway Engineering asphalt and asphalt test
procedures"(JTJ 052–2000) established in 2000, then the mixing temperature of modified
asphalt will reach 200 °C and excessively high temperature will bring the asphalt aging,
energy waste and other issues, which is clearly inappropriate.
In foreign countries, Yetkin Yildirim, H.U. Bahia and Aroon Shenoy have done some
research about how to define mixing and compaction temperature for modified asphalt mix-
ture. Yildirim simulated modified asphalt viscosity state in the mixing and compaction process
by considering the effect of shear rate in the viscosity measurement process. Bahia determined
the modified asphalt’s mixing and compaction temperature according to the curvilinear rela-
tionship between the zero shear viscosity and temperature. Shenoy believed that when investi-
gating the mixing temperature of the mixture, we should take both the rheological properties
445
of asphalt and the impact of asphalt mastic into consideration. However, these methods are
still in deficiencies (Yildirim 1999; Bahia 2000; Shenoy 2001). This paper aims to study the
adhesion work between five types of asphalt and two types of aggregates (Guangxi diabase,
Yunluo limestone) under a high temperature based on the theory of surface free energy so as
to evaluate the adhesion properties of asphalt and aggregate, while setting the temperature
which can generate the maximum adhesion work (optimum wetting state) as the reference for
asphalt mixing (Kwok, Neumann 1999; Arno et al 2006; Murat & Rifat 2014).
2 BACKGROUND
Adhesion process is that liquid contacts with solid and then the surface of the liquid and
solid disappears while forming a new liquid-solid interface. Contact angle (θ ) measurement
as described by Thomas Young in 1805 remains at present the most accurate method for
determining the interactive energy between liquid and solid (van Oss 1994):
γl θ = γ s γ sl (1)
Where γ l represents the surface energy(or surface tension) of the liquid, γ s stands for the
surface energy of the solid and γ sl accounts for the interficial tension between liquid and
solid. In Eq.(1) the γ l and cosθ are known, and γ sl and γ s are the unknown parameters.
The Dupre equation represents the free energy of interaction between solid and liquid
(Fowks 1963):
Gsl = γ sll − γ l − γ s
ΔG (2)
Where ΔG Gsl represents the free energy of interaction between liquid and solid. Combining
Eqs.(1) and (2) results in the Young-Dupre equation (Chaudhury 1984):
Gsl = γ l 1 + cos
−ΔG cosθ (3)
Where the Wsl represents the work of adhesion between liquid and solid.
From the equation (4), it’s clearly indicated that we can obtain the work of adhesion
between the asphalt and the aggregate at the determination of the surface tension (γ l ) of
asphalt and the contact angle (θ ) of the asphalt on the aggregate.
3 EXPERIMENTS
446
3.1 Surface tension measurement on pendant drop
If a drop of liquid is hanging from a syringe needle then it will assume a characteristic shape
from which the surface tension can be determined. The force of gravity on the drop and
depending on the particular height compensates the Laplace pressure, which is given by the
curvature of the drop contour at this point. The Laplace pressure results from the curvature
standing vertically upon one another in the following way:
⎛ 1 1⎞
Δp = γ ⋅ ⎜ + ⎟ (5)
⎝ R1 R2 ⎠
Where γ is the surface tension of the liquid. This equation describes the difference
between the pressure below and above a curved section of the surface of a drop with the
principal radii of curvature R1 and R2. The pressure difference Δp is the difference in pres-
sure between the outside of the drop and its inside.
This work has measured the surface tension of asphalt on DSA (Drop Shape analyzer,
Figure 1) device under a high temperature range based on pendant drop. The DSA instru-
ment (with a temperature control system) is fully automated and can be controlled by com-
puter with the provide software.
Figure 1. The DSA instrument (a) and surface tension measurement (b).
Figure 2. Diabase after cutting with precision cutting instrument (a) and contact angle measurement
on sessile drop (b).
447
distilled water and ethylene glycol. After the cleaning process, the rock specimens were put
inside an oven at 105 ± 5°C for 2h. The samples were then allowed to cool to room tempera-
ture in a desiccator with anhydrous calcium sulfate crystals.
Contact angle measurement influenced by temperature, the asphalt and aggregate should
be put under the same test temperature for 1h and then begin the test. Three measurements
were carried out per asphalt sample and temperature, and the average value of them was used
to calculate the adhesion work.
* SD = standard deviation
448
Table 3. The slope of the line (K) and the linear correlation coefficient (R2).
Temperature °C
* SD = standard deviation
Temperature °C
Modified
Aggregate asphalt 140 150 160 170 180 190
Contact Diabase I-D Avg. 58.9 45.2 35.1 28.7 22.8 19.5
angle (°) SD 1.3 0.9 0.9 0.3 0.1 0.9
I-DS10 Avg. 55.7 45.6 37.0 29.8 22.0 17.4
SD 0.7 0.8 0.6 1.0 1.1 0.5
Limestone I-D Avg. 53.4 42.0 33.7 27.3 22.5 18.85
SD 0.4 0.9 1.2 0.8 0.8 0.9
I-DS10 Avg. 54.7 45.3 38.0 31.3 23.2 18.3
SD 0.7 0.2 1.3 0.8 1.0 0.6
* SD = standard deviation
449
Table 5. The work of adhesion between asphalt and aggregate data (a).
Temperature °C
Table 5. The work of adhesion between asphalt and aggregate data (b).
Temperature °C
Modified
Aggregate asphalt 140 150 160 170 180 190
Figure 4. The relation curve between the work of adhesion and temperature (a) and The viscosity-
temperature curve of 30# asphalt (b).
The surface tension and the contact angle of asphalt on aggregate surface decrease with
increasing temperature, and these two factors affect the size of the adhesion work based on
the Eq. (4). So the adhesion work increases firstly and then decreases as the temperature
increases.
Figure 4 shows the relationship between the work of adhesion and temperature, the work of
adhesion between 30# asphalt and diabase, for example. The relationship between the work of
adhesion and temperature is in accordance with quadratic equation. All curve equations and
the correlation coefficient are given in the Table 6. Table 7 also lists the maximum adhesion
work and the corresponding temperature calculated based on the relation curve equation.
In Table 7, compared with the maximum adhesion works in different temperature, the
maximum work of adhesion between the same asphalt and the two aggregates is almost equal,
and the maximum work of adhesion between modified asphalt and aggregate is smaller. The
adhesion between asphalt and aggregate involves complex physical and the chemical proc-
esses, and adhesion work under high temperature is hard to represent the real adhesive prop-
erties under the actual temperature of asphalt pavement. But it reflects the optimal wetting
state of liquid asphalt on the surface of aggregate under the corresponding temperature.
This study also measures the viscosity of these five types of asphalt at five temperature
points, and the data are shown in Table 8 while the viscosity-temperature curves are illus-
450
Table 6. The curve equations of adhesion work varies with temperature and the correlation coefficient
(R2).
Table 7. Maximum adhesion work between asphalt and aggregate and the corresponding temperature.
Viscosity η/Pa⋅s
trated in Fig .4. According to the viscosity-temperature curve equations, the mixing tem-
perature of these five types of asphalt can be obtained (the viscosity of asphalt at 0.17 ±
0.02 Pa•s), as shown in Table 9.
In order to make asphalt wet aggregate adequately, the appropriate mixing temperature
should be selected so that the adhesive strength and the stability of asphalt mixture can
451
be enhanced. The mixing temperature of modified asphalt obtained from the viscosity-
temperature curve can be up to 200 °C, which has far exceeded the specified construction
temperature. Comparing the data in Tables 7 and 9, the optimal temperature of asphalt
wetting aggregate is almost the mixing temperature obtained by viscosity-temperature
curves, and the maximum adhesion work is also in line with the requirements of construc-
tion temperature. So it is considerable to set the temperature which maximum adhesion
work (optimum wetting state) can be obtained as the asphalt mixing temperature, and this
method proposed in the paper is more convenient and practical than those suggested by
Yildirim, Bahia and Shenoy.
5 CONCLUSION
According to the wetting equation derived from the theory of surface free energy, the adhe-
sion properties between asphalt and aggregate can be evaluated quantitatively, and the optimal
temperature of asphalt wetting aggregate can be set as the mixing temperature. Neverthe-
less, the adhesion between asphalt and aggregate involves complex physical and the chemical
processes, and it is influenced by mineral powder, asphalt mastic, aggregate gradation. There-
fore, further research needs to be done on defining mixing and compaction temperature for
modified asphalt mixture.
REFERENCES
Arno, W.H., Amit, B., Dallas, N.L. 2006. Bitumen surface energy characterization using a contact angle
approach [J]. Journal of materials in civil engineering 18(6): 759–767.
Bahia, H.U., Hanson D.I., Zeng M. 2001. Characterization of modified asphalt binders in superpave
mix design. National academy press.75.
Chaudhury, M.K. 1984. Short range and long range forces in colloid and macroscopic systems. Ph.D.
dissertation. State Univ. of New York, Buffalo, NY.
Fowkes, F.M. 1963. Additivity of intermolecular forces at interfaces. i. determination of the contribu-
tion to surface and interfacial tensions of dispersion forces in various liquids1[J]. The Journal of
Physical Chemistry, 67(12): 2538–2541.
Kwok D Y, Neumann A W. 1999. Contact angle measurement and contact angle interpretation [J].
Advances in colloid and interface science, 81(3): 167–249.
Masad, E., Zollinger, C., Bulut, R., Little, D.N. & Lytton, R.L. 2006. Characterization of HMA mois-
ture damage using surface energy and fracture properties[C].Association of Asphalt Paving Technol-
ogists-Proceedings of the Technical Sessions 2006 Annual Meeting.
Murat, K. & Rifat, B. 2014. Assessment of a sessile drop device and a new testing approach measur-
ing contact angles on aggregates and asphalt binders [J]. Journal of materials in civil engineering
26: 391–398.
Shenoy, A. 2001. Determination of the temperature for mixing aggregates with polymer modified
asphalt. International journal of pavement engineering 2(1): 33–47.
van Oss, C.J. 1994. Interfacial forces in aqueous media, Marcel Dekker, New York.
Wasiuddin, N.M. 2007. Effect of additives on surface free energy characteristics of aggregates and bind-
ers in hot mix asphalt. Ph.D. dissertation, Univ. of Oklahoma, Norman, OK.
Yiklirm, Y. 1999. Mixing and compaction temperatures for modified asphalt binder. National superpave
news 1(3): 6.
452
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: In recent years, due to the influence of the global warming, channeliza-
tion and overloading traffic, all sorts of pavement distresses have occurred on highway
asphalt pavement. Especially, the permanent strain of asphalt pavement appears easily in
high temperature season. A new type of HR rubber powder composite modified asphalt
mixture was researched and developed to solve the problem above. In order to improve
the ability of resistance to permanent deformation of asphalt pavement, various influence
factors on the ability to resist permanent deformation were studied here. In this study,
servo type material dynamic testing system (UTM-25) was used for dynamic creep test,
exploring the influences of constituent materials, including mineral aggregate gradation,
thickness of rubber powder, fiber types, HR content and the type of asphalt cement, as
well as external conditions covering temperature and load, on ability to resist permanent
deformation of it. Results indicate that when using SMA, 40 power and 4‰ HR optimum
content, HR rubber powder composite modified asphalt mixture can get superior ability to
resist permanent deformation with the high performance-to-price ratio. Besides, the ability
of resistance to permanent deformation of it has the small correlation with the temperature
and load, indicating the lesser sensitivity of the performance. Thus, it completely can be
used to solve the problem of insufficient ability to resist permanent deformation of asphalt
pavement.
1 INTRODUCTION
In recent years, there has been pavement damage of high-grade highway in greater degree, of
which the vast majority is disease of high-temperature stability. The high-temperature stabil-
ity of asphalt mixture generally refers to the ability of asphalt mixture to resist permanent
deformation under traffic loading. To some extent, this problem has been alleviated because
of the emergence of asphalt-rubber mixture, and pollution caused by waste tires has been
solved effectively. Studies show that, in terms of the ability to resist permanent deformation,
compared with matrix asphalt mixture, asphalt-rubber mixture has improved too much. But,
the level of increase is limited, which can not completely adapt to the current road condi-
tions, especially under the conditions of highway traffic channelization and heavy traffic. At
the same time, there is long distance far from SBS modified asphalt mixture, which is most
widely used (Zhao, 2008; Huang, 2002).
In this article, drying mix method was used to add HR complex additives into asphalt-
rubber mixture. Dynamic creep test was conducted to study influence of constituent materi-
als, including aggregate gradations, thickness of crumb rubber, types of fiber, HR content
and types of asphalt binders, and other external conditions, covering temperature and stress.
Thus, a novel material named HR crumb rubber composite modified asphalt mixture has
been put forward, and the outstanding ability to resist to permanent deformation has been
confirmed.
453
2 MATERIALS
Performances MA SBS AR
Crushing value Polished value Abrasion value needle and plate content
Units % BPN % %
Basalt 11.4 53 10.6 6.9
Criterion ≤ 28 ≥ 42 ≤ 30 ≤ 15
454
Table 5. Properties and composition of HR.
Items Polymer Fiber asphalt Metallic element Mortar Density Melting point Grain size
Units % % % % % g⋅cm–3 °C mm
HR ≥ 85 ≤ 10 ≤ 10 ≤2 ≤ 1.5 0.95 130~150 4
Mass passage rate of all screen meshs (standard test sieve, mm) (%)
Grading range 16.0 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075
AR-SMA13 100 94.1 67.2 29.7 26.1 22.1 16.3 14.4 13.5 9.8
AR-AC13 100 94.1 61.6 32.9 21.2 15.0 10.5 7.2 4.7 2.4
455
Figure 2. Pictures of specimen preparation.
3.1.2 Coring
Core samples 2# with a diameter of 100 mm were cored from samples1# through core
machine.
3.1.3 Cutting
Cutting machine was used to obtain dynamic creep test samples 3# by removal of two ends.
Finally, the size of dynamic creep test samples 3# is 100 mm (Diameter) × 150 mm (Height).
In order to study the influence of composition materials and external conditions on ability
to resist permanent deformation in HR rubber powder composite modified asphalt mixture,
16 groups of dynamic creep test was analyzed. The results were shown in Table.8.
456
Table 8. Dynamic creep test results of asphalt mixture.
Minimum
Flow creep
Aggregate asphalt Stress number rate
gradations binders Addition materials Temperature/°C /MPa /cycles /(με⋅cycle-1)
Notes: Crumb rubber, CR; Matrix asphalt, MA; Asphalt-rubber, AR; SBS modified asphalt, SBS;
Basalt fiber, BF; Lingo-cellulose, LC.
BF and LC were selected for the dynamic creep test. And the results were shown in
Figure 4. By analyzing test results of different fibers in Figure 4, mixing LC would dramati-
cally reduce the accumulative permanent strain of HR rubber powder composite modified
asphalt mixture. It is because CR itself has the effect of oil absorption and increasing viscosity.
Verifying test indicates adding LC into asphalt mixture can make the actual asphalt-aggregate
ratio much less than the optimum proportion of asphalt. So, the mixing of LC is unnecessary.
Meanwhile, mixing BF would just increase the accumulative permanent strain of HR rubber
powder composite modified asphalt mixture a little, which was because BF acted for reinforce-
ment. However, the cost performance of mixing BF is very low. So, the mixing of BF is also
unnecessary.
In Figure 5 and 6, with the increasing of CR mesh, accumulative permanent strain (Flow
number) of asphalt mixture decreased first and then increased (increased first and then
decreased). This suggested that asphalt-rubber mixture should adopt 40 CR, from the aspects
of improving ability to resist permanent deformation and reducing construction cost.
458
Figure 7. Creep curves under HR content. Figure 8. Relationship between flow number
and HR content.
was still superior to other types of mixture. On the one hand, test temperature was inferior to
its softening point. On the other hand, until the asphalt mixture was solidified, high polymer
in HR had formed a high strength of solidified structure, which could resist softening of
asphalt at high temperature. Meanwhile, some fibers existing in HR could alleviate the reduc-
tion of friction in asphalt mixture at high temperature.
In Figure 12 and 14, with the stress increasing, the cumulative permanent strain and
minimum creep rate (Flow number) were almost upward (downward) trend. As other kinds
of asphalt mixtures, the ability to resist permanent deformation of HR powder composite
modified asphalt mixture declined under the condition of high stress. But, its ability to resist
permanent deformation was still superior to other types of mixture.
With the increase of axial pressure, void fraction of asphalt mixture decreases gradually to
the limit void fraction. And, once exceeding the limit void fraction, free asphalt inside asphalt
mixture is unable to get enough space for expansion, under the high pressure condition.
Then, the original well connected aggregate in asphalt mixture has been squeezed and scat-
tered. Then, the shear resistance would also be reduced. What’s more, the liquidity of asphalt
in high temperature has been enhanced. So, the interior of asphalt mixture would be easy to
go into a state of flow. Thus, the whole mixture strength would be reduced.
In 0.7 MPa, HR crumb rubber modified asphalt has not reached the stage of creep dam-
age. While in 0.9 MPa it has reached the stage. However, the permanent deformation resist-
ance of the later is significantly better than other types. One main reason is that a single HR
particle has a unique cavity surface structure, which can provide partial expansion space for
free asphalt under high pressure. The other is that, containing more than 85% of the high vis-
cosity polymer, HR can enhance the link between aggregates. Therefore, its improvement of
resistance to permanent deformation is superior to mixtures by using process polymer modi-
fied asphalt. Comparing with other types of mixtures, HR crumb rubber modified asphalt
mixture has stronger capacity to resist heavy traffic and permanent deformation.
459
Figure 11. Creep curves under different Figure 12. Creep curves under different stress.
temperature.
Figure 13. Relationship between flow number (minimum creep rate) and temperature.
Figure 14. Relationship between flow number (minimum creep rate) and stress.
5 CONCLUSIONS
There are five key influence factors on ability to resist permanent deformation, including
mineral aggregate gradation, thickness of crumb rubber, fiber types, HR content and the
type of asphalt cement. Results indicate that when using AR-SMA, 40 power and 4‰ HR
optimum content, HR rubber powder composite modified asphalt mixture can get superior
ability to resist permanent deformation with the high performance-to-price ratio under the
condition of high temperature and stress.
Besides, external conditions covering temperature and load also have great effect on
asphalt mixture. The ability of resistance to permanent deformation of HR rubber powder
composite modified asphalt mixture has the small correlation with the temperature and load,
indicating the lesser sensitivity of the performance. Thus, it can be used to solve the problem
of insufficient ability to resist permanent deformation of asphalt pavement.
460
REFERENCES
AliKhodaii, A.M. 2009. Evaluation of permanent deformation of unmodified and SBS modified
asphalt mixtures using dynamic creep test. Construction and Building Materials, 237.
Gao, J.H. 2011. SHRP method used to evaluate the performance of rubber powder and powder compos-
ite modified asphalt. Journal of China & Foreign Highway, 05: 250–252.
Gotz, U.A. & Handge, M.P. 2009. Influence of e-beam irradiation on the dynamic creep and fatigue
properties of poly (aliphatic/aromatic-ester) copolymers for biomedical applications. Polymer, 5023.
Herda, Y.K. & Mohd, R.I. 2015. Evaluation of Permanent Deformation of Unmodified and Rubber-
Reinforced SMA Asphalt Mixtures Using Dynamic Creep Test. Advances in Materials Science and
Engineering, 2015.
Huang, X.M. & Wu, SP. 2002. Asphalt and asphalt mixtures. Nanjing: Southeast University Press.
Li, H. & Huang, XM.2006. High temperature stability of asphalt mixture under uniaxial static creep
test. Journal of Henan University of Science and Technology, 03: 48–51.
Meor, O.H. & R, P. 2009. Effects of Temperature and Binder Type on the Dynamic Creep of Asphaltic
Concrete. Modern Applied Science, 37.
Nuha, S.M. & Mohamed, R.K. 2013. Evaluation of Permanent Deformation of CRM-Reinforced SMA
and Its Correlation with Dynamic Stiffness and Dynamic Creep. The Scientific World Journal.
Qi, F. 2009. The creep test used to evaluate the high temperature stability of asphalt mixture. Chang’an
university, Xi’an.
Shu, W.G. & Zhan, P.Y. 2009. A simple stepwise method to determine and evaluate the initiation of terti-
ary flow for asphalt mixtures under dynamic creep test. Construction and Building Materials, 2311.
Wei, M. 2004. High temperature creep properties research of rotating compacted specimen. Journal of
Chongqing Jiaotong University (Social Sciences Edition), 23 (5):55–58.
Zhao, J. 2008. The performance study of rubber powder and powder composite modified asphalt.
Changsha University of Science and Technology, Changsha.
461
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: To investigate the effect of fines on the aging of bitumen, a set of fourteen
different mastics was prepared and aged with a modified Rolling Thin Film Oven Test.
The unmodified paving grade bitumen of 70/100 penetration was used. The bitumen
was extracted both from the unaged and aged mastic and analyzed with Dynamic Shear
Rheometer and Fourier Transform Infrared (FT-IR). The fines were analyzed for the Loss
on ignition, as well as by the FT-IR. The extracted bitumen was found to be softer than
the original one in some of the blends, implying rejuvenation. However, based on aggregate
analysis, the precipitation of bitumen into the aggregate fines during extraction was noticed.
Conversely, for some bitumen specimens, the FT-IR spectra revealed the presence of fines,
which in many cases convoluted the area of sulfoxides peak. The rejuvenation observed by
rheological testing was the effect of extraction artifacts rather than deoxidation.
1 INTRODUCTION
The techniques used to characterize bitumen were developed on straight run products.
Typically, the analysis of bitumen and aggregate is conducted in a forward design mode, where
the properties of raw materials are correlated to the performance of the asphalt mixture.
As the recycling of asphalt pavements becomes more and more popular, extraction and
recovery of the binder from aged asphalt is applied to determine the properties of the sub-
strate in the recycling process. In such backward analysis, typically rheological parameters
of choice are investigated. In some cases, this involves the determination of the PG grade or
other simpler rheological parameter, e.g. Penetration or Softening Point. These rheological
characteristics are often utilized in the quality control of the process such as success in the
rejuvenation.
The feedback from paving contractors, however, suggests that the reliability of such back-
ward analysis is questionable. The extraction and recovery processes are blamed for the dis-
crepancies and there may be general lack of understanding about the sources of deviations
within one’s own results. Some contractors have reported exceptionally high binder stiffness
values after extraction and some are complaining that binder extracted from mixtures being
softer than the bitumen originally used.
The purpose of this study was to investigate the effect of extraction and recovery on the
bitumen properties and influence of the mineral filler in the recovery processes.
Bitumen, filler and the solvent used for recovery may interact in multiple ways. Firstly,
the extraction depends on the diffusion of solvent into the material and consuming the same
amount of solvent and time of extraction in the stable temperature conditions should yield
different extracts if the stiffness of sample varies.
Secondly, due to the varying mineralogy, aggregates have variable preferences towards dif-
ferent chemical components of bitumen. Plancher et al. (1977) have reported that bitumen
suspended in a solvent over aggregates divides into two phases: a strongly aggregate adsorb-
ing phase and the phase remaining in the solution. Additionally, they showed that different
aggregates attracted different functional groups, e.g. carboxylic acid, ketones and sulfoxyls
463
and that the effect was related to the surface area of the aggregate. Incidentally, those are the
groups forming in the bitumen upon aging, and being the most polar components of bitu-
men they are suggested mostly being responsible for the increase in stiffness of the material.
We are postulating that perhaps due to increased amount of functional groups capable of
being strongly absorbable into the aggregate, the amount of bitumen retained in the aggre-
gate would increase and lead to the softer bitumen than that of the reference binder.
Additional factors affecting rheological characteristics of the extracted binder are the pres-
ence of filer impurities or residual solvent. Lastly, the precipitation of bitumen components
into low concentration solutions is explained by the loss of the mutual solubility between
bitumen components upon the increase in distance between molecules. The components of
the lowest solubility will precipitate. In larger scale, this is observable in the form of pre-
cipitation of asphalthenes into crude oil pipelines. However, this phenomenon has also been
proven by Marczewski & Szymula (2002) to increase over iron oxides rather than over silica
and thus governed by the interaction of asphalthenes in solvent with the third component.
2.1 Materials
A set of seven different mastics with two filler concentrations (37.5% and 45% by volume)
was prepared using one bitumen of penetration grade 70/100. The filler materials used pass-
ing 0.125 mm sieve were granite (KK, r = 2.67 Mg/m3), granodiorite (TE), limestone filler
(KF), fly ashes (FA and FAW) and the blends of those materials (EXP = KK + FA + KF and
REP = KK + KF). More detailed information about the filler characteristics and the expla-
nation of the reasons behind the blending ratios of EXP and REP blends can be found in the
work of Makowska & Pellinen (2015).
Table 1. The matrix of prepared combinations of mastics—varying concentration and varying type
of filler.
KK 2.67 x x x x
KF 2.71 x x x x
TE 2.72 x x x x
FA 2.25 x x x x
FAW 2.24 x x x x
REP 2.71 x x x x
EXP 2.61 x x x x
464
The prepared samples were then conditioned in RTFO oven at 163ºC with 5 l/min air flow
for a period of 25 minutes. After the aging procedure samples were placed into the desicca-
tors and cooled down to the room temperature. Mass loss was not measured due to possible
influence of oil bath on mass changes.
The amount of bitumen was reduced to avoid the mastic crawl out of the bottles and the
time of aging was reduced from the standard procedure to correspond to the amount of bitu-
men used. However, it turned out that the crawl took place only for the mastics with relatively
low softening point and never reached the top of the bottle. This suggests that the time of
aging could be increased in the experiment, if desired.
3 TEST METHODS
3.2 Rheology
DSR tests were conducted using temperature range of 30–120°C with 25-mm diameter
parallel plate geometry and shear rates given in Table 2. Heating rate was kept constant at
0.5 °C/min during the whole experiment starting from the lowest temperature. The obtained
parameter was Zero Shear Viscosity (ZSV) at steady state condition.
3.3 FT-IR
The Nicolet™ iS™50 FT-IR Spectrometer equipped with Pike GladiATR accessory was
used for the measurements. Fines after drying were tested as powder in normal laboratory
465
temperature (under 30 °C). In case of bitumens, the stage was heated up to 60 °C and a back-
ground was collected, the sample was placed on the detector and allowed to melt in order to
obtain a proper contact. Spectra was collected when sample was in a form of a liquid drop
and 32 scans with resolution 4 cm−1 and ATR correction were applied during collection of
spectra.
The ATR mode of FT-IR would be the most suitable for field laboratories as it doesn’t
require special preparation station and additional training for sample preparation.
Due to the presence of impurities in our extracts, the determination of the aging index pro-
posed by Marsac et al. (2014) or CO + SO intensity parameter proposed by Qin et al. (2014)
was abandoned in this work. However, we followed the intensity of peak at 1700 cm−1 (ICO),
which seemed to be the only unaffected marker of oxidative aging in our study (Fig. 1).
Figure 1. An example spectra used for the qualitative calibration of mastics containing a) granite,
b) fly ash and c) limestone at various mass concentrations [wt.%]. The region typically used in the
analysis of bitumen is marked with arrows, and so are the calibration bands used for different material
combinations.
466
3.4 Quantitative calibration and filler content in recovered bitumens
We observed that characteristic peaks for fillers are noticeable in the recovered bitumen
spectra. Limestone expressed most visible change at 875 cm−1, granite and granodiorite at
588 cm−1 and fly ashes contributed to the increase of signal at 400–500 cm−1. We used those
bands in qualitative calibration. For the fly ashes we used the band at 464 cm−1, which is seem-
ingly located on the shoulder of a peak with the maximum expressed in the far-IR region.
In order to determine how the filler contributes to the rest of the spectra as well as to deter-
mine the content of the impurities, a calibration was conducted. A set of mastic was prepared
for FA and KK, where 0, 1, 5, 10 and 15% concentrations were used. In case of KF blends
of 0, 0.25, 0.5, 0.75, 3, 15 and 27% were used. For FA and KK the intensity of marker peaks
increased linearly with increasing concentration. In case of KF, the dependence was linear
until 15% but not above. Interpolation between points was conducted for the higher signals.
From the increase in intensities between recovered binders and blank sample, we have calcu-
lated filler weight concentration in the recovered mastic. This information, applying apparent
volume model prediction of stiffening of bitumen presented in the work of Makowska & Pellinen
(2015), was used to estimate the increase in softening point due to the reinforcement by impurities.
The amount of filler found was between 2 and 23.79 wt.%, according to the calibration curves,
which translated into calculated increase in softening point between 0.95 and 3.2ºC.
Transfer of fines into the bitumen was larger in _45 mastics, suggesting that too large
amount of fines was a substantial load for the filter during extraction (cake thickness affects
to the pressure and may lead to filter rupture). This plays important role when analyzing field
samples of unknown mix design and volumetric ratio, and should be taken account.
Upon examination of the calibration graphs, it becomes evident that presence of silica in
the filler, as determined on the basis of works by Miller & Willkins (1952), and its transfer to
the recovered binder will increase the amplitude and area of the sulfoxyl region. The presence
of limestone slightly increases the signal in the sulfoxyl region but also affects to the area and
amplitude of the peaks of CH2, suggested for use for ICO determination (1470, 1360 cm−1) by
Marsac et al. (2014). It was observed in here that perhaps the peak at 2920 cm−1 should be
considered as normalization peak, as it is not affected by the presence of fillers and is also
related to CH vibration in sp3 hybrydisation, which belong to the group assumably unaf-
fected by the aging, similar to the peaks observed at 1460 cm−1.
467
Figure 2. Relative change in Loss on Ignition value (a) recalculated into bitumen loss into the fines
(not extracted) (b). The negative values are consistent with strong filler signal in the FT-IR and substan-
tial transfer of filler into the bitumen.
The analysis of the Loss on Ignition of the fines before contact with bitumen and after
extraction, revealed observable change as seen in Figure 1b. The effect was bidirectional. Part
of the filler enhanced the extracted bitumen and part of the bitumen stayed in the filler.
Figure 4. Correlation between rheological characteristics and infrared spectroscopy: (a) The relation-
ship between calculated softening point and increase in intensity of carbonyl peak absorption—raw
data, (b)—softening point corrected by the increase in ring and ball softening point due to the presence
of filler, (c) an example of zero shear viscosity characteristics in whole temperature range of extracted
binders as compared to references, (d) relationship between change of adsorption intensity in carbonyl
region with change of binder softening point before and after aging for a corresponding mastic.
469
We can observe from Figure 4 c that the rheological characteristics across the temperature
range are different between reference and recovered binders. For example, FA45 extracted
binder is softer at low temperatures and harder at higher temperatures than the reference
binder, while KF37 extracted binder behaves differently.
4.3 FT-IR
The EN-standard for extraction and recovery encourages laboratories to test residual solvent
presence in obtained extract, but in practice it is not conducted because it is not required.
Upon our growing experience in analysis of bitumens, not only prepared for this research
project but also supplied by various third parties for analysis, we conclude that FT-IR evalua-
tion of bitumen is of vast importance before further analysis of bitumen. So far, authors have
identified presence of fillers, presence of solvents and even signals of residual paint mark-
ings in the extracted bitumen. Authors believe that for a meaningful rheological analysis of
extracted bitumens, an utmost effort should be placed on fixing the extraction and recovery
process and its practices to improve the repeatability and quality control of the processes.
The presence of impurities may affect the scatter of results and mask proper correlation
with other binder properties. Focusing on improving extraction and recovery process should
improve correlations with rheological characterization techniques.
The peak at 2923 cm−1 should be considered for normalization rather than the peak at
1460 cm−1, as suggested by Marsac et al. (2014). In the presence of even small filler impurities,
the spectra in 1460 cm−1 region may be affected by signal coming from filler. When the vol-
ume of sample is held constant, as is in this case, and filler amount increases, the signal from
bitumen should continue to decrease. This is observed at 2923, but after initial drop between
0% and 0.25% in the 1460 region, we observe increase in the peak amplitude.
Qin et al. 2014 suggest that carbonyl (1700 cm−1) and sulfoxyl (1030 cm−1) peaks height be
used in correlation with rheological parameters. Again, because of the convolution of the
spectra by typical silica containing aggregates, in the case of impure sample, the relation to
bitumens rheological behavior may be not apparent.
Additionally it is quite hard to identify, beyond reasonable doubt, presence of filler impu-
rities on the level below 1%wt., by looking at the spectra. Perhaps an additional step should
be included in the extraction and recovery EN-standard, which would require testing for
presence of filler particles by other means.
5 CONCLUSIONS
As long as the signal from the filler is present, the relation between rheology and FT-IR
spectra is convoluted. Caution should be used in characterization of extracted bitumen. The
determination of residual filler, as well as solvent, presence in extracted bitumens before the
analysis of bitumen’s FT-IR spectra is conducted is advised. With the increasing amount of
filler in the original mastic, increased transfer of the filler into the extract was observed. The
fillers with silica increased the signal in the sulfoxyl region, while the calcium carbonate peaks
increased the signals in areas typically used for the normalization of spectra for the carbonyl
index determination as well as in the sulfoxyl region. Exceptional stiffening (false positive of
aging) was demonstrated taking place due to the presence of impurities in the form of filler
transferred during the filtration stage into the extract. We demonstrated how this effect can
be deconvoluted by applying mastic stiffening models.
Caution should be used when one discusses aging prevention (rejuvenation) in the cases
where extraction is a processing step. It was demonstrated that due to the extraction process
alone, a softening of the extracted and recovered bitumen is taking place, which was linked
with the precipitation of bitumen into the filler rather than with the rejuvenation.
We propose that FT-IR analysis of extracts should be a required practice after the extrac-
tion and recovery of the bituminous binder from bituminous mixtures to detect the presence
of solvent and filler particles. Currently this practice is only encouraged but not required.
470
ACKNOWLEDGMENTS
This work was funded by the Finnish Transport Administration. Authors would like to thank
M.Sc. Olli-Ville Laukkanen for execution of the rheological measurements, as well as Heli
Nikiforow and Aiju Heinonen for the help with sample preparation and characterization.
REFERENCES
Han R., Jin X. and C.J. Glover. 2013. Oxygen Diffusivity in Asphalts and Mastics, Petroleum Science
and Technology, 31:15, 1563–1573, DOI:10.1080/10916466.2011.559506.
Makowska, M., T. Pellinen, P. Olmos Martinez, and OV. Laukkanen. 2014. Analythical Methodology to
Determine the Composition of Filler Used in Hot-Mix Asphalt. Case Study. Transportation Research
Record: Journal of the Transportation Research Board 12–20.
Makowska, M, and T Pellinen. 2015. Etchable iron content (FETCH) proposed as the missing param-
eter for the better prediction of asphalt mastic stiffening. Construction and Building Materials 93:
528–541. DOI:10.1016/j.conbuildmat.2015.05.099.
Marczewski, AW, and M. Szymula . 2002. Adsorption of asphaltenes from toluene on mineral surface.
Colloids Surf A 208:259–66.
Marsac, Paul, Nathalie Pierard, Laurent Porot, Wim Van den Bergh, James Grenfell, Virginie Mouillet,
Simon Pouget, et al. 2014. Potential ad limits of FTIR methods for reclaimed asphalt characterisa-
tion. Materials and Structures 47: 1273–1286.
Miller, F.A., and C.H. Wilkins. Infrared Spectra and Characteristic Frequencies of Inorganic Ions. Ana-
lytical Chemistry, Vol. 24, No. 8, 1952, pp. 1253–1294.
Mirza, M W, and M W Witczak. 1995. Development of a Global Aaging System for Short and Long
Term Aging of Asphalt Cement. Journal of the association of Asphalt Paving Technologists 64:
393–430.
Petersen, J. Claine. 2001. A review of the fundamentals of asphalt oxidation. Chemical Physiochemical,
Physical Property, and Durability relationship. Transportation Research Circular E-C140. Washing-
ton, DC: Transportation Research Board.
Plancher, H, S.M. Dorrence, and J.C. Petersen. 1977. Identification of chemical types in asphalts
strongly adsorbed at the asphalt-aggregate interface and their relative displacement by water. Pro-
ceeding Aassociation of Asphalt Paving Technologists. San Antonio, Texas: Aassociation of Asphalt
Paving Technologists. 151–175.
Qin, Qian, John Schabron, Ryan Boysen, and Michael Farrar. 2014. Field aging effect on chemistry and
rheology of asphalt binders and rheological predictions for field aging. Fuel 121: 86–94.
Ray Brown, E., J E Haddock, and Campbell Crawdord. 1996. Investigations of Stone Matrix Asphalt
Mortars. Transportation Research Record 1530 (Materials and Construction: Recycled Rubber,
Aggregate and Filler in Asphalt Paving Mixtures): 95–102.
471
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
Asphalt binders produced from straight run distillation are not always suitable to fulfill the
specifications. Therefore, asphalt modifiers may be required. Typical modifiers are polymers,
chemical additives, or a combination of the two. Polyphosphoric acid (PPA) is a chemical
modifier and has been used for this purpose in North America for over thirty years (Alex-
ander 1973). Typical dosage of PPA ranges between 0.25 wt% and 1.0 wt% (based on the
weight of the binder). An estimated 100 to 400 million tons of asphalt mix modified with
PPA have been utilized on US highways in the recent years.
In the US PPA is used in various applications which involve preparation of paving or
roofing grades bitumens. These applications involve direct modification of bitumen binder
with PPA (Maldonado 2006, Masson 2007, Maldonado 2007), preparation of PPA-modi-
fied binders in combination with SBS polymers (Baumgardner 1998, Baumgardner 2000,
Engberg 2000, and Puzic 2002), the application of PPA as a catalyst in preparation of
bitumen modified with reactive terpolymers (Prejean 2007), and as an accelerator during
air-blowing process (Fort 1963, Falkiewicz 2008). In neat paving bitumen, polyphosphoric
acid increases the high-temperature Performance Grade (PG) rating of the bitumen while
maintaining the low-temperature properties. Significant improvements in the water-sen-
sitivity of mixes are also obtained. In polymer-modified bitumen, the use of polyphos-
phoric acid provides these same benefits and also allows for a significant reduction in the
level of polymer required to meet elastic recovery requirements. The mechanism by which
polyphosphoric acid interacts with bitumen to improve its rheology and overall properties
is still under investigation. One theory that has been put forward suggests that polyphos-
phoric acid reacts with various organic functional groups in bitumen (Baumgardner 2005)
473
Figure 1. Possible mode of action of Polyphosphoric Acid in asphalt modification.
breaking up asphaltene agglomerates and allowing the individual asphaltene units to form
a better dispersion in the maltene phase. The dispersed individual asphaltene units are
relatively more effective in forming long-range networks and in turn contribute to elastic
behavior, Figure 1.
The chemical composition of bitumens varies and depends on the source of the crudes.
For example, bitumens obtained from Venezuelan crudes are typically characterized by high
asphaltene content, whereas bitumens obtained from Russian or Chinese crudes have high
wax content. The influence of the composition of asphalt has been recognized for many
years, as being an important factor in controlling the performance of PPA-modified bind-
ers (Orange 2004). In general, the higher the level of asphaltenes, the more effective is the
modification with PPA. Recently, we investigated the effect of PPA on the physical and rheo-
logical properties of bitumens obtained from different parts of the world (Shulga 2012). The
present paper is a continuation of the previous work, and involves in-depth investigation of
the effect of PPA on the properties of Chinese bitumen. In particular, we were interested in
development of a prototype polymer modified bitumen formulas demonstrating the benefits
of PPA co-modification within the limits of specification compliance for Chinese Specifica-
tion JTGT40, 1-D and standard PG requirements.
2. EXPERIMENTAL
2.1 Materials
Base bitumen PG 64–22 from Sinopec (China) was used throughout the study. 105%
polyphosphoric acid from ICL Performance Products, SBS polymers D 1101 from Kraton,
and elemental sulfur from Sigma Aldrich were used for binder modifications.
2.2 Methods
All asphalt samples were first modified with Hydrolyene® 90T (Holly Frontier) to achieve
properties required to achieve polymer modified binder compliance. Table 1 shows repre-
sentative samples and the amounts of each modifier.
All binders were evaluated with respect to the Chinese JTGT-40 I-D specifications and the US
AASHTO M320 (Table 1 & 2) specifications. The various test methods performed on the bitu-
mens according to US and Chinese standards are presented in Table 2. The effect of short and
long-term aging on PPA-modified bitumens was studied by conditioning samples using rotating
thin film oven (RTFO) and pressure aged vessel (PAV) oven. Low-temperature performance was
measured by the Bending Beam Rheometer and Direct Tension Device. For all lab tests the bitu-
men samples were heated to 275 °F (135 °C) and 105% PPA was added while under low shear.
The samples were then allowed to mix for additional 30 minutes under low shear.
Modification of the base binder with SBS polymer was controlled by monitoring the dissolu-
tion process using fluorescence photomicrography. Table 3 shows an example of fluorescence pho-
tos of several SBS-modified bitumen samples A, B, and E (see Table 1 for sample composition).
Typically, shorter mixing time is required in the presence of sulfur crosslinker as shown in Table 3.
474
Table 1. Bitumen samples and the amounts of modifiers used in the study.
Concentration
Ingredient Control A B C D E
95% bitumen/5% 90T Hydrolene® oil,% 100 95.5 96.5 95.0 95.0 96.94
SBS D 1101,% --- 5.0 4.0 4.0 3.0 2.5
105% PPA,% --- --- 0.5 1.0 2.0 0.5
Elemental sulfur --- --- --- --- --- 0.06
Table 3. Fluorescence photos of SBS-modified bitumen mixed for various periods of time.
Compliance of the base Chinese bitumen (Sinopec) was evaluated according to the US and
Chinese specification, AASHTO M 320 and JTGT 40, respectively. Table 4 shows a variation
of bitumen properties for neat binder and binder modified with 3% SBS polymer.
Table 4 shows that neat and SBS-modified Chinese binder achieved a PG of 64–22 and PG
70–16, respectively according to US specifications. For Chinese specifications the neat and
SBS-modified binders show a compliance with JTGT 40 specifications except penetration at
25 C which was found to be 46 dmm for SBS-modified bitumen. It was determined that the
base binder, although exhibiting excellent polymer compatibility, was too stiff/hard to com-
ply with JTGT 40, Type I-C. Furthermore, addition of PPA typically results in increase of the
475
bitumen stiffness. Consequently, it was decided to soften the base binder with hydrocarbon
oil which is a common practice across the world.
It is known that addition of PPA to SBS-modified bitumen allows for substantial reduc-
tion of polymer content without a sacrifice in performance of the binder which leads to a
significant production and economic benefits. Table 5 shows the full Chinese specification
requirements and compliance data according to JTGT 40 I-D for several prototype samples.
As seen from Table 5 neat (unmodified) bitumen sample does not show a compliance with
JTGT 40 specification. Modification of the binder with 5% SBS polymer leads to overall improve-
ments in properties; however the prototype binder is still not passing separation specification.
A reduction of SBS concentration by 40% (Sample B) and replacement with 0.5 wt% of PPA
leads to some improvement in separation, however, not enough to pass the specified value. Also,
it should be noted that ductility at 5 C decreases from 46 cm for sample A to 26 cm for sample
B. Sample B marginally passes the ductility specification. This effect is typically observed in
PPA-modified asphalts. As it was mentioned earlier, PPA provides stiffening effect by chemically
reacting with asphaltene compounds. As concentration of PPA increases further to 2.0 wt%
(sample D), this effect becomes more pronounced and it is reflected in decrease of penetration
and ductility values. The value of penetration decreases from 58 dmm for sample B to 39 dmm
for sample D, whereas ductility decreases from 26 cm to 17 cm. Therefore, it was decided to
accept sample B as a model and improve separation with the aid of elemental sulfur used as a
crosslinker. Sample E containing 2.5 wt% SBS polymer, 0.5 wt% PPA, and 0.06 wt% of sulfur
shows full compliance with JTGT 40 I-D specifications with separation values being well below
specified value of 2.5 C. The application of crosslinker also improved ductility which was found
to be 27.5 cm. Sample E also provided excellent viscosity result which benefits pavement instal-
lation by making the preparation of mix easier and achieve the required pavement density.
Viscoelastic properties of Chinese binders were also investigated according to the US
specifications AASHTO M320. Table 6 shows a comparison of the rheological properties
for samples B and E.
Evaluation of polyphosphoric acid in bitumen systems in the presence of SBS polymers and sul-
fur crosslinker demonstrates complementary performance. Table 5 shows results for a system where
a portion of SBS is substituted with a relatively smaller amount of 105% polyphosphoric acid. The
total percent additive is reduced and all performance parameters were found to remain within
specifications. Further, the systems with polyphosphoric acid shows a significant wider effective
PG range, with improvement in both high-temperature and low-temperature performance.
Table 4. Characterization of neat and SBS modified bitumen according to AASHTO M320 and
JTGT 40 specifications.
Neat+
Property Spec Neat 3% SBS
476
Table 5. Compliance with China polymer modified bitumen specifications JTGT 40 I-D.
Results
Table 6. Viscoelastic properties of PPA and SBS-modified bitumens according to AASHTO M 320.
Dynamic Shear (G*/sin δ), 10 rad/sec, kPa 76 C 1.0 min 0.87 1.42
Phase Angle @ 76 C, δ report 79 69.9
RTFOT Residue
Dynamic Shear (G*/sin δ), 10 rad/sec, kPa 76 C 2.2 min 2.64 3.94
82 C 1.43 2.37
Phase angle @ 82 C 75.1 60.2
Elastic recovery @ 25C 75 min 75 82.5
PRESSURE AGING RESIDUE (100 C, 300 psi, 20 h)
Creep Stiffness Stiffness, MPa (60sec) −12 C 300 max 106 118
m Value 0.300 min 0.325 0.328
Stiffness, MPa (60sec) −18 C 300 max 232 191
m Value 0.300 min 0.276 0.321
Performance Grade, PG 70–22 76–28
Useful temperature Range, C 103.8 110.7
4 CONCLUSIONS
We developed prototype polymer modified bitumens demonstrating the benefits of PPA co-
modifications using a Chinese binder from Sinopec. The resulting prototype shows a compli-
ance with Chinese specifications JTGT 40 I-D and meets all standard PG requirements. With
the aid of PPA it was possible to achieve near 50% reduction of SBS content which may have
a significant reduction of raw material cost. In general, the application of PPA in bitumen
modification provides the following benefits:
• Compliance with Chinese and US specifications
• Increase in the high temperature Performance Grade (PG) rating with no loss of low tem-
perature properties.
• Increase in the adherence of the bitumen to a moisture-sensitive aggregate, resulting in anti-
stripping benefits similar to those obtained from liquid amine-based agents and hydrated lime.
477
• No detrimental effects when used together with hydrated lime and two types of liquid anti-
strip agents.
• Significant reduction in rutting.
Due to the differences in bitumen chemistry and as indicated in our data the properties of
the bitumen must be carefully monitored during formula selection process.
ACKNOWLEDGEMENTS
REFERENCES
478
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: This paper presents a laboratory study in TLA (Trinidad Lake Asphalt)
modified asphalt with different TLA content including the high temperature performance
and microstructure. The softening point, viscosity-temperature characteristics, failure tem-
perature and temperature sensitivity were adopted to characterize the high temperature
performance of TLA modified asphalt. Meanwhile, the microstructure was investigated
by Scanning Electron Microscopy (SEM), Differential Scanning Calorimeter (DSC) and
infrared spectroscopy (IR). The results show that adding TLA could improve the high
temperature performance and temperature sensitivity of TLA modified asphalt. With the
increase of TLA content, the softening point, viscosity and failure temperature increase.
By considering above properties, the optimum content of TLA is 25%–35%. With increase
of TLA content, phase morphology is converted from a continuous asphalt phase with
dispersed TLA phase to two twisted continuous phases. And the Tg of modified asphalt
increases with the increase of TLA content, which is unfavorable for the low temperature
performance of asphalt. Moreover, the chemical reaction was not happened after adding
TLA.
1 INTRODUCTION
Rencently, asphalt is widely used in road pavement and airport construction as the binder
of aggregates(Shingo et al.2001, Sengoz et al. 2007). Unfortunately, further applications of
asphalt are restricted by some disadvantages, such as high-temperature rutting and low tem-
perature cracking (Navarro et al. 2001). In order to improve the quality of asphalt, it is gener-
ally modified by various modifers such as polymers, crumb rubber and anti-rutting agent and
so on (Airey et al. 2004, Becker et al. 2003).
The Trinidad Lake Asphalt (TLA) is a natural asphalt, produced in the coastal Plateau
of Trinidad that located in northeastern South America (Wang 2012). Engineering practice
proved that TLA as a modifier generally performed similar to above polymer additives in
that they can increase the performance of asphalt, and then improve the ability of pavement
(Liao et al. 2014). Moreover, the component of TLA includes asphaltenes, resins, oil fraction
and insoluble composition, which are similar to that of base asphalt (Tong 2008). And that
indicates better compatibility between asphalt and TLA, which is benifical for the perform-
ance of TLA modified asphalt.
Although the TLA are successfully used in many engineeing practices, the influence of
TLA content on the high temperature and change of microstructure of modified asphalt is
not clear. So, the objective of this paper is to investigate the influence of TLA content on
the hgh temperature properties and microstructure of TLA modified asphalt. Experimental
methods such as the rotational viscosity tests, temperature sweep tests, temperature sensitiv-
ity tests, and SEM, DSC and FTIR Spectroscopy tests will be conducted and analyzed for
this purpose.
479
2 EXPERIMENTS AND METHODS
2.1 Materials
The virgin asphalt with 60/80 penetration grade produced at the Ssangyong asphalt plant
of South Korea was used. The properties of the virgin asphalt are listed in Table 1. And the
Trindad Lake asphalt (TLA) was selected as a modifier to modify asphalt and its properties
are presented in Table 2.
480
Figure 1. The flow chart of samples preparation and experimental program.
481
components and inorganic component could improve the high temperature performance of
asphalt.
In addition, it could be concluded that the increasing amplitude of softening point increases
slightly with the TLA content from 5% to 25%, while increasing amplitude of softening point
increases dramatically. By considering the softening point, the optimum content of TLA is
25%–35%.
482
Figure 3. The viscosity-temperature curve of TLA modified asphalt with different TLA content.
Figure 4. The resistance rutting factor G*/sinδ of TLA modified asphalt with different TLA conten
before and after RTFOT.
great linear relationship between log(G*/sinδ) and log T, which is consistent with the find-
ings in previous studies (Yu et al. 2013). Furthermore, the values of G*/sinδ of TLA modi-
fied asphalt are higher than that of non-modified asphalt at the same temperature. In other
words, the high temperature performance of asphalt are improved after adding TLA. With
the increase of TLA content, the resistance rutting factor G*/sinδ of TLA modified asphalt
become higher to some extent. Moreover, the values of G*/sinδ of original samples are higher
than that of RTFOT samples.
The failure temperature corresponding to a G*/sinδ of 1.0 kPa and 2.2 kPa for TLA mod-
ifed asphalt before and after RTFOT could be obtained from the results of Figure 4. And
the failure temperatures of TLA modified asphalt with different TLAcontent are listed in
Table 3. It is obviously found that the failure temperatures of TLA modified asphalt are
483
Table 3. The failure temperature of TLA modified asphalt with different content
before and after RTFOT.
Failure temperature/°C
TAL content/% G*/sinδ = 1.0 kPa G*/sinδ = 2.2 kPa
0 69.32 66.20
5 71.21 67.73
15 71.92 69.15
25 72.70 71.50
35 80.53 77.08
Figure 5. Logarithm of storage modulus and loss modulus of TLA modifed asphalt with different
TLA content.
higher than that of non-modified asphalt. Meanwhile, the failure temperature of TLA modi-
fied asphalt increase gradually with the increase of TLA content. Meanwhile, there is a big
increase amplitude of failure temperature from 72.70°C to 80.53°C, when the TLA content
increase from 25% to 35%.
Furthermore, after RTFOT the failure temperature of TLA modified asphalt is lower than
that of the original samples at the same TLA content. And the lower the TLA content is,
the smaller the amplitude reduction of failure temperature is. Simliar to the original samples,
the failure temperature increases as the TLA content increasing. The failure temperature is
71.50°C when the TLA content is 25%, and the high temperature performance grade is 70,
which is more than that of TLA modified asphalt with the TLA content lower than 15%.
What’s more, when the TLA content is 25% the failure temperature is up to 77.08°C, and the
high temperature performance grade is 76. In other words, the high temperature performance
grade of modified asphalt with 25% and 35% are higher than that of non-modified asphalt.
484
Table 4. Absolute slope values for the regression lines in Figure 5.
Absolute slope value for lgG’ Absolute slope value for lgG”
TLA content/% regression equation regression equation
0 59.307 43.307
5 57.234 43.095
15 56.731 42.893
25 56.214 42.517
35 56.037 42.045
Figure 6. SEM image of phase morphology of TLA modified asphalt with different TLA content.
temperature, respectively. And then the statistical linear regression equations for these rela-
tionships are obtained by linear fitting. Furthermore, the slopes and absolute values of each
regression line were calculated and are listed in Table 4.
As is depicted in Table 4, the sensitivity of the storage modulus and the loss modulus of
the TLA modified asphalt decrease gradually with the increase of the foaming water content.
In other words, the temperature sensitivity of the TLA modified asphalt increases as the
foaming water content is increasing.
3.5 Morphology
For characterization of modifed asphalt morphology, the Scanning Electron Microscope was
used to the microstructure of TLA modified asphalt. And the state of TLA dispersion in the
base asphalt can be determined. As depicted in Figure 6, the images show a clear change in
morphology of the TLA modified asphalt with different TLA content.
As shown, the TLA modifier could be dispersed in a continuous asphalt phase and the
particle size of TLA is small and uniform when the TLA content is 5% and 15%. However,
when the TLA content is up to 25%, the asphaltene and ash cluster together to some extent
485
Figure 7. DSC drawing of LTA modified asphalt with different TLA content.
TLA content/% 0 5 15 25 35
Tg/°C −23.66 −19.71 −17.49 −16.87 −16.29
for the limited maltha content which is unfavorable for absorbing asphaltene and ash. What’s
worse, the morphology change from a continuous asphalt phase with dispersed asphaltene
phase to two twisted continuous phases. And the two interlocked phases form a network
structure which could decrease the properties of the asphalt.
486
Figure 8. FTIR spectra of TLA modified asphalt with different TLA content.
4 CONCLUSIONS
In this study, the high temperature performance including softening point, viscosity-tempera-
ture property and failure temperature, temperature sensitivity, morphology of non-modified
and TLA modified asphalt and change of thermal stability and chemical reaction before and
after adding TLA were characterized by conventionla properties, rheology, SEM, DSC and
FTIR spectroscopy. The influence of TLA content on the above mentioned properties was
systematically investigated. The results show that the TLA content plays an important role in
the high temperature and microstructure of TLA modified asphalt.
The addition of TLA to base asphalt leads to a significant increase in softening point,
viscosity and failure temperature, and improve the temperature sensitivity. With the increase
of TLA content, the softening point, viscosity and failure temperature increase. Adding TLA
could improve the high temperature performance and temperature sensitivity of TLA modi-
fied asphalt. By considering above properties, the optimum content of TLA is 25%–35%.
There is a significant variation in morphology of TLA modified asphalt. With increase of
TLA content, phase morphology is converted from a continuous asphalt phase with dispersed
TLA phase to two twisted continuous phases. And the Tg of modified asphalt increases with
the increase of TLA content, which is unfavorable for the low temperature performance of
asphalt. Meanwhile, the chemical reaction was not happened after adding TLA.
ACKNOWLEDGMENTS
This work was supported by the National Natural Science Foundation of China (51278173), the
Natural Science Foundation of Jiangsu Province for Distingished Young Scholar (BK20150038),
and“the Fundamental Research Funds for the Central Universities”(2015B21614).
REFERENCES
AASHTO T 316-04. Standard Method of Test for Viscosity Determination of Asphalt Binder Using
Rotational Viscometer.
487
Airey, G.D. 2004. Styrene butadiene styrene polymer modification of road bitumens. Journal of Materi-
als Science 39:951–959.
Becker, I.M. Muller, A.J. & Rodriguez, Y. 2003. Use of rheological compatibility criteria to study SBS
modified asphalts. Journal of Applied Polymer Science 90:1772–82.
Bowers, B.F. Huang, B.S. Shu, X. Miller, B.C. 2014. Investigation of Reclaimed Asphalt Pavement
blending efficiency through GPC and FTIR. Construction and Building Materials 50: 517–523.
Liao, M.C. & Chen, J.S. 2014. Gordon Airey, Shi-Jing Wang. Rheological behavior of bitumen mixed
with Trinidad lake asphalt. Construction and Building Materials 66: 361–367.
Navarro, F.J. Martinezboza, F.J. Partal, P. & Gallegos, C. 2001. Effect of processing variables on the
linear viscoelastic properties of SBS-oil blends. Polymer Engineering and Science 2001; 41(12):
2216–25.
Sengoz, B. & Isikyakar, G. 2007. Evaluation of the properties and microstructure of SBS and EVA poly-
mer modified bitumen. Construction and Building Materials 22:1897–905.
Shingo, K. Shigeru, T. Zhang, X.M. Dong, D.W. & Inagaki, N. 2001. Compatibilizer role of styrene-
butadiene-styrene tri-block copolymer in asphalt. Polymer Journal 33(3):209–213.
Tong, L. 2008. The compatibility of TLA modified asphalt. Changsha: Changsha University of Science
& technology.
Wang, L.Z. & Wei, J.M.2008. Analysis and discussion on the index of temperature susceptibility for
asphalt binders. Science Technology and Engineering 18(121):5793–8.
Wang, N. 2012. The study on the mechanism and application of TLA modified asphalt. Chongqing:
Chongqing Jiaotong University.
Yu, X. Sun, H.W. & Luo, Y.L. 2013. Research on the evaluation index of temperature sensitivity of
CRMA. Journal of Building Materials 16(2):265–70.
488
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
G.A. Leegwater
Delft University of Technology, Delft, The Netherlands
TNO, Delft, The Netherlands
A. Scarpas
Delft University of Technology, Delft, The Netherlands
S.M.J.G. Erkens
Rijkswaterstaat, Utrecht, The Netherlands
Delft University of Technology, Delft, The Netherlands
ABSTRACT: Asphalt concrete has the advantageous ability to heal autonomously, however
the mechanisms behind this are not fully understood. To increase insight in the healing mecha-
nism, the healing model used in polymer science is adopted. It interprets healing as the sum of
wetting and intrinsic healing. The presented work introduces a new test set-up, which is designed
to investigate the relative contribution of wetting and intrinsic healing by measuring the strength
gain when two pieces of binder are brought into contact. Results obtained show that for a soft,
pure binder, wetting is the dominant process. This can be concluded from the fact that at least
50% of the observed healing can be attributed to wetting. Wetting is highly dependent on both
the load level and the duration of load application. Consequently, it is shown that the level of
healing observed in asphalt concrete is very dependent on the boundary conditions.
1 INTRODUCTION
Asphalt concrete has the advantageous ability to heal autonomously. During rest periods
damage present in the material is restored to a certain extent. This healing can be observed in
the regain of stiffness and/or strength of the material after rest periods.
The fact that damage in asphalt concrete can be restored during rest periods has first been
reported by Bazin and Saunier (Bazin and Saunier 1967). From this moment on the phenom-
enon is called healing and it is studied ever since e.g. (Bonnaure, Huibers, and Boonders 1982;
Kim, Little, and Benson 1990; Bhasin, Palvadi, and Little 2011). It has been demonstrated
that the level of healing increases with longer resting periods and higher healing tempera-
tures. Next to this, it has been shown that healing requires a minimum level of force perpen-
dicular to the damage and small damages show more healing compared to large macroscopic
damages (Qiu 2012). However, research shows that the specific test method used has a strong
influence of the observed healing. For instance, force controlled tests show more healing
compared to displacement controlled tests (Francken 1998). Thorough attempts have been
undertaken to explain the observed difference in healing between the different test methods,
however no validated explanation has been found up until now. From this lack of insight it
can be concluded that the mechanisms behind healing of asphalt are not fully understood,
which impedes focused exploitation of this material property. The research presented in this
paper attempts to unravel healing behavior in more detail.
There is a variety of approaches that can be chosen when studying healing. Healing is the
inverse of damage, consequently the damage that is taken as the starting point of a heal-
ing test is an important factor when designing a healing test. The current approach in most
pavement design guidelines is such that healing performance is an integral part of the fatigue
489
performance. Therefore a large part of the reported healing research is executed on speci-
mens loaded in fatigue, while introducing rest periods in different ways (Qiu 2012; Van den
Bergh 2012). A complicating factor of this approach is that damage development during
fatigue tests on asphalt is also not fully understood (Mangiafico et al. 2015). Consequently,
if fatigue damage is taken as a starting point for healing tests, uncertainty is introduced by
the unknown damage level. So, although measuring healing using fatigue tests has practical
relevance, the undefined damage level impedes fundamental understanding of the healing
processes.
In this study an attempt is made to measure healing using a very well defined level of
damage, more specifically the damage is a physical discontinuity of known shape and size. A
test set-up is designed to measure the strength gain when two separate pieces of bitumen are
brought into contact under controlled conditions. Bommavaram et al. have already tried this
approach in the past (Bommavaram, Bhasin, and Little 2009). In their test set-up pieces of
bitumen where brought together in the DSR and healing was measured by testing the gain
of shear stiffness versus time. This research has led to valuable insights, however real insight
into the mechanisms was not obtained. Complicating factors with their test set-up were the
complicated stress state and the lack of information on strength gain.
When interpretation the results obtained with the new designed test method, the model for
healing of polymers proposed by (Wool and O’ Connor 1981) is used to gain more insight
in the healing mechanisms. In this model two processes are discerned, first surfaces have to
come into contact, next the areas in contact have a certain capacity to transfer loads. The
model has already been adopted by asphalt researchers in the past to try and explain the
mechanisms behind healing (Kim, Little, and Benson 1990; Little and Bhasin 2007).
In the model, the first part of the process, where two surfaces come in contact, is referred
to as wetting. Real contact between surfaces develops at a very small scale, at this level all
material surfaces show a certain level of roughness. As two surfaces approach there will be an
increasing number of points between them, the surface area in contact at each of these con-
tact points will grow over time, as shown in Figure 1. As a result, the development of contact
area between two surfaces is a three dimensional process. Influencing factors of wetting are;
the external load that brings the surfaces together; the geometry (roughness of the surface);
the viscoelastic properties of the material and the adhesion of the material (Johnson, Kend-
all, and Roberts 1971; Kendall 2001). Of these four factors, for bitumen three are influenced
by temperature; the geometry, the viscoelastic properties and the adhesion.
The second process in recognized by the proposed healing model, is the ability of surfaces
in contact to carry loads. Bitumen is sticky and will adhere to other material or itself (Kendall
2001). This initial adhesion, that is present from the moment the surfaces touch, is named the
instantaneous intrinsic healing. Next to this there is also a time dependent part. In polymer
science it has been shown that molecules can bridge over crack interfaces due to molecular
motion (de Gennes 1971), this process results in the disappearance to the interface over time.
This process is illustrated in Figure 2. Although bitumen molecules are very diverse and
distinctly differ from polymers, it is assumed that some kind of reptation or self-diffusion in
bitumen will also bridge crack interfaces over time (Kim, Little, and Benson 1990). Influenc-
ing factors of intrinsic healing are chemistry of the bitumen molecules (polarity, molecular
Figure 1 a. Side view of immerging contact points between two surfaces approaching. b. Top view of
increasing contact area; area in contact is shown in grey, nucleating from the center point and growing
radially indicated by the arrow (r).
490
Figure 2. Two pieces of bitumen will adhere to each other as a consequence of the sticky nature of
bitumen. Over time, due to molecular motion bitumen molecules will “cross” the interface of two sur-
faces in contact, resulting in a gradual disappearance of the interface.
2 TEST METHOD
491
Figure 3. Schematic impression of steps in the test set-up.
Figure 4. Silicon mold, with stainless steel rings in place during preparation of bitumen specimens.
2.2 Materials
All samples tested were made of a pengrade 70/100 bitumen from Kuwait Petroleum (Q8).
Bitumen with a pengrade of 70/100 is relatively soft. This type of bitumen is regularly used
for road construction in The Netherlands and consequently also for research purposes. Dutch
design guidelines specify that asphalt concrete made with bitumen of pengrade 70/100 has
good healing properties.
492
2.3 Sample preparation
The bitumen and the mold with the metal rings in place are both heated for one hour at
165°C, after which the bitumen is poured in the mold. The mold containing the bitumen sam-
ples is left to cool for 30 minutes at room temperature and subsequently for 1 hour at −24°C.
In the mold the bottom side of the sample is facing up. After cooling the excess bitumen is
removed, using a heated knife, making sure that the bitumen exactly fits the ring. The sample
is left to cool again at −24°C until demolding. Demolding is done under controlled tempera-
ture conditions (14°C +/−1°C). After demolding the stainless steel rings are cleaned using
methylene chloride to improve handling during healing and testing. Tests have shown that
the bitumen needs some time to stabilize after production in order to obtain more repeatable
results, therefore the samples are stored in the temperature controlled room (14°C +/−1°C)
for 48 hours, prior to assembly.
Figure 6. a. A bitumen sample in a storage container, with the silicon paper fixing the contact area
showing on top b. An assembled test specimen with a small weight placed on top.
493
Figure 7. Specimen that has been assembled in the storage room, mounted for testing in the DSR.
The assembled specimen is taken out of the storage container and fixed in the DSR using
clamps (Figure 7). The temperature is kept for 10 minutes at 10°C to allow for temperature
equilibrium in the sample. The test is displacement controlled, applying a strain rate of
0.5%/s.
In the second assembly method, the DSR is used to bring the two pieces of bitumen
together. This means the healing occurs inside the temperature chamber of the DSR. First
the samples are taken out of the freezer (−24°C) and stored at −2 °C +/−2°C. One sample is
mounted in the bottom clamp of the DSR and one in the top clamp. The clamps with the
specimens are placed in the climate chamber of the DSR, and the temperature is brought to
14°C +/−0.2°C. This temperature is maintained for 10 minutes to gain temperature equilib-
rium in the specimen. This temperature equilibrium time can be short as the samples are very
small. Specimens are assembled by moving the two clamps of the DSR towards each other
with a speed of 0.01 mm/s, until a certain force level is realized. After reaching the desired
contact level, the displacement is kept constant. Due to relaxation the force reduces to 0 N
within a time span of seconds, the exact relaxation time depends on the magnitude of the
force.
In order to break the specimens, the temperature is changed to 10°C +/−0.2°C. It takes
16 minutes for the temperature to reach 10°C, including 10 minutes equilibrium time. Con-
sequently the specimens are subjected to a healing time of 16 minutes. The tensile test is dis-
placement controlled applying a strain rate of 0.5%/s.
494
3 RESULTS
The results of the tests are presented in this paragraph, they will be discussed in the next
paragraph.
In Figure 8 the results are presented from tests done on specimens that have been assem-
bled in the temperature room and have been allowed to heal for various periods of time. This
graph also presents the strength of the reference samples at the age of one week. The tests
on the assembled samples have been run three times and variation in the results is expressed
in the graph by error bars showing the standard deviation. Two of three samples tested at
168 h fell and broke, consequently this value is based on a single measurement. It can be seen
that up until a healing time of 24 hours the tensile strength increases significantly, after this
the measured tensile strength does not increase. The observed tensile strength after a healing
period of 6 minutes (0.1 hour) is around 50% of the strength observed after 24 hours and
168 hours. It can also be noted that the strength of the reference sample is less compared to
the healed specimen.
Figure 9 presents the healing strength observed for the samples assembled in the DSR,
while varying assembly force. It can be seen in the graph that, as the assembly force increases,
the tensile strength of the assembled specimen increases. The tests have been run twice and
variation in the results is expressed in the graph. The variation in the test results of assem-
bly at 0.05 N and 0.5 N is so limited, that the error bars have become invisible. The tensile
strength at 1 N does show some variation, it is unclear if this is caused by an artefact during
the measurement or if this variation is inherent to the measurement.
In order to assess the test results from an energy perspective, the amount of work required
to assemble and break the specimens is shown in Figure 10. Work is plotted on a logarithmic
scale versus the assembly force. Again the observed variation in the results is limited in case
when the assembly load is 0.05 N and 0.5 N, however for assembly at 1 N the variation is
larger.
4 DISCUSSION OF RESULTS
495
Figure 9. The development of the tensile strength versus the assembly force.
Figure 10. The amount of work required for assembly and breaking, versus the assembly force.
as it is in the same order of magnitude as the reference strength. Intuitively 24 hours seems
fast for full healing, however one should realize that a 70/100 bitumen is a soft binder and the
pure binder is expected to heal much faster compared to asphalt concrete.
The slight difference between the tensile strength at the plateau value compared to the
strength, is most likely due to the difference in sample geometry. The reference strength can
be used to calculate the healing ratio, giving the strength of a healed specimen relative to the
original strength in a factor between 0 and 1. In this case the reference strength is lower than
the ultimate strength found in the tests, as a healing ration of more than 1 is not possible a
conversion factor is needed. If the strength after 24 hours is assumed to be full healing, this
strength should correspond to a healing factor of 1, consequently a conversion factor of 0.9
is needed to calculate the healing factor from the measured tensile strength.
496
and a part that develops over time. After only 6 minutes of healing, the contribution of time
dependent processes governed by molecular motion are expected to be small.
Consequently, the healing observed after 6 minutes can largely be attributed to a combina-
tion of wetting and adhesion. Current results do not provide any quantitative information
on the level of wetting after 6 minutes, however full wetting is the natural upper limit. From
these two observations, it can be concluded that for the pure binder studied in this research
at least 50% of the macroscopic observed healing can be attributed to wetting. Indicating
that the realization of contact area is more important for healing, compared to the physical
and chemical processes that take place once contact is established. To put this in terms of the
presented healing model, for pure soft binders wetting is the dominant process.
4.4 Comparison of energy needed for assembly to energy needed for tensile failure
In order to evaluate the amount of energy needed in making and breaking contact, the amount
of work required for assembly and failure is plotted versus the assembly force (Figure 10). The
results show that the work required to break the specimens is ten times larger compared to the
work done during assembly. The fact that more energy is needed to break the specimens indi-
cates that energy is gained during assembly. This observation matches the principles of surface
energy, stating that energy is required to create surfaces and is gained when surfaces disappear.
In this experiment, the energy put into the assembly is most likely transferred into elastic
and permanent deformation of the surfaces. These deformations are needed to bring the
micro textured surfaces in real contact. The level of intrinsic healing is assumed to be con-
stant in the experiment as a consequence of the conditions (fixed temperature and healing
time), this implies that the difference in work required to separate surface is related to the level
of wetting. The fact that the energy needed to break the well wetted specimens is 100 times
higher, shows the importance of wetting for regaining resistance against fatigue loading.
The ratio of required work of 10 between assembly and break at 0.05 and 0.5 N could be
interpreted as follows: the amount of energy that is invested into the creation of contact is
proportional to the amount of wetting that is observed. The fact that this relationship is lost
a higher load level might be explained by other sources of energy loss. It is likely that, at the
497
highest load level, next to permanent deformation of the surface, energy is also lost in per-
manent deformation of the whole specimen.
When using the presented healing model, wetting is the dominant process for the pure, soft
binder studied in this research. This can be concluded from the fact that at very short healing
times already 50% healing is observed.
The presented data also shows that the process of wetting for bitumen is highly dependent
on the load perpendicular to the damage. Both the level and the duration of the load appli-
cation have a large influence on the observed macroscopic healing, showing that a higher
load level present for a longer time results in more healing. This implies that the boundary
conditions during the rest period, are very important for the healing behavior of asphalt.
This conclusion seems trivial, however much work when studying healing has been focused
on varying time and temperature, while no attention was paid to the boundary conditions
during healing. Only incidentally boundary conditions were varied in a very simplistic man-
ner (vertical or horizontal storage). For future research in healing it is recommended to also
address boundary conditions explicitly during healing.
The results presented in this paper are part of a bigger investigation into healing of bitu-
men and asphalt mortar. The processes that influence wetting are the next focus of the
research program. Here the influence of bitumen properties like stiffness and phase angle
will be studied and attempts will be undertaken to vary the surface texture of bitumen sam-
ples brought into contact.
REFERENCES
Bazin, P, and J Saunier. 1967. “Deformability, fatigue and healing properties of asphalt mixes.” In Intl
Conf Struct Design Asphalt Pvmts.
Bhasin, Amit, Sundeep Palvadi, and Dallas N Little. 2011. ‘Influence of aging and temperature on
intrinsic healing of asphalt binders’, Transportation Research Record: Journal of the Transportation
Research Board, 2207: 70–78.
Bommavaram, Ramamohan Reddy, Amit Bhasin, and Dallas N Little. 2009. ‘Determining Intrin-
sic Healing Properties of Asphalt Binders: role of the dynamic shear rheometer ‘, Transportation
Research Record: Journal of the Transportation Research Board, 2126: 47–54.
Bonnaure, FP, AHJJ Huibers, and A Boonders. 1982. ‘A laboratory investigation of the influence of
rest periods on the fatigue characteristics of bituminous mixes’, Journal of the Association of Asphalt
Paving Technologists, 51: 104–28.
de Gennes, Pierre-Giles. 1971. ‘Reptation of a polymer chain in the presence of fixed obstacles’, The
journal of chemical physics, 55: 572.
Francken, Louis. (ed.)^(eds.). 1998. RILEM Report 17: Bitumious Binders and Mixes (London).
Johnson, KL, K Kendall, and AD Roberts. 1971. “Surface energy and the contact of elastic solids.”
In Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering Sciences,
301–13. The Royal Society.
Kendall, Kevin. 2001. Molecular adhesion and its applications: the sticky universe (Springer).
Kim, Y Richard, Dallas N Little, and Fred C Benson. 1990. ‘Chemical and mechenical evaluation on heal-
ing mechanism of asphalt concrete’, Journal of the Association of Asphalt Paving Technologists, 59.
Little, Dallas N, and Amit Bhasin. 2007. ‘Exploring Mechanism of Healing in Asphalt Mixtures and
Quantifying its Impact’, Springer Series in Materials Science, 100: 205.
Mangiafico, S, C Sauzéat, H Di Benedetto, S Pouget, F Olard, and L Planque. 2015. ‘Quantification of
biasing effects during fatigue tests on asphalt mixes: non-linearity, self-heating and thixotropy’, Road
Materials and Pavement Design, 16: 73–99.
Qiu, Jian. 2012. ‘Self-Healing of Asphalt Mixtures ‘, Technical University Delft.
Van den Bergh, W. Van de Ven, MFC. 2012. ‘The Influence of Ageing on the Fatigue and Healing Prop-
erties of Bituminous Mortars’, Procedia-Social and Behavioral Sciences, 53: 256–65.
Wool, R.P., and K.M. O’ Connor. 1981. ‘A theory crack healing in polymers’, Journal of Applied Phys-
ics, 52: 5953–63.
498
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: In this paper crumb rubber modified asphalt (CRMA) is investigated, and the
appropriate crumb rubber powder content of CRMA is determined. The high temperature
performance, the low temperature performance and the fatigue resistance of CRMA were
compared with that of the base asphalt and SBS modified asphalt. Furthermore, the micro-
structure of CRMA is characterized by the scanning electron microscope (SEM), component
analysis, differential scanning calorimetry analysis (DSC) and infrared spectroscopy (IR).
The results show that the performance of CRMA is best at a crumb rubber powder content
of more than 18% by weight, powder particle size of 20, mixing temperature 180–200°C and
45 min of mixing time. The high and low temperature performance of CRMA is better than
that of the base asphalt and SBS modified asphalt. It is indicated that the phase morphology
with the crumb rubber powder being the disperse phase and the asphalt the continuous phase
was formed obviously. Moreover, complex chemical reactions did not happen in the system.
1 INTRODUCTION
Nowadays asphalt pavements are more and more widely used. However, the quality of ordinary
asphalt is still far from satisfying the needs of highway construction. Due to the environmental
temperature impact, the repeated wheel loads and the overlap of changes of stress-strain state,
the permanent deformation of asphalt mixtures accumulates day by day, which affects the pave-
ment smoothness and driving comfort (Adhikari B et al. 2000, Morris GR et al. 1976, J. Baek
et al. 2009). As a result, modified asphalt binders become more and more popular, and it has
become the main means to improve the performance of ordinary asphalt and asphalt mixtures.
Compared with SBS polymer modified asphalt mixture and SMA mixture, crumb rubber
modified asphalt (CRMA) mixture has a similar performance but lower costs. Producing
CRMA will cost $120-$150 per ton less than SBS modified asphalt (Feng Zhang et al. 2015).
Meanwhile, crumb rubber powder contains an admixture which has potential benefits on the
properties of the mixture binder.
Crumb rubber tires are very intractable polymer materials and difficult to degrade. If we
leave them alone or deal with improper measures, such as landfill and incineration, they will
waste land and arise environmental problems (Kezhen Yan et al. 2015). The serious envi-
ronmental impacts caused by a large number of waste tires has raised wide social concerns.
Therefore, the ideal treatment of discarded tires, which should transfer them into recycle
resources, has become a big issue (Al-Qadi et al. 2011).
The good news is that CRMA can deal with the problem of waste rubber tires, and it also
provides a new opportunity to improve the performance of asphalt pavements. Recently, a
number of scholars are committed to study the performance of CRMA pavements, but a sys-
tematic research on CRMA is still missing. In this study, the influence of the crumb rubber
content on the performance of CRMA is investigated. The high temperature performance,
the low temperature properties and the fatigue resistance of CRMA are compared with those
of a base asphalt and SBS modified asphalt. Furthermore, the scanning electron microscope
(SEM), component analysis, differential scanning calorimetry analysis (DSC) and infrared
spectroscopy (IR) are applied to further reveal the microstructure of CRMA.
499
2 MATERIAL PREPARATION AND RESEARCH METHODOLOGY
2.4 DSR
The MCR 302 Dynamic Shear Rheometer (Anton Paar, Austria) was used to obtain the
shear flow at very low frequency of the base asphalt, SBS modified asphalt, and CRMA,
under the conditions of temperature sweep mode at 60°C, plate-plate geometry, and sample
dimensions of 25 mm diameter and 2 mm thickness.
2.5 BBR
CRMA with the rubber powder content of 16%, 18%, 20%, 22% and 24% (% by weight),
respectively, was tested in the bending beam rheometer (BBR, American, CANNON) to
Testing Item Relative density Moisture content (%) Metal content (%) Tailover
500
evaluate the low temperature properties at the temperature of −12°C. The SBS modified
asphalt and the base asphalt were tested in the same way.
2.6 SEM
To better understand the effect of the different crumb rubber powder contents (0, 16%, 18%
and 20% by weight) in the CRMA, the morphology of the sample fracture was observed with
a scanning electron microscope instrument (SSX-500, Shimadzu). Before observation, the
fracture surfaces were sputter-coated with a thin layer of gold under 5 kV.
2.8 DSC
The crumb rubber powder (mesh number 20), the base asphalt, and CRMA with 18% (by
weight) crumb rubber powder were tested using the Differential Scanning Calorimeter (DSC)
at temperatures ranging from −60°C to 160°C.
501
Table 4. Viscosity of asphalt with different amounts of crumb rubber powder.
15% 1.51
20% 3.20
25% 4.08
30% 10.02
3.2 Viscosity
Viscosity was the main index used to indicate the construction workability of the asphalt
mixture. Figure 2 illustrates that at 80°C, 100°C, 120°C, 135°C, 160°C and 180°C the viscosity
values of crumb rubber modified asphalt are all obviously larger than those of SBS modified
asphalt and 70# base asphalt. Under isothermal conditions it is well known that the greater
viscosity results in a better high temperature stability, so it can be concluded that the high
temperature performance of CRMA is the best among the three types of asphalt binder.
3.3 G*/sinδ
The Dynamic Shear Rheometer (DSR) was used to evaluate the rheological properties at the
temperature and loading frequency as specified in SHRP (Strategic Highway Research Pro-
gram), which was conducted by the US Federal Highway Administration. Besides, the rut-
ting factor G*/sinδ was selected as the evaluation index of the high temperature performance
for asphalt (Ghazi G., Al-Khateeb et al. 2015). It is worth noting that G*/sinδ was chosen to
evaluate the high temperature performance instead of the penetration or viscosity, because
it is measured at a specific deformation rate (angular rate is 10 rad/s), which to some degree
takes into account the viscoelastic property of asphalt through the phase angle (Chiu CT
et al. 2007, Liu ST et al. 2009).
Figure 3 shows the log(G*/sinδ) versus T (temperature) relation for the three asphalt bind-
ers, which was established after RTFOT for the temperature range of 25°C to 80°C. Besides,
Table 5 gives the results of the fail temperature, which was calculated after RTFOT when G*/
sinδ is equal to 2.2 kPa.
It can be seen in Figure 3 that among the three types of asphalt binder the fail temperature
of CRMA is the highest, SBS modified asphalt is the middle, and 70# base asphalt is the
lowest. Consequently, the high temperature performance of CRMA is the best among the
three types of asphalt binder.
3.4 BBR
When the crumb rubber powder content increases, the stiffness modulus of CRMA gradu-
ally decreases, reflecting the improvement of the anti-cracking performance. Meanwhile, the
502
Figure 2. Viscosity of different asphalt binders.
SBS modified
70# base asphalt asphalt CRMA
m-value of the creep rate gradually increases, reflecting that the rate of stress relaxation and
anti-cracking performance also increase (X.H. Chen et al. 2011). When the crumb rubber
powder content is beyond 22%, the m-value of CRMA is more than that of SBS modified
asphalt and the stiffness modulus of CRMA is clearly less than that of 70# base asphalt.
When the crumb rubber powder content is beyond 16% the stiffness modulus of CRMA is
less than that of SBS modified asphalt, illustrating that the low temperature performance of
CRMA is superior to that of the 70# base asphalt and SBS modified asphalt (see Table 6).
3.5 G*⋅sinδ
The value of the fatigue factor G*⋅sinδ is used to evaluate the index of fatigue properties for
asphalt binders (Constr. F.G. Praticò et al. 2013). G*⋅sinδ is required to be not more than
5 MPa (Joe W. Button. Robert L. Lytton 2007, Phoenix et al. 2000). DSR tests were con-
ducted upon the asphalt binders aged through RTFO and PAV tests, and the curve of G*⋅sinδ
versus T (temperature) was obtained (Figure 4). Then, according to the Superpave specifica-
tion, the threshold temperature corresponding to a G*⋅sinδ value of 5.0 MPa was defined
503
as the ultimate fatigue temperature FTf (Superpave A-410). The FTf-values of the different
binders are calculated by the linear regression formula and the results are shown in Table 7.
It is known that fatigue failure mainly occurs at lower temperatures. A lower FTf-value
indicates that the fatigue resistance is better. Consequently, the fatigue resistance of CRMA
is better than that of SBS modified asphalt and 70# base asphalt.
3.6 SEM
In order to understand the appearance of CRMA with different polymer content, the base
asphalt and the CRMA with 16%, 18% and 20% crumb rubber powder were subjected to
the SEM analysis. As shown in Figure 5, when powder was blended with asphalt, a certain
grid system was formed. With the increase of the crumb rubber powder content a two-phase
system of rubber powder and asphalt was more obvious as the formation of the grid system
was more intensive.
The crumb rubber powder and asphalt were mixed under high temperature conditions. By
mixing the powder particles, swelling of the light fraction occurred after absorbing asphalt.
Lightweight components in the asphalt were absorbed by the rubber powder, the swelling vol-
ume increased and the desulfurization reaction took place. As a result, the powder becomes
loose and soft and the asphalt restores its plasticity and viscosity (Ming Lianga et al. 2015).
However, in the CRMA the crumb rubber powder particles connect the base asphalt mol-
ecules through the gel so as to form a continuous phase system and eventually even a network
structure (J. Borah et al. 2012, Norhidayah Abdul Hassan et al. 2015).
When the powder particle has the limits of the colloid, a regular table interface active
agent is difficult to play a role in the stability of the CRMA system. The proportion of the
solid phase and the liquid phase is the key factor for the stability of CRMA. As shown in
Figure 5, the proportion of solid phase and liquid phase becomes larger with the increase
of the amount of rubber particles. Anyhow, the powder dosage should not be excessive to
ensure that the CRMA tends to be stable.
504
Table 7. Fatigue limit temperature FTf.
3.8 DSC
The DSC test results of the rubber powder, base asphalt and CRMA are shown in Figure 6.
Comparison of the DSC diagrams for the CRMA and the base asphalt shows that the
thermal effect on the asphalt changed after the rubber powder is absorbed. First, the area
of the endothermic peak changed. Second, the peak position and shape of the endothermic
peak also changed (Colbert et al. 2012). This suggests that during the modification the waste
tire crumb rubber powder has changed the existing forms and quantities of the crystalliza-
tion as well as the phase transformation of the composition. At the same time, the melting
temperature of some components changed, which led to the change of the thermodynamic
properties of the asphalt (Luis Loria et al. 2008). What can be seen from the perspective of
surface chemistry, the CRMA becomes a multiphase system after the crumb rubber was
mixed to change the structure of the group including asphalt colloid. From Figure 6 it can
be concluded that the powder at low temperature (−40°C to −20°C) and at high temperature
(80°C to 120°C) exhibits an obvious endothermic peak. For CRMA there is no phase tran-
505
sition between −40°C and −20°C which indicates that some small molecules react with the
asphalt. Due to the large molecular substance produced by the powder in the endothermic
phenomenon, some fluctuation in the high temperature range occurs.
4 CONCLUSIONS
CRMA is mixed into the base asphalt by physical blending so as to improve the quality of
the asphalt. CRMA has great flexibility, low temperature sensitivity as well as good high and
sample Saturation point (%) Aromatic (%) Colloid (%) Asphaltene (%)
506
Figure 7. IR spectrum of crumb rubber powder.
low temperature performance. In addition, the CRMA has good stability and anti-fatigue
performance with lower penetration and higher viscosity and softening point. Besides, the
intensity of the mixture is significantly improved.
Considering the technical and economic performance of CRMA, 20 mesh powder obtained
by grinding truck tires is used. The production temperature of CRMA should not be lower
than 180°C, and the mixing time should be more than 45 min.
Study on the performance of CRMA using the penetration test and the SHRP system
shows that the high temperature performance, the low temperature performance and the
anti-aging properties of CRMA are all superior to those of SBS modified asphalt as well as
the base asphalt.
According to the results of the electron microscope test, proximate analysis, DSC and
infrared spectrum analysis of the CRMA, the crumb rubber powder has no chemical reac-
tion with the base asphalt. Improvement of the performance of CRMA is mainly due to the
changed rubber powder’s and asphalt’s physical state after mixing.
ACKNOWLEDGMENTS
This research was supported by the National Natural Science Foundation of China
(51278173), the Natural Science Foundation of Jiangsu Province for Distinguished Young
Scholars (BK20150038), and the Fundamental Research Funds for the Central Universities
(2015B21614).
507
REFERENCES
Adhikari B, De D, Maiti S. 2000. Reclamation and recycling of waste rubber. Progress in Polymer Sci-
ence 25(7):909∼948.
Binh Vaong. 1992. Influence of density and moisture content on dynamic stress-strain behavior of a low
plasticity crushed rock. Road & Transport Research, 1(2):109∼112.
Chiu CT, Lu LC. 2007. A laboratory study on stone matrix asphalt using ground tire rubber. Construc-
tion and Building Materials 21(5):1027–33
Colbert, Baron W. 2012. The performance and modification of recycled electronic waste plastics for the
improvement of asphalt pavement materials. Michigan Technological University.
Constr. F.G. Praticò, R. Vaiana. 2013. A study on volumetric versus surface properties of wearing
courses, Construction and Building Materials 38: 766–775.
Feng Zhang, Changbin Hu, 2015. The research for structural characteristics and modification mechanism
of crumb rubber compound modified asphalts. Construction and Building Materials 76:330–342.
Ghazi G. Al-Khateeb,Khaled Z. Ramadan. 2015. Investigation of the Effect of rubber on Rheological Prop-
erties of Asphalt Binders using Superpave DSR, KSCE. Journal of Civil Engineering 19(1):127–135.
Imad L. Al-Qadi, Jongcun Baek, 2011. Sand Mix Interlayer to Control Reflective Cracking in Hot-Mix
Asphalt Overlay. TRB.
J. Baek, I.L. Al-Qadi, and W.G. Buttlar, 2009. Numerical Modeling and Life Cycle Cost Analysis of
Hot-Mix Asphalt Overlay with Steel Netting Interlayer System for Airport Pavements, TRB.
J. Borah and C. Wang. 2012. Morphological and flame retardant behaviors of rubber-modified asphalt.
Key Engineering Materials, 501:532–537.
J.L. Feiteira Dias, L.G. Picado-Santos, S.D. Capitão. 2014. Mechanical performance of dry process fine
crumb rubber asphalt mixtures placed on the Portuguese road network. Construction and Building
Materials 73 (2014) 247–254.
Joe W. Button, Robert L. Lytton, 2007. Guidelines for Using Geosynthetics with Hot Mix Asphalt
Overlays to Reduce Reflective Cracking, TRB.
Kezhen Yan, Hongbin Xu, Lingyun You, 2015. Rheological properties of asphalts modified by waste tire
rubber and reclaimed low density polyethylene. Construction and Building Materials 83:143–149.
Ling TQ, Xiao C, Xia W, et al. 2010. Characteristics of asphalt–rubber mortar and aggregate gradation
optimization based on high temperature condition. J Civ Archit Environ Eng 32(5):47–52.
Liu ST, Cao WD, Fang JG, Shang SJ. 2009. Variance analysis and performance evaluation of different
crumb rubber modified (CRM) asphalt. Construction and Building Materials 23:2701–8.
Luis Loria, Peter E. Sebaaly, Elie Y. Hajj. 2008. Long-Term Performance of Reflective Cracking Mitiga-
tion Techniques in Nevada. TRB.
Ming Lianga, Xue Xin. 2015. Viscous properties, storage stability and their relationships with micro-
structure of tire scrap rubber modified asphalt. Construction and Building Materials 74:124–131.
Morris GR, McDonald CH. 1976. Asphalt-rubber Stress Absorbing Membranes:Field Performance
and State of the Art. Transportation Research Record (595):52∼58.
Norhidayah Abdul Hassan, Gordon D. Airey. 2016. Microstructural characterisation of dry mixed rub-
berised asphalt mixtures. Construction and Building Materials 82:173–183.
Phoenix, Arizona. 2000. Standard Specifications for Road and Bridge Construction. Arizona Depart-
ment of Transportation.
Research Report MD-10-SP608B4D, 2010. Maryland Department of Transportation State Highway
Administration, Evaluation of Laboratory Tests to Quantify Frictional Properties of Aggregates.
SHRP superior Performing asphalt Pavements (Superpave): The Product of SHRP asphalt research Pro-
gram. Strategic Highway Research Program A-410, National Research Council, Washington. DC,
1994.
Thomas Bennert, Ali Maher. 2008.Field and Laboratory Evaluation of a Reflective Crack Interlayer in
New Jersey. TRB.
Xiang Shu, Baoshan Huang. 2014. Recycling of waste tire rubber in asphalt and portland cement con-
crete: An overview, Construction and Building Materials 67:217–224.
Xiao C, Ling TQ, Qiu YJ. 2013. Optimization of technical measures for improving high temperature
performance of asphalt–rubber mixture. J Mod Transport 21(4):273–80.
X.H. Chen, D.W. Wang. 2011. Fractal and spectral analysis of aggregate surface profile in polishing
process S.A. Senior, C.A. Rogers, Laboratory tests for predictin, Wear 271:2746–2750.
508
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: This study investigated the thermal behaviour of Crumb Rubber Modifier
(CRM) extracted from hot Asphalt Rubber (AR) and warm AR binders through Differen-
tial Scanning Calorimeter (DSC) testing and Thermal Gravimetric Analysis (TGA). Three
methods were implemented to extract CRM from the modified binders, which resulted in
the following findings: 1) the interaction between CRM and base binder is a component
exchange process: i.e., natural rubber and styrene-butadiene rubber are released from CRM
and blended with asphalt molecules; 2) different Warm Mix Asphalt (WMA) additives have
different effects on the interaction between asphalt and rubber: wax based additives penetrate
into CRM while the liquid chemical additive rarely interacts with the rubber modifier; and
3) the incorporation of WMA additives, especially the wax based additives, promotes the
component exchange process between rubber and asphalt.
1 INTRODUCTION
Recycling waste vehicle tires into Asphalt Rubber (AR) pavement is a sustainable technology
which provides not only satisfying mechanical properties but also noise reduction function
(Azizian et al. 2003; Shen et al. 2009; Paje et al. 2010). Specifically, AR is defined as a blend of
asphalt cement, Crumb Rubber Modifier (CRM) and other additives, with a rubber content
of no less than 15% by weight of total blend (Bahia and Davies 1994). The field-blended AR
is prepared by high shear mixing CRM and base asphalt at high temperature (usually above
170°C). The performance of AR may be influenced by various factors, such as base asphalt
properties, rubber size, rubber content and mixing condition (Ghavibazoo & Abdelrahman
2013; Shu and Huang 2014). In general, AR has been reported to have superior rutting,
cracking and fatigue performance. However, the incorporation of CRM also leads poorer
workability and faces the criticisms of higher construction temperature, extra energy cost as
well as higher construction emission (Thodesen et al. 2009; Shu and Huang 2014).
Recently, various warm-mix asphalt (WMA) additives have been applied to improve the
workability of high-viscosity asphalt binders like AR (Xiao et al. 2012; Oliveira et al. 2012;
Zhao et al. 2012). The WMA additives can be generally classified into three categories: foaming
additives, organic additives and chemical additives. Foaming additives provide better work-
ability by causing volume expansion of asphalt binder, while organic additives and chemical
additives achieve this goal by decreasing the binder viscosity and acting as a surfactant at the
asphalt-aggregate interface, respectively (Rubio et al. 2102; Oliveira et al. 2012). Extensive
studies have been conducted to investigate the effects of various non-foaming WMA addi-
tives on AR binder’s rheological performance (Lesuer 2009; Xiao et al. 2012; Leng et al. 2013;
Yu et al. 2013; Jamshidi et al. 2013), but very few of them have looked into the interaction
mechanisms of different components within warm AR binder. To help fill this gap, this study
aims to investigate the effects of asphalt and different WMA additives on the composition
of crumb rubber through thermal analysis. To achieve this objective, three warm AR binders
were prepared with Sasobit (organic additive), Evother-DAT (chemical additive), and 56#
paraffin wax (organic additive), which are denoted as ARS, ARE, and ARW, respectively.
509
Then, the crumb rubber particles in these warm AR binders were extracted using organic
solvents, and characterized through two thermal analysis techniques: thermo-gravimetric
analysis (TGA) and differential scanning calorimeter (DSC) analysis. The decomposition,
fusion and crystallization temperatures were finally determined as the fingerprints of various
components in crumb rubber, which provided qualitative and quantitative information on
the interaction levels of CRM with base binder and WMA additives.
2 EXPERIMENTAL DESIGN
2.1 Materials
The AR binder was prepared by blending 18% of 40 mesh crumb rubber by the total weight
of AR with the penetration grade 60/70 base asphalt using a high shear mixer. The mixing
was conducted at 176 °C for one hour at a shear rate of 4000 rpm/min. Three WMA addi-
tives, including Evotherm-DAT, Sasobit, and 56# paraffin were added to AR immediately
afterwards, with the percentages s of 5%, 3% and 1.5%, respectively, by weight of AR, based
on the manufactures’ recommendations and preliminary tests. The properties of each WMA
additive are shown in Table 1, and the rheological properties of prepared binders are shown
in Table 2.
From Table 2, it can be observed that compared to base asphalt, AR provides better per-
formance on rutting, fatigue and low-temperature cracking resistance. All WMA additives
were effective in reducing the high-temperature viscosity of AR, but their effects on other
rheological properties varied. All WMA additives negatively affected the low-temperature
crack resistance of AR, while Sasobit showed positive effects on the rutting and fatigue per-
510
formance of AR, and Evotherm-DAT showed obvious negative effect on the high-tempera-
ture performance of AR.
511
Figure 1. DSC and TG curves of original CRM.
temperature. The TG curve exhibits two main regions of weight loss, i.e., 100–130 °C and
300–500 °C. The DSC curve shows three main endotherms and one exotherm. These ther-
mal features are caused by various components in the tested CRM, including NR, SR and
minor constituents of oil, plasticizer and water. The first mass loss and endothermic peak are
attributed to the evaporation of water, followed by the decomposition of oil, plasticizer and
other additives at lower temperature range (<300 °C). As the temperature was further raised
to above 300 °C, two peaks can be noticed at 329 °C and 410 °C in the DSC curve, which are
ascribed to the degradation of NR and SR, respectively.
Figure 4. DSC and TG curves of extracted CRM extracted from different warm AR binders.
513
thermic peak can be noticed for the rubber extracted from ARS and ARW, even though no
obvious mass loss can be observed. The exothermic peaks at this location are indicators of
material crystallization. It is know that the asphalt components such as aromatic and satu-
rated hydrocarbons do not crystallize at this temperature (Benbouzid & Hafsi 2007). Thus,
the crystallization phenomenon should be caused by n-alkanes (paraffin) in CRM mainly
from the wax-based WMA additives. In other words, n-alkanes from the wax based additives,
i.e., Sasobit and 56# paraffin wax, not only mix with asphalt components but also penetrate
in to particles. In addition, the heat capacity variation of 56#paraffin wax was found more
significant than that of the Sasobit according to the DSC results. The relatively poor heat
capacity of wax leads to poor temperature susceptibility of asphalt with high wax content.
As a result, the low-temperature cracking resistance of the ARW is the worst among the three
warm ARs.
The TG curves illustrate that crumb rubber from ARW has more oily components than
the others (nearly 5.4% of mass loss before 300 °C), which may be attributed to the penetra-
tion of some oily components from paraffin wax into the crumb rubber during the mixing
process. In comparison, the commercial wax based additive, Sasobit, had little effect on the
oily component ratio of CRM. This difference may be caused by the different preparation
techniques of these two additives. It has been reported that the carbon chain length of Saso-
bit is much longer than that of regular wax, which leads to a higher melting point and better
temperature stability (Tasdemir 2009; Jamshidi et al. 2013). The incorporation of the chemi-
cal additive, Evotherm-DAT, seems to have limit influence on the exotherm and endotherm
behavior of CRM. One possible reason is that some components had already evaporated
during the mixing process, and the residual surfactants were evenly dispersed among the
carbon chains of base asphalt without penetrating into crumb rubber. Since Evotherm-DAT
had little interaction with CRM, its effect on the rheological properties of AR should be
mainly attributed to its liquid nature and the interaction with asphalt molecules.
Table 3 summarizes the percentages of different components in the extracted CRMs from
AR, ARE, ARS, ARW and AR2 (with 70 minutes mixing time, same as warm ARs). The
component concentrations in AR and AR2 are very close to each other, indicating that the
rubber has almost reached the maximum dissolution level under the mixing condition adopted
in this study. In comparison, the NR and SR contents of CRM from warm ARs are lower
than those of CRM extracted from AR, possibly due to the mixing condition. The dissolution
level of CRM also varied obviously within different warm ARs. It seems that SR is easier to be
released from CRM in less viscous asphalt, especially when wax based additives are used.
4 CONCLUSIONS
In this study, the TGA and DSC tests were performed to characterize the pyrolysis and reac-
tion mechanisms of crumb rubber extracted from hot AR binder and warm AR binders with
various WMA additives. The following points summarize the main findings of this study:
• The interaction between CRM and asphalt affects the composition of the extracted crumb
rubber from AR. NR in CRM is easier to be released and mixed with asphalt molecules
than SR.
• Asphaltene does not penetrate into CRM during the mixing process of AR.
Table 3. Component concentrations of extracted CRM samples from AR and warm ARs.
514
• Wax based additives penetrate into CRM during the mixing process of warm AR, and the
conventional wax with shorter carbon chain length is easier to interact with rubber than
the commercial wax-type additive, Sasobit.
• Evotherm-DAT was found to have limited interaction with crumb rubber.
• The incorporation of WMA additives, especially wax based additives, promotes the inter-
action between CRM and base asphalt.
In summary, this study has shown the high potential of applying thermal analysis meth-
ods to investigate the interaction mechanism between asphalt and various modifiers. Future
study is recommended on developing the quantitative relationship models between the ther-
mal test results and the rheological properties of modified asphalt.
ACKNOWLEDGEMENT
The authors sincerely acknowledge the funding support from the Hong Kong Research
Grants Council (Project Number: 539113).
REFERENCES
Abdelrahman, M., & Carpenter, S. 1999. Mechanism of interaction of asphalt cement with crumb
rubber modifier. Transportation Research Record: Journal of the Transportation Research Board,
(1661), 106–113.
Azizian, M.F., Nelson, P.O., Thayumanavan, P., & Williamson, K.J. 2003. Environmental impact of
highway construction and repair materials on surface and ground waters: Case study: crumb rubber
asphalt concrete.Waste management, 23(8), 719–728.
Benbouzid, M., & Hafsi, S. 2008. Thermal and kinetic analyses of pure and oxidized bitumens. Fuel,
87(8), 1585–1590.
Bahia, H.U., & Davies, R. 1994. Effect of crumb rubber modifiers (CRM) on performance related prop-
erties of asphalt binders. Asphalt paving technology, 63, 414–434.
Chen, F., & Qian, J. 2003. Studies of the thermal degradation of waste rubber. Waste Management,
23(6), 43–47.
Fernández-Berridi, M.J., González, N., Mugica, A., & Bernicot, C. 2006. Pyrolysis-FTIR and TGA
techniques as tools in the characterization of blends of natural rubber and SBR. Thermochimica
Acta, 444(1), 65–70.
González, V., Martínez-Boza, F.J., Gallegos, C., Pérez-Lepe, A., & Páez, A. 2012. A study into the
processing of bitumen modified with tire crumb rubber and polymeric additives. Fuel Processing
Technology, 95, 137–143.
Jamshidi, A., Hamzah, M.O., & You, Z. 2013. Performance of warm mix asphalt containing Sasobit®:
State-of-the-art. Construction and Building Materials, Vol. 38, 2014, pp. 530–553.
Jeong, K.D., Lee, S.J., Amirkhanian, S.N., & Kim, K.W. 2010. Interaction effects of crumb rubber
modified asphalt binders. Construction and Building Materials, 24(5), 824–831.
Leite, L.F.M., & Soares, B.G. 1999. Interaction of asphalt with ground tire rubber. Petroleum science
and technology, 17(9–10), 1071–1088.
Leng, Z., Gamez, A., & Al-Qadi, I.L. 2013. Mechanical property characterization of warm-mix asphalt
prepared with chemical additives. Journal of Materials in Civil Engineering, 26(2). 304–311.
Loeber, L., Muller, G., Morel, J., & Sutton, O. 1998. Bitumen in colloid science: a chemical, structural
and rheological approach. Fuel, 77(13), 1443–1450.
Martínez, J.D., Puy, N., Murillo, R., García, T., Navarro, M.V., & Mastral, A.M. 2013. Waste tyre
pyrolysis–a review. Renewable and Sustainable Energy Reviews, 23, 179–213.
Oliveira, J.R., Silva, H.M., Abreu, L.P., & Fernandes, S.R. 2013. Use of a warm mix asphalt additive
to reduce the production temperatures and to improve the performance of asphalt rubber mixtures.
Journal of Cleaner Production, 15(41), pp. 15–22.
Paje, S.E., Bueno, M., Terán, F., Miró, R., Pérez-Jiménez, F., & Martínez, A.H. 2010. Acoustic field
evaluation of asphalt mixtures with crumb rubber. Applied Acoustics, 71(6), 578–582.
Peralta, J., Hilliou, L., Silva, H.M., Machado, A.V., & Williams, R.C. 2014. Rheological changes in the
bitumen caused by heating and interaction with rubber during asphalt–rubber production. Rheo-
logica Acta, 53(2), 143–157.
515
Ragab, M., & Abdelrahman, M. 2014. Effects of Interaction Conditions on Internal Network Structure
of Crumb Rubber-Modified Asphalts. Transportation Research Record: Journal of the Transporta-
tion Research Board, (2444), 130–141.
Rubio, M.C., Martínez, G., Baena, L., & Moreno, F. 2012. Warm mix asphalt: an overview. Journal of
Cleaner Production, 24. 76–84.
Shen, J., Amirkhanian, S., Xiao, F., & Tang, B. 2009. Influence of surface area and size of crumb rub-
ber on high temperature properties of crumb rubber modified binders. Construction and Building
Materials, 23(1) 304–310.
Ghavibazoo, A., & Abdelrahman, M. 2013. Composition analysis of crumb rubber during interaction
with asphalt and effect on properties of binder. International Journal of Pavement Engineering, Vol.
14, No.5, pp. 517–530.
Shu, X., & Huang, B. 2014. Recycling of waste tire rubber in asphalt and Portland cement concrete: An
overview. Construction and Building Materials, 67, 217–224.
Tan, Y., & Guo, M. 2014. Interfacial thickness and interaction between asphalt and mineral fillers.
Materials and structures, 47(4), 605–614.
Tasdemir, Y. 2009. High temperature of wax modified binders and asphalt mixtures. Construction and
Building Materials, 23(10), 3220–3224.
Thodesen, C., Xiao, F., & Amirkhanian, S.N. 2009. Modeling viscosity behavior of crumb rubber modi-
fied binders. Construction and Building Materials, 23(9), 3053–3062.
Xiao, F, Punith, V, and Amirkhanian, S.N. 2012. Effects of non-foaming WMA additives on asphalt
binders at high performance temperatures. Fuel, 94. 144–155.
Yu, X., Leng, Z., & Wei, T. 2013. Investigation of the rheological modification mechanism of warm-
mix additives on crumb-rubber-modified asphalt. Journal of Materials in Civil Engineering, Vol. 26,
No.2, 2013, pp. 312–319.
Yu, H., Leng, Z., Xiao, F., & Gao, Z. 2016. Rheological and chemical characteristics of rubberized
binders with non-foaming warm-Mix additives. No.16–5268, Transportation Research Board 95th
Annual Meeting (accepted).
Zhao, W., Xiao, F., Amirkhanian, S.N., & Putman, B.J. 2012. Characterization of rutting performance
of warm additive modified asphalt mixtures. Construction and Building Materials, 31. 265–272.
516
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Meng Guo
National Center for Materials Service Safety, University of Science and Technology Beijing, Beijing,
China
Harbin Institute of Technology, Harbin, China
The University of Texas at Austin, Austin, USA
Yiqiu Tan
Harbin Institute of Technology, Harbin, China
Amit Bhasin
The University of Texas at Austin, Austin, USA
Jianming Wei
National Institute of Clean-and-Low-Carbon Energy, Beijing, China
Xiong Yang
Harbin Institute of Technology, Harbin, China
Yue Hou
National Center for Materials Service Safety, University of Science and Technology Beijing,
Beijing, China
ABSTRACT: The interfacial interaction between asphalt and aggregate plays an important
role in performance of asphalt mixtures. Molecular dynamics simulation was used to build
models of four asphalt components and five typical mineral aggregate crystals. The interface
adsorption behavior was studied using these models. The effectiveness of these models was
analyzed by comparing the physical properties of the model to the measured values. Results
show that the model properties were close to the measured values. The simulated results
show that Van der Waals force is dominant in the work of adhesion compared to coulomb
electrostatic force. The rank order for the influence of mineral aggregate composition on the
work of adhesion is: CaO>MgO>SiO2>Al2O3>Fe2O3. The rank order for the influence of four
asphalt components on the work of adhesion is: aromatics>asphaltene>resin>saturate. The
work of adhesion between the four asphalt components and mineral aggregates improved
with an increase in temperature.
1 INTRODUCTION
An asphalt mixture is a composite of asphalt binder and aggregate and therefore has abun-
dant binder-aggregate interfaces. The properties of these interfaces directly affect the per-
formance of the asphalt mixture. The failure of interface is generally initiated by defects at
a nanometer to micrometer length scale. It is very difficult and time consuming to obtain
information of interface cracking at a nanoscale using experimental methods (Tan & Guo
2014). Compared to experimental methods, finite element simulation method can be much
more effective for a parametric study (Hou, Wang & Yue et al. 2014 ). However, finite element
simulations are applicable for continuums and cannot appropriate to reflect the behavior due
to molecular interactions or chemical bonds. These limitations make it difficult to evaluate
517
the mechanism of interface failure driven by nanoscale interactions between asphalt binder
and mineral aggregate. This challenge can be overcome by the use of simulations based on
molecular dynamics.
In the area of road engineering, currently molecular dynamics simulation is mainly applied
to study the asphalt binder. Jennings et al. proposed a representative asphalt molecular struc-
ture during the Strategic Highway Research Program (SHRP) (Jennings, Pribanic & Desando
1993). Pauli et al. further investigated the rationality of these molecular structures by using
Atomic Force Microscopy (AFM) (Pauli, Grimes & Huang et al. 2003). Zhang and Green-
field simplified the composition of asphalt binders by proposing the use of representative
asphalt molecular models to simulate the composition of asphalt binder: asphaltene, resin,
naphthenic aromatics and polar aromatics. They further used molecular dynamics simula-
tion with these model molecular structures to understand the behavior of asphalt binders at
a molecular length scale. In addition, they studied the influence of adding polystyrene chains
on the comprehensive performance of asphalt binders, such as the coefficient of thermal
expansion and bulk modulus. They also analyzed the temperature dependence of relaxation
time and diffusion coefficient of asphalt components (Zhang & Greenfield 2007 a, b, c).
Molecular dynamics simulation is also widely used in the study of composite modification
of asphalt material, deicing mechanism of asphalt pavement, photo-oxidative degradation of
asphalt binder, etc. (Ding, Tang & Zhang et al. 2015).
Previous studies have used molecular dynamics simulations to study the interfacial behav-
ior of polymer molecules. Clancy and Mattice used molecular dynamics to study the inter-
face and thin surface of polyolefin. They obtained the theoretical value of cohesive energy
density and solubility parameter. They also found that molecular relaxation can be used to
explain the interactions in an interfacial network (Clancy & Mattice 1999). Subsequently,
Deng et al. used molecular modeling and simulation to study the interfaces of two common
coupling agents and matrix polymers in composites. They analyzed the interface strength and
mechanical performance at the nanometer length scale (Deng, Tan & Tay 2004).
Although the research on interfacial interaction in asphalt mixtures is limited, some
studies have explored the adsorption and diffusion of heavy crude oil on soil surface.
Murgich et al. calculated the interaction energy of asphaltene and resin on the surface
of kaolin under vacuum. They found that van der Waals interaction (60%–70%) was
more important than Coulomb interaction (20%–30%). The hydrogen bond accounts for
less than 10% of the interaction (Murgich, Rodriguez & Izquierdo et al. 1998). Most
recent research studies use a single component to represent the molecular structure of an
asphalt binder. But this is different from the real molecular structure of an asphalt binder,
which consists of many complex components. In particular, the interaction between dif-
ferent components is likely to influence the interfacial behavior and must be considered.
Norinaga, Andrews and He et al. used molecular dynamic simulation to study the dif-
fusion coefficient and concentration distribution of asphalt binders. They found that
increasing the number of simulated molecules can improve the accuracy of the simulation
(Norinaga, Wargardalam & Takasugi et al. 2001; Andrews, Guerra & Mullins et al. 2006;
He, Li & Wu et al. 2013).
The objective of this research is to build a proper interfacial model between asphalt binder
and mineral aggregate and study the factors that influence interfacial adhesion.
2 MODEL CONSTRUCTION
Figure 1. Three-dimensional diagram of the model of four asphalt components (Black: C; Yellow: S;
Blue: N; White: H; Red: O). (Murgich, Rodriguez & Aray 1996; Verstraete, Schnongs & Dulot 2010;
Zhang & Greenfield 2007).
Figure 2. Asphalt binder layer (Black: asphaltene; Red: resin; Blue: Aromatic; Green: Saturate).
519
components of aggregate on adhesion between asphalt binder and aggregate by using X-ray
photoelectron spectroscopy, energy dispersive X-ray spectroscopy and peel tests. They found
that SiO2, Al2O3, CaO, MgO, Fe2O3 are the main minerals that affect adhesion (Horgniesa,
Darque-Ceretti & Fezai et al. 2011). Therefore, this research mainly focused on the crystal
structure of the five mineral components, and analyzed the interfacial behavor between them
and asphalt binder.
1. Construction of a crystal unit cell.
A crystal unit cell is the smallest unit in a crystal. Its size and shape can be characterized
by the edge length and crossing angle. The lattice constants used in this research are shown
in Table 1.
2. Interception and optimization of lattice planes.
In order to describe the structure of crystal unit cell clearly, a plane was first intercepted.
For example, the (001) lattice plane of SiO2 was intercepted and its thickness was set as 10 Å.
The energy minimization procedure was used to optimize the structure, followed by con-
structing the supercell. Finally, a vacuum layer with 10 Å thickness was added.
Lattice constant
Oxide type a b c Α β γ
520
3 VERIFICATION OF MODEL EFFECTIVENESS
Table 2. The comparision of simulation data and test data of asphalt in this research.
521
Therefore, it is believed that the simulated mineral crystal model can be used as a reasonable
representation of the real mineral aggregates for the purposes of a parametric analysis.
522
5 RESULTS AND DISCUSSION
In this research, the adhesion between asphalt binder and mineral aggregate was used to
evaluate the adsorption strength. The work of adhesion was calculated using Equation 1.
ΔE E A +EB E AB (1)
ΔE ΔE vdw + ΔE
ΔEcoulomb (2)
Where, ΔEvdw——the work of adhesion due to Van der Waals force interaction (kcal/mol);
ΔEEcoulomb——the work of adhesion due to Coulomb electrostatic force interaction (kcal/
mol);
5.1 The work of adhesion between the four asphalt components and SiO2 surface
The work of adhesion between the four asphalt components and SiO2 surface is shown in
Table 4.
It can be seen from Table 4 that the interaction due to the Van der Waals force was greater
than the interaction due to the Coulomb electrostatic force. When the temperature was at
25 °C, the work of adhesion between saturates and SiO2 surface was greatest, while the work
of adhesion between the other three components and SiO2 surface had negative values. This
indicates that the main component affecting adhesion between the asphalt binder and SiO2
at 25 °C condition are the saturates. With the temperature increased, only the adhesion work
between saturate and SiO2 decreased, while the adhesion work between the other three com-
ponents and SiO2 increased. It must be noted that in a real system the binder and aggregate
begin their interactions at a high temperature during mixing and production.
5.2 The work of adhesion between asphalt four components and Al2O3 surface
The work of adhesion between the four asphalt components and Al2O3 surface is given in
Table 5.
Table 4. The adhesion work between four asphalt components and SiO2 (Positive value means attrac-
tive force, and negative value means repulsive force in this paper).
523
Table 5. The adhesion work between four asphalt components and Al2O3.
It can be seen from Table 5 that the adhesion work between asphalt binder and Al2O3 was
10 times larger than SiO2. Similar to SiO2, the Van der Waals force played a more important
role than Coulomb electrostatic force. The aromatic had strongest work of adhesion with
Al2O3. The asphaltene and saturate had similar work of adhesion with Al2O3, which was
nearly half of the aromatic. The resin had the smallest work of adhesion with Al2O3. When
the temperature was within 65 °C, the work of adhesion increased dramatically with the
temperature increased. This indicates that increasing temperature is helpful for enhancing
the interaction between asphalt binder and Al2O3. However, when the temperature increased
from 65 °C to 165 °C, only the work of adhesion between the asphalt binder and the saturate
increased, whereas it did not change much for the other three components. This was likely
due to the fact that the saturate had the lowest molecular weight, which made it more sensi-
tive to the temperature.
5.3 The adhesion work between asphalt four components and CaO surface
The work of adhesion between the four asphalt components and CaO surface is presented
in Table 6.
It can be seen from Table 6 that compared to SiO2 and Al2O3, CaO had greater work of
adhesion with the four asphalt components, which was nearly 10 times of Al2O3. The work
of adhesion between the four asphalt components and CaO ranked as follows: aromatic
>asphaltene>resin>saturate. The van der Waals force played a more important role than Cou-
lomb electrostatic force in forming the adhesion force between asphalt binder and CaO. With
an increase in the temperature, the work of adhesion between the four asphalt components
and CaO increased significantly. Work of adhesion for the resin was the most sensitive to
temperature.
5.4 The work of adhesion between the four asphalt components and MgO surface
The work of adhesion between the four asphalt components and MgO surface is shown in
Table 7.
The results in Table 7 indicate that the work of adhesion between the four asphalt compo-
nents and MgO was a little bit lower than CaO, but still much higher than SiO2 and Al2O3.
Table 4 – Table 7 showed that CaO and MgO had greater work of adhesion with asphalt four
components, which were nearly 10 times of SiO2 and Al2O3. This partially agreed with Horgnies
et al.’s study. They used the peeling test and X-ray photoelectron spectroscopy method to inves-
tigate the interfacial adhesion performance between different minerals and asphalt binders.
They found that the granite aggregate consisting of SiO2 and Al2O3 had poorer adhesion per-
formance with asphalt binder. However, the dolomite consisting of MgO and CaO had better
524
Table 6. The adhesion work between four asphalt components and CaO.
Table 7. The adhesion work between four asphalt components and MgO.
Table 8. The adhesion work between four asphalt components and Fe2O3.
adhesion with asphalt binder, which made the most failure of peel tests happen in the asphalt
binder rather than the interface between asphalt binder and dolomite. This indicated that MgO
and CaO had stronger adhesion work with asphalt binder (Amorosi & Rampello 2007).
On the basis of the results in Table 7, it is found that the work of adhesion between MgO
and asphalt four components ranked as follows: aromatic>asphaltene>saturate>resin. The
525
work of adhesion increased with the temperature increased within 65 °C. The van der Waals
force played a main role in forming the work of adhesion.
5.5 The work of adhesion between the four asphalt components and Fe2O3 surface
The work of adhesion between the four asphalt components and Fe2O3 surface is shown in
Table 8.
It can be seen from Table 8 that compared to SiO2, Al2O3, CaO and MgO, the work of
adhesion between Fe2O3 and asphalt four components was very low. The interaction due to
Coulomb electrostatic force was slightly stronger than the interaction due to van der Waals
force. The higher the temperature, the greater was the work of adhesion.
6 CONCLUSIONS
In order to study the interfacial adhesion between asphalt binder and mineral aggregates,
molecular dynamic simulation was used to build an interface model. The effectiveness of the
model was verified and a dynamic simulation process was conducted. The adhesion proper-
ties between different asphalt components and minerals were obtained. The following are the
main findings from this study:
• CaO and MgO had greater work of adhesion with the four asphalt components (Saturates,
Aromatics, Resins and Asphaltenes), which was nearly 10 times of SiO2 and Al2O3. The
work of adhesion between Fe2O3 and the four asphalt components was smallest.
• The interaction due to van der Waals force played a more important role than Coulomb
electrostatic force in forming the adhesive bonds.
• The aromatic and asphaltene fractions had stronger work of adhesion with the mineral
surface compared to resins and saturates.
• The work of adhesion increased with an increase of temperature.
The limitation of molecular dynamics simulation used in this research is the small size of
the model and the simulated short time may only partially represent the real service situation.
The interaction between different aggregate minerals has to be considered in the next step.
REFERENCES
Allinger, N.L. 1976. Calculation of molecular structure and energy by force-field methods. Advances in
Physical Organic Chemistry, 13: 1–82.
Amorosi, A. & Rampello, S. 2007. An experimental investigation into the mechanical behaviour of a
structured stiff clay. Geotechnique, 57(2): 153–166.
Andrews, A.B., Guerra, R.E. & Mullins, O.C. et al. 2006. Diffusivity of asphaltene molecules by fluores-
cence correlation spectroscopy. Journal Physical Chemistry A, 110(26): 8093–8097.
Badre, S., Goncalves, C.C. & Norinaga, K. et al. 2006. Molecular size and weight of asphaltene and
asphaltene solubility fractions from coals, crude oils and bitumen. Fuel, 85(1): 1–11.
Brooks, B. & Karplus, M. 1983. Harmonic dynamics of proteins: normal mode and fluctuations in bovine
pancreatic trypsin inhibitor. Proceedings of National Academic Science, USA, 80: 6571–6575.
Brown, R.A. & Case, D.A. 2006. Second derivatives in generalized born theory. Journal of Computa-
tional Chemistry, 27 (14): 1662–1675.
Chilingar, G.V. & Yen, T.F. 1994. Asphaltenes and asphalts. Amsterdam. New York: Elsevier Science,
67–85.
Clancy, T.C. & Mattice, W.L. 1999. Computer simulation of polyolefin interfaces. Computational and
Theoreial Polymer Science, 9(3–4): 261–270.
Deng, M., Tan, V.B.C., Tay, T.E. 2004. Atomistic modeling: Interfacial diffusion and adhesion of poly-
carbonate and silanes. Polymer, 45(18): 6399–6407.
Ding, Y.J., Tang, B.M. & Zhang, Y.Z., et al. 2015. Molecular dynamics simulation to investigate the
influence of SBS on molecular agglomeration behavior of asphalt. Journal of Materials in Civil
Engineering, 27(9): 04014248.
526
Groenzin, H. & Mullins, O.C. 2000. Molecular size and structure of asphaltenes from various sources.
Energy & Fuels, 14(3): 677–684.
Groenzin, H., Mullins, O.C. 2001. Molecular size and structure of asphaltenes. Petroleum Science and
Technology, 19(1–2): 219–230.
He, L., Li, X. & Wu, G. et al. 2013. Distribution of saturates, aromatics, resins and asphaltenes fractions
in the bituminous layer of Athabasca oil sands. Energy & Fuels, 27: 4677–4683.
Horgniesa, M., Darque-Ceretti, E. & Fezai, H. et al. 2011. Influence of the interfacial composition on
the adhesion between aggregates and bitumen: Investigations by EDX, XPS and peel tests. Interna-
tional Journal of Adhesion & Adhesives, 31: 238–247.
Hortal, A.R., Hurtado, P. & Martinez-Haya, B. et al. 2007. Molecular-weight distributions of coal
and petroleum asphaltenes from laser desorption/ionization experiments. Energy & Fuels, 21(5):
2863–2868.
Hou, Y., Wang, L. & Yue, P. et al. 2014. Modeling mode I cracking failure in asphalt binder by ssing
nonconserved phase-field model. Journal of Materials in Civil Engineering, 26(4): 684–691.
Huang, E., Lin, J.F. & Xu, J. et al. 1999. Compression studies of gibbsite and its high-pressure poly-
morph. Physics and Chemistry of Minerals, 26(7): 576–583.
Jennings, P.W., Pribanic, J.A. & Desando, M.A. et al. 1993. Binder characterization and evaluation by
nuclear magnetic resonance spectroscopy. Rep. No. SHRP-A-355, Strategic Highway Research Pro-
gram, National Research Council, Washington DC: 3–12.
Mo, Y.Z., Xu, J.C. 2014. Studies on mechanical properties and optimization model of PI/SiO2 nanocom-
posite based on Materials Studio. Advanced Materials Research, 1049–1050: 54–57.
Mullins, O.C. 2007. Asphaltenes, heavy oils, and petroleomics. New York: Springer. Xxi: 669–670.
Mullins, O.C., Martínez-Haya, B. & Marshall, A.G. 2008. Contrasting perspective on asphaltene molec-
ular weight. This comment vs the overview of A.A. Herod, K.D. Bartle, and R. Kandiyoti. Energy
& Fuels, 22(3): 1765–1773.
Murgich, J., Rodriguez, J. & Aray, Y. 1996. Molecular recognition and molecular mechanics of micelles
of some model asphaltenes and resins. Energy & Fuels, 10: 68–76.
Murgich, J., Rodriguez, J. & Izquierdo, A. et al. 1998. Interatomic interactions in the adsorption of
asphaltenes and resins on kaolinite calculated by molecular dynamics. Energy & Fuels. 12: 339–343.
Norinaga, K., Wargardalam, V.J. & Takasugi, S. et al. 2001. Measurement of self-diffusion coefficient
of asphaltene in pyridine by pulsed field gradient spin-echo H-1 NMR. Energy & Fuels. 15(5):
1317–1318.
Ozkahraman, H.T., Isik, E.C. 2005. The effect of chemical and mineralogical composition of aggregates
on tensile adhesion strength of tiles. Construction and Building Materials, 19: 251–255.
Pauli, A.T., Grimes, W. & Huang S.C. et al. 2003. Surface energy studies of SHRP asphalts by AFM.
225th National Meeting of the American-Chemical-Society, New Orleans, Louisiana, 225: 422.
Schneider, M.H., Andrews, A.B. & Mitra-Kirtley, S. et al. 2007. Asphaltene molecular size by fluores-
cence correlation spectroscopy. Energy & Fuels, 21(5): 2875–2882.
Speight, J.G. 2007. The chemistry and technology of petroleum (4th edition). Boca Raton: CRC Press/
Taylor & Francis, 945–946.
Sultana, S. & Bhasin, A. 2014. Effect of chemical composition on rheology and mechanical properties
of asphalt binder. Construction and Building Materials. 72: 293–300.
Tan. Y. & Guo. M. 2014. Interfacial thickness and interaction between asphalt and mineral fillers. Mate-
rials and Structures, 47(4): 605–614.
Verstraete, J.J., Schnongs, P.h. & Dulot, H. et al. 2010. Molecular reconstruction of heavy petroleum
residue fractions. Chemical Engineering Science, 65: 304–312.
Wu, G.Z., He, L. & Chen, D.Y. 2013. Sorption and distribution of asphaltene, resin, aromatic and satu-
rate fractions of heavy crude oil on quartz surface: Molecular dynamic simulation. Chemosphere,
92: 1465–1471.
Zhang, L. & Greenfield, M.L. 2007a. Molecular orientation in model asphalts using molecular simula-
tion. Energy & Fuels, 21(2): 1102–1111.
Zhang, L. & Greenfield, M.L. 2007b. Relaxation time, diffusion, and viscosity analysis of model asphalt
systems using molecular simulation. Journal of Chemical Physics, 127(19): 1–13.
Zhang, L., Greenfield, M.L. 2007c. Analyzing properties of model asphalts using molecular simulation.
Energy & Fuels, 21: 1712–1716.
Ziegel, K.D. & Romanov, A. 1973. Modulus reinforcement in elastomer composites. I. inorganic fillers.
Journal of Applied Polymer Science, 17: 1119–1131.
527
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Current binder investigations focus on the behaviour of the material from
the perspective of failure and do not address the behaviour of the material in the temperature
regime between such extremities (20 to 60°C). In this regime, the response of the binder tran-
sitions from a viscoelastic solid to a viscoelastic fluid. In this investigation, the transition was
quantified using two scales of measurement, one at the microscopic scale and another at the
macroscopic scale. At the macroscopic scale, temperature sweep and frequency sweep experi-
ments were performed while at the microscopic scale, FTIR spectroscopy was performed. The
transition temperature was estimated for an unmodified and a plastomer modified bitumen
in unaged and short—term aged conditions. The transition regime captured from rheologi-
cal techniques was much more sensitive to the influence of modification process and aging
condition, while such sensitivity was not seen in the FTIR measurements.
1 INTRODUCTION
The two main distresses in a pavement are rutting and fatigue cracking. Rutting occurs dur-
ing the initial stages in the pavement life while fatigue cracking occurs during the later stages.
While the critical temperature at which rutting occurs is of the order of 60°C, for fatigue
cracking it is around 20°C and below. Though all the layers in a pavement contribute to
these distresses, the contribution of binder in the bituminous layer has been observed to be
significant (Kennedy et al., 1994). Most binder specifications focus on the performance of
bitumen at such failure temperature regimes. Little attention is paid to its performance in the
intermediate temperature regime of 20–60°C. In this temperature regime, one can observe
an interesting phenomena called as viscoelastic solid—fluid transition wherein the response
of the binder undergoes transition from a viscoelastic solid to a viscoelastic fluid. For vari-
ous applications in bitumen related to the distress prediction, it is required to identify its
viscoelastic solid–fluid transition. When bitumen exhibits a viscoelastic solid behaviour, the
stiffness of the binder can be considerably high making it susceptible to fatigue cracking.
Similarly, the accumulation of strains leading to rutting is rapid when the material exhibits
a viscoelastic fluid response. However, one cannot specify a sharp transition for bitumen
considering its multi-constituent nature.
The viscoelastic solid—fluid transition in bitumen has been observed to occur over a
broad temperature range and hence it is called as the transition regime rather than a tran-
sition temperature as defined for most polymers (Padmarekha and Krishnan, 2013). This
transition regime is a mushy region wherein the viscoelastic solid or viscoelastic fluid com-
ponent can exhibit predominance depending on the testing conditions. One can say that the
response in the transition regime is a mixture of the viscoelastic solid and viscoelastic fluid
response. Such broad transition regime observed for bitumen can be attributed to the multi-
ple constituents present in it. Each of these constituents can undergo transition at different
temperatures resulting in a broad transition regime.
529
The transition temperature for a material can be determined using different techniques.
Differential Scanning Calorimetry is a commonly used technique to identify the transitions
occurring due to the temperature variation. This technique can identify the transitions in the
material but cannot be used to identify the constituents sensitive to such transitions. Spectro-
scopic techniques such as FTIR spectroscopy have been used to estimate the transition tem-
perature for simple synthetic polymer systems focusing on specific functionalities. For such
polymers, the chemical composition and the microstructure are well known which enables
such precise measurements. For many other polymers comprising of multiple constituents,
rheological experiments performed in the time domain or frequency domain are commonly
used to evaluate the transition regime. Such techniques are useful when one has limited infor-
mation related to the microstructure of a material.
It should be noted that different techniques adopted to estimate the transition temperature
for polymers measure the same at different scales. The transitions captured by differential
scanning calorimetry correspond to specific compounds in the material. FTIR spectroscopy
focuses more at the ‘microscopic’ scale capturing transition in a material through the vibra-
tion of molecular bonds and the environment in which they are present in. Such techniques
can be used when one needs to monitor specific functionalities in a material. The rheological
tests can be considered to capture the overall response of the material and are hence associ-
ated with a ‘macroscopic’ scale. One may or may not obtain a common transition tempera-
ture from the different scales of measurement. For single constituent materials such as simple
synthetic polymer systems, all scales of measurement can be expected to elicit similar infor-
mation and hence one may expect close ranges of transition temperatures.
Identification of a suitable technique to assess the temperature dependent properties of
bitumen especially at the microscopic scale is a challenging task. The technique should allow
measurement of the properties of bitumen at different temperatures and also should be able
to identify the functionalities sensitive to variation in temperature. Considering some of the
suitable techniques in this regard such as DSC, GC-MS and spectroscopic techniques, FTIR
spectroscopy was chosen in this study to capture the changes in microstructure in the tem-
perature range of interest. The salient feature of this technique is that, it is sensitive to the
changes in chemical composition as well as other changes such as the variation in conforma-
tion and hydrogen bonding. However, application of techniques such as FTIR spectroscopy
has to be used with caution for materials such as bitumen. One can only choose specific
constituents and analyze its response in the transition regime rather than using it as an inde-
pendent technique to ascertain the transition regime.
The two important factors that influence the transitory response of bitumen are modi-
fication and aging. Modifiers are added to bitumen to improve its resistance against rut-
ting and fatigue cracking. The modifier can undergo different interaction mechanisms with
bitumen and this can be diffusion of fractions from one constituent to another termed as
‘physical’ interaction or a chemical interaction resulting in the formation of new compounds.
The nature of modifier and its interaction mechanism together play a significant role in
altering the transition regime. Aging is an inevitable phenomenon that causes a change in
the chemical composition of bitumen in the presence of environmental conditions. Though
aging manifests in several forms, here it is used to denote the aging caused due to oxidation.
The influence of aging on modified bitumen is even more interesting as parallel aging mecha-
nisms can occur in bitumen and the modifier. The influence of modifier on the transition
regime and the effect of such aging on modified bitumen have been least studied.
To summarize, the viscoelastic solid—fluid transition in bitumen occurs in the intermedi-
ate temperature regime of 20–60°C. Different techniques available to evaluate the transition
regime measure the response of the material at different scales. In this study, the transition
regime was evaluated for an unmodified bitumen and plastomer modified bitumen. Rheolog-
ical experiments were performed to evaluate the transition temperature based on its mechani-
cal response. To understand the corresponding changes in the underlying microstructure,
FTIR spectroscopy was performed and specific constituents were monitored at different tem-
peratures. The effect of aging on such transition regime was also assessed.
530
2 EXPERIMENTAL INVESTIGATION
2.1 Materials
Two types of materials were used for the rheological investigations: an unmodified binder of
VG30 grade (as per IS 73:2013) and this binder modified with a modifier which is a function-
alized polyethylene. This modified bitumen is classified as PMB(P)-60 as per IS15462: 2004.
Though the modifier contains some reactive functional groups, it is generally grouped as a
plastomer due to the predominant plastomeric nature of the polyolefin backbone. Hence it
will be referred to as plastomer and plastomer modified bitumen (PMB-P) henceforth in this
study. The plastomer modified bitumen was prepared by blending 3.5% of plastomer with
bitumen at 155°C for 1.5 hours. The performance of these materials was assessed under two
aging conditions, namely, unaged and short—term aged.
531
Figure 1. Torque from temperature sweep test.
of tan δ in the transition regime stated as Winter—Chambon criteria (Chambon and Winter,
1987) were used to estimate the transition temperature.
534
n-alkanes from 735 to 715 cm−1 was chosen as it is distinct with minimal overlapping from
other vibrations. Two parameters, the peak position and intensity of the methylene rocking
region were calculated from the FTIR spectra.
The peak position and intensity of the methylene rocking region has been used to estimate
the transition temperature for various polymers such as polymethylene, polyethylene and
polyacrylonitrile (Ogura 1971, Snyder et al., 1986, Hageman et al., 1989). These polymers
are semi-crystalline in nature and the n-alkanes constituting the crystalline fraction were
observed to be more sensitive to temperature in all these materials. These C-H vibrations
namely, stretching, bending and rocking have been commonly used to evaluate the tempera-
ture dependent properties of different polymers. A shift in peak position or the variations in
peak intensity have been associated with the change in conformation of the long chain mol-
ecules. When polyethylene moves from a solid state to a liquid state, a shift in peak position
of about 3–4 cm−1 has been observed. Similarly, the peak intensity is higher in the solid state
of the material by about 50% compared to that in a liquid state.
Figure 6 shows the methylene rocking region for VG30 and PMB-P in unaged condition. A
broad peak is observed in the region between 730 and 720 cm−1. Two peaks are expected in this
regime, one at 721 cm−1 and another at 730 cm−1. The peaks at 721 and 730 cm−1 are attributed
to the rocking of methylene chain with more than four carbon atoms in a row and less than
four carbon atoms in a row respectively. For convenience, (CH2)n > 4 and (CH2)n < 4 will be called
as long chain and short chain molecules henceforth. For the unmodified bitumen shown in
Figure 6(a), both these peaks have coalesced together into one peak observed at 724 cm−1. The
second derivative of this region is shown in Figure 7 for VG30 in unaged condition. The two
Figure 7. Second derivative of methylene rocking region for VG30 in unaged condition.
535
peaks at 721 and 730 cm−1 can be clearly seen in this figure and hence the peak at 724 cm−1 has
to be deconvoluted for further analysis. For PMB-P it is interesting to see that, the two peaks
observed in this region, are distinct. The modifier is expected to possess more short chain
molecules which contribute to the distinct peak at 730 cm−1 for PMB-P.
Figure 8(a) shows the variation in position of the methylene rocking region with tempera-
ture. The peak corresponding to long chain molecules on deconvolution is considered for
analysis. The position of this peak is observed to be considerably lesser for PMB-P compared
to VG30. This can be attributed to the chemical interaction of plastomer with the function-
alities attached to the methylene chains in bitumen causing a constraint on these vibrations in
the modified bitumen (Nivitha and Krishnan, 2015). The nature of interaction mechanism is
seen to be reflected in this parameter. From this figure, one can also see that the variation of
peak position has two slopes, one from 25–55°C and another from 55–75°C. This is clearly
seen in the unaged condition in Figure 8(a). Such point of change in slope has been associ-
ated with the change in conformation of the methylene chains (Hageman et al., 1989).
Figure 8(b) shows the intensity variation of the methylene rocking region. This parameter
is seen to be higher for the short—term aged condition compared to the unaged condition
for both the materials. This parameter is more sensitive to the aging condition of the mate-
rial rather than the nature of modifier and interaction mechanism. To estimate the point of
change in slope of the peak intensity, the normalized variation in intensity of the 721 cm−1
peak is calculated as shown in Figure 9. This parameter also shows a clear point of change in
slope from 45–55°C indicating a change in conformation of the material.
Figure 9. Normalized intensity variation with temperature for the rocking of long chain molecules.
536
Table 1. Transitory regime for unmodified and modified bitumen (°C).
4 CONCLUSION
Two parameters, torque and normal force were analyzed from a temperature sweep test to
estimate the transition temperature. These two parameters indicated identical transition
regime with a slight deviation in the values in the unaged condition. In a frequency sweep
test, the G’ and G” crossover was used to estimate the transitory regime while the Winter—
Chambon criteria was not satisfied for both the materials. In the rheology experiments, the
testing conditions play a significant role in influencing the transition regime. The effect of
shear history in combination with the choice of test parameter has been observed to shift the
transition regime by about 20°C. The FTIR spectra indicated a change in conformation in
the temperature range of 45–55°C irrespective of the nature of material and the aging condi-
tion. Such change in conformation of the n-alkanes can be considered as one of the reason
for viscoelastic solid—fluid transition in bitumen.
The effect of modifier was seen distinct only in some of the parameters considered for
analysis at both scales of measurement. The normal force variation was identical for these
two materials indicating that there are no significant temperature dependent changes in the
microstructure due to the effect of modification. This is also reflected in the peak intensity
variation of the rocking of long chain molecules. However, the peak position showed the
effect of modification through a shift in peak position observed for the PMB-P compared to
537
VG30. The G’ and G” crossover from the frequency sweep test was also able to clearly distin-
guish the effect of aging and modification through a shift in the transition regime.
ACKNOWLEDGEMENT
The authors thank Department of Science and Technology for funding this investigation.
The grant number is DST/TSG/STS/2011/46.
REFERENCES
Claudy, P., Letoffe, J.M., King, G.N. & Planche, J.P. 1992. Characterization of Asphalt Cements by
Thermomicroscopy and Differential Scanning Calorimetry: Correlation to Classic Physical Proper-
ties. Fuel Science and Technology International, 10(4–6): 735–765.
Giavarini, C., Mastrofini, D. and Scarsella, M. 2000. Macrostructure and Rheological Properties of
Chemically Modified Residues and Bitumens. Energy & Fuels, 14: 495–502.
Hagemann, H., Snyder, R.G., Peacock, A.J. & Mandelkern, L. 1989. Quantitative Infrared Methods for
the Measurement of Crystallinity and Its Temperature Dependence: Polyethylene. Macromolecules,
22: 3600–3606.
Harrison, I.R., Wang, G. & Hsu, T. 1992. SHRP-A/UFR-92–612 - A Differential Scanning Calorim-
etry Study of Asphalt Binders. Strategic Highway Research Program, National Research Council,
Washington DC.
IS73:2013. 2013. Indian Standard for Paving Bitumen - Fourth Revision. Bureau of Indian Standards,
New Delhi.
IS15462:2004. 2004. Indian Standard for Polymer and Rubber Modified Bitumen. Bureau of Indian
Standards, New Delhi.
Kennedy, T., G.A. Huber, E.T. Harrigan, R.J. Cominsky, C.S. Hughes, H.L. Von Quintus & J.S.
Moulthrop. 1994. SHRP-A-410: Superior Performing Asphalt Pavements (Superpave): The Product
of the SHRP Asphalt Research Program. Strategic Highway Research Program, National Research
Council, Washington DC.
Matejka, L. 1991. Rheology of Epoxy Networks Near the Gel Point. Polymer Bulletin, 26: 109–116.
Nivitha, M.R., Prasad, E. & Krishnan, J.M. 2015. Ageing in Modified Bitumen Using FTIR Spectros-
copy. International Journal of Pavement Engineering, Published Online, DOI:10.1080/10298436.20
15.1007230.
Ogura, K., Kawamura, S. & Sobue, H. 1971. Infrared Spectroscopic Approaches to the Polymer Transi-
tions. I. The Multiple Transitions in Poly( acrylonitrile) (PAN). Macromolecules, 4: 79–81.
Padmarekha, A. & Krishnan, J.M. 2013. Viscoelastic Transition of Unaged and Aged Asphalt. Journal
of Materials in Civil Engineering, 25(12): 1852–1863.
Snyder, R.G., Maroncelli, M., Strauss, H.L. & Hallmark, V.M. 1986. Temperature and Phase Behaviour
of Infrared Intensities: The Poly(methylene) Chain. Journal of Physical Chemistry, 90: 5623–5630.
538
Pavement Performance & LCCA
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: This paper analyzed the heat transfer process within the layers of asphalt
pavement materials, and provided better understanding on the temperature changes between
shadowed and unshadowed pavement surfaces. Natural environment and tree canopies have
been found to have influence on the heat energy absorption of pavement surfaces. The shades
of leaves are believed to have an impact on the solar radiation filtration. The purpose of this
research is to systematically determine the surface temperature changes of shadowed and
unshadowed pavements caused by solar radiation. Finite Element Method (FEM) and heat
transfer theory were performed to analyze solar radiation and temperature variations. Filed
observations and temperature collections were conducted to compare the analysis results.
The paper concludes that the temperature difference between shadowed and unshadowed
pavements varying from 4.47°C to 1.58°C in a cloudy day and varying from 24.06°C to
32.38°C in a sunny day. The vegetation effect on pavements by trees can significantly reduce
pavement temperatures.
541
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Zhijun Cheng
National University of Defense Technology, Changsha, China
Yaning Qiao
University of Nottingham, Nottingham, UK
1 INTRODUCTION
543
Frontline Systems’ Risk Solver Platform add-in for Microsoft Office Excel to discuss how
an annual highway maintenance budget needs to be allocated or determined to achieve the
District’s value proposition for various scenarios. Geoffrey[10] investigated an optimizing
decision model on the best combination of preventive maintenance treatments and timings in
the resurfacing life-cycle for a given highway pavement section. The key infrastructure man-
agement concepts of treatment-specific triggers, performance jump models and performance
regression models were incorporated in an optimization procedure to form a viable tool to
support scheduling decisions for highway preventive maintenance. In addition, in recent years
a large portion of relative studies in genetic algorithms have been performed to investigate
the pavement maintenance and rehabilitation trade-off problem at the network-level. Tack
and Chou [11] showed that a genetic algorithm based optimization proved to be beneficial in
determining multi-year maintenance programs and proved its advantages on dealing with the
problem with a substantial number of variables and constraints in terms of its flexibility and
scalability. Cheu’s subsequent study [12] further supported this conclusion.
However, the efficient management of these treatments is a complex work, which varies
with the type of pavement and the effectiveness of different treatment technologies with
proven efficiencies for a specified region. In this paper, we model the life-cycle deteriora-
tion of a pavement segment as a stationary Wiener process and develop a Markov decision
model to minimize the life-cycle costs of the pavement. The proposed deterioration model
is based on the Combined Condition Index (CCI) of the pavement at the beginning of the
planning horizon. Although maintenance actions such as chip seal or overlay cannot restore
the performance to the total new state, they can improve the performance and extend the
life. Therefore the deterioration model should be updated if specific interventions have been
taken after a regular inspection. Next, we integrate the predictive distribution of the degrada-
tion performance within the Markov decision process model to derive the optimal preventive
maintenance policy for a pavement segment. We show that the optimal policy under the
expected finite-horizon discounted cost criterion is a monotonically non-decreasing control
limit policy over time that optimally balances the cost of reconstruction, the cost of preven-
tive maintenance, and the cost of collecting data. Finally, we present a case study based on
real data from a specified pavement and study the performance of the policy under different
cost settings.
The remaining of this paper is outlined as follows. In Section 2 a deterioration model for the
pavement and the corresponding preventive maintenance strategy are introduced, based on real
data from Virginia Department of Transportation (VDOT). Section 3 presents the associated
continuous-state Markov decision model and the policy iteration algorithm to get the optimal
results with the discretization on the continuous state in terms of the calculation complexity. A
case study is investigated in Section 4 to illustrate the implementation and effectiveness of the
model. Finally, concluding remarks and future extensions are outlined in Section 5.
X (t ) = X ( ) μt + σ B (t )),, μ ≥ 0 (1)
where B(.) is a standard Brownian motion. μ is the slope of the linear drift and σ >0 is the
diffusion coefficient. It is stationary that the mean degradation is linearly increasing with t
. The pavement is announced to no use if X (t ) reaches a pre-specified degradation thresh-
old for the first time. X (t ) has the following properties in terms of the characteristics of the
Wiener process [14]:
1. X ( ) 0 almost surely.
2. For any time sequence{ q }1≤ q ≤ n,q,q ∈Z+ with0 < t1 < t2 < < tn , the random increments X (t1 ) ,
X (t2 t1 ), , X (tn tn 1 ) where X (t j t j ) X (t
(t j ) X (t j ) , are independent, and any 794806-
paper-E017.eps794806-paper-E018.eps follows N ( ,σ 2 t − r ).
3. The paths of X (t ) are continuous with probability one.
To characterize the segment-to-segment variability, μ and σ can be assumed to be ran-
dom parameters, which are specified as w 1 / σ 2 Gam
G (a, b ) d μ | w N (d , c / w ) .Where w
545
a gamma distributed variable with mean a b and variance a b2 , and μ is a normally distrib-
uted variable with mean d and variance c w given the value of w.
(
⎛ w Δx − μΔt ) ⎞⎟
2
k
w1 / 2
| μ,w ) = ∏ exp ⎜ −
j
L( (2)
Δt
1: k
j =1 2πΔt ⎜⎝ 2 ⎟⎠
( ) ( )
Wherew 1 / 2 ,t0 = 0 Δx j x t j x t j 1 and Δt t j − t j −1, for j k . According to
the Bayesian method, the joint posterior PDF for ( , μ ) can be formulated as
L( | w, μ )π (w, μ )
π ( , μ | X1:k ) = ∞ ∞
1: k
(3)
∫ ∫ L(
0 −∞
1: k | w, μ )π (w, μ d μdw
546
The details of the parameters’ calculation and updation can be found in reference [15].
Then, if the new deterioration data have been collected by regular inspection, the distribu-
tion of w, μ would be updated with Equation (2)-(3). X (t ) is correspondingly determined by
sampling from the deterioration model with updated random parameters.
V (X k ) = {
cr V n 1 ( ), X k Lc
min{W (donothing ),minW (ai )} X k Lc
(4)
Where
⎧⎪ fX k x t ) X k U (ai )
fY x ) = ⎨ +∞ (7)
⎪⎩∫−−∞ fX k y u t ) fUai u )du
)du X k U (ai )
Where fX′k x,t ) is determined by the updated deterioration model with the maintenance
action ai fUa (u ). is the PDF of residual damage variable U (ai ) . Moreover we assume that if
i
the observed signal exceeds the threshold Lc between the inspections and returns to a value
below this threshold at the next inspection time, this does not constitute a failure. Unfortu-
nately the solution of the optimization equation is difficult to get because of the complexity
of the continuous-state assumption.
547
space of the system is S ′ = { , Δ, 2 Δ, , Lc − Δ, Lc }. The corresponding value equation is
given by:
⎧
⎪⎪cr V ( ) X k Lc
V (X k ) = ⎨min{cin + α ∑ X p(X k | X k )V (X k +1 ),
k +1
⎪
⎪⎩ min{c(ak ) (cinin + ∑ X p′(X k + | U ((aak ))V (X k ))}} X k ≤ Lc
k +1 (8)
At each decision epoch k , given that the observed signal does not exceed the PM thresh-
old Lp , the transition probabilities to possible degradation states at the next epochk + 1can
be estimated by the distribution of the degradation level without updating the deterioration
model. Let FX k denote the distribution ofXX k at epochk .
When considering the effect of maintenance actions, the degradation level isYk X k + U (ai )
and the distribution of Yk is determined by updating the deterioration model and residual
damage model. Then,
P′′(X k + | X k a = ai ) = FYk + (X
(X + + Δ ) − FYk +1 (X k +1 ) (10)
where
+∞ y x
FYk ( y ) = ∫ fX k +1 x )dx ∫ fUa u )ddu (11)
∞ −∞
− i
A policy iteration algorithm is proposed by analyzing the special form of the control limit
structure to decrease the computational burden. Let pδ and rδ be the transition probability
and rewards given the policy δ which can be calculated by equation(7)-(11). Then the policy
iteration algorithm is given by
Step 1: Initialization: set k = 0. Start with an arbitrary initial policyδ .
Step 2: Policy evaluation: evaluate the current policy for all by i, j S ½solving:
vδ (i ) = rδ (i ) + α ∑ j Pδ v( j )
(12)
Step 3: Policy improvement: check whether shifting the control limitL La up or down results
in an improvement:
Fori La, if
cin + ∑ j
pijv( j ) < in{c(ai ) + (ciin + ∑ j pujv( j ))} (13)
max{c(( i ) + ( in ∑ j u
uj v( j ))} in ∑ j ij v( j ) (14)
SetLLa La − Δ, updating the parameters of the deterioration model. Then for all
i [ La , Lc ]
Go to step2.
Otherwise, go to step4.
548
Table 1. The values of Combined Condition Index (CCI) before and after overlay
and mill&fill.
Overlay Mill&fill
1 77 89 12 52 98 46
2 48 92 44 34 93 59
3 23 93 70 26 93 67
4 35 89 54 18 95 77
5 34 88 54 11 96 85
6 37 88 51 33 98 65
7 48 94 46 13 98 85
8 61 94 33 15 100 85
9 53 79 26 29 98 69
10 38 87 49 43 100 57
Step 4: Check for optimality: if k N , stop. Set optimal policy δ * δ . Otherwise, incre-
ment k by 1, and go to step2.
4 CASE STUDY
The case study is based on the CCI data from Figure 1 and the values of CCI before and after
two actions are given in Table 1.
As mentioned in section 2, the parameters of the deterioration process can be evaluated
and updated. μ ( 22 1 19)
(2 1// σ 2 ~ Ga ( .64, 0.044 ) . The set of maintenance actions
is A {a1, a2 ,aa3 , a4 } representing do nothing, overlay, mill&fill and reconstruction. Generally
reconstruction can restore the pavement to the ‘as good as new’ state while the effect of the
other two actions can be described by residual damage variables as discussed before. The
PDF of them can be fitted with the data given in Table 1 and the results are given in Figure 2
with U (a2 ) ~ N (12 9681,1 ) U (a
), (a3 ), N ( .0151, 0.6438).
Since the real cost data are not available, the test instant is used to verify the decision model
with the following data: cin $ , c1 = $ , c2 $ , c3 = $300 . A discretization scheme with
h = 100 is used, then Δ = 1, ′ = {0,1, 2,...,100}. The discount factor is assumed to be 0.9. Then
the optimal policy is taking overlay when the value of CCI exceeds 45 with a computed opti-
mal cost of V * ( ) = 148.68 .
Actually the optimal policy is the balance of cost and effect. Figure 3 gives the decision
results with the different cost ratios. If the cost ratio of two preventive actions is obvious, the
more consideration will focus on costs even if the effects are different. The more cost-effective
action overlay will be taken as shown in the first diagram of Figure 3. When the cost ratio
is small for two actions, the decision is made on the effect of them as shown in the second
diagram of Figure 3.
Moreover, when the cost of preventive maintenance is close to the cost of reconstruction,
the optimal policy is more tolerant to delay the maintenance to worse performance. In other
words, the control limit is closer to the failure threshold (reconstruction). For example, when
the cost of mill&fill is $300, which is equal to the cost of the reconstruction, it’s prone to take
the maintenance actions at the worse performance.
5 CONCLUSION
The life-cycle management of pavements has put much attention on the decisions regard-
ing the timing and the type of preventive maintenance when their performance is still good.
In this paper we have presented a model for optimal maintenance of such infrastructure.
549
Figure 2. The cumulative density function curves of residual damage of two maintenance actions.
550
REFERENCES
[1] Northwest pavement management systems users group & Kay. R. 1992. Pavement surface condition
rating manual. Washington state transportation center, Washington state printing office.
[2] Johnson, A.M. Best practices handbook on asphalt pavement maintenance. 2000. Minnesota Tech-
nology Transfer Center, University of Minnesota.
[3] Gary H., Stephen B. & David G.P. 2000. Selecting a pavement maintenance treatment for flexible
pavements. Foundation for pavement preservation.
[4] Mike J.& John C. 2002. Pavement maintenance manual. Nebraska department of roads.
[5] Abaza K. & Ashur S. 1999. Optimum decision policy for management of pavement maintenance
and rehabilitation. Transportation Research Record 1655: 8–15.
[6] Yang J., Lu J.J., Gunaratne M. & Dietrich B. 2006. Modeling Crack Deterioration of Flexible
Pavements: Comparison of Markov Chains and Artificial Neural Networks, Pavement Manage-
ment, Monitoring, Evaluation and Data Storage and Accelerated Testing. Transportation Research
Record 2586:18–25.
[7] Syed W H, Karim C. & Gilbert Y B. 2012. Long-term pavement performance effectiveness of preven-
tive maintenance treatment using Markov Chain Algorithm. Engineering Journal 16(4):149–159.
[8] Samuel L. & Kumares C.S. 2005. Life-cycle evaluation of pavement preventive maintenance. ASCE
Journal of Transportation Engineering 131 (10):744–751.
[9] Jesus M. Sercan A. Doug R.B. & Denise A.K. 2011. Network-level optimization of pavement
maintenance renewal strategies. Advanced Engineering Informatics 25: 699–712.
[10] Geoffrey L, Samuel L. & Zong L. 2008. Decision support for optimal scheduling of highway pave-
ment preventive maintenance within resurfacing cycle. Decision Support System 46: 376–387.
[11] Tack J. & Chou E. 2002. Multi-year pavement repair scheduling optimization by pre-constrained
genetic algorithm. Transportation Research Record 1816: 3–8.
[12] Cheu R. L, Wang Y. & Fwa T.F. 2004. Genetic algorithm-simulation methodology for pavement
maintenance scheduling. Computer-Aided Civil and Infrastructure Engineering 19 (6): 446–455.
[13] Giustozzi, F. & Crispino, M. 2012. Preventive Maintenance on road pavements: performance and
environmental assessment of strategies. SIIV Roma MMXII- 5th International Congress, 1–11.
[14] Alaa H.E., Nagi Z.G. & Lisa M. 2011. Structured Replacement Policies for Components with
Complex Degradation Processes and Dedicated Sensors. Operations Research 59:684–695.
[15] Xiaolin W., Ping J., Bo G. & Zhijun, C. 2014. Real-time Reliability Evaluation with a General
Wiener Process-based Degradation Model. Quality and reliability Engineering International 30(2):
205–220.
551
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Ye Yu
Southeast University, Nanjing, Jiangsu, China
Lu Sun
Southeast University, Nanjing, Jiangsu, China
Catholic University of America, Washington DC, USA
Haoran Zhu
Jiangsu Transportation Institute, Nanjing, Jiangsu, China
1 INTRODUCTION
Road authorities are facing the challenges of aging highways and deteriorating networks
with limited resources (Haider and Dwaikat, 2013, Simpson et al., 2013, Santos et al., 2015).
It is widely recognized that pavement management, which is implemented using pavement
management system (PMS), is vital in helping road authorities to manage their own road
networks effectively and efficiently (Fernandes and Neves, 2014, Dong et al., 2015). There
are generally two major administrative levels in pavement management: the network level
and the project level (Haas et al., 1994, Chu and Chen, 2012). The network level deals with
the pavement network as a whole and is generally concerned with high-level decisions relating
to network-wide planning, policy and budget. The project-level pavement management deals
with smaller constituent sections within the network and is intended to predict pavement
deterioration, select the appropriate preservation activity, and develop the optimal preserva-
tion schedule for a specific pavement segment (Zhang et al., 2013, Wu and Flintsch, 2009).
In order to aggregate pavement performance data of many pavement segments for net-
work-level pavement management, mean and standard deviation of specific performance
indicator are often used practically. However, summarizing these data with their mean and
standard deviation is far from enough to characterize network-level pavement performance
data. Different pavement segments may be coped with preventive maintenance in differ-
ent years and in different ways. The empirical distribution of pavement performance data
553
collected from many pavement segments often exhibits multimodality as pavement perform-
ance across pavement segments is heterogeneous. In this situation, a unimodal distribution
such as normal distribution may no longer be capable of capturing the characteristics of
network-level performance data in a satisfactory manner. Since the empirical distribution
of network-level pavement performance data deviates considerably from normal distribu-
tion, aggregating these data only with their mean and standard deviation cannot characterize
network-level pavement performance accurately.
Mixture models, which is a combination of more than one distribution, have the advantage
for producing a sufficient goodness-of-fit for empirical modeling of multi-modal distribu-
tion and for providing information about each sub-population in the dataset (Frühwirth-
Schnatter, 2006). In this study, Gaussian Mixture Model (GMM) is proposed to aggregating
network-level pavement performance data to account for performance heterogeneity across
many pavement segments. Comparing to a unimodal distribution such as normal distribu-
tion, the benefits of using mixture models to aggregate performance data are twofold. On
one hand, accounting for performance heterogeneity across pavement segments can provide
a better fit to empirical performance data and therefore capture the overall performance
characteristics of multiple pavement segments more accurately. On the other hand, mixture
model can classify pavement segments into multiple groups according to their performance
condition. This can be beneficial because appropriate countermeasures for performance
improvement and distress mitigation can be prepared for pavement segments in each group,
respectively.
The primary objective of this study is to propose a Gaussian mixture model to aggregate
pavement performance data for network-level pavement management which can account
for performance heterogeneity across the pavement segments. In this study heterogeneity is
defined as the performance difference across multiple pavement segments in a road network.
In practice, heterogeneity is often ignored when aggregating pavement performance data for
multiple pavement segments. However, ignoring performance heterogeneity can lead to biased
and inaccurate results. Instead of summarizing performance data with their mean and stand-
ard deviation directly, the proposed method aggregates pavement performance data with mix-
ture weights, means, and standard deviations of the fitted Gaussian mixture models.
The remainder of the paper is organized as follows. The next section presents the meth-
odology of Gaussian mixture model. The third section gives a brief description of the data
set used in this study and reports the empirical findings. Concluding remarks are provided in
the last section.
2 METHODOLOGY
where p(x |λ) = probability density of a Gaussian mixture model; x = D-dimensional contin-
uous-valued data vector; αi = mixture weight for component i; and g(x | μi, Σi ) = component
Gaussian density for component i.
Each component density is a D-variate Gaussian function of the form
1 ⎧ 1 ⎫
g ( x | μi , Σ i ) = exp ⎨− ( x μi )′ Σ i 1
(x μi ) ⎬ (2)
( )D / | i |
/2
⎩ 2 ⎭
where g(x | μi, Σi) = probability density of component i; μi = mean vector of component i; and
Σi = covariance matrix of component i.
A Gaussian mixture model is parameterized by the mixture weights, mean vectors, and
covariance matrices (or standard variations for one-dimension data) from all components. It
can be represented by the compact notation
{α i μi , Σ i } i = 1, ..., M (3)
Q( , (i )
) E ⎡⎣ log p( Y Θ ) Χ, Θ(i( i ) ⎤⎦ (4)
where Θ(i-1) = current parameters estimates used to evaluate the expectation and
Θ = new parameters that will be optimized in Step 2 to increase Q.
Step 2 (M-step) Obtain new parameters estimates Θ(i) which maximize the expectation
computed in Step 1.
Θ( ) g Q (Θ, Θ( )
) (5)
Θ
For Gaussian mixture model, the estimates of the new parameters in terms of the current
parameters estimates are given in Equation (6)-(8) (Bilmes, 1998).
1 N
αi = ∑ p i xj
N j =1
( g
) (6)
∑ x p (i x
j =1
j j
g
)
μi = N
(7)
∑ p i xj
j =1
( g
)
N
∑ p (i x )( ) (x )
T
j Θg x j − nnew
i
new
j − nnew
i
j =1
Σi = N
(8)
∑ p i x j Θg
j =1
( )
where Θg = current parameters estimates and αi, μi, and Σi = new parameters estimates for
component i.
3 EMPIRICAL STUDY
The proposed method is tested on the rutting data of Freeway G30 in Jiangsu, China. Freeway
G30 in Jiangsu is also called Lianxu Freeway. It is about 240 km long and has 4 lanes (2 lanes in
each direction). Road width is 28.0 m. It runs from Lianyungang, Xinyi to Xuzhou. All of these
cities are considered key areas for the economic development in the northern part of Jiangsu
Province. Since rutting is one of the main distresses on Lianxu Freeway, this study aims at aggre-
gating rutting data on Lianxu Freeway based on the proposed method. Rutting has increasingly
become one of the main concerns for highway agencies in China because traffic volume has
increased dramatically in the past two decades (Wang et al., 2009). Rutting depth data used in
this study were collected in both directions in 2011 and 2012 using advanced laser profiler. The
survey was carried out over a period of 2 days from October 23 to October 24 of the same year.
Outer lane in each direction is chosen as the inspection lane. Average rutting depth data for each
1-km segment in each direction were recorded in Pavement Management System (PMS) and
were retrieved for aggregating rutting data for network-level pavement management.
Table 1 shows the mean and Standard Deviation (SD) of the rutting depth data in each direc-
tion. The mean of eastbound rutting depth is larger than westbound rutting depth in 2011, indi-
cating that the eastbound overall rutting condition is worse than the westbound. The mean rutting
depth increased from 2011 to 2012 for both directions. However, the mean of eastbound rutting
depth grew quicker than westbound rutting depth. In 2012, the mean of eastbound rutting depth
556
Table 1. Summary for rutting depth data.
Mean SD Mean SD
mm mm mm mm
became smaller than westbound rutting depth. In terms of standard deviation, it seems that the
standard deviation of rutting depth data improved from 2011 to 2012 for both directions. In addi-
tion, the eastbound rutting depth has a larger variation than westbound rutting depth.
While the mean and standard deviation of these data can provide a general overview of
pavement performance, they cannot capture the characteristics of pavement performance
data accurately. In pavement performance data, heterogeneity can come from many sources.
Different pavement segments are exposed to different traffic loads, different climate and dif-
ferent environmental conditions. In addition, pavement segments are probably coped with
different maintenance or rehabilitation treatments during their service life.
To account for the performance heterogeneity in rutting data, Gaussian mixture models
are fitted using Expectation-Maximization (EM) algorithm based on average rutting depth
data for each directions in 2011 and 2012, respectively. Pavement segments of the eastbound
and the westbound direction are considered separately in this study. Bayesian Information
Criterion (BIC), which is the most commonly used penalized maximum likelihood method
for mixture models, is adopted to select the optimum number of components in Gaussian
mixture models (Frühwirth-Schnatter, 2006). The results shows that two-component Gaus-
sian mixture models can produce the lowest BIC values for all the four cases (pavement seg-
ments in the eastbound direction in 2011 and 2012, pavement segments in the westbound
direction in 2011 and 2012) in this study, indicating that two-component GMMs perform
better than GMMs with more components. As a result, two-component GMMs are adopted
for the following analysis. Table 2 presents the parameters of the four GMMs.
Figure 2. Observed density of average rut depth overlaid with the fitted 2-component GMM for east-
bound direction in 2012.
proportions are 0.402 and 0.598, respectively. The parameters for the components of the
2-component GMM are μ1 = 7.593 (σ1 = 1.887), and μ2 = 4.200 (σ2 = 1.181), respectively.
Hence, summarizing these data with their mean (5.565) and standard deviation (2.249) does
not do justice to their nature.
What is interesting in the results is that the component proportions (0.395 and 0.605) of
the fitted 2-component GMM for 2012 is almost the same as those (0.402 and 0.598) for 2011.
This suggests that pavement segments in the eastbound direction may be classified into two
categories. In addition, the proportion of each component in the data is relatively stable. The
source of heterogeneity may be further analyzed by investigating the features of pavement seg-
ments in each category. It is important to note that the category with worse performance is of
particular interest to pavement management because this category will trigger the thresholds
for maintenance or rehabilitation before the other category. When forecasting pavement per-
formance, it is more reasonable to generate prediction for the two categories separately.
The first category, which accounts for about 40% of the data, has a mean of 7.593 mm in
2011 and a mean of 7.840 mm in 2012. The second category, which accounts for nearly 60%
of the data, has a mean of 4.200 mm in 2011 and has a mean of 4.484 mm in 2012. The rate
of degradation for the first and the second category (in terms of rut depth) is 0.247 mm/
year and 0.284 mm/year, respectively. It seems that the rutting condition of both categories
decreased from 2011 to 2012. This phenomenon is also illustrated in Figure 3. Probability
densities of the two components shift to the right as time goes by. This result can be explained
558
Figure 3. Component distributions of the fitted GMMs for eastbound direction in 2011 and 2012.
Figure 4. Overall rutting depth distribution for eastbound direction in 2011 and 2012.
by the fact that pavement segments deteriorate under the combined effects of traffic loading
and the environment. Similarly, as shown in Figure 4, the overall mixture density of rutting
data in the eastbound direction shows a trend of shifting to the right as time goes on. From
Figure 3 and Figure 4, one can draw the conclusion that the overall rutting conditions in the
eastbound direction decreased from 2011 to 2012.
Figure 6. Observed density of average rut depth overlaid with the fitted 2-component GMM for west-
bound direction in 2012.
Figure 7. Component distributions of the fitted GMMs for westbound direction in 2011 and 2012.
performance) deteriorated faster than the second category. This may be explained by the fact
that in general pavement deterioration becomes faster as pavement Remaining Service Life
(RSL) decreases. What’s more, the proportion of the first category, which is the category with
relatively bad condition in each year, improved from 12.1% to 16% from 2011 to 2012.
560
Figure 8. Overall rutting depth distribution for westbound direction in 2011 and 2012.
Mean rut depth in 2011 Mean rut depth in 2012 Rate of degradation
mm mm mm/year
Eastbound
Category 1 7.593 7.840 0.247
Category 2 4.200 4.484 0.284
Westbound
Category 1 8.077 8.808 0.731
Category 2 4.972 5.580 0.608
Thirdly, it can be seen from Figure 8 that the overall rutting condition of the westbound
decreased from 2011 to 2012. The overall mixture density of rutting data in the westbound
direction shows a trend of shifting to the right as time goes on.
4 CONCLUSION
Network-level pavement performance data which is collected from many pavement segments
often exhibits heterogeneity. Performance heterogeneity across pavement segments may come
from various sources, such as difference in maintenance or rehabilitation treatments, traffic
loads, environment, survey condition, equipment, period of monitoring, and presence of
worksites. Summarizing these data with their mean and standard deviation is far from enough
to characterize network-level pavement performance data. In order to account for performance
heterogeneity, this study proposes a Gaussian Mixture Model (GMM) to aggregate pavement
performance data for network-level pavement management. The Expectation-Maximization
(EM) algorithm is adopted for parameter estimation. Bayesian Information Criterion (BIC)
is employed to select the optimum number of components in Gaussian mixture models. The
proposed method is tested on the rutting data of Freeway G30 in Jiangsu, China.
561
The empirical study shows that the observed densities of rutting data deviate considerably
from normal distribution, indicating that performance heterogeneity does exist across the pave-
ment segments in network-level pavement performance data. The results suggest that Gaussian
mixture models can capture the characteristics of heterogeneous performance data and provide
an accurate summary for network-level pavement management. Pavement performance data
can be aggregated using mixture weights, mean values, and standard deviations of the fitted
Gaussian mixture models. The aggregated performance data of different years can be easily
compared using parameters or mixture density plot of corresponding fitted GMMs. In a word,
the proposed method is able to accurately aggregate heterogeneous pavement performance data
and it can be used to analyze overall performance trend of many pavement segments.
In addition, Gaussian mixture models can classify pavement segments into multiple groups
according to performance condition. The category with relatively bad performance is of par-
ticular interest to pavement management since pavement segments in this category will trigger
the thresholds for maintenance or rehabilitation before the other pavement segments. Further
studies will investigate the source of heterogeneity in the dataset used in this study and the
reason for degradation rate difference between the eastbound and westbound direction.
REFERENCES
Bilmes, J.A. 1998. A gentle tutorial of the EM algorithm and its application to parameter estimation for
Gaussian mixture and hidden Markov models. International Computer Science Institute, 4, 126.
Chu, J.C. & Chen, Y.J. 2012. Optimal threshold-based network-level transportation infrastructure life-
cycle management with heterogeneous maintenance actions. Transportation Research Part B-Meth-
odological, 46, 1123–1143.
Dempster, A.P., Laird, N.M. & Rubin, D.B. 1977. Maximum likelihood from incomplete data via the
EM algorithm. Journal of the Royal Statistical Society. Series B (Methodological), 1–38.
Dong, Q., Huang, B.S. & Richards, S.H. 2015. Calibration and Application of Treatment Performance
Models in a Pavement Management System in Tennessee. Journal of Transportation Engineering, 141.
Fernandes, A. & Neves, J. 2014. Threshold values of pavement surface properties for maintenance pur-
poses based on accidents modelling. International Journal of Pavement Engineering, 15, 917–924.
FR Hwirth-Schnatter, S. 2006. Finite mixture and Markov switching models, Springer Science & Business
Media.
Haas, R., Hudson, W.R. & Zaniewski, J.P. 1994. Modern pavement management, Malabar, Krieger Publishing.
Haider, S.W. & Dwaikat, M.B. 2013. Estimating Optimum Timings for Treatments on Flexible Pave-
ments with Surface Rutting. Journal of Transportation Engineering-Asce, 139, 485–493.
Hyun, K., Hernandez, S., Tok, A. & Ritchie, S.G. 2015. Truck Body Configuration Volume and Weight
Distribution Estimation by Using Weigh-in-Motion Data. Transportation Research Record, 103–112.
Jun, J. 2010. Understanding the variability of speed distributions under mixed traffic conditions caused
by holiday traffic. Transportation Research Part C-Emerging Technologies, 18, 599–610.
Lao, Y.T., Zhang, G.H., Corey, J. & Wang, Y.H. 2012. Gaussian Mixture Model-Based Speed Estima-
tion and Vehicle Classification Using Single-Loop Measurements. Journal of Intelligent Transporta-
tion Systems, 16, 184–196.
Park, B.J., Zhang, Y.L. & Lord, D. 2010. Bayesian mixture modeling approach to account for heteroge-
neity in speed data. Transportation Research Part B-Methodological, 44, 662–673.
Santos, J., Ferreira, A. & Flintsch, G. 2015. A life cycle assessment model for pavement management: meth-
odology and computational framework. International Journal of Pavement Engineering, 16, 268–286.
Simpson, A., Rada, G., Visintine, B. & Groeger, J. 2013. Evaluating Pavement Condition of the National
Highway System. Transportation Research Record: Journal of the Transportation Research Board, 50–58.
Wang, H., Zhang, Q.S. & Tan, J.Q. 2009. Investigation of Layer Contributions to Asphalt Pavement
Rutting. Journal of Materials in Civil Engineering, 21, 181–185.
Wu, Z. & Flintsch, G.W. 2009. Pavement Preservation Optimization Considering Multiple Objectives
and Budget Variability. Journal of Transportation Engineering-Asce, 135, 305–315.
Zhang, H., Keoleian, G.A. & Lepech, M.D. 2013. Network-Level Pavement Asset Management System
Integrated with Life-Cycle Analysis and Life-Cycle Optimization. Journal of Infrastructure Systems,
19, 99–107.
Zou, Y.J. & Zhang, Y.L. 2011. Use of Skew-Normal and Skew-t Distributions for Mixture Modeling of
Freeway Speed Data. Transportation Research Record, 2260, 67–75.
562
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
B.A. Feyissa
Ingenieurbüro Abesha—Engineering Office, Freiburg, Baden-Württemberg, Germany
ABSTRACT: The objective was to model rutting of long life asphalt concrete pavement
subjected to non-uniform traffic loading and ever-changing temperature cycles during its
service life. Critical stress states within the pavement had been evaluated using ABAUQS
finite element program based on long life asphalt concrete concept. A Burger-based rut pre-
diction model had been developed and programmed by taking into account the prevailing
traffic loading spectrum and the fluctuation of the temperature likely to happen across the
depth of pavement. The finite element analyses showed that critical vertical and shear stresses
are developed within the surface and binder layer, respectively. An increase the E-moduli of
the binder yielded reduction in vertical stresses at the mid-depth of the binder and base lay-
ers. The rut prediction model indicated that the rate of rutting had been influenced by the
viscous parameter and temperature of the asphaltic layers, wander effect and speed of the
vehicles.
1 BACKGROUND
The repetitive traffic loading that the road experiences during its service life combined with
environmental factors causes permanent deformation, fatigue cracking, instability and other
forms of damage on pavement structures. Estimation and prediction of the maximum dis-
tresses within the pavement layer is vital for proper design of the pavement. Constitutive
modeling of the deformation behavior of asphalt concrete enables better understanding of
the deterioration mechanism and assists to design most economical and long lasting roads.
At the same time, better understanding of materials behaviors and structural responses of
asphalt concrete allows for optimization of asphalt pavement thickness and material choice.
Many models developed so far estimate the number of repetitions to failure in the rut and/
or fatigue mode. But basic fundamental traffic characteristics such as speed, loading spec-
trum, wander effect and temperature fluctuation rather have not been included in damage
prediction model. Because of such reasons a pavement deformation model has been devel-
oped which aimed mainly at rut accumulation over a period of time due to time dependent
mixed traffic flow as well as daily and seasonal temperature variations. An attempt has been
made to evaluate the permanent deformation with different material parameters, temperature
fluctuation, structural geometry as well as traffic characteristics.
2 TRAFFIC CHARACTERISTICS
Excessive deformation resulting from frequent repetitions of vehicle loads in heavy duty
asphalt-concrete pavements is the main concern. The simplest pavement structural model
asserts that each individual load inflicts a certain amount of unrecoverable damage. How-
ever, trucks have a greater share and wheel loads of heavy vehicles are considered as primary
contributors.
563
The major traffic input parameters for analysis and modeling are the magnitude and dis-
tribution of the wheel loads on the pavement surface. Most design standards require quanti-
fication of various loading levels that a pavement encounters over its design life. A possible
solution is to use the number of Equivalent Single Axle Loads (ESAL) which converts wheel
loads of various magnitudes of the mixed traffic to a single load. The other solution is the
load spectra approach that uses the actual traffic data without converting into ESAL, which
is adopted in this research.
On the other hand, many of the existing mechanistic pavement response models do not
include vehicle speed and wander effect as an important factor. But evidences show that the
magnitude of the calculated pavement strain response decreased with increased vehicle speed
(Siddharthan et al. 2002). In addition, the distribution of vehicles path in the transverse direc-
tion plays a significant role in pavement performance (White et al. 2002). At the beginning
of the service life, the evenness of the pavement may give the drivers psychological freedom
to maneuver the vehicles within the lane space. The wander effect plays here a constructive
role in minimizing the rut formation. After some period of time the gradual appearance of
rutting along the wheel track would rather make drivers to adhere to the center of lane which
ultimately accelerate the rut formation along the wheel track.
In this research, a load spectrum approach has been adopted to characterize the traffic
input. About 15 types of vehicles with corresponding wheel load, vehicle proportion, hourly
flow, operating speed and wander effects have been taken into account.
The temperature distribution in a pavement is directly affected by the thermal property and
environmental conditions to which it is exposed (Yavuzturk 2005). The ability to accurately
predict asphalt pavement temperature is helpful to perform back-calculations of materials’
parameters. Due to heat exchange, the change of the pavement temperature is delayed com-
pared with air temperature (Huang 2008); the deeper the location is the later each portion of
pavement reaches its highest temperature. The temperature distribution within a pavement
investigated (Wellner & Kayser 2008) indicated that the thermal propagation depends on the
conductivity of the material, the amplitude of periodical temperature change, cycle time and
other thermal behavior of the system.
It has been indicated (Werkmeister et al. 2003) that the maximum temperature occurs at
the superstructure relative to substructure. This highly affects the viscoelastic characteristics
564
Figure 2. Rut formation along the wheel track.
and performance of the asphalt concrete layers. The stresses applied on asphalt pavements at
high temperature cause relatively higher damage (Wellner & Kayser 2008).
Temperature plays dominant role in the variation of modulus of asphalt concrete (AC)
layers. The variation is typically related to the average temperature of a layer, which in turn
depends on the temperature gradient within the layer (Nazarian & Alvarado 2006). The tem-
perature gradient across the depth of the pavement is schematically represented as shown in
Figure 1.
4 PAVEMENT RUTTING
Most failure criteria have been attributed mostly to the weak pavement structure (subgrade)
in terms of the vertical compressive strain to the number of cycles to failure. However, rutting
has been observed in some thicker asphaltic layers due to the fact that pavements had been
constructed over a strong and non-yielding substructures that led rutting to be developed and
confined in the AC layers as shown below in the Figure 2. Pavement constructions with thick
asphalt layers contributed lesser or no risk of rutting in the granular layer even at higher
number of load application (Werkmeister et al. 2003), implying that only asphalt layers are
liable to rutting for thicker AC pavements.
Rutting is a major distress form when the ambient temperature is high such as in hot
tropical climate regions or during the summer months of temperate zones. At elevated tem-
peratures, AC mixes become softer and exhibit markedly different volume-change and shape-
distortion deformation modes in terms of their sensitivity to temperature and rate of loading
(Havey et al. 2009). At high temperature or under slow moving loads, hot mix asphalt exhib-
ited purely viscous flow, which resulted in severe rutting (Elseifi et al. 2006).
Modeling and analysis of long life asphalt concrete pavement have been advocated for the
past several years. Until recently, asphalt concrete roads have been designed to serve 20–25
years. In fact, some roads which had been constructed with the same design life have been
serving beyond the design limit without excessive damage. The long-life pavement structures
consist of but not limited to impermeable, durable, and wear resistant top layers; a stiff and
565
thick rut-resistant intermediate layer; and a flexible fatigue-resistant bottom layer resting on
a stable and high-strength foundation (Walubita et al. 2008). Of course, pavements that are
built above certain threshold strength would have sustainable structural life. Decreasing the
tensile strain at the bottom of the Hot Mix Asphalt (HMA) layers results in an increasing
in fatigue life of a pavement structure (Shen & Carpenter 2005). The use of high modulus
asphalt mix is an alternative option to reduce the tensile strain at the bottom of AC layer
(Lee et al. 2007).
Researches indicated that major distress forms are associated with the thickness of the
pavement structure. It has been found (Nunn & Ferne 2001) that high rates of rutting were
associated with thin asphalt pavements, whereas, thicker sections showed a slow rate of rut-
ting which was confined to the top layer of the pavement. It was suggested (Park et al. 2005)
that the ability of AC pavements to carry heavy axle loads was achieved by improving the
stiffness of the asphalt structure and adjusting the thickness.
An effort has been made to stretch the life of the pavement through mitigation (Epps 2000)
of the source of distress and use of more resistant materials. It has been reported (Lee et al.
2007a, Lee et al. 2007b) that high modulus asphalt binders and mixes showed higher dynamic
modulus and lower rutting depths than those of the conventional mix at high temperatures.
The low voids content and high stiffness of hot mix asphalt concrete (HMAC) also provided
protection to the base course and great resistance to rutting (Corté 2001).
There has been a proliferation of computerized pavement analysis systems which use elastic,
viscoelastic and/or viscoplastic theories for analysis and life prediction (Cebon 2009). In this
research, the primary responses of AC pavements are evaluated by ABAQUS Finite Element
(FE) program. The stress states responsible for rutting are vertical compressive (Fig. 3a) and
shear stresses (Fig. 3b), which are proportional to the applied tire pressure on the surface of
the pavement. For thicker pavements, the maximum vertical compressive stress is developed
in the wearing layer and confined within the wheel foot print of the AC layers. There is a
significant drop in magnitude of this stress at lower depths (at subbase and subgrade layers),
however, the influence area gets wider. On the other hand, the maximum vertical strain is
observed at the top of the subgrade (as evidenced in many literatures) due to its lower stiff-
ness characteristics. Relatively, the vertical strain within the intermediate layers (binder, base
and sub base) is lower than both the wearing and subgrade layers because of higher verti-
566
cal stress within the wearing course and higher elasticity modulus in both binder and base
layers.
The structural properties have significant role on pavement responses. A 2D pavement
model has been analyzed with different layer modulus using ABAQUS software. As shown in
Figure 3, the maximum vertical and shear stresses developed at the surface and intermediate
layer, respectively. The analysis indicated that increasing the binder E—modulus decreases the
magnitude of the vertical stress developed within each layers. An increase the E-moduli of the
binder twice yields 4.9% and 14.3% reduction in vertical stresses developed at the mid-depth
of the binder and base layers, respectively, just directly below the point of load application.
While increasing the same by forth fold yields as well 11.1% and 27.6% reduction in these lay-
ers. There is a reduction in vertical and shear strains and an increase in lateral stress within the
wearing course with higher E-values wearing layer. It is clearly indicated from the analysis that
the strength of the asphaltic layers is in favor of the subbase and subgrade layers in decreasing
the stress-strain states. Higher elasticity modulus in the binder course rather increase the shear
stress developed within the binder itself and decreases the same in the base course and vise
versa. The shear strain, in both cases reduces with increasing the E-moduli of the layers.
The structural thickness greatly affects the response of the pavement. The analysis indi-
cated that the stress-strain state in general decreases with increasing the asphaltic layer thick-
nesses. There is a significant reduction in the shear stress and strain states as well as reduction
in vertical strains below the asphaltic layers with increasing AC thickness.
Accurate performance predictions are possible only if the nature of the material is properly
modeled. The temperature, strain rate as well as the state of stress dependency and the mate-
rial responses cannot be neglected.
An attempt made to evaluate the damage accumulation with FE demanded huge memory
space and high computation-time to simulate prevailing traffic and climatic/temperature/
change simultaneously. A rut prediction has been developed and programmed (with C pro-
gramming language) based on Burger material model. The model evaluates the accumulation
of rutting during the presumable design period of the pavement under the prevailing traffic
and temperature ranges for every hour of a day throughout the year. The pavement responses
under various loading, material properties and pavement structural geometries are extracted
from FE analyses.
⎡1 t 1 ⎤
ε( ) = σ0 ⎢ + + ( E η
)⎥ (1)
E
⎣ 1 η1 E 2 ⎦
For time t > t0
⎡t 1 ⎤
ε ( ) σ 0 ⎢ 0 − e −t E2 η2 ( E η
)⎥ (2)
η
⎣ 1 E 2 ⎦
where ε(t) = strain at time t; σ0 = stress; E1 = immediate elastic modulus; E2 = delayed elastic
modulus; η1 = immediate Burger’s viscous component; η2 = delayed Burger’s viscous compo-
nent; t0 = maximum loading time.
567
In case of a relaxation test where constant strain is applied, the time dependent modulus
of elasticity is expressed as:
-t -t
E(t) = E1 e t1 + E2 e t2 or (3)
⎛ −E t⎞ ⎛ −E t⎞
⎜ 1⎟ ⎜ 2
⎟
⎝ η1 ⎠ ⎝ η2 ⎠
E(t) = E1 e + E2 e (4)
where ∆P = total rut depth; ε pi =average permanent strain in the ith asphalt layer; ∆Zi = thick-
ness of the ith asphalt layer; n = total number of sub layers; m = total number of wheel load
during the service life.
Figure 4. Burger model strain response during loading and unloading cycle.
568
The total rut depth is obtained by summing the products of the average permanent strain
at mid depth and corresponding thickness of each layer using Equation 5 above.
The transient temperature variance of pavement and timing of loading/unloading have
been taken into account. This would guarantee a more reasonable simulation, closer to the
field situation and more effective to address the rutting behavior.
7.2 Results
The initial traffic, growth rate, composition, proportion and slow moving heavy vehicles
highly affected the results. At later time of the service year, the number of vehicles increases
rapidly (depending on the rate of traffic growth) which demands relatively longer load-
ing time (due to traffic congestion and lower operating speed) and/or shorter unloading
time which resulted in severe rut formation. Keeping material and structural properties the
same, the deformation curve has shown a rapid increment at later years with 7.5% growth
rate which suggests that design standards have to thoroughly select applicable rate to rea-
sonably address associated rutting and avoid over or under estimation of the pavement
structure.
On the other hand, the following analysis is made on a maximum temperature ranging of
–15 to 50 0C at 4.5% growth rate with different initial traffic volumes (800 commercial vehicles
per day (cv/d) to 1500 cv/d or 27.89 to 52.30 million ESAL80kN, respectively, at 40 years).
The material properties and FE results at different temperature are shown in Figure 6.
As shown in the Figure 7, the structural geometry and material characteristics described in
Figure 6 can serve for more than 35 years when the forecasted traffic is below 26.55 million
ESAL under the temperature range mentioned above.
At the same time, an attempt has also been made to evaluate the effect of material charac-
teristics. The analyses indicated that, enhancing the viscous component brings a change even
at relatively higher temperature. The immediate viscosity parameter, particularly at higher
temperature, plays a significant role.
The same structural geometry can sustain an ESAL of 40–50 million ESAL for about 40
years at a maximum temperature range of 0–50 0C when the immediate Burger’s viscous
parameter (η1) in Figure 6 is increased by double. The deformation behavior and temperature
ranges are shown in Figure 8b.
Several results indicated that best performance can be achieved if the viscous parameter
at higher temperature is relatively higher. It has been obtained that thicker AC layers with a
minimum of 2 × 103 GPa.s immediate viscous parameter (η1) at higher temperature would
give a promising long life to the pavement structure.
569
Figure 7. Permanent deformation under different traffic volume and temperature.
It is concluded that apart from the applied traffic load and the structural geometry; the tem-
perature and the viscoelastic materials properties play leading role on rutting. Higher vis-
cous parameters of the asphaltic layers guarantee lower rutting rate and better performance.
Thicker asphaltic layers reduce the distress developed within the substructure (Subbase and
subgrade) and let rutting to be confined within the superstructure. Providing better quality of
materials with reference to viscous parameter would insure durable and better performance
road. It is highly regarded and economical to consider the design of pavements on the bases
of long service life which can be achieved by thoroughly identifying the complex response of
a pavement developed due to a combination of different load and environmental factors.
It is recommended that the influence of the shear stress is better modeled and accommo-
dated in rut analysis model to refine the work. A comprehensive model is required which include
fatigue, damages caused by moisture and other modes of failure in order to mitigate the complex
nature of possible damage modes and associated factors. The program contains subroutines
which deal with the traffic characteristics and the daily and annual fluctuation of temperature
which can be integrated and implemented in fatigue and low temperature cracking model to
evaluate such modes of damages during the service life of the road. In addition, calibrations of
the results with field result and/or previous works are most beneficial. Finally, incorporating pos-
sible sources of uncertainties and quantifying them statistically will be helpful.
ACKNOWLEDGEMENT
571
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
Rutting is major distress of asphalt pavement, which mainly results from deformations
of asphalt mixtures under high temperature and heavy traffic loads. In China, large num-
bers of highways suffer seriously from rutting five years after coming into use, especially
in southern provinces like Guangdong Province. Gravel Asphalt Concrete (GAC) mix-
ture is proposed and designed under this circumstance. The principles for designing GAC
mixtures are to maximize the percentage of course aggregates and use less fine aggregates
than HMA mixtures. Xian Hongwei developed a new method to design the gradation of
GAC-20 and performed experiments to ensure that the newly designed GAC-20 is more
stable in high-temperature performance than traditional asphalt mixtures (Xian Hongwei
2010). Zhang Shoufu utilized GAC-25 to make repairs of pavements in southern Xinjiang
Province, one of the hottest regions in China in summer, and the tests after rehabilitation
proved that GAC-25 is suitable for high-temperature regions (Zhang Shoufu 2011). Yang
Qi compared different asphalt mixtures such as SMA, GAC and SUP and confirmed that
GAC-20 was more suitable to prevent rutting than other options, especially in Guang-
dong Province (Yang Qi 2009). Lv Rui confirmed that GAC has advantage in resisting
rutting in hot regions based on tests on specimens from Jiangmen-Zhaoqing Expressway
(Lv Rui 2013).
Meanwhile, the index used for evaluating high-temperature performance and perform-
ance under heavy loads of GAC in hot regions still needs to be discussed. Li Zhi developed
designing method for gradation of asphalt mixtures under heavy loads and confirmed that
this modification on gradation was successful in resisting rutting (Lv Rui 2013). F.P.Pramesti
utilized four points bending test to predict fatigue cracking of GAC and developed a new
calibration factor to describe fatigue life of GAC mixtures (F.P.Pramesti 2013). Based on
this method, Zhang Yongsheng suggested that the index for evaluating high-temperature
performance and performance under heavy loads of GAC used in Wu Shen Expressway in
Guangdong Province should be reconsidered (Zhang Yongsheng 2015).
573
The objective of this paper is to develop an evaluation index to describe anti-rutting
performance of GAC mixtures under high temperature and heavy loads. First, the Chinese
Wheel Grinding (CWG) test and Accelerated Asphalt Pavement (AAP) test were conducted
under different temperatures to determine the crucial temperature of anti-rutting perform-
ance of GAC mixtures. Then the correlation between results from CWG test and AAP test
was confirmed. The relationship between dynamic stability (DS) and viscosity of asphalt
mixtures as well as traffic loads were also conducted. Lastly, the fitting results of DS are
confirmed based on temperature, traffic loads and viscosity of modified asphalt used in
GAC mixtures.
The rutting tests performed in this paper were as follows: Rutting test based on Chinese
Wheel Grinding (CWG) test methods of asphalt mixtures and Accelerated Asphalt Pavement
(AAP) test (CHEN Huaxin 2009).
(t2 t1 ) × N
DS = × c1 × c2 (2)
d2 d1
where
d1 – Deformation at 45 min (mm);
d2 – Deformation at 60 min (mm);
N – Loading times (42/min);
C1 – Modifying coefficient for the size of specimen. 1.0 was selected here;
C1 – Modifying coefficient for the type of experimenting equipment. 1.0 was selected here;
T1 – 45 min;
T2 – 60 min.
The results of test under different temperatures are summarized in Table 1.
A linear function was generated to fit the relationships between DS and experimental tem-
perature. The coefficient in the function was calculated by using Microsoft Office Excel 2013.
The function and the coefficient are presented as Equation (3)
574
DS = a1T + b1 (3)
Temperature (°C) 55 60 65 70
575
Table 2. Results of DS between 65 °C and 70 °C.
Temperature (°C) 66 67 68 69
N2 N1
S1 =
DS (4)
d2′ d1′
where
d1′ – Deformation when loaded 6000 times (mm);
d2′ – Deformation when loaded 8000 times (mm);
N1 – Loading time, which is 6000;
N2 – Loading times, which is 8000;
The results of DS1 according to experimental temperature are shown in Tab. 3.
Data from Table. 3 show that the result conducted from AAP is similar to that from
CWG test. A linear function was also generated to fit the relationship between DS1 and
experimental temperature. The coefficient in the function was calculated by using Micro-
soft Office Excel 2013. The function and the coefficient are presented as the following
equation:
S1 = a 2T + b2
DS (5)
576
Table 3. Results of AAP test.
Temperature (°C) 55 60 65 70
Temperature (°C) 66 67 68 69
were conducted to see whether the AAP test experiments can replace CWG test to determine
anti-rutting performance of GAC mixtures.
Based on Correlation Analysis from Microsoft Office Excel, the relationships of results
from two rutting tests were examined. The fitting function is shown as Equation (6), while
Fig. 5 shows the two DS-Temperature curves.
where R2 is 0.9999.
577
Figure 5. DS-Temperature curves of two tests.
The results of two tests have good correlations with each other, so it is reasonable to per-
form rutting tests using AAP to evaluate anti-rutting performance of GAC mixtures under
different temperatures and loading pressures. The following experiments are based on this
assumption.
The viscosity of modified asphalt using GAC mixtures decreases as the temperature becomes
higher because of the decrease of the strength of asphalt mixtures (CHEN Huaxin, ZHOU
Yan, WANG Binggang 2009). The asphalt used in GAC mixtures is mainly SBS modified
asphalt. So it is necessary to perform experiments to discover the relationship between the
viscosity of SBS and temperature.
In order to imitate the temperature in southern areas of China like Guangdong Prov-
ince, the experimental temperatures were set as 55 °C, 60 °C, 65 °C and 70 °C. The modi-
fied asphalt tested was SBS(I-D). The equipment used for the test was the widely used
Vacuum Capillary Viscometer Tester (VCVT) utilized in previous research (LI Lihan,
GENG Han, SUN Yan-na 2010). Note that three different models of specimen were
used in this experiment. The results of viscosity of SBS(I-D) in relation to temperature
is shown in Table. 5.
It is indicated from Table. 5 that the viscosity of modified asphalt has a fierce decrease
when experimental temperature increases.
Then the results of rutting test were combined with the results in Table. 5 to determine
the relationship between DS and viscosity of modified asphalt. The results of CWG test and
AAP test are shown in Fig. 6.
It can be indicated from two curves that the viscosity of modified asphalt decreases when
DS is decreasing along with the increase of temperature. The relationship between DS and
viscosity of modified asphalt can be described as the following equation:
DS = a3 b3 (7)
578
Table 5. Correlation between the viscosity of SBS (I-D) and temperature.
Viscosity Average
Temperature °C Model Times Pa * s Pa * s
Figure 6. Relationship between DS and modified asphalt: (a) CWG test (b) AAP test.
Experiments aiming to evaluate anti-rutting performance of GAC mixtures under heavy load-
ing pressures were performed. The size of specimens was the same as which was used in AAP
in previous experiments. Four groups of tests were conducted with 3 specimens in each group.
In order to imitate the heavy loading pressure, the pressures of the experiments were set as
0.7 MPa, 0.9 MPa, 1.1 MPa and 1.3 MPa while the temperature was set as 65 °C. Note that
0.7 MPa was the standard axle load set in Specifications (JTG E20-2011). The number of load-
ing time was 8000 and the loading speed was the same. Table. 6 shows the amounts of loading
pressures and Table.7 shows the results of DS according to different loading pressures.
579
The relationship between DS and loading pressure can be fitted as the following
equation:
DS = a4e 2 6467 P
+ b4 (8)
580
Table 8. Relationship between DS and temperature and loading pressure.
Temperature °C
DS
(turn/mm)
Pressure MPa 0.7 0.9 1.1 1.3 Pa * s
tively. When comparing with the drop of DS due to the increase of experimental temperature,
it is concluded that the drop of DS of GAC mixtures mainly results from the rise of tempera-
ture (5 °Ceach time), and the rise caused by changes of temperature is much less than that
caused by the rise of loading pressure (0.2 MPa each time).
ln( ) = a4 ln(
l ( ) + b4 ln(( 4 4 ) + m ln ln( ) + n (9)
5 CONCLUSIONS
In this paper, experiments were conducted to evaluate anti-rutting performance of GAC mix-
tures under high temperature and heavy traffic loads. The tests based on both CWG test and
AAP test were performed, and results from these two tests were compared. The conclusions
are drawn as follows:
1. The results from both two rutting tests suggest that the anti-rutting performance of GAC
mixtures decrease remarkably when increasing temperature or loading pressures. The
results of correlation analysis indicate that AAP test can replace the CWG test to evaluate
anti-rutting performance.
2. The comparison of influences of temperature and loading pressures on DS suggests that
the drop of DS resulting from increase of temperature by 5 °C is much larger than that
resulting from increase of loading pressure by 0.2 MPa.
3. The fitting results of DS under high temperature and heavy loads suggest that both the
increase of temperature and loading pressure can lead to fierce decrease of anti-rutting
performance of GAC mixtures.
REFERENCES
Cao Dongwei, Liu Qingqun, Tang Guoqi. Porous Asphalt Pavement [M]. 2010.
Chen Huaxin, Zhou Yan, Wang Binggang. Dynamic Mechanics Performance of Aged SBS Modified
Asphalt. Journal of Chang’an University (Natural Science Edition). Vol.29, No. 1, Jan. 2009.
Li Lihan, GENG Han, SUN Yan-na. Evaluation Method and Indicator for Viscosity of High-Viscosity
Asphalt. Journal of Building Materials. Vol.13, No. 3, Jun. 2010.
Lv Rui, Research on Application of Regional Grading GAC-16C Asphalt Concrete in Jiangmen-
Zhaoqing Expressway, Technology of Highway and Transport Oct. 2013 No. 5.
581
Pramesti, F.P., A.A.A. Molenaar, M.F.C. van de Ven, The Prediction of Fatigue Life Based on Four
Point Bending Test, The 2nd International Conference on Rehabilitation and Maintenance in Civil
Engineering, Procedia Engineering 54 (2013): 851–862.
Pramesti, F.P., A.A.A. Molenaar, M.F.C. van de Ven, Fatigue Cracking of Gravel Asphalt Concrete:
Cumulative Damage Determination, 7th RILEM International Conference on Cracking in Pave-
ments, pp. 739–749.
Standard Test Methods of Bitumen and Bituminous Mixtures for Highway Engineering 2011.
Xian Hong-wei, Research of Grading Design Method for Asphalt Mixture of GAC-20, Science
Technology and Engineering, Vol. 10, No. 26, Sep. 2010.
Yang Qi, Selection of Scheme for Cement Concrete Pavement Major Maintenance of Fokai Express-
way, Guang Dong Gong Lu Jiao Tong, vol. 3, No. 29, Mar. 2009.
Zhang Shoufu, Application of GAC-25 Graded Asphalt Mixtures under Heavy Loads. China Highway.
2011, 3: 118–120.
Zhang Yongsheng. Research on Asphalt Pavement Structure Design of WuShen Expressway In Guang
Dong Province[D]. Research Institute of Highway M.O.T. 2015, 4.
582
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Rutting is one of the most serious distresses of semi-rigid asphalt pavements.
In this research AASHTOW are pavement ME design software was used for rutting prediction
of semi-rigid pavements. Local calibration was done first by taking into account local materials,
traffic information, and environmental conditions. A total of 32 sites representing all common
mixture types of Jiangsu province were selected to complete the calibration and validation.
Then the calibrated software was used to identify sensitive factors for rutting prediction. Criti-
cal inputs needed for the calibration and sensitive analysis were tested or collected from local
project level PMS database. Modulus of semi-rigid base was back calculated from deflection
basin data. Calibration coefficients for each typical structure and the optimal coefficients for
this region were identified respectively. Results also reveal that thickness of HMA layers and
traffic inputs are more sensitive to the rutting prediction than void content of HMA mixtures.
1 INTRODUCTION
The Mechanistic Empirical Pavement Design Guide (MEPDG) was developed under the
National Cooperative Highway Research Program (NCHRP) Project 1–37 A (NCHRP
2004) to overcome the deficiencies and limitations of the empirical design procedures in the
1993 AASHTO guide for design of pavement structures. New build (2.2) for the ME Design
system has released on August 12th 2015. In this release new semi-rigid pavement is included.
A semi-rigid pavement is composed of a flexible layer (e.g., HMA) and a rigid layer (e.g.,
Cement-Treated Base [CTB], cement stabilized base [CSB], Rolled-Compacted Concrete
[RCC], or lean mix concrete). This type of pavement is widely applied in China. According
to the performance monitoring results, permanent deformation, or rutting, also commonly
appeared on semi-rigid pavement, as flexible base HMA pavement. As to semi-rigid pave-
ments, rutting is mostly confined in the asphalt layers while in flexible pavements rutting is
the total permanent deformation of the unbound or bound base layers and upper asphalt
layers. The empirical permanent deformation transfer functions used in the ME Design are
nationally calibrated using design inputs and distress data largely from the national Long-
Term Pavement Performance (LTPP) database. Although this effort was comprehensive, fur-
ther calibration and validation studies to suit local conditions are highly recommended by
the NCHRP Project 1–37 A as a prudent step in implementing a new design procedure that
is so different from the previous procedures. Thus, it is necessary to calibrate the pavement
ME design performance models for implementation in semi-rigid pavement by taking into
account local materials, traffic information, and environmental conditions.
Many research studies have been conducted to demonstrate the local calibration procedure
after the release of each version of the MEPDG Software. Over 13 states in the US have
conducted or plan to undertake local calibration studies for their own pavement conditions,
including Indiana, Wisconsin, North Carolina, Texas and so on.
Limited studies on semi-rigid pavement performance prediction model calibration.
A significant finding derived from previous studies which is relevant to this study is that
rutting predictions for flexible pavement could be improved through local calibration.
583
The primary objective of this research study is to improve the accuracy of Pavement
ME Design pavement performance predictions for Jiangsu pavement systems through local
calibration of pavement ME Design Version 2.2 performance prediction models, and then
use the calibrated software to identify the factors mostly influencing the prediction results.
2 SITE INFORMATION
To develop the database for conducting pavement ME Design local calibration, representa-
tive pavement sites across Jiangsu province, China were selected considering traffic levels and
typical structure types. A total of 8 semi-rigid pavement sections were selected from a list
of potential highway segments. The total mileage of the sections is 107.44 kilometers. The
rutting is surveyed twice per year every 10 meters interval. All data are saved in pavement
management system of Jiangsu.
Figure1 presents the average annual daily truck traffic (AADTT) distributions for the site.
Figure 2 shows the pavement structures of selected 8 sections. In Figure 2, numbers (13,
20, and 25) represents nominal maximum aggregate sizes; SMA is the abbreviation of stone
matrix stone; AC is the abbreviation of asphalt concrete; SUP is the abbreviation of super-
pave; CTB is the abbreviation of cement treated base. The star represents that the binder used
in the mixture is modified asphalt.
584
3 RUTTING PREDICTION MODEL
The approach presented in the Design Guide is based upon incremental damage. The damage
or rutting is estimated for each subseason at mid-depth of each sublayer within the pavement
system. To estimate the permanent deformation of each individual sublayer, the system veri-
fies the type of the sublayer, and computes the plastic strain accumulated at the end of each
subseason. The overall permanent deformation for a given season is the sum of permanent
deformation for each indifidual layer and is mathematically expressed as Equation (1).
n
RD = ∑ε pi h i (1)
i =1
εp
= a1Ta2 N a3 (2)
εr
εp = Accumulated plastic strain at N repetitions of load (in/in); εr = Resilient strain of the asphalt
material as a function of mix properties, temperature and time rate of loading (in/in); N = Number
of load repetitions; T = Temperature (deg F); ai = Non-liner regression coefficients.
While statistical relationships evaluated from laboratory repeated load tests on asphalt
mixture, were found to be reasonable; field calibration factors, βri, were necessary to ascertain
the final field distress model. The final asphalt rutting equation implemented in the Design
Guide is expressed as Equation (3).
εp
= k1β 110 3 15552
T1.734 βr 2 N 0.39937 βr 3 . (3)
εr
In this equation it can be observed that a depth parameter “k1” has been introduced to
provide as accurate a rut depth prediction model as possible.
The calibration was done in a step-by-step basis. First the two-power calibration factors
(βr2 βr 3 ) on the asphalt model were found from the simulation runs. The default value
were βr2 = 0.9, and βr3 = 1.2 . The CTB base is supposed to generate no permanent deforma-
585
tion. The last calibration factor on the asphalt (which is a direct multiplier) βr1 was obtained
through optimization. From the previous research, the calibration factors are very close to
each other and that the difference in the granular base factor does not significantly impact
the asphalt calibration factor to any significant degree.
According to equation (4), it is obvious that βr1 is a direct multiplier, which indicates that
this factor has liner effect on the permanent deformation prediction model. It also indicates
that βr1 can be calibrated outside the software. The other two calibration factors (βr2 βr 3 )
need to be calibrated by running the software. The factors of βr2 βr 3 represent the influ-
ences of temperature and axle loading on pavement systems, and thus they need to be cali-
brated by running the prediction process.
The calibration process is divided into two steps: first using ME design to predict the per-
manent deformation under different combination of βr2 βr 3 . Second optimize the value
of βr1 to minimize the sum of standard error (SSE) between predicted and measured values.
The combination of βr2 βr 3 are initially determined by their possible ranges. It is
widely agreed that βr2 βr 3 vary within [0.8, 1.2]. If increasing 0.1 for every try, there are
totally 25 combinations, shown in Table 1.
The value of βr1 was kept constant for each combination while βr2 βr 3 was changed
for each calculation process. Programming solver was utilized to seek the best βr1 for each
combination of βr2 βr 3 . As a result, 25 combination of ( βr1 βr 2 and βr 3 ) were initially
extracted. The selected 8 sections belong to 4 main highways in Jiangsu, namely Yanjing
Highway, Lianxu Highway, Huning Highway and Ninghang Highway. For each highway, the
optimal combination of ( βr1 βr 2 and βr 3 ) was also programming solved to minimize the SSE
between predicted and measured values. The summary of calibration results is presented in
Table 2.
The selected 8 sections (belonging to 4 highways) gathered most representative pavement struc-
tures applied in Jiangsu. There were totally 81 rutting points picked for the rutting calibration,
and thus there were 25*81 = 2025 couples of measured and predicted rutting values. Program-
586
Figure 3. Comparison of calibration results.
ing solving all these data together, a combination of ( βr1 βr 2 and βr 3 ) could be obtained as the
calibration factors for Jiangsu district. The calibration results are exhibited in Figure 3. These
results show that SSE of district factors combination is much larger than SSE of single highway,
but it is still smaller than 20 mm2. βr1 of district combination is smaller than any single highway
combination, while βr2 of district combination is larger than any single highway combination.
Two highways adopt the default 0.9 as the βr2 value. The value of βr3 is close to each other.
5 SENSITIVITY ANALYSIS
Previous researches show that rutting depth is sensitive to axle loading, Temperature, void
contents and thickness of asphalt layer. This part is to research how sensitive the calibrated
rutting model is to the mentioned elements.
587
Figure 4. Predicted values and measured values.
588
Table 4. Traffic levels in Jiangsu province.
Low <150
Light 150∼400
Medium 400∼1200
Heavy 1200∼3000
Extra heavy >3000
Factors
1 Medium 3 12
2 Medium 5 18
3 Medium 7 24
4 Heavy 3 18
5 Heavy 5 24
6 Heavy 7 12
7 Extra heavy 3 24
8 Extra heavy 5 12
9 Extra heavy 7 18
1 5.13 8.96
2 2.47 4.09
3 1.60 2.54
4 4.78 8.13
5 2.62 4.47
6 12.29 22.09
7 5.13 8.08
8 17.76 29.11
9 6.99 10.93
the common thicknesses of asphalt layer in Jiangsu are near 180 mm, not exceeding the range
of [160 mm, 200 mm], and thus the calibration factor βr1 is not feasible enough for over thin
or over thick conditions.
589
Table 7. Intuitive analysis results of the orthogonal experiment.
A B C Bias A B C Bias
depth prediction, while void contents of asphalt mixture has the smallest influence, and traf-
fic loading has a medium impact.
6 CONCLUSIONS
This paper studied the permanent deformation model of semi-rigid pavement via ME Design
software. The nationally calibrated permanent deformation model is re-calibrated for Jiangsu
province, China. In the process of calibration, 8 sections, belonging to 4 different highways,
were selected and another 4 highways in Jiangsu were chosen to validate the calibrated
results. The local calibration factors of rutting prediction model were identified to improve
the accuracy of model prediction. Combinations of calibration factors for selected highways
and for Jiangsu district were gained respectively. The validation results using district fac-
tor combination show that the precision of calibrated prediction model has been promoted
to an acceptable level. The sensitivity of rutting prediction model was then analyzed using
590
orthogonal experiment. The test results of orthogonal experiment were intuitive analyzed
and variance analyzed to identify which factor influence the rutting prediction most within
traffic loading, void contents and thickness of asphalt concrete layer. Both the intuitive anal-
ysis and variance analysis show that thickness of asphalt layer most significantly affects the
rutting prediction. Traffic loading has a weaker influence in rutting prediction than thickness
of asphalt layer, but it still has a closer relation to rutting prediction than void contents of
asphalt mixture.
REFERENCES
Chen, A., Feng Y. 2001. Multi-factor simulation sensitivity statistical analysis technique. Statistical
Research (in Chinese). 2001(9):54–56.
Hoegh, Kyle, Lev Khazanovich, and Maureen Jense. “Local calibration of mechanistic-empirical pave-
ment design guide rutting model.” Transportation Research Record: Journal of the Transportation
Research Board 2180.1 (2010): 130–141.
Kim, S., Ceylan, H., Ma, D. and Gopalakrishnan, K. 2014. Calibration of Pavement ME Design and
Mechanistic-Empirical Pavement Design Guide performance Prediction Models for Iowa Pavement
Systems. Journal of Transportation Engineering. 140(10): 1–13.
Li J, Pierce L M, Uhlmeyer J. 2009. Calibration of flexible pavement in mechanistic-empirical pavement
design guide for Washington state. Transportation Research Record: Journal of the Transportation
Research Board. 2095(1): 73–83.
Muthadi, Naresh R., Kim, Y.R. 2008. Local calibration of mechanistic-empirical pavement design guide
for flexible pavement design. Transportation Research Record: Journal of the Transportation Research
Board 2087.1 131–141.
591
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Hao Tang
Southeast University, Nanjing, Jiangsu, P.R. China
Haidong Kuai
Highway Maintenance Technology, National Engineering Research Centre, Beijing, P.R. China
Xiaoming Huang
Southeast University, Nanjing, Jiangsu, P.R. China
ABSTRACT: Chinese highway maintenance technologies were researched from the perspec-
tive of energy consumption. The boundary of maintenance life cycle was defined as follow:
minerals extraction as the beginning and the reopening to traffic as the end. The energy con-
sumptions of highway maintenance were studied with the application of Life Cycle Assess-
ment (LCA), which was divided into 4 stages: raw materials production, mixture production,
on-site construction and materials transport. This paper put emphasis on researching calcula-
tion model of mixture production and on-site construction associated with Chinese highway
maintenance. The Energy Consumption Index of Maintenance (ECIM) was put forward in
accordance with Chinese standards GB/T 2589–2008. Different kinds of Chinese maintenance
technologies were calculated with the model and the results were compared with European
achievement. It shows that the ECIM disparity between Chinese maintenance technologies
and European ones is within 4% on condition that subjective factors are left out.
1 INTRODUCTION
The main mission of the Chinese highway industry has changed from “both construction and
maintenance” to “maintenance first”. Certainly, a great quantity of energy consumption will
take place in the large and medium-sized maintenance projects associated with highway pave-
ment and base. It is necessary to establish practical energy consumption models and evaluation
indexes of Chinese highway maintenance technologies, which can help evaluate the energy-
saving effect of highway maintenance technologies and optimize the selection of scheme. For
decades, researchers in developed countries have carried out a large number of studies on the
energy consumption associated with highway maintenance. In these studies, energy consump-
tion models and indexes were established in terms of the whole life cycle of maintenance
projects (Chehovits & Galehouse 2011). Stripple (2001) made detailed analysis on the energy
consumption of frequently-used material for highway construction and maintenance during
the production stage, forming a pretty distinct data inventory, and compared the energy con-
sumptions and carbon emissions of two pavements (asphalt & cement), both of which applied
cold or hot production technologies. Based on the pre-existing research achievements, Dor-
chies et al (2005) using the authoritative data from 5 research agencies as basic parameters, set
up the whole life cycle energy consumption calculation model for highway construction and
maintenance, which was widely acknowledged and quoted by Western scholars.
Most of Chinese researchers made quantitative investigation on energy consumption of
certain stage of highway construction or maintenance. Cheng (2010) calculated the mixture
production energy consumption by checking key temperatures and oil consumption during
593
production of WMA and HWA respectively. To study on the energy consumption during the
highway construction and maintenance, several Chinese researchers, for the first time, had
used the Life Cycle Assessment (LCA) approach in recent years. Pan (2011) made the quanti-
tative analysis of energy consumption of expressways within its whole life cycle associated with
3 different kinds of pavements, with which the calculation model used focuses on the energy
consumption of construction stage. By using the quota method, Yang (2012) put forward
energy consumption calculation models for the asphalt pavement construction and mainte-
nance projects, which is convenient but not accurate. In this paper, the energy consumption
calculation models and evaluation indexes of maintenance technologies for China’s highway
were established based on the process-oriented LCA. In comparison with the research achieve-
ments of western developed countries, the correctness of the calculation model was proved.
2 RESEARCH METHOD
Eog
EU = ×A +E f (1)
a Po + Pg fg a r so
594
where EUa is the comprehensive energy consumption per unit mass of standard asphalt in
China; Eog is the comprehensive energy consumption in China’s oil and gas fields during a
year; Po is China’s oil output during a year; Pg is China’s natural gas output during a year; fg
is the factor for natural gas conversion into crude oil, and its recommended value is 1/1255;
Aa is the distribution coefficient of daughter products; Er is the extraction energy consump-
tion per unit mass of asphalt, and the standard oil is set as the uniform unit; fso is the factor
for standard oil conversion into the standard Coal Equivalent (SCE).
In comparison with standard asphalt, there are two more stages in the whole life cycle
of modified asphalt and emulsified asphalt: the production of additives and processing like
shearing. The relevant formulas were derived and respective results were got with the same
principle as the first two stages, which may not be detailed for brevity.
Cement is widely used as hydraulic materials cementing mixture to increase strength in the
maintenance of Chinese highway base. The production stages of cement are divided into:
(1) the production of cement raw meal, (2) the production of clinker, and (3) the production
of cement. At present, the energy consumption analysis on life cycle of portland cement,
which is commonly used in highway projects, is pretty mature in China. In fact, the Chinese
standard, The Norm of Energy Consumption per Unit Products of cement, has set the quota
of energy consumption in each stage. The paper use the data from China Energy Statistical
Yearbook, which is provided by China cement association.
There are a variety of aggregates, produced in different ways, using in China’s highway
projects. In general, the production stage of aggregate includes two phases, the first one is
stone quarrying and the second includes crushing and screening, with chemical energy and
electrical energy consumed respectively. On the basis of surveys conducted in a large number
of stone mining and processing plants in China, the comprehensive energy consumption per
unit product of aggregate was obtained through calculation.
Parameters were collected from the big data (Department of Energy Statistics of PRC
2012, Ministry of Environmental Protection of PRC 2008) and field surveys, and then the
numerical values of comprehensive energy consumptions for all kinds of raw materials were
obtained. Figure 1 shows the result (The unit kgce refers to kilogram of standard coal equiva-
lent). The energy consumptions for raw materials’ production shall be equal to the sum of all
kinds of raw materials.
595
2.2.2 Production of mixture
The production of mixture means processing the raw material to produce mixture in line with
the requirements of highway maintenance projects. According to field surveys in Jiangsu
Province and Beijing, the production of mixture can be divided into three stages: (1) the
pretreatment of Reclaimed Asphalt Pavement (RAP), which includes the milling of original
pavement; (2) heating of old material and raw material; (3) the mixing of mixture. To achieve
the modularization of models, the mixture production of in-place recycling technology is
specified to belong to the on-site construction.
It is noticeable that to pursue statistical significance, the average annual comprehensive
heating oil consumptions per thousand tons in mixing plants at the standard heating temper-
ature were adopted as the field research data. However, there is great difference between the
actual heating temperatures of a variety of technologies, such as warm mix technology and
hot mix technology. Therefore, the concept of correction factor Kf for heating temperature
of comprehensive heating oil consumption was introduced to take the impacts of different
heating temperatures on oil consumptions into consideration. Assuming the specific heat
capacity of raw material as the constant value during the process of temperature change and
with the application of the definition equation of specific heat capacity, the formula of this
correction factor was derived:
t −t
Ki = i n (2)
t −t
si n
where Ki is the correction factor for heating temperature of the type i raw material; tn is the
normal temperature of raw materials before putting into the mixing plants, generally set to
25oC or 20oC; ti is the actual heating temperature of the type i raw material; tsi is the standard
heating temperature of the type i raw material.
In the process of the mixing, the energy consumption in the mixing plant is generated from
the energy conversion, that is, the chemical energy of fuels and the electrical energy were con-
verted into heat energy and mechanical energy. Hence, the formula for energy consumption
of mixture production was derived:
⎛ P ⎞
N
R
Emp ∑ ⎝ Ki × Fsii × fhi × 100i ⎠ + Ems × fee + E pt × 1003
i =1
(3)
where Emp is the comprehensive energy consumption per unit mass of mixture during the
mixture production process; N is the type number of raw materials input in the system; Fsi
is the comprehensive heating oil consumption per unit mass of the type i raw material at the
standard heating temperature; fhi is the factor for the fuel’s conversion into SCE, and the fuel
is consumed by the type i raw material during the heating phase; Ems is the comprehensive
electricity consumption per unit product of mixture that produced in mixing plant; fee is the
factor for electric energy’s conversion into SCE; Ept is the comprehensive energy consumption
per unit mass of the pretreatment of RAP; R3 is the proportion of RAP in the mixture.
596
F pav
E pav = (4)
v pav ×w pav ×t pav ×dm
where Epav is the comprehensive energy consumption per unit mass of mixture during the
paving process; Fpav is the comprehensive energy consumption of paver or in-place recycling
machines; vpav is the actual operational speed of paver or in-place recycling machines; wpav is
the actual paving width of the paver or in-place recycling machines; tpav is the actual paving
thickness of paver or in-place recycling machines; dm is the density of mixture.
It is difficult to count the energy consumption during the rolling process. In the light
of China’s technical specification for construction (Ministry of Transport of China 2004)
and the engineering practices, the calculation model for energy consumption of rolling was
established according to paving conditions. Specifically, the model made the assumption that
synchronizing the beginning and ending time of the paving and rolling so as to avoid the
complicated counting on the numbers of rolling, thus improving the computational efficiency
and accuracy of the calculation model. The specific way is to calculate the energy consump-
tion per unit on the basis of the time of paving, the number of rollers and the comprehensive
fuel consumption per unit time of each roller. The calculation model is as follow:
M
∑ ( Fri × fri )
Ero = i =1 (5)
v pav ×w pav ×t pav ×dm
where Ero is the comprehensive energy consumption per unit mass of mixture during the roll-
ing process; M is the number of rollers; Fri is the comprehensive fuel consumption per unit
time of the type i roller; fri is the factor for diesel’s or petrol’s conversion into SCE, and the
diesel or petrol was consumed by the type i roller.
Haul distance
Transport km
597
paper selected multiple maintenance projects across the nation as samples and unified the
values of three main haul distance. The specific way is counting the average value of a certain
haul distance for calculation. Table 1 shows the result.
Another important influence factor on the energy consumption during transportation is the
comprehensive energy consumptions that were generated by vehicles. This paper researched
the data of load of vehicles and their comprehensive fuel consumption, and then made sta-
tistical analysis, which showed that: (1) Generally, transport vehicles powered by diesel, the
comprehensive fuel consumption of which mainly affected by the load and running speed,
and thus made a large difference. Take the speed of 60 km/h for instance, the fuel consump-
tion per hundred kilometers increases 6.4 L with an increment of every 10 tons’ load; (2)
Overloading is common in the vehicles transporting in China and thus their comprehensive
fuel consumptions is higher than the nominal ones, but the comprehensive fuel consumption
per unit load is lower. To sum up, this paper did normalization of the comprehensive energy
consumptions of vehicles: assign a middle value to the load of vehicles as 50 tons and the
comprehensive fuel consumption is close to 50 L/100 km, and thus the comprehensive energy
consumption of transporting vehicles is 0.013 kgce/ton-km. Through field monitoring the
transport of materials at various highway maintenance projects, the universal formula for
energy consumption of materials’ transport is as follow:
Etr
⎛2 D
Fveh × ⎜
( D )×( −R ) R3 + 2D3 (1 R2 ÷100)×( − ÷ ) + D ⎞⎟
100 ×( + R ÷ ) (+ ÷ )
(6)
⎜ 4⎟
⎝ ⎠
where Etr is the comprehensive energy consumption per unit mass of mixture during the
transport of materials; Fveh is the comprehensive energy consumption of vehicles; D1 is the
haul distance between construction site and mixing plant; D2 is the haul distance between
material yard and mixing plant; D3 is the haul distance between new materials and material
yard; D4 is the haul distance between mixing plant and construction site; R1 is the asphalt-
aggregate ratio; R2 is the cement content; R3 is the RAP ratio.
After measurement and calculation, it is found that the energy consumption of subsidiary
transport and total energy consumption differ by four orders of magnitude. From this, for
the purpose of increasing the computational efficiency of calculation model, energy con-
sumptions of subsidiary transports including the operation of road sprinkler and the trans-
port of modifiers are omitted.
598
Table 2. Factor for energy resources conversion into SCE.
To verify the correctness of calculation models, this paper did the energy consumption quan-
tification on the large and medium-sized maintenance projects of China’s high-grade high-
ways in recent 3 years, and the energy consumption database of maintenance technologies in
China was established. Figure 3 shows the quantitative result and European research achieve-
ment (C represents Chinese, E represents European).Summaries were obtained after compar-
ing Chinese and European energy consumption databases of maintenance technologies.
1. With regard to the same highway maintenance technology, the ECIM value of Chinese
project differ from that of European one because of subtle distinctions including energy
599
Figure 3. The energy consumption of Chinese and European highway maintenance technologies.
4 CONCLUSIONS
The analyses reveal that the results of calculation models proposed in this paper coincide with
those of European database, which indirectly testifies the correctness and practicability of the
600
calculation model; and that maintenance technologies all over the world have general character
in the area of whole-life-cycle energy consumption, and thus decision-makers and construction
enterprises from different countries can adopt more advanced maintenance technologies to
reduce energy consumption by boosting academic exchanges and learning from each other.
The energy consumption calculation model for Chinese highway maintenance were estab-
lished using the process-oriented LCA, which makes the model accurate. Besides, algorithm
was optimized according to practical condition of engineering in China. The database for
energy consumption of raw materials used in Chinese highway maintenance was established
in this paper, which could be served as a reference for domestic and overseas counterparts to
carry out relative studies. The concept of Energy Consumption Index of Maintenance (ECIM),
introduced under the Chinese standard in this paper, is practical in evaluating the energy con-
sumption intensity of highway maintenance technology. The results of energy consumption
model in this paper approach to those of European researches, especially in terms of the energy
consumption distribution. In the further research, carbon emission and harmful gas emission
of highway maintenance could be studied based on energy consumption calculation model.
REFERENCES
Chehovits, J. & Galehouse, L. 2010. Energy usage and greenhouse gas emissions of pavement preservation
processes for asphalt concrete pavements. First International Conference on Pavement Preservation.
Newport Beach CA, USA: California Department of Transportation.
Cheng, L., Yan G.J. & Chen D.Z. et al. 2010. Quantitative investigation on energy conservation and
emission reduction related to warm mix asphalt (WMA). Chinese Journal of Environmental Engineer-
ing 4(9): 2151–2155.
Dorchies, P.T., Chappat, M. & Bilal, J. 2005. Environmental Road of the Future: Analysis of Energy
Consumption and Greenhouse Gas Emissions. Fiftieth Annual Conference of the Canadian Technical
Asphalt Association (CTAA). Victoria British Columbia, Canada: Polyscience Publications.
Jiang, Y. & Yang, X. 2010. Electricity Equivalent Application in Energy Analysis. Energy of China 32(5):
5–11.
Lu, H.T. & Yang, W.A. 2011. Statistical Model of Highway Life-cycle Energy Consumption. Journal of
Wuhan University of Technology (Transportation Science & Engineering) 35(5): 1044–1048.
Ministry of Environmental Protection of the People’s Republic of China. 2008. Cleaner Production
Standard Petroleum Refinery Industry (Semi-Asphaltic Flux) (HJ 443–2008). Beijing: China Environ-
ment Science Press.
Ministry of Transport of the People’s Republic of China. 2004. Technical Specifications for Construction
of Highway Asphalt Pavements (JTG F40–2004). Beijing: China Communications Press.
National Bureau of Statistics of the People’s Republic of China. 2012. China Energy Statistical Year-
book. Beijing: China Statistics Press.
National Development and Reform Commission & Standardization Administration of the People’s
Republic of China. 2008. General Principles for Calculation of Total Production Energy Consumption
(GB/T2589–2008). Beijing: Standards Press of China.
National Development and Reform Commission of the People’s Republic of China. 2013. The Norm
of Energy Consumption per Unit Products of Cement (GB 16780–2012). Beijing: Standards Press of
China.
Pan, M.P. 2011. The Methodology Research and Application on Energy Consumption and Carbon
Emissions of Highway Bases on the Life Cycle Assessment. Master’s thesis, South China University
of Technology.
Santero, N.J., Masanet, E. & Horvath, A. 2011. Life-cycle Assessment of Pavements. Part I: Critical
Review. Resources Conservation & Recycling 55(9): 801–809.
Stripple, H. 2001. Life Cycle Assessment of Road: A Pilot Study for Inventory Analysis (Second Revised
Edition). Sweden: Swedish Environmental Research Institute IVL.
Wang, X.F. 2013. Study of Asphalt Pavement Preventive Maintenance Timing and Strategy Based on
Life Cycle Analysis. Master’s thesis, Lanzhou Jiaotong University.
Yang, B. 2012. Quantitative Method and Evaluation System of Energy-Saving and Emission Reduction
for Asphalt Pavement. PhD Diss., Chang’an University.
601
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
The political arena of South Africa has been pursuing the implementation of Carbon Tax
following South Africa’s commitment to reduce greenhouse gas (GHG) emissions with
34 percent by 2020 and 42 percent by 2025, below business as usual. The Carbon Tax is
aligned with the 2011 National Climate Change Response Policy, as outlined in South Afri-
ca’s Intended Nationally Determined Contributions. This has been submitted to the United
Nations for the Conference of Parties 21 of the United Nations Framework Convention on
Climate Change, held in Paris in 2015. It seems certain that this implementation will proceed
and National Treasury having published the Draft Carbon Tax Bill for public comment, fol-
lowing on the announcement made by the Minister of Finance to implement this through
government policy by early 2016.
Despite lobbying from certain interest groups, that this will drive a green-economy, con-
cerns have been raised about the vulnerability of South Africa if exposed to international
carbon taxation practices. Alton et al. (2014) has demonstrated that national welfare and
unemployment will decrease by 1.2 and 0.6 percent, respectively, but these analysis hinges
on many factors including no change in partner trading behaviour over the implementa-
tion phases of carbon tax and if recycled carbon tax is distributed to promote growth and
equity.
Similar to other developing countries such as India, the drive for taxation of fossil fuels
might be more related to revenue and macro-economic considerations and concerns about
losing competitiveness, especially in our export markets seems justified. This is exasperated
by the fact that South African labor-inertia revolves around energy intensive-sectors with
higher energy prices having the largest effect on poorer households.
Notwithstanding the arguably conflicting consequence the Bill could have on the socio-
economic impact of a developing country, the absence of transparency and clarity allows
for discord in the implementation of the Draft Bill. Louw (2013) highlights the prominent
issues, but the most confounding in its absence is guidance in the accounting and auditing of
the quantification of the emissions on which tax will be payable. In addition, the Draft Bill
603
also alludes to the provision of a carbon offset allowance, but, if the intended objective is to
reduce South Africa’s emissions, these offsets lead to contradiction, as large GHG emitters
will be free to emit into perpetuity.
The absence of a holistic comprehensive accounting of emissions is exemplified by refer-
ring to one of the Bills aims in reducing production of carbon intensive products. It reflects
the underestimation in understanding the complexity of the issue (including a full account-
ing of the emission cycle of the product, not necessarily only in the production phase) and
makes viewing the Bill close to implementation-ready, unrealistic.
In the transportation industry, the last matter is the most concerning in lieu, that globally,
the sector is estimated to contribute to about 25% of carbon dioxide (CO2) emissions whilst
as part of this sector, road transport contributes 80% (WRI 2010). It has also been reported
by AASHTO (2011), that the transportation sector contributes to 22% of global energy con-
sumption and 25% of fossil fuel use. As one of the greatest emitters towards environmental
burdens, it is clear that the road industry in South Africa has a pivotal role to play in guiding
government in a way that would avoid the current pitfalls and implement risk mitigation.
It is these issues that underpin the objectives of this paper in order to understand the
South African Road Sector’s way forward in sustainable accounting. To achieve this goal,
the objectives of this paper is to illustrate the importance in comprehending the full environ-
mental burden (in this case specifically CO2 emissions), but at the same time contextualize
the validity of prioritizing which environmental burdens are most important in the South
African sphere. Lastly, a comparison of Life Cycle Analysis (LCA) and rating tools are made
including the way forward for sustainable measurement discussed.
2.1 The risks in not understanding the full life cycle of environmental burdens
Figure 1(a) graphically shows the phases (materials extraction and production, construc-
tion, use, maintenance, use and end of life) of the pavement life cycle as reported by many
researchers. The activities to be measured, considering output flows causing environmental
burdens, have been adapted from Santero (2011a, b), through the inclusion of additional
activities, various inputs and outputs linked to the phases, and the interlinking of transporta-
tion activities, an important aspect to consider when measuring environmental loads.
One consideration for the measurement of sustainability is LCA and includes the account-
ing of these environmental input and output flows through a defined scope and system goals
as governed by the ISO14040standard. This is in contrast with the essence of the Bill which
compartmentalises certain aspects of the life cycle for the product under measurement.
604
The misunderstanding in measuring environmental burdens is not unique in reference to
the Bill as shown by other examples such as the reporting by the South African Institution of
Civil Engineers (2010) saying: “that the road surfacing industry responded to a presidential
call to reduce the country’s greenhouse gas emissions by vigorously pursuing bituminous
emulsions rather than hot mix materials”. Lower production temperatures and reduced bitu-
men content are certainly beneficial attributes considering the high environmental burden
outflows hot mix asphalt (HMA) produce during the production phase, but ignores entirely
the consideration that this might lead to varying rehabilitation measures or possibly change
road roughness degradation curves, consequently ending in increased fuel consumption.
This could lead to a potential expensive environmental miscalculation as studies have
reported a reduction of up to 10% in fuel consumption with the decrease in surface texture
(Beuving et al. 2010, NCAT 2015).
To articulate the risks associated with measuring environmental burdens of isolated parts
of the life cycle, three typical South African pavement structures are analysed as a rudimen-
tary example. Research has shown that the emissions outflow in the operational phase has a
significant role to play. It is noteworthy to mention that the South African situation might
differ as a result of its design philosophy which includes a thin asphalt layer (normally 30 mm
to 50 mm thick), a base from crushed stone and a cemented subbase. This is different to many
developed countries, where pavement designs typically include surface/base layers consisting
of 150 mm HMA layers or even thicker. It is well documented that the environmental burden
of bitumen significantly contributes to the addition of CO2 emissions in the production and
construction phase, meaning that a significantly reduced pavement layer thickness could alter
the comparative outcomes of these results.
A systematic approach is followed in this example and quantification of the activity (1)
production and construction (2) maintenance and (3) fuel consumption in the use phase, with
aggregation of the CO2 emissions outflow of these activities summated in Figure 2.
Figure 2. Accounting of CO2 emissions in the various phases with (a) including three South African
pavement structures (b) the process data needed for accounting and (c) the total CO2 emissions of the
three reported activities.
605
axles (consisting of 80 kN single wheel axle configurations), respectively. For this example a
road width of 10.0 m and 1.0 km in length was assumed.
For accounting of the emissions, the individual materials during manufacturing and con-
struction (crushed and borrow-pit aggregates, cement, bitumen, emulsion and water) inven-
tory results as reported by Chappat and Bilal (2003) have been utilised and further processed
to constitute the various pavement materials used in South Africa, referred to as G1-G7
(ranging from high to low quality granular materials), C3 and C4 (cement stabilised materi-
als) and asphalt (containing 5% bitumen by mass). For further processing inventory results
for laying/placing values by Patrick et al. (2010), Meil (2006) and Chehovits et al. (2010) was
used as their construction activities are similar to that of South Africa.
Special mention is made of Scenario 1 in Figure 2(a), in which double seals are used for
surfacing, as the construction of surfacing seals represents 80% of the current paved network
in South Africa (Van Zyl n.d.). For the accumulation of these construction materials a spread
rate of 22 kg/m2 for chips, a cold binder application rate of 3.0 l/m2 and a fog spray of 60%
emulsion at an application rate of 1.1 l/m2 was assumed.
A typical value of 60 km was used for transporting crushed aggregates/unbound materials
(G1-G4) and a value of 30 km for transport of borrowpit materials (G5-G7). The analysis
included 14 ton trucks (0.12 CO2 kg/ton/km) for transportation of materials from borrow-
pits and 20 ton trucks (0.09 CO2 kg/ton/km) for cement, HMA and crushed aggregates from
quarries. The recent Euro standards for vehicle emissions are considerably lower than the
inventory values of Chappat and Bilal (2003) used in this example, but consideration was
given the typical average age of the vehicle fleet in South Africa which exceeds 10 years. The
authors are aware that these results are sensitive to the distance materials are transported,
but deem the reported distances as plausible. A 20 year life cycle was assumed in which two
surface maintenance activities incorporating a double seal application was included.
In the use phase the vehicle volumes for each of the three pavement structures in Figure 2(a)
are ES01: between 20–75 vehicles per day, ES03: 75–220 and ES30: 220–700 as recommended
in the South African Pavement Engineering Manual (2014). The emission factors used for
light vehicles are 0.57 CO2e kg/km and 2.41 CO2e kg/km for heavy vehicles, respectively. As
can be seen from Figure 2(c), two situations were considered i.e. at the lower end, if only
light vehicles are considered to use the road and at mid-range, if the road use includes 80%
light vehicles and 20% heavy vehicles. The respective values illustrated as part of Figure 2(c)
(ranging between values of <7% and <22%) demonstrates how much the CO2 emissions in the
construction and maintenance period is that of the total GHG emissions when all the phases
for a particular scenario is summated.
It is interesting to note, that even for the respective scenarios representing the South Afri-
can case (where seemingly less HMA is used in the pavement layers when compared to many
other countries), result in considerable emissions as a result of fuel consumption in the use
phase. Mitigating actions here would have a considerable effect on the output flows. Even in
consideration of lower volume roads (Scenario 1) the CO2 emissions in the construction and
maintenance phase only contribute to 22% of the emissions in the accumulation of the three
activities that has been considered. A decrease to 7% of the overall contribution is observed
when heavy vehicles are included in vehicle volumes.
In Figure 3, the results are further processed in order to compare the CO2 emissions of a
realistic 5% and 10% saving in fuel consumption (as a result of a lower surface texture) to the
total emissions in the construction and maintenance phase.
The results reveal that the potential saving in fuel consumption because of a reduction of
the surface texture is significant, especially when heavy vehicles are considered as part of the
vehicle fleet. On the other hand these savings are small for the first scenario which typically
presents a low volume road.
This example certainly demonstrates the importance of considering the full life cycle.
Whilst it does not advocate the total lack of regard for the construction phase, it does show
that for some cases other phases of the life cycle result in higher environmental burdens.
The risk of miscalculation of the environmental burdens as implied by the Bill is apparent
when considering the analysis of these pavement structures.
606
Figure 3. CO2 emissions in the construction and maintenance phase when compared to the results of
5 and 10% saving in fuel consumption.
Global 29.3 16.0 11.0 32.0 35.0 Water Water use 47.0
warming
Ozone 2.1 5.0 11.0 5.0 8.0 Eutrophication
depletion
Acidification 3.0 5.0 9.0 6.0 5.0 Acidification
Eutrophication 6.2 5.0 9.0 13 4.0 Toxicity*
Human 5.4/7.5 11.0 6.0 5/8 8.5 Air Acidification 12.0
toxicity
Ecological 7.5 11.0 6.0 5/8 4.0 GWP
toxicity
Fossil fuel 9.5 5.0 7.0 – 11.0 Ozone Depletion
depletion
Habitat 6.1 16.0 6.0 – – Ozone Creation
alteration
Smog 3.5 6.0 9.0 – – Toxicity**
Criteria air 8.9 6.0 10.0 – – Land Acidification 20.0
pollution
Indoor air 3.3 11.0 17.0 – – Toxicity***
quality
Other – – – 29 24.5 Land Use****
Mined Minerals 21.0
*Human and aquatic ** Human ***Human and terrestrial Abiotic Energy
****Transformed and occupied.
poverty line and unemployment estimated to be higher than 25% (Statistics South Africa
2014). Income inequality remains a critical challenge with it being imperative in address-
ing this issue for a sustainable approach. Also, as a consequence the main activities in the
country deriving from cultural activities is agricultural with 86% surface area classified as
such (Cowling & Hilton Taylor 1996). It means the remaining surface area results in a high
concentration of threatened natural plant groups. South Africa also lacks important arte-
rial rivers or lakes and is susceptible to prolonged periods of droughts (part and parcel to
its average rainfall of 497 mm/year being well below the world average) in various parts of
the country.
These issues make the immediate situation more pressing and could have extremely nega-
tive repercussions which reveal themselves as quickly as or more quickly than those involv-
ing environmental burdens manifesting over a longer time period. It is not argued that the
concept of climate change should be disregarded by a developing country, but taking into
account the limited resources to our disposal in enforcing environmental goals, a realignment
of priorities are necessary. South Africa contributes on average of only around 1.1% of the
total world’s carbon emissions, compared to China and the USA who contributes 20.2% and
19.1%, respectively (Baer et al. 2007). In this regard, should greater responsibility not lie with
industrialized countries to that of developing countries?
The issues highlights the dissimilarity of social, economic and environmental aspects to
that of European countries and underpins the fact that sustainability measurement is country
specific. In the South African case, the impacts on the three natural environmental resource
groups Water, Land, and Mined Abiotic Resources are prioritised, but at the same time a
balance between the socio-economic developments should be struck.
608
3 THE WAY FORWARD FOR SOUTH AFRICA
Recognising that the sustainability challenges for South Africa are unique, the necessity for the
South African road industry to lead the movement towards sustainable decision making must
be addressed. This will enable implementation of comprehensive and systematic approaches
including the mitigating of potential risks. The road industry specifically has recognized the
need to embrace sustainable practices and concomitant to the developments on government
level, an initiative to measure and rate the sustainability of road projects is currently underway
in South Africa. The “Greenroads South Africa Rating System” is being developed by road
industry volunteers (via the establishment of the Green Roads Council of South Africa) who
are key stakeholders in the country’s transportation sector. From the establishment of the body
in 2011, its development process has included affiliation with the Green Roads in USA rating
system. It became apparent that the rating systems in developed countries, with low unemploy-
ment, overlooks socio-economical aspects prevalent in poor countries. The consideration of
critical elements for the South Africa milieu would be better suited by the development of our
own sustainability tool and in 2015 this body established themselves as an independent entity.
Amidst these events a wider range of approaches have been considered and the question
remains, which route towards sustainability measurement must be followed. How would we
measure the sustainability of decision making in projects? Do we develop and implement a rating
tool which is recognised as a viable and acceptable norm for measuring the sustainability of such
roads, or is the answer to develop a holistic approach such as a LCA framework for industry?
One could argue that LCA provides a quantitative scientific basis for decision-making
whilst a rating-based method is comprised of a qualitative approach. In short, LCA is cer-
tainly a more attractive option. It comprises of a developed systems approach to provide
decision support for questions regarding the environmental impact of industrial processes
and products, potentially avoiding the paradox of decisions that could unintentionally cause
greater harm elsewhere. On the other hand, rating systems are point based and is vague with
respect to actual environmental impact reduction. This could lead to pursuing the achieve-
ment of a point rather than understanding the weighting significance of the issue, resulting
in a binary application of a very complex matter.
On the other hand, despite the huge momentum gained in LCA of pavements, many uncer-
tainties in modelling LCA of pavements still remains. In this regard, critical aspects to be
addressed in order to have a full view of a holistic LCA in pavements have been highlighted.
These aspects include inconsistency of functional units, improper system boundaries, limited
inventory results, asphalt fume toxicity uncertainty, feedstock energy inclusion, traffic delay
incorporation, maintenance implementation, use- and end of life indicators (Santero 2011a,
b, Harvey et al. 2015). Table 2(a) reports on these issues with additional aspects pertaining
LCA identified in this paper.
It should be clarified that many of these issues do not originate from the uncertainty in the
science of LCA, but rather the complexity of pavement structures. Therefore as highlighted in
Table 2, not necessarily a challenge unique to LCA. For example, the issue of inconsistent func-
tional units and improper boundary conditions is a product of the various type of pavements
(type, width, materials, design strategy etc). Regionalization, impact category weighing bal-
ances, and maintenance implementation strategy challenges all manifest as a result of articulat-
ing conclusions across regional boundaries. In addition, differing data sources for the life cycle
inventory exist. These dissimilarities are throughout the lifecycle, originating from pavement
designs and practices, available materials, differing electricity mixes, local climate, environmen-
tal conditions, traffic volumes, design life and other project specific external factors.
The inclusion of feedstock energy on the other hand is not strictly as a result of pavement
complexity but rather in the adherence to strict accounting procedures, as prescribed by ISO
14040 (LCA standard). The debate includes that materials such as bitumen could theoretically
be utilised to produce lighter fuel products and therefore accounting of its stored energy
should be included in the environmental flows. This implies a value of 40 MJ/ton rather
than 0.6 MJ/ton (Butt 2012) to be used, potentially swaying results due to the significantly
609
Table 2. (a) Unresolved issues in LCA of pavements (b) Proposed addition of categories to be consid-
ered for developing countries as part of the LCA methodology.
increased environmental burdens. Realistically, bitumen will not be used for energy recovery
when it has already been placed in the road layers.
Issues emanating from LCA uncertainty includes limited inventory results and impact cat-
egories weighting. An example of this is bitumen and considering that in its quantification
the origin of the source of crude oil influences results significantly. In fact, Gerdes & Skone
(2009) has shown a 60% increase in GHG emissions when foreign crude oil was transported
from other countries as opposed to domestic supply. Often a challenge in LCA is the imbal-
ances in characterization and weighting factors needed to complete the analysis. Certainly
for the South African context, and specifically regional diversity will not be correct as impact
categories, normalisation and weighting results are based on the impacts of ecosystem qual-
ity. In addition, water and land quantity impacts are usually not addressed in current LCA’s,
or incorrect for the South African context.
The most predominant uncertainty, internationally, is the elements resulting in environ-
mental burdens in the use phase. Reference is made to factors such as the structural response
of the pavement on the fuel economy rolling resistance, albedo, carbonation (cementitious
materials), lighting, leachate, and tire wear, work zone congestion, stormwater runoff, and
emissions and climate change effects (Harvey 2015). Here, even factors considered as insig-
nificant i.e the use of an air conditioner in a hot climate could have a significant effect on the
fuel economy and sway the results otherwise. This is exasperated by the often absent track-
ing of performance of postulated environmental burdens. These uncertainties reflect that
although LCA provides a more reliable base for measurement, it also has its own shortfalls.
By far the greatest challenge however is the integration of socio-economic impacts, a field
even more prevalent of uncertainties than that of its environmental counterpart.
Capability to identify developing country contextual problem areas on the basis of a scien-
tific analysis are important. Consequently the best alternatives can be developed using LCA
theory, but realism in the alignment of policy and strategy orientation is critical.
610
In order to address these challenges and developments still to be made, it seems realis-
tic that a two-phased approach should be implemented, including: (1) The roads industry
developing a sustainable rating tool to address the short term needs, but at the same time (2)
initiate the process based on scientific fundamentals for consideration in integration of a full
LCA (environmental and socio-economic) for the future.
4 CONCLUSIONS
The approach of the South African government towards the Carbon Tax Bill has reflected the
knowledge and development still needed in understanding the accounting and articulation
of environmental burdens. It is apparent that specialists in this field is needed for guidance.
611
This will enable alignment of sustainable strategies with the critical environmental and social
burdens associated with the needs and context of a developing country.
The alignment of such legislation such as the Bill should pursue objectives and principles
as set out by national legislation and includes consideration of the National Environmental
Management Act, 1998 and the Constitution which refers to the right to an environment that
is not harmful to health or well-being.
In the development of a sustainable tool the argument has been made for the incorpora-
tion of socio-economic impacts which should have equal measure to that of environmental
burdens in a developing country. As a significant emitter of environmental burdens and huge
potential for socio-economic development uptake, the responsibility of the roads sector to
guide government and develop their own sustainability measurement tool, aligned with these
issues are imperative. Through discussing the options of such a tool, LCA and rating systems
is considered, and a two phased approach for practical reasons are proposed. This includes
the short-term development of a rating system concomitant to the development of a method-
ical LCA incorporating environment and social impact measures. Contextual categories for
both environmental and social impacts have been proposed.
Considering the limited resources and challenges that the South African government is
facing, the way forward is to not only design and implement an effective sustainability tool,
but also to strike a careful balance between environmental burden, goals and economic devel-
opment through socio-economic equity.
REFERENCES
AASHTO. 2011. AASHTO Comments Concerning FHWA’s Sustainable Highways Self- Evaluation
Tool. Washington, D.C.
Alton, T., Arndt, C., Davies, R., Hartley, F., Makrelov, K., Thurlowc & J., Ubogu, D. 2014. Introducing
Carbon Taxes in South Africa. Applied Energy 116(2014): 344–354.
Baer, P., Athanasiou, T., Kartha, S. & Kemp-Benedict, E. 2008. The Greenhouse Development Rights
Framework -The Right to Development in a Climate Constrained World. 2nd Edition. Heinrich Böll
Foundation, Christian Aid, EcoEquity and the Stockholm Environmental Institute, Berlin.
Beuving, E., De Jonghe,T. Lindahl, T. & Stawiarski, A. 2004. Eurobitume Environmental Impacts and
Fuel Efficiency of Road Pavements. Industry Report. Joint EAPA/Eurobitume Task Group Fuel
Efficiency, Europe.
Brent, A. 2004. Development of a Life Cycle Impact Assessment procedure for Life Cycle Management
in South Africa. Ph.D thesis, University of Pretoria, South Africa.
Butt, A. 2012. Life Cycle Assessment of Asphalt Pavements including the Feedstock Energy and Asphalt
Additives. Ph.D thesis, University of California, Davis.
Chappat, M., Bilal, J. 2003. The Environmental Road of the Future. Colas Group, France.
Chehovits,J., Galehouse, L. 2010. Energy Usage and Greenhouse Gas Emissions of Pavement Preserva-
tion Processes for Asphalt Concrete Pavements. Compendium of Papers from the First International
Conference on Pavement Preservation.
Clark, W.C.; Norberg-Bohm, V. 1992. International Comparisons of Environmental Hazards: Develop-
ment and Evaluation of a Method for Linking Environmental Data with Strategic Debate Manage-
ment Priorities for Risk Management. Centre for Science & International Affairs, John F. Kennedy
School of Government, Harvard University, Cambridge.
Cowling, R.M., & Hilton-Taylor, C.1994. Patterns of plant diversity and endemism in Southern Africa:
An overview, In: Huntley, B.J. (Ed.), Botanical diversity in Southern Africa: 31–52, National Botanical
Institute, Pretoria.
Gerdes, K. and Schone, T. 2009. Consideration of Crude Oil Source in Evaluating Transportation Fuel
GHG Emissions. National Energy Technology Laboratory, USA.
Huang, Y., Bird, R., & Heidrich, O. 2009. Development of a life cycle assessment tool for construction
and maintenance of asphalt pavements. Journal of Cleaner Production 17(2): 283–296.
Huppers, G. & van Oers, L. 2011. Evaluation of Weighting Methods for Measuring the EU-27 Overall
Environmental Impact. Report Nr. EUR 24985 EN – 2011. European Commission Joint Research
Centre and Institute for Environment and Sustainability.
Louw, R. 2013. Promethium Carbon comments on the Carbon Tax Policy Paper of 2 May 2013. Pro-
methium, South Africa.
612
http://www.promethium.co.za/wp-content/uploads/2013/08/2013–08–02-Promethium-Carbon-tax-
comments-rl.pdf, Accessed 15 December 2015
Meil, J. 2006. A Life Cycle Perspective on Concrete and Asphalt Roadways: Embodied Primary Energy
and Global Warming Potential. Portland Cement Association. Skokie, Illinois.
http://www.mrmca.com/paving/athena.pdf, Accessed 12 December 2015
Patrick, J. and Arampamoorth, H. 2010. Quantifying the Benefits of Waste Minimisation in Road Con-
struction. NZ Transport Agency research report 406. NZ Transport Agency, Opus International
Consultants, New Zealand.
Santero, N.J., Masanet, E., & Horvath, A. 2011a. Life-cycle assessment of pavements Part II: Filling the
research gaps. Resources, Conservation and Recycling, 55(9–10), 810–818.
Santero, N.J., Masanet, E., & Horvath, A. 2011b. Life-cycle assessment of pavements. Part I: Critical
review. Resources, Conservation and Recycling, 55(9–10), 801–809.
Schooten, M., Vancly, F., & Slootweg, M. 2003. Conceptualising social change processes and social
impacts. In H.A Becker and F Vanclay (Eds.) The International Handbook of Social Impact Assess-
ment: Conceptual and Methodological Advances. Cheltenham, Edward Elgar Publishing.
South African Department of Environmental Affairs and Tourism. State of the Environment Reports.
http://www.environment.gov.za/, accessed on 27 December 2015.
South African Department of Environmental Affairs and Tourism. 1997. White paper on the conserva-
tion and sustainable use of South Africa’s biological diversity. Government Gazette No. 18163, Gov-
ernment Printer, Pretoria.
South African Institution of Civil Engineers. 2010. Road Surfacing Industry Suggests Greening its
Footprint for Sustainable Development. Magazine of the South African Institution of Civil Engineers,
South Africa.
South African National Roads Agency Limited. 2014. South African Pavement Engineering Manual
(SAPEM), 1st ed. SANRAL, Pretoria, South Africa.
Sowerby, C., Langstraat J, Harmer, C., Folkeson, L. & Gudmundsson, H. 2014. SUNRA - a sustainabil-
ity rating system framework for National Road Administrations. Transport Research Arena, Paris
Statistics South Africa. 2014. Quarterly labour force survey - Quarter 4, 2014. Pretoria, South Africa.
United Nations Convention to Combat Desertification (UNCCD), The Secretariat of the United
Nations Convention to Combat Desertification http://www.unccd.int/, Accessed on 27 December
2015.
Van Zyl, G. n.d. Seal and Thin Surfacing Technology. TRB Workshop 153-South African Highway
Technology Workshop, Washington DC, USA.
World Resources Institute. 2010. Climate Analysis Indicators Tool (CAIT). Washington, DC Willis, J.,
Robbins, M. & Thompson, M. 2015. Effects of pavement properties on vehicular rolling resistance: A
literature review. NCAT Report 14–07. National Centre for Asphalt Technology, Auburn University,
Auburn, Alabama.
http://www.nra.co.za/content/TRB3.pdf, Accessed on 27 December 2015
US Environmental Protection Agency, Science Advisory Board. 1990. Reducing risk: Setting priorities
and strategies for environmental protection, SAB-EC-90-021. EPA, Washington, DC.
Yang, R. 2014. Development of a pavement life cycle assessment tool utilizing regional data and intro-
ducing an asphalt binder model. Masters Thesis, University of Illinois, Urbana-Champaign.
613
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
With the significant increase of traffic volume and traffic load, many early cement treated
base asphalt pavements need to be rehabilitated. Building asphalt overlay is an efficient way
to increase the integral strength as well as to extend the service life. Therefore, overlay design
is of great importance. At first, overlay design mainly relies on experience. By the 1960s,
with the advent of finite element, mechanical analysis was applied to design overlay (Esben
Byskov,1970). Today, there are three ways to design asphalt overlay. The first one is based
on the efficient thickness of old pavement, which is well-developed. But it is hard to choose
appropriate parameters of overlay. The second one is based on the deflection of old pave-
ment, which is widely used in some states in USA and Canada. The third one is based on
mechanical analysis as well as experience. Compared to the other two methods, this one is
more efficient. However, this method needs to collect a large amount of data. As a result, it
is not available in some area where the data is not enough.
However, all these three methods did not take pavement performance into account. In
order to solve this problem, one of the most famous methods was established by a Canadian
scientist J.J. Hajek. The reason why this method is good is that it take different factors like
traffic volume, thickness of overlay and patching into consideration. The disadvantage still
exists. This model cannot forecast the deterioration of pavement performance and can only
be used in certain condition. In China the most common method is established by Lijun Sun.
His method is based on pavement performance. This method takes traffic volume, deflection
of old road and material into consideration. The advantage of this method is that it estab-
lished the relationship between performance and efficient numbers of old pavement. There
are two shortages of this model, first, this method only take pavement condition index (PCI)
into consideration and cannot forecast the deteriorate trend of main distress types of pave-
ment; second, this method is no longer efficient when design overlay based on old overlay. In
order to solve these problems, a new method is established.
This paper introduces a new method based on pavement performance. Compared to
Hajek’s and Lijun Sun’s models, this model can forecast the trend of main distress types and
615
is suitable for all conditions in seasonal frozen region in China. Also, when design overlay on
the old overlay, this method can be used. In this method, one forecast model and two attenu-
ation models have been explored. This method not only combined advantages of different
methods but also able to design overlay after the pavement has been repaired. These advan-
tages make this method more reasonable compared to the other two methods, especially in
China.
The main purpose of asphalt overlay is to improve the pavement structural strength and
pavement performance. Compared to new built pavement, asphalt overlay confronts severe
condition because of the complex foundation. In order to make the asphalt overlay design
more reasonable, the main distress types of asphalt overlay should be investigated first. To
find out the distress types, different sections of Jingha freeway with asphalt overlay in sea-
sonal frozen region have been investigated. Jingha freeway starts at Beijing and take in Liaon-
ing province, Jilin province, Heilongjiang province and ends up in Harbin. The location of
Jingha freeway makes it an excellent choice for this research. By analyzing the collected data,
the main distress types can be decided. Once the main distress types have been decided, the
most important performance can be determined then the most important evaluation indexes
can also be determined. Table 1 shows the proportion of distress types of typical sections of
Jingha freeway and cracking, rutting and patching are the main distress types in seasonal
frozen region. According to the Chinese guide, rutting should be discussed separately.
The result of the phenomenon in Table 1 is that the structural strength of freeways of
Shenshan freeway and Tiesi freeway is much higher than other freeways where transverse
cracking is more likely to occur. And patching is mainly aimed at maintain cracking. So
transverse cracking is the main distress type.
Table 2 shows rutting depth of 4 sections of Jingha freeway where had been repaired.
The most common rutting in Jingha freeway is flow rutting, which is formed in high-tem-
perature period (Eissenmann, 1987). In the slow lane of ShenShan freeway, overload is also a
616
critical reason for rutting. The rutting problem can be solved by maintenance. However, after
3 to 5 years after maintenance, rutting will occur again. Therefore, rutting should be taken
into consideration during the overlay design.
Since transverse cracking and rutting are the major distress types in seasonal frozen region
in China, in the following research, rutting depth index (RDI) and cracking index (CI) will
be the crucial evaluation indexes.
In order to establish a new method of overlay design, models of important evaluation indexes
should be established. First, based on the main overlay distress types of overlay in seasonal
frozen region, RDI and CI should be considered. Second, the models are influenced by dif-
ferent factors like maintenance scheme, traffic volume and structure of pavement, so it is
reasonable to divide freeways into more accurate sections and Jingha freeway is divided into
nine sections. In the end, with the help of mathematical statistics and scientific methods,
models can be established.
617
Figure 1. RQI of Shenshan freeway.
CI = 0.075 [ −T ha + t S+ S d] (1)
a( T1 − .7 ha S + b T + ΔRH )lg( Eb / )
CI = b
(2)
1 + exp(c d ⋅ t e )
618
where, ΔT2 = the difference value between the highest temperature and the lowest temperature
during the year; ΔRH = the difference value between the highest humidity and the lowest
humidity; Eb = resilience modulus of cement treated base material; σ is splitting strength of
cement treated base material; a, b, c, d, e are parameters of this model: for cement treated
base: a = 0.15,b = 0.01,c = 2.4,d = 0.38,e = 0.8; for flexible base: a = 0.12, b = 0, c = 3.0,
d = 1.18, e = 0.8; The other parameters have the same meaning as equation.
In this model, a(T1 .7 ha ) g S is the amount of cracking caused by low temperature and
b( 2 ) g (Eb ) is the amount of reflective cracking, there is no reflective cracking
for flexible material base. Compare to the model established by Hongyan Ma, this model
is more accurate, the correlation coefficient of this model is 0.92, much higher than that of
frozen region.
It can be seen in Figure 2 that there is a big difference value between Hongyan Ma’s model
and measured value. The difference between this new model and measured value is smaller.
As a result, this mew model is more efficient when it comes to the forecast of CI in seasonal
frozen region.
⎛ −⎜ ⎟ ⎞
⎛α ⎞
β
PPI = PPII 0 ⎜1 − e ⎝ y⎠
⎟ (3)
⎜⎝ ⎟⎠
where PPI = pavement performance index; PPI0 = the initial figure for PPI; y = the service
year of the pavement; α, β are parameters.
Based on the model of PPI, the regression result of PSSI in different sections of Jingha free-
way can be seen in Table 4. It can be noted from Table 4, in different sections, parameters are
different. In order to establish the relationship between PSSI and structural strength of old pave-
ment, the efficient structural number (SNeff) is needed (Romanoschi, 1999). In this paper, the
method to calculate SNeff is based on AASHTO guide for design of pavement structure (1993).
All the data are collected from the sections whose traffic levels are the same. Taking those sec-
tions whose AADT range from 4000–2000 for example, the result can be seen in Table 5.
It can be seen in Table 5 that there is correlation between SNeff and parameters. Using
regression analysis, the formula of PSSI attenuation model in sections whose AADT range
from 4000–2000 is shown in the below equation:
⎛ −⎜
( Nefff ))
⎛ 18 / 1+ exp(10.07 −1.82 SN 2⎞
⎟
⎞
PSSI = ( (+ ( − SN
Nefff )) + ) ⎜
× ⎜1 − e
⎜
⎝
y ⎟
⎠ ⎟
⎟ (4)
⎜⎝ ⎟⎠
where SNeff = the efficient structural number of old pavement. The correlation coefficient of
α is 0.82.
It is the same to establish PSSI attenuation model when AADT or maintenance scheme
has changed.
Section name
Parameter SS-M1 SS-M2 SS-M3 SS-L1 SS-L2 SS-L3 SS-L4 ST-M1 TS-M1
PSSI0 93.49 92.37 92.90 92.57 93.10 93.44 93.55 94.59 96.13
α 11.00 14.56 12.40 10.43 19.29 11.07 12.98 14.86 23.85
β 1.07 1.11 1.05 1.01 0.84 0.76 1.12 0.88 0.67
R2 0.99 0.90 0.90 0.70 0.90 0.95 0.99 0.98 1.00
N 6 6 6 6 6 6 6 4 4
Parameter
Building asphalt overlay is an efficient way to increase RDI. After building asphalt overlay,
the same decrease trend can be seen. The regression result of RDI in different sections of
Jingha freeway can be seen in Table 6.
620
Figure 4. RDI of Jingha freeway.
Section name
Parameter SS-M1 SS-M2 SS-M3 SS-L1 SS-L2 SS-L3 SS-L4 ST-M1 TS-M1
PPI0 94.23 94.73 94.25 93.97 92.59 94.01 91.19 94.68 94.454
α 10.37 12.38 11.04 3.35 4.53 3.78 4.44 13.97 10.8
β 0.72 0.62 0.71 1.15 1.10 1.08 1.28 0.59 0.66
R2 0.98 0.99 0.99 0.94 0.94 0.98 0.90 0.92 0.98
N 6 6 6 5 5 5 5 6 5
Parameter
It can be seen that in these figures the trend of RDI is similar to that of PSSI. The way
to establish the RDI attenuation model is the same as that of PSSI attenuation model. The
regression result of RDI in different sections of Jingha freeway can be seen in Table 6. It can
be noted from Table 6 that parameters in different sections are different. Since the environ-
ment temperature remains stable at 33°C, the most important factor that influences RDI
is axle-load. This model can shows the relationship between accumulated axle-load acting
number Ne and RDI. The result can be seen in Table 7.
It can be seen in Table 7 that there is correlation between Ne and parameters. The param-
eter α is influenced by the type of lane. The formula of this model is divided into two parts:
one is aimed at the main lane while the other is aimed at the low lane. Using regression analy-
sis, RDI attenuation model is shown in the below equations:
621
⎛ ⎛ (− Ne + )⎞
2.22 α −0.48
⎞
−⎜ ⎟
RDII M RDI ⎜1 − e ⎝ y ⎠ ⎟
⎜ ⎟
⎝ ⎠
(5)
⎛ ⎛ (− Ne + )⎞
2.22 α −0.48
⎞
−⎜ ⎟
RDII L RDII 0 ⎜1 − e ⎝ y ⎠ ⎟
⎜ ⎟
⎝ ⎠
(6)
where Ne = accumulated axle-load acting number. Equation 5 is for the main lane and equa-
tion 6 is for the low lane. The correlation coefficients of α for the main lane and the low
lane are 0.68 and 0.85 respectively. The correlation coefficient of β is 0.9. In this model, life
parameter α is the deterioration rate of pavement performance while shape parameter β is
the relative deterioration rate of pavement performance during certain period.
The multi-index design method is based on attenuation models. The PSSI attenuation model
is used to calculate the structural number, while the other four indexes (CI, RDI, RQI, SRI)
are used to check the overlay design meet the requirement.
622
The steps are as follows:
1. According to the specification for design of highway asphalt pavement of China, the traf-
fic level should be determined;
2. Analyzing the old pavement and divided the freeway into different sections;
3. Determining the main distress type in each section through investigation and calculating
the efficient structural number of old pavement;
4. Calculating the minimum objective structural number of each section;
5. Calculating the difference value between the efficient structural number of old pavement
and objective structural number;
6. Calculating the thickness of new overlay;
7. Choosing suitable pavement structure and design the material;
8. Using CI forecast model and RDI attenuation model to check whether the structure meet
the requirement;
9. If the structure meets the requirement, checking the pavement structure by mechanical; if
the structure does not meet the requirement, adjust the thickness of overlay until it meets
the requirement;
10. Comparing different kinds of overlay design and choosing the best design.
CONCLUSION
This paper established a new method of overlay design in seasonal frozen region. This method
can forecast the performance of pavement and design a new overlay based on deteriorate
models. Compared to other method, this method is more useful.
1. In seasonal frozen region, rutting and transverse cracking are the main kind of distress
types of freeways’ overlay and the reason of this phenomenon has been discussed in this
paper.
2. In order to evaluate pavement performance, PSSI and RDI are evaluation indexes. Based
on the data of investigation of Jingha freeway, PSSI and RDI attenuation models are
established.
3. Based on the investigation of Jingha freeway and model of Hongyan Ma, CI forecast
model is established.
4. Based on all these models and other overlay design methods, the multi-indexes design of
overlay in seasonal frozen region is established.
ACKNOWLEDGEMENTS
This study was supported by National Natural Science Foundation of China (No. 51478152).
The authors would like to acknowledge their financial support.
REFERENCES
623
Haas, R. &Meyer, F. & Assaf, G. & Lee, H.1988. A comprehensive study of cold climate airport pave-
ment cracking. Proceedings, AAPT. Vol. 57. 1988:290∼319.
Hiltumen, D.R. & Roque, R. 1994. A mechanics-based prediction model for thermal cracking of asphalt
concrete pavements. Journal of AAPT, Vol. 63, 1994:81∼108.
Hongyan Ma & Decheng Feng & Dongsheng Wang. 2010. A model forecast model of crack index of
asphalt pavement.
J.J.Hajek & Phang W A & Prakash A. 1986. Estimating the Life of Asphalt Overlays Using Long-Term
Pavement Performance Data[R]. The Research and Development Branch Ontario Ministry of Trans-
portation and Communications.
Romanoschi, S, & Metcalf, J B. 1999, Simple Approach to Estimation of Pavement Structural Capacity,
TRR 1652. Transportation Research Board, Washington DC.
SHELL International Petroleum Co. Ltd. SHELL Pavement Design Manual-Asphalt Pavements and
Overlays for Road Traffic, 1978. London.
624
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
K. Remya Varma
Department of Civil and Environmental Engineering, University of Wisconsin-Madison, Madison,
Wisconsin, USA
J. Murali Krishnan
Department of Civil Engineering, Indian Institute of Technology Madras, Chennai, India
H.U. Bahia
Department of Civil and Environmental Engineering, University of Wisconsin-Madison, Madison,
Wisconsin, USA
ABSTRACT: A parameter termed ‘Normalised Modulus’ (NM) is used as the basis for
developing the ASTM: D7460 (2010) standard for determining the fatigue of compacted
asphalt concrete subjected to repeated flexural bending. The parameter (NM) is calculated
and the cycle corresponding to the peak of NM curve is referred as the fatigue life of the
material. However, it is observed that the material response need not always follow the
assumptions related to NM.
In the present investigation, four point beam bending tests were conducted on asphalt mix-
tures using unmodified and modified binder. The control binder and modified binder used for
the study were Viscosity Grade 30 (VG30) and Polymer Modified Binder 40 (Elastomer) as
per Indian standards. Experiments were carried out on beam samples of size 380 × 63 × 50 mm
fabricated with 4 ± 0.5% air voids at 10 Hz frequency for five strain levels (200, 400, 500, 600
and 800 micro-strain) at 20 and 0°C. While for some testing conditions, a clear peak value was
seen when the data corresponding to normalised modulus was analysed, for different strains and
materials, such peak was not seen. In fact, four different types of NM curve trends were observed
and each such trend was quantified using the evolution of the stress-strain-time curve.
1 INTRODUCTION
1.1 Background
The concept of cumulative damage in fatigue was originally proposed by Miner (1945) for the
uniaxial tests conducted on aluminum alloys. According to this concept, the damage could be
expressed in terms of the ratio of number of cycles applied to the number of cycles to failure
at a given stress level. Failure occurs when the summation of these increments of damage at
several stress levels equals unity. The failure is considered as the inception of crack. Miner’s
hypothesis can be expressed as:
ni
∑N =1 (1)
i
where ni is the number of load cycles applied at stress level i and Ni is the number of load
cycles to failure at stress i. As per this hypothesis, each load contributes a certain amount of
625
damage to the material. Such linear cumulative damage hypothesis assumes that the total life
of the material can be estimated by adding the percentage of life consumed by the material at
each stress level. An important assumption here is that the definition of failure at any strain/
stress level is always identified by a “single process”.
Miner’s hypothesis has been used widely for the metals even though many deviations have
been observed (Fatemi and Yang, 1998). It is expected that the validity or otherwise of the
Miner’s hypothesis be verified by detailed experimental investigations (See Barenberg (1972),
Soussou and Moavenzadeh (1974) for early investigations on hot mix asphalt).
Asphalt pavements are subjected to different stress levels during the life as a result of
varying traffic conditions. Miner’s hypothesis related to cumulative fatigue damage needs a
re-look within the context of increased understanding developed related to the evolution of
material micro-structure, sophisticated equipment currently being used and of course, the
use of modified binders. For instance, when the fatigue damage is quantified using four point
bending, one sees three different stages of damage evolution. In the first stage, micro-cracks
are initiated. In the second stage, the micro-crack coalesce to form macro-crack and in the
third stage, the sample disintegrates. It should be noted that for the validity of the Miner’s
equation to hold, it is necessary that one needs to stay within one “single process” defining
the failure point (cycles to failure: Nf) and not the three or more stages of damage evolution
as is normally seen for asphalt mixtures.
Hopman et al., (1989) carried out investigations on asphalt mixes using four point bend-
ing tests to check the validity of Miner’s hypothesis. Using the framework of the three stage
fatigue damage process, Hopman et al., (1989) concluded that the concept of linear cumula-
tive damage cannot be applied to all the three stages, but is limited to the first stage. It should
be noted that Hopman et al., (1989) quantified damage using dissipated energy. Weise (2012)
investigated the validity of Miner’s hypothesis for the asphalt mixes using cyclic indirect ten-
sile test with different loading configuration. It was observed that the Miner’s hypothesis
is dependent on the test temperature, loading frequency and also on the order of different
stresses and strains. The validity or otherwise of such hypothesis was found to be dependent
on the sequence of loading. The aim of this paper is to show that if one uses Miner’s equation
in a realm beyond its limit, one is likely to get erroneous results and it is illustrated using the
concept of normalised modulus.
To use Miner’s hypothesis for asphalt, the number of cycles to failure at each stress or
stress level has to be determined using a common definition of failure. The peak of the nor-
malised modulus has been used as the basis for developing the ASTM: D7460 (2010) stand-
ard for determining the fatigue failure of compacted asphalt concrete subjected to repeated
flexural bending. The parameter ‘normalised modulus’ (NM) is calculated and the cycle cor-
responding to the peak of NM curve is referred as the fatigue life of the material. However, it
is observed that the materials need not always follow the idealized NM curve. This investiga-
tion re-looks at the validity of Normalised Modulus for typical bituminous mixtures tested
in four point bending at two different temperatures.
2 MATERIALS
The asphalt mixture selected for the study is bituminous concrete (BC) with unmodified and
modified binders. The control binder and modified binder used for the study were Viscosity
Grade 30 (VG30) and Polymer Modified Binder 40 (Elastomer) as per Indian standards IS:
73–2013 (2013) and IS: 15462–2004 (2004) respectively. The corresponding mixes are desig-
nated here as VG30 and PMB40 (E) in this study for ease of reference.
3 EXPERIMENTAL INVESTIGATION
In the present investigation, four point beam bending tests were conducted on asphalt mix-
tures using VG30 and PMB40 (E). Beam samples were fabricated using PReSBOX shear
626
compactor (ASTM: D7981 (2015)). Experiments were carried out on beam samples of size
380 × 63 × 50 mm with 4 ± 0.5% air void. Tests were performed in a controlled displacement
mode using sinusoidal waveform at 10 Hz frequency for five different strain levels (200, 400,
500, 600 and 800 micro-strain) at 20 and 0°C test temperature. The load and displacement
data were recorded at every 0.001 sec using which stiffness modulus and normalised modulus
were calculated. Detailed analysis of the results is covered in the following section.
The stiffness modulus (SM) is calculated by solving the equations of simple bending theory
assuming the beam as a linearised elastic material. Not withstanding the fact that such analy-
sis has limitations since asphalt mixtures behave like viscoelastic material, considering the
ease of computations such approximations are valid. The stiffness modulus values calculated
for the asphalt mixture with modified and unmodified binder are shown in Figures 1 and 2.
The traditional criteria for determining the fatigue life of the material considers the cycle cor-
responding to 50% of the initial stiffness modulus as the fatigue life of the material.
ASTM: D7460 (2010) suggests equation 2 to calculate the normalised modulus. As per this
equation, the product of stiffness modulus and number of cycles are normalised with respect
to the reference cycle and stiffness modulus at the reference cycle.
Si Ni
NM = (2)
S0 N0
where, NM is the normalised modulus; Si is the flexural beam stiffness at cycle i, Ni is the ith
cycle, S0 is the flexural beam stiffness at reference cycle and N0 is the reference cycle (gener-
ally, 50th cycle). As per this approach, the fatigue life is considered as the number of cycles
corresponding to the peak of the normalised curve. The main idea here is that the first devia-
tion from the Miner’s equation is considered as the onset of fatigue damage. Such assump-
tion rules out the possibility of additional fatigue life present in the material after the onset
of such peak value. It should be also pointed out that reduction in the beam stiffness modulus
is taken as the parameter to quantify the damage.
¥
Normalised modulus curve mentioned in this paper is same as the Normalised Complex
Modulus × Cycles versus Cycles as per the ASTM: D7460 (2010) standard. But for simplicity,
it will be referred as Normalised Modulus (NM) curve in this paper.
Figure 3 shows the normalised modulus curves for the samples at 200, 400 and 600 micro-
strain tested at 20 and 0°C and at 10 Hz frequency. It is evident that the normalised modulus
Figure 1. Stiffness modulus variation with the loading cycles at 20°C and 10 Hz.
627
Figure 2. Stiffness modulus variation with the loading cycles at 0°C and 10 Hz.
trend changes with the strain amplitude, temperature and material used. At lower strain lev-
els, the normalised modulus value increases with the number of cycles and shows the linear
trend. This is the clear case where the Miner’s hypothesis of cumulative damage exhibits the
linear trend. Every loading cycle adds identical amount of damage to the material quantified
here in terms of reduction of stiffness modulus. From Figure 3a and 3b, it is observed that at
628
lower strain level, 200 με, both BC with VG30 and BC with PMB40 (E) exhibited the same
NM curve trend. At 400 με (Figures 3c and 3d), the temperature influences the NM curve
of BC with VG30 whereas at 600 με (Figures 3e and 3f), NM curve of both BC with VG30
and BC with PMB40 (E) are identically influenced by the temperature. The asphalt mixture
with modified binder shows better fatigue performance compared to the unmodified binder.
Based on the various trends of the normalised modulus curves, the response were classified
into four categories. To extract meaningful information of the trend, it will be interesting to
look at the stress-strain behaviour of the material for each such category and such approach
is followed here.
4.1 Type I
Type I response exhibit normalised modulus curve with a linear trend in which the material
follows Miner’s hypothesis of cumulative damage. The normalised modulus value increases
with the cycles and the peak value is not attained within the testing cycles. This generally
happens when the material is subjected to lower strain amplitude (200 με in this case). To
extrapolate type I trend, Weibull distribution is used. Equation 3 represents the Weibull func-
tion used for extrapolation of fatigue life (Tsai et al., 2002).
ln( ln ) = γ × ln(
l ( ) + ln( λ ) (3)
where, SR represents the stiffness ratio given as the ratio of the stiffness modulus at any given
cycle to the reference stiffness modulus; N is the cycle number; γ is the slope of the line,
ln ( − ln ( )) versus ln ( λ ) and ln ( λ ) is the intercept of the line, ln ( − ln ( )) versus ln ( λ ) .
The fatigue life of the material can be estimated by solving equation for the value of N where
value of SR is 0.5.
Figure 4 shows the stiffness modulus and normalised modulus curve for Type I response
along with the stress-strain curve. The reduction in the stiffness modulus indicates the incre-
ment of material damage. From Figure 4a, it is noted that the changes in the stiffness modu-
lus value is small as the cycle proceeds. The stiffness is reducing at a lower rate indicating the
reduced damage accumulation signifying higher fatigue life of the material. Figure 4b shows
the normalised modulus curve of Type I response. Figure 4c shows the stress-strain curve for
the points A and B in the normalised curve. From Figure 4c, it is observed that the stress-
strain behaviour hardly changes even when the material is subjected to 600,000 cycles. This
indicates that the material is not damaged. It is also clear from the elliptical shape that the
response of the material is viscoelastic in nature (Padmarekha et al., 2013).
4.2 Type II
Type II response show the NM curve in which the NM value increases linearly, attains a peak
and then decreases. However after attaining the peak of normalised modulus, the stiffness
reduces at a higher rate compared to the increase in the number of cycles. This indicates the
onset of material damage and the number of cycles corresponding to the peak value is consid-
ered as the initiation of fatigue damage. In some cases, a secondary increase in NM value is also
observed following this reduction stage. However, this secondary increase is not considered for
the analysis since it represents the material in damaged condition (Rowe et al. (2012)).
Figure 5a shows the stiffness modulus and normalised modulus curve for Type II response.
It clearly shows the connection between the cycles corresponding to the peak of the normal-
ised modulus curve and the initiation of third stage of the stiffness modulus curve which
occurs almost at the same number of cycles. Figure 5c and 5d show the stress-strain curves
for the points A, B, C and D in the normalised curve. The shape of stress-strain curve at B
still indicates that the material is in good condition. From the stress-strain curve at the points
C and D, it is observed that the material is completely damaged, stiffness modulus is almost
negligible and the material can crack at any time and hence the region is not considered for
the analysis.
629
4.3 Type III
Figure 6a shows the stiffness modulus and normalised modulus curve for the Type III
response. Type III response exhibit an increasing normalised modulus trend initially, attains
a peak value and continue to be in that peak value range. This is shown in Figure 6b. There
is no drastic reduction stage following the peak value and also the normalised modulus does
not increase further with the loading cycles. The stiffness modulus trend at this region shows
the decrease at a much lower rate with a constant slope as the loading cycles increases. Fig-
ure 6c shows the stress-strain curves for the cycles A and B marked on the NM curve. It is
noted that at A and B, stress-strain curve is a straight line which shows that the material is
elastic in nature. The distortions of stress-strain curve shown in Figure 6d at C and D indi-
cates the material damage. Type III trend is generally observed at 0°C. This shows that the
failure mode is completely different for the material at 20 and 0°C and the associated NM
criterion of the peak value may not be valid for Type III response. From Figure 5d and 6d, it
is observed that the shape of the stress-strain graphs of BC with PMB40 (E) at 20°C and BC
with VG30 at 0°C are found to be the same once the material is damaged irrespective of the
temperature and the material response (elastic vs. viscoelastic).
630
Figure 6. Type- III material—VG30 at 0°C, 10 Hz and 400 με.
4.4 Type IV
Type IV response shows linearly increasing trend initially, attains a peak and either maintain
this peak for some cycles before starting a secondary increasing trend or increase with a dif-
ferent slope, lesser than the initial slope. This trend is mostly evinced at 0°C when the material
response is elastic.
Figure 7a shows the sample stiffness modulus and normalised modulus curve for Type
IV response. The deviation of stress-strain graph is used to quantify the onset of damage.
Figure 7b shows the normalised modulus curve. In case of materials where the secondary
increasing trend is observed, the slope of the second stage increase is generally lower com-
pared to the first stage. This indicates that most of the damage that happens to the material
is at the first stage and the rate of stiffness reduction is more during this stage. The stress-
strain curve deviation increases with the increase in the number of cycles even when there is
an increasing trend for the NM curve after the first peak as shown in Figure 7c and 7d. It has
been noted that the stress-strain graph forms a straight line as in the case of Type III response
indicating the elastic nature of the material. It is not clear whether one can ascribe onset of
damage at cycle C. From Figure 5 and 7, it can be concluded that the same material can have
different mode of damage at different temperature. This is attributed to the changes in the
material from viscoelastic nature at 20°C to elastic at 0°C.
It is interesting to note that substantial changes take place to Type- III and IV trends
after the primary NM increasing stage. These issues are not addressed in the ASTM: D7460
(2010). It is observed that the stress-strain response of the material is different at different
temperatures and strain levels and the usage of NM is seen to be ineffective for quantifying
the fatigue damage of the asphalt mixtures.
Table 1 shows the classification of materials based on the NM curve trends. From Table 1, it
is clear that at 20°C, asphalt mixtures with unmodified and modified binders follow the trend
which can be analysed using the NM. It should be pointed out here that even ASTM D7460
(2010) stipulates 20°C though an effective test temperature for equivalent fatigue damage is
also defined. However the same materials follow different trends at 0°C depending on the
strain amplitude. The trends are scattered and this makes the analysis of the material using
NM inconsistent. The damage accumulation of asphalt mixture with modified binder is com-
pletely different when compared to that of unmodified binder at 0°C. Except at 200 με, NM
curves of modified and unmodified binders did not match in the trend of any higher strain
levels. This also raises issues related to quantifying the advantages of modified binders with
unmodified binder. The asphalt mixture with modified binder exhibits better fatigue perform-
ance compared to unmodified binders at all strain levels as evident from the stiffness modulus.
However it is not clear whether one can use the NM approach to differentiate the same 0°C
since one sees different trends of damage accumulation for the same applied strain level.
631
Figure 7. Type- IV material—PMB40 (E) at 0°C, 10 Hz and 600 με.
Strain (με)
VG30 20 I II II II I*
0 I III II III IV
PMB40 (E) 20 I II II II II
0 I II IV IV II
*Test was run only till 50% of the stiffness value and within the testing conditions NM did not show
any peak.
Table 2 shows the fatigue life of the asphalt mixtures with unmodified and modified bind-
ers at various strain levels and test temperatures. From Table 2, it is clear that the asphalt
mixture with modified binder shows better fatigue performance compared to unmodified
binders and the order of performance varies depending on the temperature and strain lev-
els. As the strain level increases, the difference in the fatigue life of asphalt mixtures with
unmodified and modified binders becomes less. This may be because at higher strain levels,
the influence of damage due to the coalescence of micro-cracks to form macro-cracks is more
prevalent than the viscoelastic nature of the material.
In case of the material where NM curve peak value is not attained, Weibull function is used
to calculate the fatigue life as explained under the section 4.1 of Type I response. However
such assumption of Weibull damage process needs to be verified and it has always been non-
unique (Prowell et al., 2010)). Also one needs to have proper understanding on how to select
the cycle corresponding to the peak value depending on the various NM curve trend.
632
5 CONCLUSION
ACKNOWLEDGEMENT
The first and second authors thank the Department of Science and Technology, Govt.
of India for funding this investigation. The grant number is DST/TSG/STS/2011/46. The
authors acknowledge the technical assistance provided by M/s IPC Controls, Australia dur-
ing the conduct of the experiments.
REFERENCES
ASTM: D7460–10. 2010. Standard Test Method for Determining Fatigue Failure of Compacted
Asphalt Concrete Subjected to Repeated Flexural Bending. ASTM International. Pennsylvania,
United States.
ASTM: D7981–15. 2015. Standard Practice for Compaction of Prismatic Asphalt Specimens by Means
of the Shear Box Compactor. ASTM International. Pennsylvania, United States.
Barenberg, E.J. 1972. Incorporating Fatigue Test Results into Pavement Design Processes. Fatigue of
Compacted Bituminous Aggregate Mixtures. ASTM STP508. ASTM International. Pennsylvania,
United States. 85–101.
Hopman, P.C., Kunst, P.A.J.C. & Pronk, A.C. 1989. A Renewed Interpretation Method for Fatigue
Measurements: Verification of Miner’s Rule, 4th Eurobitume Symposium, 4–6 October 1989. Madrid,
Spain. 1:557–561.
IS73:2013. 2013. Indian Standard for Paving Bitumen - Fourth Revision. Bureau of Indian Standards,
New Delhi.
IS15462:2004. 2004. Indian Standard for Polymer and Rubber Modified Bitumen. Bureau of Indian
Standards, New Delhi.
633
Miner, M.A. 1945. Cumulative Damage in Fatigue. Journal of Applied Mechanics, 12(3): A159–A164.
Padmarekha, A., Kanmani Chockalingam., Saravanan, U., Abhijit, P. Deshpande & Murali Krishnan,
J. 2013. Large Amplitude Oscillatory Shear of Unmodified and Modified Bitumen. Road Materials
and Pavement Design. 14:12–24.
Prowell, B.D., Brown, E.R., Anderson, R.M., Daniel, J.S., Swamy, A.K., Quintus, H.V., Shen, S., Car-
penter, S.H., Bhattacharjee, S. & Maghsoodloo, S. 2010. Validating the Fatigue Endurance Limit for
Hot Mix Asphalt. NCHRP: 646. Transportation Research Board, Washington, D.C., USA.
Rowe, G.M., Blankenship, P., Sharrock, M.J. and Benner, T. 2012. The Fatigue Performance of Asphalt
Mixtures in the Four Point Bending Beam Fatigue Test in accordance with AASHTO and ASTM
analysis methods, 5th Eurasphalt & Eurobitume Congress, 13–15 June 2012. Istanbul.
Soussou, J.E. & Moavenzadeh, F. 1974. Statistical Characteristics of Fatigue Damage Accumulation in
Flexible Pavements. Fatigue and Dynamic Testing of Bituminous Mixtures. ASTM STP 561. ASTM
International, Pennsylvania, United States. 3–11.
Tsai, B.W., Harvey, J.T. & Monismith, C.L. 2002. High Temperature Fatigue and Fatigue Damage
Process of Aggregate-Asphalt Mixes. Journal of the Association of Asphalt Paving Technologists,
71:345–385.
Weise, C. 2012. Investigation of the Validity of Miner’s Law for Asphalt Mixes, 5th Eurasphalt & Euro-
bitume Congress, 13–15 June 2012. Istanbul.
634
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
W. Visser
AgileAssets Inc., Austin, Texas, USA
Shell Research, Amsterdam, The Netherlands
W.R. Hudson
AgileAssets Inc., Austin, Texas, USA
Transportation Engineering, University of Texas, Austin, USA
ABSTRACT: The development of modern pavement design took a major step in the sixties
with the use of mechanistic methods. A similar trend occurred with the development of pave-
ment evaluation techniques and the characterization of pavement materials and subgrades
in mechanistic terms.
The development of Pavement Management Systems [PMS] started in the USA in the
seventies.
From the start there was a trend to keep pavement management separate from pavement
design. Most transportation agencies had different divisions for these two activities, even
though pavement evaluation techniques were sometimes used by both. Road Maintenance
was also usually in a separate division.
The performance of a road pavement depends on its components, on subgrade, layer
thicknesses, workmanship, loading conditions, climate, but also maintenance actions (pre-
ventive and corrective), rehabilitation activities and, obviously, available funds. That means
that design, evaluation and maintenance management should be completely integrated which
can be accomplished in a modern PMS.
1 INTRODUCTION
In the last 50 years major progress was made in pavement design, pavement evaluation, pave-
ment maintenance, and pavement management. In addition to the effects of climate and traf-
fic, pavement performance depends on all these activities but in some transportation agencies
these are not properly integrated. The main purpose of this paper is to review the progress
made and to recommend a better integration of all required pavement activities which can be
facilitated with a comprehensive PMS.
635
PONOS computer program that can be used to establish the stiffness values for various bitumi-
nous binders at a range of temperatures and loading times (De Bats 1973). A logical extension of
that effort was the testing of a range of asphalt mixes to determine their mechanical properties
with bending tests at high frequencies: stiffness modulus, strain at break and fatigue properties
(Heukelom 1966). This program continued with similar and additional testing methods includ-
ing creep tests at long loading times. Other mix characteristics were also tested in a fretting
machine, a wheel tracking test and a large laboratory test track in a temperature-controlled envi-
ronment. That work produced a nomogram and a computer program for asphalt mix properties
and made it possible to predict the mechanical properties of asphalt mixes for a range of mix
compositions, temperatures and loading times (Bonnaure et al. 1977).
The ability to express the properties of asphalt mixes in engineering terms made it pos-
sible to calculate stress and strain patterns in flexible pavements. Research and experience
have demonstrated that the concept of the pavement as a linear elastic multi-layer system
corresponds with practical performance as long as the correct values for material properties,
temperature and loading conditions are used. BISTRO, a related computer program, was
developed in 1967 and the BISAR program (De Jong et al. 1973), that also can handle shear
stresses, followed in 1973. These programs are still used at several Universities. They were used
also to produce the Shell Pavement Design Manual (Claessen et al. 1977), introduced in 1978.
This manual contains many design charts, data tables and worksheets. BISTRO and BISAR
give more detailed computations than actually used in the design charts. This was done on
purpose because the greater accuracy is unrealistic owing to the natural scatter in material
properties, doubt regarding future traffic, environmental conditions etc. Consequently it will
always be necessary to exercise engineering judgement when designing the pavement.
About 1800 copies of the Shell Manual have been distributed around the world. In 1985
an addendum was published that allowed the user to incorporate safety factors of his own
choice (Gerretsen & Koole, 1987).
In the meantime many agencies have developed their own pavement design systems. In the
USA an extensive research program sponsored by AASHTO resulted in 2004 in the Mecha-
nistic-Empirical Pavement Design Guide [MEPDG] (AASHTO 2004). Since this guide also
covers rigid pavements, its features will be discussed in section 2.3.
636
Whatever pavement design system is used, it is essential that the specific main parameters
of the design are entered into the Pavement Management System of the agency. In that way
the results of the first structural evaluation of the as-built pavement can be compared with
the design data and long term performance data can be channelled back to the designers in
order to compare them with the design predictions. It should always be remembered that
pavement design uses material properties that are rather variable and the only way to check
the structural performance of a pavement is to monitor it, using proper evaluation techniques
reported in a PMS.
% of Time Answers
Right Wrong
1 Variable 95% 5%
4 Variables 81% 19%
8 Variables 66% 34%
16 Variables 44% 56%
32 Variables 19% 81%
The uncertainty and variability of most design parameters also means that it is prudent
to test the performance of the pavement (bearing capacity) soon after completion and check
whether the results conform to the predictions made during the pavement design process. It
is essential that the pavement design team is fully involved at this stage and has access to the
data. The same holds for the maintenance team. In current practice the maintenance team is
often only involved when the pavement performance levels have dropped below a critical level.
For preventive maintenance to be successful it is important to evaluate the pavement every
one or two years and to have the performance data available for the maintenance team and
the design team (the latter should be able to check whether the original design life chosen pans
out). A modern pavement management system is able to accommodate those needs as long as
it is kept up-to-date and all parties have proper access to the data that are relevant to them.
Measuring the surface characteristics like skid resistance, roughness [IRI], degree of cracking
and amount of rutting [in flexible pavements] is mostly done at speeds of up to 40–50 mph, but
measuring the structural properties of the pavement layers—needed to check pavement or over-
lay design—could previously only be done by closing road sections and either cutting cores and
measure in a lab or by measuring the response of a pavement to a test load. In the fifties Shell
Research developed the first Road Vibration Machine [RVM] (Van der Poel 1953) which ten years
later was followed by a heavy RVM which generated sinusoidal vertical forces with a maximum
peak value that was representative of heavy traffic. Both the deflections and the wave propaga-
tions were measured and it was possible to deduct the dynamic moduli for each pavement layer
and the subgrade. For the top layers normally a light high frequency vibrator was used to calcu-
late a dynamic modulus from the wave velocity, Poisson’s ratio and density. A drawback of the
RVM was that only four locations could be tested in a day. In the seventies commercial vibrators,
such as the Road Rater, were introduced and many are still being used successfully.
Measuring the deflection under a static load is another way to get information about the
bearing capacity of a pavement. The Benkelman Beam, developed in 1952 in the USA by
WASHO is still being used. The pavement responds to a static load which is not represent-
ative for dynamic traffic loading. The French developed the LaCroix deflectograph as an
automated Benkelman Beam at very low traffic speeds in the sixties. In several countries
experiments were done with “Rolling Deflectographs” with speeds up to 18 km/h.
In the seventies the first Falling Weight Deflectometers [FWD] were developed in Den-
mark and the Netherlands (Claessen et al. 1976). The main advantage of the FWD is its sim-
plicity whilst it has a loading signal with a similar shape and magnitude as under a big tire on
a commercial vehicle. When the pavement thickness, the temperature and loading signal are
638
known it is possible to back-calculate the actual structural properties of the pavement layers
which subsequently can be compared with the original or design properties. It is possible to
carry out some 400 measurements per day, but it does interfere with traffic.
Recently the Rolling Wheel Deflectometer [RWD] was introduced in the U.S.A. to support
existing non-destructive testing techniques by providing a screening tool at the network level.
The RWD consists of a 15 m long semitrailer applying a load of 9,000 kg by means of a
regular dual-tire assembly over the rear single axle. The deflections are collected at near traf-
fic speeds. Another recent development is a somewhat similar Traffic Speed Deflectometer
[TFD] with a rear axle load of 10,000 kg with which data are recorded at survey speeds of up
to 80 km/h at a rate of 1000 Hz, i.e. a 20 mm spacing of the raw measurements.
No doubt these developments will continue so that it becomes easier to obtain reliable data
about the behaviour of the pavements. It is equally important that all key players have access
to these data. For instance, when a new pavement has been designed and constructed it is
important that the evaluation data can be seen by the pavement designers to check whether
all design parameters are met. The same holds true for an overlay or a reconstruction of an
existing road.
In 1957–1961, the AASHO Road Test (Highway Research Board 1962), a $500 million con-
trolled experiment [in today’s dollars] was conducted “to solve once and for all” the problems
of pavement design and construction. The results of this test were widely analyzed and incor-
porated into the very first pavement design guide ever produced by AASHTO. It was used in
Texas and other states from 1962 but was only accepted by AASHTO as an interim guide in
1972 (Haas et al. 1994). The Guide includes design equations for rigid and flexible pavements.
These design principles were widely used in the design of interstate and primarily highways
in the United States and around the world since 1965.
In spite of these advances in design, testing methods and construction a great many failures
continued to occur in the interstate and primary highway system of the United States and
Canada. As a result the NCHRP and a Canadian group funded projects to solve this problem.
These studies showed that pavement “design” alone is not adequate to produce the required
pavement performance, although good design is the essential first step. As a result of these
studies the concept of pavement management was born. A complete history of this process is
given in reference (Haas, Hudson et al. 1994) and updated in (Haas, Hudson et al. 2015).
The results of these independent studies were dubbed “Pavement Management Systems”
which is best illustrated in Figure 2 (Haas, Hudson et al. 1994). As shown the design models,
whether empirical, mechanistic, or mechanistic-empirical, are predicting behavior, not perform-
ance. This is discussed in many references including (Yoder & Witczak 1975) and in the devel-
opment of the MEPDG (Pierce & McGovern 2014) where it became necessary to empirically
modify the mechanistic equations to make reasonable predictions of pavement behavior.
Behavior is an early indicator of future distress (potential cracking, deformation, and disin-
tegration) of the pavement. It is therefore a leading indicator of future “pavement response.”
Behavior taken to a limit creates such distress. These distresses accumulate as a function
of traffic and create roughness in an otherwise smooth road. Smoothness/Roughness is the
characteristic most important to pavement users and driver reaction and was dubbed Serv-
iceability by Carey and Irick (1960).
In a time-sequence, higher than expected deflection is an early indicator that distress will
begin to occur and increasing distress indicates that the serviceability level, PSI may begin
to decrease, see Figure 2. This rate of decrease is an indicator of performance or future
pavement failure. Some people have developed behavior indexes; distress indexes, roughness
indexes (IRI), and remaining service life equations. All are useful in different ways and at
different times in pavement life for predicting damage and indicating the need for pavement
strengthening. Distress indexes define needed preservation, maintenance, or rehabilitation
actions. Roughness, expressed as PSI, lists the “quality” of a pavement at any given point in
639
time and thus over time shows the PSI history. PSI is superior to IRI as a measure of perform-
ance since it expresses the users’ opinion and desires. The area under the PSI curve is a true
measure of “Pavement Performance.”
Pavement design predicts the design life as a function of serviceability. In reality, most of
the actual observed performance is below the expected performance as shown in the dashed
line in Figure 4. Actual pavement life will depend not only on the design, but also on mainte-
nance, rehabilitation, and reconstruction.
This process of premature failure is why pavement management was developed and the
solution to this problem is the integration of pavement design back into pavement manage-
ment. Design the pavement in the best possible way and then evaluate, measure, maintain and
rehabilitate the pavement as necessary to provide the required performance and keep a record
of all these activities in a comprehensive and accessible PMS.
In the past 40 years, pavement management has vastly improved (Haas et al. 2015). In the
USA nearly every state has some type of pavement management system and at least 40 of
the states have a modern functional computer based PMS that integrates design and all other
parts of the puzzle. Worldwide pavement management systems are also being used and their
number is growing.
640
Figure 4. Typical observed pavement performance compared with expected performance.
Since about 10–15 years, a lot of complementary systems have been developed such as main-
tenance management, bridge management, safety management, and comprehensive asset man-
agement systems. All these systems can be fully integrated with PMS and with each other.
Most people don’t realize that pavement management was actually developed as a result of
problems with pavement design. It originally was intended to be a project-level activity that
could help determine how a pavement that is designed could provide the desired perform-
ance over the design period. Like many new ideas this idea was picked up by only a few
states in the USA some 40 years ago: Arizona, Kansas, Washington State, and Texas. The
implementation of a pavement management system involves cooperation and integration
between design, maintenance, construction, planning, and traffic. These units are usually
silos or independent divisions within a DOT and historically did not cooperate well with each
other. One well-known state highway engineer, when presented with the concept of pavement
management in 1962 said, “I’m not going to use a chart that has maintenance and construc-
tion involved with designing my pavements.” Such thoughts detracted from the implementa-
tion of project-level pavement management. Secondly the design divisions in many of these
DOTs felt like they had the design problem under control. Even though their pavements were
deteriorating prematurely, they did not make changes. On the other hand there were innova-
tive people in some departments who recognized the benefits of pavement management for
evaluating projects and a network or group of projects to determine what the overall needs
were for pavement maintenance and rehabilitation.
As a result the pavement management systems in the original four states were modified to
do Network-Level Pavement Management. These same people shared their information with
others around the country and within 10 years there were more than 20 state DOTs involved
with Network-Level Pavement Management. But few of them were using pavement manage-
ment as a part of design.
Originally there was also a trend to keep pavement design separate from pavement man-
agement in knowledge sharing conferences. Findings of research and various road tests for
pavement design were reported during international “Ann Arbor Meetings” whilst progress
with pavement management was regularly discussed during Pavement Management Confer-
ences. Many transportation agencies still have different divisions for these two activities, even
today, Road maintenance is also usually in a separate division without access to PMS data.
641
Figure 5. Interrelationship of Pavement Design, Evaluation, and Management.
In an ideal world the organization of each transportation department should use and
maintain a comprehensive PMS and provide pavement designers, pavement evaluators and
maintenance engineers access to the PMS to interface with their activities with the obligation
to keep all information up-to-date. Figure 4 shows the important interfaces of the necessary
activities. That does require assigning trained and capable pavement managers at an organi-
zational level near the upper department management. A proper integration of design, evalu-
ation, and maintenance requires such an organization.
6 CONCLUSIONS
642
– Measure the feedback from the pavement at least annually from the start, including deflec-
tion, cracking, rutting, roughness, serviceability, and, if needed skid resistance,
– Provide preventive maintenance when the pavement begins to deteriorate below the desired
performance level,
– Provide maintenance or major maintenance when needed to extend serviceability,
– Rehabilitate the pavement when it becomes clear that the performance levels are dropping
too rapidly for maintenance to be economical,
– Adjust the performance models and prediction to simulate observed performance as
needed,
– Continue this cycle for as long as the highway needs to remain in service, usually 50 years
or more.
– Aim for an organizational structure of the transportation department where all above
activities can be integrated by keeping the required performance data in one database that
is accessible and maintained by pavement designers, pavement evaluators, and mainte-
nance engineers.
REFERENCES
AASHTO. 1972. Interim Guide for Design of Pavement Structures, American Association of State
Highway and Transportation Officials (AASHTO), Washington, D.C.
AASHTO. 2004. Mechanistic-Empirical Design of New and Rehabilitation Pavement Structures, Final
Report, NCHRP.
Bonnaure, F., Gest, G. & Ugé, P. 1977. A new method of predicting the stiffness of asphalt paving mix-
tures, Proceedings AAPT.
Carey, W.N. & Irick, P.E.. 1960. The Pavement Serviceability-Performance Concept. HRB Bulletin 250,
Highway Research Board.
Claessen, A.I.M., Valkering, C.P. & Ditmarsch, R. 1976. Pavement Evaluation with the Falling Weight
Deflectometer, Proceedings AAPT, vol. 45.
Claessen, A.I.M, Edwards, J.M., Sommer, P. & Uge, P. 1977. Asphalt Pavement Design – The Shell
Method, Proceedings Fourth International Conference on the Structural Design of Asphalt Pave-
ments, Ann Abor, Michigan.
De Bats, F.Th. 1973. A computer simulation of Van der Poel’s nomograph, Journal of Applied Chem-
istry and Bio-Technology, vol. 23.
De Jong, D.L., Peutz, M.G.F. & Korswagen. A.R. 1973. Computer Program BISAR, Layered systems
under normal and tangential surface loads, Koninklijke/ Shell-Laboratorium, Amsterdam, External
Report AMSR.0006.73.
Gerretsen, A.H. & Koole, R.C. 1987. Seven Years’ Experience with the Structural Aspects of the Shell
Pavement Design Manual, 6th International Conference on the Structural Design of Asphalt Pave-
ments, University of Michigan, Ann Arbor.
Haas, R., Hudson, W.R. & Zaniewski, J.P. 1994. Modern Pavement Management, Krieger Publishing,
Florida.
Haas, R., Hudson, W.R. & Cowe Falls, L. 2015. Pavement Asset Management, Wiley Publishing.
Heukelom, W. 1966. Observations on the rheology and fracture of bitumens and asphalt mixes, Pro-
ceedings AAPT.
Highway Research Board 1962. The AASHO Road Test: Report 5-Pavement Research, HRB Special
Report 61-E.
Hudson, W.R. & Scrivner, F.H. 1962. AASHO Road Test Principal Relationships – Performance Versus
Stress, Rigid Pavements, Highway Research Board Special Report 73.
Pierce, L.M. & McGovern, G. 2014. Implementation of the AASHTO Mechanistic-Empirical Pavement
Design Guide and Software, NCHRP Synthesis 457, Washington D.C.
Van der Poel, C. 1953. Vibration research on road constructions, American Society Testing Materials,
Publ. 156.
Van der Poel, C. 1954. A general system describing the visco-elastic properties of bitumens and its rela-
tion to routine test data, Journal of Applied Chemistry, vol.4.
Westergaard, H.M. 1927. Theory of Concrete Pavement Design, Proceedings, Highway Research
Board,
Yoder, E.J. & Witczak, M.W. 1975. Principles of Pavement Design, John Wiley and Sons.
643
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Naiyu Yin
Northeast Electric Power Design Institute Co., Ltd. of China Power Engineering Consulting Group,
Changchun, China
Dehua Ma
Nanjing University of Science and Technology, Nanjing, China
1 INTRODUCTION
Under the effect of environmental factors and traffic load, the damages of cracking, upheaval,
pit and rutting etc. will emerge in the asphalt pavement (Serfass et al., 1996). Compared with
other forms of damages, rutting occupies largest proportion of early damages of asphalt
pavement (Cooper et al., 2004). Asphalt in asphalt mixture will be softened when the temper-
ature is high. The damage of rutting will then form under the of traffic loads, which causes
adverse effects to the pavement. Firstly, it will affect the road surface roughness, resulting in
significantly reduced driving comfort and safety. Secondly, pavement deformation leads to
poor road surface drainage and the ability of pavement skid resistance, meanwhile, acceler-
ates water damage. In addition, the asphalt layer will become thick and stress will concentrate
at the wheel track, which will weaken the overall strength of the surface layer even the whole
pavement structure. It is likely to cause other diseases and increase the difficulty of the road
repair and maintenance (Vavrik et al., 2002, Benjamin et al., 2001, Andriescu et al., 2004).
Researchers have taken a lot of measures to prevent or slow down the generation of rutting
such as increasing the aggregate quality, improving mixture gradation, controlling volume
targets, reducing the amount of asphalt, reducing asphalt label, using modified asphalt, and
strengthening the compaction process, etc. (Najd et al., 2005). Although these measures can
improve the high temperature stability of asphalt mixture, the effect is not ideal. Thus, anti-
rutting agent appears as a modifier for improving asphalt mixture rutting resistance (Lim
et al., 1994). The main component of the anti-rutting agent is polymer. Through adsorption
of asphalt, swelling effect and its own cement glue reaction, the film thickness of asphalt
645
surface can be increased and the overall asphalt colloid strength is improved. Therefore sig-
nificantly the anti-asphalt pavement permanent deformation capacity is improved (Tayfur
et al., 2007, Jasso s et al., 2013). At the beginning, the anti-rutting agent was employed in
Europe, such as the French anti-rutting agent PR, Germany anti-rutting agent LF8000.
These anti-rutting agents have been widely used in the projects and have achieved sound
effects (Xue et al., 2013).
In China, the use of anti-rutting agent in the engineering is rare and the related research
begins recently. With the strong dependence on imported mature products, it has a serious
impact on the application and development of anti-rutting agent in China’s road engineering
(Xu et al., 2010, Xiao et al., 2006). Therefore, based on the advanced technology of simi-
lar products abroad, Jilin Province Department of Transportation developed anti-rutting
agent PCF independently. In this study, compared with anti-rutting agent PR, firstly the high
temperature stability, low temperature cracking resistance and water stability of AC-16 and
SMA-16 grading asphalt mixture modified by PCF was tested. Then, using uniaxial static
creep test, the creep property of asphalt mixture adding different anti-rutting agents were
analyzed. Finally, a reliable uniaxial static creep testing model was established based on finite
difference method.
Granulometric Class (mm) 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075
3
Apparent Density (g/cm ) 2.831 2.716 2.805 2.659 2.623 2.631 2.614 2.689 2.667
Water Absorption ωx (%) 1.385 1.20 0.81 - - - - - -
646
Figure 1. The grading curves of two different kinds of asphalt.
647
⎡1 t 1 ⎛ − ⎞⎤
t
ε = σ0 ⎢ + + 1 − e τr ⎟ ⎥ (1)
⎢⎣ E1 η1 E2 ⎝ ⎠ ⎥⎦
where σ 0 is the constant stress imposed in creep test and the unit is MPa; E1 is elasticity
modulus of spring and the unit is MPa. t is load period and the unit is s; η1 is the viscosity of
series connection glue pot and the unit is MPa⋅s; E2 is elasticity modulus of delayed element
spring and the unit is MPa⋅s; τr is zero shear stress and the unit is MPa.
The test was conducted using Cooper multi-purpose servo asphalt material testing machine
produced in England, as shown in Figure 4(a). The specimen was the standard Marshall
specimen molded by compaction method. The voidage ranges within 4% ± 0.2%, as shown
in Figure 2 (b). Test conditions are as shown in Table 5.
The numerical modeling for uniaxial static tests conducted based on finite difference prin-
ciple using FLAC3D software. The model was built in Figure 3.
Heat Test
Soaking preservation Stress level Pre-load stress temperature Load Relaxation
time (h) pattern (MPa) (MPa) (°C) time (s) time (s)
24 Air Bath 0.5 0.025 40/50 3600 300
648
The geometric model size was in accordance with standard Marshall specimen. The diam-
eter of the model is 101 mm and the height is 63 mm, as shown in Figure 4. When defining
the material parameters of the model, elastic modulus E and dynamic viscosity η were from
uniaxial static creep test. The bulk modulus and shear modulus were obtained from the fol-
lowing formula.
Bulk modulus
E
K= (2)
3(1 − 2 μ )
Shear modulus
E
G= (3)
2(1 + μ )
649
Figure 5. Pavement performance results.
formance is slightly inferior to that of anti-rutting agent PR. The anti-rutting agent PCF
makes AC-16 and SMA-16 grading asphalt mixture performance at low temperature increase
by 3.95% and 2.90%. The anti-rutting agent PR for both graded asphalt mixture perform-
ance at low temperature has increased by 5.26% and 5.16%. This shows that the anti-rutting
agent can improve the low temperature performance as well as the high temperature perform-
ance of asphalt mixture.
From the test result of freezing-thawing splitting strength, it can be seen that two anti-
rutting agents can improve the water stability of asphalt mixture specimen. However, the
effect of anti-rutting agent PCF is much better than that of anti-rutting agent PR, especially
for SMA-16 grading asphalt mixture. The anti-rutting agent PCF has the most obvious effect
on improving freeze-thaw splitting strength. Differences in water stability improving effect
are mainly decided by the adhesion of the two anti-rutting agents. The test results show that
anti-rutting agent PCF has larger adhesion and can effectively prevent the erosion of water
on the asphalt mixture.
of the material, a permanent deformation eventually formed which was residual deforma-
tion. The tests results are shown in Table 6.
As shown in Figure 7 and Table 6, after adding the anti-rutting agent, the cumulative defor-
mation of asphalt mixture and residual deformation than original asphalt mixture were much
significantly reduced, indicating that the two anti-rutting agents can effectively improve the
asphalt mixture resistance to high temperature deformation. Similar to rutting test results,
anti-rutting agent PCF has slightly poorer improving property than that of PR anti-rutting
agent, but the difference is not significant. PCF anti-rutting agent was economic efficiency,
and thus has a promising application prospect.
To calculate the asphalt mixture material parameters before and after adding the anti-rutting
agent, Burgers model was employed for static nonlinear fitting process of uniaxial creep data.
The fitting results are shown in Table 7. Part of the fitting figure is shown in Figure 8.
From the fitting results of Burgers model, when adding anti-rutting agent, Burgers model
parameters of asphalt mixture change. From the physical meaning of the parameters in Burg-
ers model, the early elasticity modulus of asphalt mixture E1 represents the non-deformability
of material in the initial force. The bigger the E1 value is, the smaller the deformation of
asphalt mixture at early stage will be (Yilmaz et al., 2011). For the calculation results, the add-
ing two anti-rutting agents can significantly improve E1 value of mixture, namely improving
the non-deformability at the early stage. η1 represents the non-deformable viscosity of asphalt
mixture at early stage under load. Higher η1 can improve non-deformability of material under
high temperature (Oruc et al., 2007). The experimental data shows that the adding anti-rutting
agent significantly increased the η1 value which verifies the fact that the anti-rutting agent can
effectively improve the asphalt viscosity and the early strength of the asphalt mixture.
The uniaxial static creep tests were analyzed using FLAC 3D. The simulation results of
asphalt mixture modified by two anti-rutting agents at 50°C are shown in Figure 9.
As can be seen from Figure 9, the uniaxial static creep tests simulation results and measured
values are close. However, it is assumed that the mixture specimen is homogeneous model in
651
Table 7. Burgers model parameters results.
Types of E1 (MPa) E2 (MPa) η1 ((MPa⋅s)) η2 (MPa⋅s)
asphalt
mixture 50°C 40°C 50°C 40°C 50°C 40°C 50°C 40°C
AC 69.03 73.88 227.39 293.02 2.09E+06 1.44E+06 0.51E+05 0.30E+05
AC+ PCF 72.45 78.42 239.61 330.45 2.21E+06 4.08E+06 0.53E+05 0.86E+05
AC+ PR 81.84 99.96 253.86 312.61 2.57E+06 3.11E+06 0.71E+05 0.75E+05
SMA 79.56 85.63 241.48 307.27 2.14E+06 2.85E+06 0.58E+05 0.47E+05
SMA+ PCF 92.43 100.73 366.17 473.36 2.91E+06 4.83E+06 0.86E+05 0.96E+05
SMA+ PR 103.59 105.73 371.69 413.15 2.52E+06 3.29E+06 0.86E+05 0.87E+05
Figure 8. Anti-rutting agent modified asphalt mixture creep curves fitting (50°C).
the simulation process. It is not exactly the same with the actual situation. Therefore, there
are still some differences between the predicted value and the measured value. By statistical
calculations, the model curve maximum predicted error is 4.2%, the minimum predicted error
is 0.3%, and the average error is 3.1%. Therefore, in this study the simulation of asphalt mix-
ture creep test is accepted and simulation results are close to experimental measured values.
4 CONCLUSION
The performance evaluation and creep properties of asphalt pavement modified by anti-
rutting agent PCF were studied in this study. It is concluded as followed:
1. Similar with anti-rutting agent PR, the anti-rutting agent PCF can significantly improve
the high and low temperature properties of the asphalt mixture as well as the water stabil-
ity of the mixture. For high temperature stability and low temperature cracking resistance,
anti-rutting agent PCF increase rate is slightly less than that of anti-rutting agent PR; but
for the water stability of the mixture, the effect of anti-rutting agent PCF is much better
than that of anti-rutting agent.
652
Figure 9. The creep simulation results of asphalt mixture modified by anti-rutting agents (50°C).
2. It can be seen from uniaxial creep static test results, anti-rutting agent PCF has slightly
poorer improving property of high temperature non-deformability than that of anti-rut-
ting agent PR, but the difference is not significant. The anti-rutting agent PCF was eco-
nomic, and thus has a promising application prospect.
3. In this study, the uniaxial static creep tests simulation results and actual measured values
are close. By statistical calculations, the model curve maximum predicted error is 4.2%, the
minimum predicted error is 0.3%, and the average error is 3.1%. Therefore, the simulation
of asphalt mixture creep test is acceptable.
REFERENCES
Andriescu A, Hesp S, Youtcheff J. Essential and plastic works of ductile fracture in asphalt binders[J].
Transportation Research Record: Journal of the Transportation Research Board, 2004 (1875): 1–7.
Benjamin J.S. Low-temperature and dynamic fatigue tou-ghening mechanisms in asphalt mastics and
mixtures[D]. Kingston: Queen’s University, 2001.
Cooper S.B., Abadie C, Mohammad L.N. Evaluation of Open-graded Friction Course Mixture[R].
Louisiana Transportation Research Center, 2004.
Dai Q, You Z. Prediction of creep stiffness of asphalt mixture with micromechanical finite-element and
discrete-element models[J]. Journal of Engineering Mechanics, 2007, 133(2): 163–173.
Jasso M, Bakos D, MacLeod D, et al. Preparation and properties of conventional asphalt modified by
physical mixtures of linear SBS and montmorillonite clay[J]. Construction and Building Materials,
2013, 38: 759–765.
Lim I.L., Johnston I.W., Choi S.K., et al. Fracture testing of a soft rock with semi-circular specimens
under three-point bending. Part 2—Mixed-mode[C]//International journal of rock mechanics and
mining sciences & geomechanics abstracts. Pergamon, 1994, 31(3): 199–212.
Marasteanu M.O., Basu A, Hesp S.A.M., et al. Time–Temperature Superposition and AASHTO MP1a
Critical Temperature for Low-temperature Cracking[J]. International Journal of Pavement Engineer-
ing, 2004, 5(1): 31–38.
Najd A, Zheng C.C. Measuring crack propagation in reinforced asphalt concretes[J]. Journal of South-
east University (English Edition), 2005, 21(3): 342–347.
653
Oruc S, Celik F, Akpinar M.V. Effect of cement on emulsified asphalt mixtures[J]. Journal of Materials
Engineering and Performance, 2007, 16(5): 578–583.
Serfass J.P., Samanos J. Fiber-modified Asphalt Concrete Characteristics, Applications and Behavior
(With Discussion) [J]. Journal of the Association of Asphalt Paving Technologists, 1996, 65.
Tayfur S, Ozen H, Aksoy A. Investigation of rutting performance of asphalt mixtures containing poly-
mer modifiers[J]. Construction and Building Materials, 2007, 21(2): 328–337.
Vavrik W, Pine W, Carpenter S. Aggregate blending for asphalt mix design: Bailey method[J]. Transpor-
tation Research Record: Journal of the Transportation Research Board, 2002 (1789): 146–153.
Xiao Qingyi, Rui Shaoquan, Wang Hang, et al. Research on the Effect of the PR PLASTS Contents on
Asphalt Mixture Performance[J]. Journal of Wuhan University of Technology, 2006, 28(7): 36–39.
Xu Yang, Guan Hongxin, Zhang Qisen, et al. Rutting Resistant Pavement Material Design of Continu-
ous Uphill Highway[J]. Journal of Wuhan University of Technology, 2010, 32(14): 62–64.
Xue Q, Feng X, Liu L, et al. Evaluation of pavement straw composite fiber on SMA pavement
performances[J]. Construction and Building Materials, 2013, 41: 834–843.
Yilmaz M, Kök B.V., Kuloğlu N. Effects of using asphaltite as filler on mechanical properties of hot mix
asphalt[J]. Construction and Building Materials, 2011, 25(11): 4279–4286.
654
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACTS: Durability and sustainability are the main objectives of the pavement pres-
ervation technology. An cost-effective strategy, known as Preventive Maintenance Treatment
(PMT), is always considered as an useful manner leading to a better pavement quality at
lower total costs by applying treatments before the pavement deteriorates to a poor condition
level requiring a corrective treatment. For the asphalt pavements in Beijing, various PMTs
were conducted during recent years to gradually replace the previous corrective maintenance
strategy. This paper presented a framework for characterizing the PMTs performance applied
at different pavement age. The data collected from the local transportation agency were ana-
lyzed to characterize the performance decay trends of new constructed pavements and further
compare the enhancements resulted from PMTs conducted at different timing scenarios. The
analysis approach presented here would be of help to select appropriate PMT to meet both
the cost-effective goal and preservation purposes in future pavement management practice.
1 INTRODUCTION
655
2004), one fundamental and important question is how to characterize and quantify the
improvement on pavement performance achieved with PMTs application in terms of treat-
ment types, maintenance timing and cost-effective analysis. Frequently, the cost-to-effective-
ness ratio is employed to compare candidate treatment strategies. A measure of effectiveness
is computed as the area under a pavement condition indicator versus the time (pavement age)
and a large area means a good effectiveness. Placing treatments at different timing scenarios
always result different improved effectiveness and cost consumption and thus, cost-effective
analysis can be conducted to identify the optimal treatment application. During the NCHRP
14–14 project, a comprehensive cost-effectiveness based methodology for identifying the
optimal timing for PMTs application was developed which consider pavement performance
contributions from multiple pavement condition indicators (Peshkin et al., 2004). A mount
of other studies based on similar approach can be found in literatures for selecting the most
cost-effective maintenance treatment strategy (Hicks et al., 1999; Mamlouk & Zaniewski,
2001; Baladi et al., 2002; Wei & Tighe, 2004; Li et al., 2010; Haider & Dwaikat, 2011; Ram &
Peshkin, 2014; Wu et al., 2015).
This paper presented a case study in Beijing for cost-effective analysis for two PMTs in
terms of microsurfacing and thin hot-mix-asphalt overlays. Different PMTs timing scenarios
were conducted during recent six years and the corresponding pavement condition data were
collected annually to measure/analysis the PMTs performance.
Currently there are totally 922 kilometers expressway constructed in Beijing area and oper-
ated by local Department Of Transportation (DOT). Previous pavement preservation merely
cover the routine maintenance that applied every year and the corrective maintenance that
conducted at highly distressed condition level. In recent years, various preventive maintenance
activities are gradually integrating into preservation program in terms of microsurfacing,
chip seal, fog seal, thin overlays et al. for improving maximum efficiency of the maintenance
costs. The preservation decision process every year is mainly relied on the annual pavement
condition assessment results as detailed below and available maintenance funding.
PCI
C 100 15DR 0.412 (1)
The PQI calculation is shown in equation (2) and the weighting factors (ω) for individual
indicator are provided in Table 1. Either using PQI or individual indicators, the pavement
condition assessment can be completed based on the grading criteria given in Table 2.
656
Table 1. Weighting factors of individual indicators for PQI calculation.
PQI ω PCI
C PCI + RQI
RQI Q + ω RDII RDI + ω SSRI SRI
RQI (2)
657
Table 3. Selected sections for PMTs performance analysis.
(ET) is quantified as the difference in area between the do-nothing area and post-treatment
area, which is labeled as the dash line area.
Step 1: Selection of Effectiveness Cutoff Value
The concept of effectiveness cutoff value is built upon an reasonable assumption that when
the pavement condition deteriorates to a certain poor level the serviceable ability from the
pavement structure is too limited to account for effectiveness calculation. Cutoff values are
identified as the performance indicator-axis boundary for effective area defined from the
measured/predicted performance curves. In the Beijing case, the expressways are operated and
maintained on a high criterion and thus, as shown in Figure 1 the effectiveness cutoff value
for PCI indicator is always taken as high as the value of 70 according to the annual assessment
report of the pavement condition form the agency of Capital Highway Development Group.
Step 2: Calculation of Service Effectiveness (EDN) in the Do-Nothing Case
Benefit from a PMT application is quantified based on the improvement in pavement per-
formance compared with the “do-nothing” case which is able to obtained from the predicted
condition performance curve fitted from the measured performance data in history. In other
words, the do-nothing alternative defines the pavement performance that would be expected
if only minor routine maintenance were conducted. In this study, the measured PCI data
were fitted with the nonlinear equation (3) using the Microsoft Excel Solver.
⎧⎪ ⎡ ⎛ a ⎞ b ⎤ ⎫⎪
PCI
C PCII 0 ⎨1− exp ⎢ −⎜ ⎟ ⎥ ⎬
PCI (3)
⎪⎩ ⎢⎣ ⎝ y ⎠ ⎥⎦ ⎪⎭
658
where y is the pavement age, a and b are regression coefficients. PCI0 is the initial PCI value
when pavement newly constructed and is set as 100 is this study.
Step 3: Calculation of Service Effectiveness (EPT) in Post-Treatment Case
For measuring the effectiveness of the PMTs, how pavement performance is improved once
the PMT applied is also important. The service effectiveness for post-treatment case (EPT) is
labeled as integration of solid line area and dash line area.
Step 4: Determination of the Treatment Effectiveness (ET) and Associated Benefits (B)
The effectiveness of the particular treatment (ET) is quantified as the difference in area
between the do-nothing area and post-treatment area, which is expressed in equation (4).
However, when comparing the ET identified from different sections, it is found that the mag-
nitude of EDN varied sometimes and impacted the ET evaluation. So a normalized indicator
benefit (B) is introduced and calculated as equation (5) to remove this EDN influence.
ET E PT − E DN (4)
ET
B= (5)
E DN
Following the characterizing approach above, all ten sections of expressways S50 and G6
were analyzed based on the available PCI performance data. The fitted PCI performance
curve as well as the area calculation of EDN and ET are respectively given in Figures 2–4 for
the three expressways. The fitting results of PCI model and cost-effective indicators ET/C and
B/C are summarized in Table 4.
It can be seen from Figure 2 and Table 4 that the EDN of two sections from S50 (up line)
are pretty close to each other whereas for S50 (down line) case the difference on EDN from dif-
ferent sections increased. When comparing the EDN from G6 sections, this difference reached
the maximum level. These indicate that even for the sections from the same expressway, on
Figure 2. Determination of EDN and ET for sections of S50 (up line) expressway.
659
Figure 3. Determination of EDN and ET for sections of S50 (down line) expressway.
660
Table 4. Cost-effective and benefit analysis results.
Pavement
age when
PMTs PCI Model
applied parameters
Expressway Section
ID ID MS TO a b EDN EPT ET B C ET/C B/C
S50 (Up Line) S1 / 13 27.1 0.9 345.2 356.9 11.7 3.4% 8.6 1.4 0.4%
S2 / / 26.8 0.9 341.4 / / / / / /
S50 (Down Line) S3 / 13 33.6 0.6 315.7 330.8 15.1 4.8% 8.6 1.8 0.6%
S4 14 / 30.2 0.8 340.1 / / / 4.1 / /
S5 11 13 33.9 0.5 285.6 314.7 29.1 10.2% 12.7 2.3 0.8%
G6 (Up Line) S6 15 18 32.0 0.39 195.8 227.9 32.1 16.4% 12.7 2.5 1.3%
S7 / 14 12.1 9.00 304.6 383.6 78.9 25.9% 8.6 9.2 3.0%
S8 / 15 17.7 1.72 329.3 433.3 104 31.6% 8.6 12.2 3.7%
S9 18 / 21.0 2.00 416.1 437.1 21.0 5.0% 4.1 5.1 1.2%
S10 18 15 19.5 1.51 346.2 395.3 49.0 14.2% 12.7 3.9 1.1%
Figure 5. Cost-effective comparison for different PMTs timing scenarios of G6 (up line) (a) ET/C
(b) B/C.
which the climate condition, construction history and traffic level are almost identical, the
pavement serviceable ability/effectiveness may also varied. Besides, section 6 and section 7
from G6 expressway showed rapid deterioration of PCI before preservation activities that is
entirely different from other sections, and it is interesting to observe that for these two cases
the PCI also dramatically deteriorated after preventive maintenance application. This result
demonstrates that the PMTs that is designed to address functional condition of a pavement
could provide little help for the highly deteriorated pavement structures which need correc-
tive treatments.
From the ET/C and B/C results in Table 4, it can be seen that for the S50 (down line) case
combination application of MS and TO treatments on section 5 slightly performed better
than section 3 where merely TO preservation was applied. Figure 5 (a) and (b) respectively
present the ET/C and B/C results of sections from G6 (up line) expressway where multiple
PMTs timing scenarios were conducted in past six years. “TO-14” indicates that at the pave-
ment age of 14 years, a thin overlay preventive maintenance was constructed and “MS-15 &
TO-18” represents two PMTs were performed respectively at the pavement age of 15 an 18
years. It can be observed that a single PMT of thin overlay at the year of 14 or 15 could con-
tributed advanced maximum levels for both ET/C and B/C, which indicates that applying the
preventive maintenance of thin overlay is cost-effective enough to keep the pavement condi-
tion within a good serviceable performance. Regarding the comparison of TO-14 and TO-15,
it is shown that TO-15 treatment could provide slight better performance improvement either
considering the effectiveness or benefit criteria.
661
5 SUMMARY
This study mainly discussed the effectiveness improvement of pavement performance resulted
from the preventive maintenance activities instead of traditional corrective treatments. The
benefit obtained from different timing scenarios were characterized and compared to select
an optimal cost-effective strategy. The specific findings of this study are:
1. For the sections from the same expressway where the climate condition, construction his-
tory and traffic level are almost similar, pavement serviceable effectiveness in “do-nothing”
case may be different enough to affect the performance evaluation of maintenance treat-
ments, so a normalization process of quantifying benefit is needed for optimal analysis.
2. Preventive maintenance that merely addressed the improvement for functional condition
of a pavement presented little benefit for the highly distressed pavement structures where
corrective maintenance is needed.
3. Microsurfacing and thin hot-mix-asphalt overlays were analyzed in this study and it is
observed that a single treatment of thin overlay is good enough to meet both the cost-
effective goal and pavement preservation purposes. Nevertheless, this is limited in some
degree and further refinement is needed due to the fact that only construction cost was
covered in this study.
Future research is covering multiple pavement condition indicator in terms of ride quality
and rutting performance for more comprehensively characterizing the effectiveness improve-
ment and associated benefit under the preventive maintenance program in Beijing.
ACKNOWLEDGEMENT
The authors would like to gratefully acknowledge the sponsorship from National Natural
Science Foundation of China (Grant No. 51478021) and Beijing Natural Science Foundation
(Grant No. KZ201510005002).
REFERENCES
Baladi, G.Y. et al. 2002. Cost-Effective Preventive Maintenance: Case Study. Transportation Research
Record: Journal of the Transportation Research Board, 1795: 17–26.
Carroll, D.A. et al. 2004. Implementing Highway Preventive Maintenance: Comparing Challenges,
Processes, and Solutions in Three States. Transportation Research Record: Journal of the Transporta-
tion Research Board, 1877: 10–16.
Galehouse, L. 1998. Innovative Concepts for Preventive Maintenance. Transportation Research Record:
Journal of the Transportation Research Board, 1627: 1–6.
Hicks, R.G. et al. 1997. Framework for Selecting Effective Preventive Maintenance Treatments for
Flexible Pavements. Transportation Research Record: Journal of the Transportation Research Board,
1597: 1–10.
Hicks, R.G. et al. 1999. Selecting a Preventive Maintenance Treatment for Flexible Pavements. Transpor-
tation Research Record: Journal of the Transportation Research Board, 1680: 1–12.
Harvey, J.T. et al. 2012. Probabilistic Approach to Life-Cycle Cost Analysis of Preventive Maintenance
Strategies on Flexible Pavements. Transportation Research Record: Journal of the Transportation
Research Board, 2292: 61–72.
Haider, S.W. & Dwaikat. M.B. 2011. Estimating Optimum Timing for Preventive Maintenance Treat-
ment to Mitigate Pavement Roughness. Transportation Research Record: Journal of the Transporta-
tion Research Board, 2235: 43–53.
Li, Q. et al. 2010. Matter Element Analysis for Optimal Timing and Preventive Maintenance of Pave-
ments. Transportation Research Record: Journal of the Transportation Research Board, 2150: 18–27.
Mamlouk, M.S. & Zaniewski, J.P. 2001. Optimizing Pavement Preservation: An Urgent Demand for
Every Highway Agency. International Journal of Pavement Engineering, 2(2): 135–148.
662
Peshkin, D.G. et al. 2004. NCHRP Report 523: Optimal Timing of Pavement Preventive Mainte-
nance Treatment Applications. Washington D.C.: Transportation Research Board of the National
Academies.
Ram, P.V. & Peshkin, D.G. 2014. Performance and Benefits of Michigan Department of Transpor-
tation’s Capital Preventive Maintenance Program. Transportation Research Record: Journal of the
Transportation Research Board, 2431: 24–32.
Wei, C. & Tighe, S. 2004. Development of Preventive Maintenance Decision Trees Based on
Cost-Effectiveness Analysis: An Ontario Case Study. Transportation Research Record: Journal of the
Transportation Research Board, 1866: 9–19.
Wu, D. et al. 2015. A Risk-Based Optimization for Pavement Preventative Maintenance with Probabil-
istic LCCA: A Chinese case. International Journal of Pavement Engineering, in press.
Zaniewski, J. & Mamlouk, M. 1999. Pavement Preventive Maintenance: Key to Quality Highways.
Transportation Research Record: Journal of the Transportation Research Board, 1680: 26–29.
Zimmerman, K.A. & Peshkin, D.G. 2003. Pavement Management Perspective on Integrating Preventive
Maintenance into a Pavement Management System. Transportation Research Record: Journal of the
Transportation Research Board, 1827: 3–9.
Zimmerman, K.A. & Peshkin, D.G. 2004. Issues in Integrating Pavement Management and Preventive
Maintenance. Transportation Research Record: Journal of the Transportation Research Board, 1889:
13–20.
663
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Pavement life includes both the structural life and the functional life of
the pavement. This study primarily investigated the functional life (i.e., the duration of the
anti-icing performance) of an anti-icing asphalt mixture with low freezing point materials.
Two materials that lower the freezing temperature were selected for producing the anti-
icing asphalt mixture in this study. The one material is a coarser granular material, whereas
the other is a finer powdery material. After designing and verifying the performance of the
two anti-icing asphalt mixtures, the anti-icing effect was tested by the solution conductivity
method. Based on the amount of precipitation in the Beijing region, the testing method
of the anti-icing effect was based on the similarity theory for an asphalt mixture with a
low freezing point material. According to the test data of the anti-icing performance for
125 days, a prediction model of functional life was established for the anti-icing asphalt
mixture.
1 INTRODUCTION
Northern China experiences very cold winters with negative air temperatures. In snowy
weather, pavement surfaces become slippery and hard due to an accumulation of snow and
freezing ice, and as a result, there is a huge increase in the traffic accident rate and occur-
rence of traffic jams. Recently, southern China has been having frequent extreme rain-snow
weather and icy rain. The icy rain covers the pavement surface, especially in sections of the
pavement that slope, turn, and form road crossings. Accordingly, the ice and snow result in a
great security risk to transportation (Zhang, 2009; Cui, 2010).
To reduce the traffic accident rate and increase transportation efficiency and capacity, it is
crucial to quickly remove any ice and snow from pavement surfaces (Tan, 2008; Zhang, 2010).
Nowadays, novel, active technologies for removing ice and snow are constantly emerging.
One such technological advancement is the anti-icing asphalt mixture technology. The indus-
try is highly concerned with this development as it is an environmentally-friendly approach
that effectively melts snow and ice, and some pavement trials have already confirmed the
effectiveness of anti-icing technology (Lu et al., 2009). However, it is important to note that
cars driving over the area and rain erosion causes some of the anti-icing material to be elimi-
nated from the pavement surface, and thus, the anti-icing effect is reduced (Sun, 2012; Sun,
2011). As a result, it is necessary to experimentally confirm the duration of the anti-icing
effect. Thus, this paper mainly focuses on a method for evaluating the functional life of the
anti-icing asphalt mixture.
2 MATERIAL PROPERTIES
The anti-icing asphalt mixture consists of aggregate, binder, mineral filler, and snow—melt-
ing filler. This mixture exhibits the same mechanical performance as normal pavement, but
665
it can actively remove ice and snow as well. Firstly, the properties of the virgin materials, the
aggregate gradation, and the pavement performance are all investigated prior to assessing the
duration of the ice and snow-melting effect.
666
Table 1. Performance parameters of anti-icing asphalt mixture AC-13.
Anti-icing Anti-icing
Test items Unit filler A filler B Requirement Test method
After verifying the asphalt mixture’s basic performance, the effectiveness and duration of the
anti-icing property of the anti-icing asphalt mixture were investigated.
close to the assumed accumulated precipitation. Thus, the cumulative annual precipitation
on the pavement is used as the most unfavorable soaking conditions. The maximum annual
precipitation in nearly a decade is defined as the most unfavorable soaking amount hmax of
the anti-icing asphalt pavement structure for a certain region. The modeling process is shown
below in Fig. 2.
According to the similarity principle and assumptions, the separating process of the anti-icing
filler in actual pavement can be investigated by soaking a specimen in water. The ratio between
the maximum volume of soaking water and the specimen volume is essentially equal to one of
the accumulated precipitation height in a year hmax and the thickness of the anti-icing asphalt
pavement hpavement. Thus, the soaking model is constructed as the following equation:
π × D 2 × hspecimen h max
Soaking model : Vmmax = (1)
4 × h pavementt × 106
where Vmax is the maximum volume of soaking water, L; D is the specimen diameter, mm;
hspecimen is the specimen height, mm; and hmax is the accumulated precipitation height on the
pavement in a year, mm; and hpavement is the thickness of the surface layer, mm.
After constructing the evaluation method for anti-icing effectiveness, the functional life can
be investigated. Based on the Beijing region, data was obtained for evaluating the anti-icing
functional life.
668
4.1 Test results
According to meteorological data, the maximum annual rainfall hmax in Beijing for a decade
is 721.1 mm. The surface layer thickness is 40 mm, and hpavement = 40 mm. The maximum vol-
ume Vmax in the laboratory can be calculated based on Equation 1. Thus, Vmax is calculated
as 9.26 L. The snow-melting coefficient K is determined for the anti-icing asphalt mixture A
and B, and the measuring period lasted 125 days. Fig. 3 displays the test results.
As seen from Fig. 3, K increases linearly with time. The relationship curve is fitted for
the snow-melting coefficient and time by the linear regression method. The linear equation
is established, as shown in Equation 2. The relation coefficient R of the equation for both
mixture A and B is greater than 97%.
K = αt + β (2)
This curve describes the snow-melting effectiveness and duration of the anti-icing asphalt
mixture. Coefficient α characterizes the separating rate of the anti-icing filler from the mix-
ture. The larger the value of α is, the more rapid the dissolution of chlorine ions is. At the
same time, the anti-icing filler is also rapidly eliminated, and the duration is short. Thus, α
is defined as the time factor of the snow-melting performance, and β is defined as the ability
factor of the snow-melting performance on behalf of the snow-melting effect. The larger the
value of β is, the more the chlorine ions that separate out are and the more obvious the snow-
melting ability is. As a result, the snow-melting effect is better. The two coefficients, α and β,
that are obtained by the above evaluation method can reflect the effect and duration of an
anti-icing asphalt mixture’s snow-melting property.
According to the fitting results, α = 0.0006 and β = 0.447 for mixture A, whereas α = 0.0012
and β = 3.088 for mixture B. As seen in Table 2, the snow-melting ability of mixture B is
higher than that of mixture A, but mixture A has a longer snow-melting duration time. As
illustrated in Figure 4, this can be explained by the difference in how the two anti-icing mix-
tures distribute into the mixture. The particle size of anti-icing filler A is finer than that of
mixture B; therefore, if the two fillers are evenly distributed in the mixture, a lesser propor-
tion of anti-icing filler A is exposed to the pavement surface than anti-icing filler B. Thus,
anti-icing filler A separates at a slower rate than anti-icing filler B, but A has a longer dura-
tion time than B.
669
Table 2. Snow-melting parameters.
Anti-icing asphalt
Mixture type Anti-icing asphalt mixture A mixture B
( K max − β )
f rediction model: Deffective =
Functional lifep (day ) (3)
α × 24
where Deffective is an effective snow-melting day, β is the ability factor, and α is the time
factor.
The Kmax values for the two mixtures were determined by the conductivity test. Kmax is 7.32
for mixture A and 11.94 for mixture B. Thus, the effective snow-melting days in Table 2 are
calculated according to Equation 3.
As seen in Table 2, the Deffective of mixture A is 477 days and 236 days for mixture B. When
90 days of rainfall and snowfall each year is adopted, the anti-icing functional life can be
obtained. The results indicated 4–5 years for A and 2–3 years for B.
Two years after completion, the in situ snow-melting performance of mixture B is assessed
by the Silver chloride AgCl precipitation method. As Figure 5 shows, after the silver nitrate
670
AgNO3 titration, there is still white precipitate on the pavement surface. This demonstrates
the presence of the anti-icing filler after two years. Thus, the predicted result is credible.
5 CONCLUSIONS
REFERENCES
Benjajnin T Green, Kerop D Janoyan. 2005. Use of Electrically Conductive Concrete Overlays for
Passive Control of Snow and Ice on Road. International Conference on Energy Environment and
Disasters, Charlotte, USA, 7: 795–801.
Cui Longxi. 2010. Research on Asphalt Mixture including salt. Chong Qing Jiaotong University:
16–17.
Hong Dan. 2010. Study on rough riding surface for antiicing. Chong Qing Jiaotong University: 9–11.
JTG E20–2011, Standard Test Methods of Bitumen and Bituminous Mixtures for Highway Engineer-
ing. (Chinese Standard).
JTG F40-2004, Technical specification for construction of highway asphalt pavement. (Chinese
Standard).
Sun Yuqi. 2011. Study on the performance of asphalt pavement with snow melt salt. Chang’an Univer-
sity: 1–2.
Lv Lili, Zhang Lijuan, Guo Yunxia. 2009. Chemical Freezing Check Pavement Application Summary.
Transport Standardization (Chinese Journal), (206):163–165.
Sheri Deicing. 2000. Mixing Proportioning, Optimizing and Properties. ACU Materials Journal, (97):
172–181.
Sun Rongrong. 2012. Develop and Evaluate of Domestic Pavement Include Salt. Harbin Institute of
Technology: 23–29.
Tan Yiqiu. Could“deicing”win“freeze”?-pavement deicing technology in the snow disaster. http://www.
sciencenet.cn/htmlnews/2008241230442200889.html.
Wang Xinfa. 2011. Research on the water stability of asphalt mixture and water damage of pavement.
Zhejiang University of Technology: 41–44.
Zhang Lei. 2009. Experiment Study and Numerical Simulation on the Characteristics of Road Ice-snow
Melting.Ji Lin University: 34–37.
Zhang Lijuan. 2010. Study of Salt Anti-freezing Asphalt Mixtures. Chang’an University: 21–25.
Zhang Chuanliang, Zhang Lijuan, Wu Xirong, Li Yafen. 2010. Chemical Freezing Check Pavement
Application in Foreign Countries. Transport Standardization (Chinese Journal)(15): 50–52.
Zhang Hongwei, Pan Wenjun, Shen Lu, Ren Quan. 2009. Study on the Laboratory test of melting
snow and ice on asphalt mixture. Journal of China & Foreign Highway (Chinese Journal), 29(01):
220–222.
671
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Based on the existing fatigue evaluation methods and evaluation indexes
of asphalt, Fatigue Factor test, Time Sweep test controlled by constant strain and Liner
Amplitude Sweep test were conducted on different recycled asphalts. Fatigue properties and
applicability of fatigue evaluation methods and indexes of recycled asphalt were investigated.
Also difference mechanism of anti-fatigue performance of recycled asphalts was analyzed by
conducting FTIR test and analyzing master curves results. Results showed that fatigue factor
test and Fatigue factor couldn’t be recommended as the fatigue evaluation method and index
of recycled asphalt. Considering definite physical meaning and sensibility to regenerant con-
tent, Nf3-fatigue performance parameter index was recommended as the fatigue evaluation
index of recycled asphalt and fatigue life could be considered to determine the regenerant
content, so an optimal and a maximum regenerant content were from 2% to 4% and 6%
respectively; FTIR results suggested that there were existing obvious distinctions between
regenerant and asphalt but no chemical reaction between them. And the difference of fatigue
property of recycled asphalt was influenced by alkanes stretching vibration absorption peak
area from 2923 cm−1 to 2924 cm−1 and alkanes bending vibration absorption peak area at
1461 cm−1 and 1671 cm−1. Also Complex modulus and phase angle significantly influenced
the fatigue property of recycled asphalt. A better fatigue life would appeare when complex
modulus was at an intermediate range, smaller phase angle at intermediate frequency seg-
ment meant a better fatigue life.
1 INTRODUCTION
So far, with the coming maintenance period of the aging road in China, the old asphalt pave-
ment’s recycling has became an important rehabilitation mean because of its many benefits
such as economical, environmental and sustainable. As a result, regeneration technique of
aging asphalt pavement has became a popular topic. As for asphalt pavement recycling tech-
nology, much more concentration was focused on the study of RAP performance (Gonzalo.
2011, Shu 2008, Maupin. 2008, Imad L. 2013, Shen 2007, Bennert 2007, Behnia 2009) because
RAP could be studied directly and conveniently without complex operation. But a lot of
complex procedures exist to conduct research on performance of recycled asphalt such as
aging asphalt extraction for a long time, aging asphalt distillation for a long time and asphalt
regenerating, these complex procedures led to little research on study of recycled asphalt
performance. Little references conducted on the fatigue properties of recycled asphalt were
found. However, the aging and damage of asphalt were the main factors that resulted in aging
and fatigue damage and some other disasters of asphalt pavement. So conducting research
on the fatigue properties of recycled asphalt was necessary.
The most widely used methods to estimate the fatigue properties of binder include
fatigue factor test, time sweep test controlled by constant strain or by constant stress and
Liner Amplitude Sweep (LAS) test (Bahia 1999, A Shenoy 2002, Bonnetti, 2002, Kim
2003, Planche 2004, Tan 2008). And a lot of indexes existe to evaluate the fatigue life of
asphalt; G*sinδ proposed by SHRP is the most widely used anti-fatigue performance index
of asphalt. The loading number gained from time sweep test, which means the modulus
reduced to half of the initial modulus, was also one of the most widely recommended
673
indexes to determine the damage of binders. Accumulative Energy Dissipation Rate
(DER) proposed by NCHRP9–10 was implemented to investigate the fatigue properties
of asphalt. Considering the damage of asphalt, according to AASHTO Designation: TP
101–14, the fatigue performance parameter-Nf3 which calculated from VECD (Kim 1988,
Lee 1996, Kim 2003) model through analyzing LAS test results has been widely adopted to
estimate anti-fatigue performance of binders. Research showed that G*sinδ had a univer-
sal applicability to evaluating the fatigue properties of basic asphalt but modified asphalt.
Time sweep test and LAS test were common tools to investigate the fatigue properties of
modified asphalt. However, there was little systematic study on the fatigue properties of
recycled asphalt. Fatigue evaluation method and fatigue evaluation index have not been
widely proposed, recognized and accepted.
Based on the above, the objective of this study was to investigate the anti-fatigue perform-
ance of recycled asphalts regenerated by different contents of regenerant and to determine
fatigue evaluation method and fatigue evaluation index of recycled asphalts. Also the dif-
ference mechanism of anti-fatigue performance of recycled asphalts was simply analyzed.
Series of tests were conducted on recycled asphalts containing different contents of recy-
cling agent in the paper. Anti-fatigue performance evaluation method and index of recycled
asphalt would be recommended. Reference value and theoretical basis would be provided
for subsequent research on how to investigate and evaluate the anti-fatigue performance
of recycled asphalt better. Also a theoretical basis would be provided for determining the
optimum regenerant content and the maximum regenerant content of recycled asphalt
through the research. The paper was of great value for shedding light on how to improve
the fatigue life of RAP.
3 MATERIALS
Aged asphalt was taken from an aged arterial road in Heilongjiang province after 2-hour
extraction process and 2-hour distillation process from RAP. Through the process, old
asphalt was completely recycled without any trichloroethylene remained in old asphalt. Then,
old asphalt regenerated by combining with three kind of regenerants and the final regenerant
produced in Liaoning province was selected according to the results of penetration, soften-
ing point and ductility of recycled asphalt. And five contents of regenerant were determined
which were at 0%, 2%, 4%, 6%, 8%, 10% respectively by the weight of old asphalt.
Based on the existing literature (Chen 2011), the regeneration condition was presented
as follows, old asphalt was reheated in the oven at 125ºC for 3 hours to make it flow easily.
Then, the regenerant was added and combined with old asphalt evenly by manual mixing for
15 minutes. Finally, the mixture was put back into the oven at 125ºC for 1 hour to make the
regenerant and old asphalt mix fully and adequately. After a series of processes, specimens
were prepared and a series of tests was conducted.
4 TEST METHODS
674
strain at 1% employing DHR instrument produced in TA company in America. Each testing
temperature was at 28ºC, 25ºC, 22ºC, 19ºC, 16ºC, 13ºC respectively. Critical temperature of
asphalt was finally achieved.
Factors Test frequency/Hz Test temperature /ºC Control mode Binder type
675
5 RESULTS AND ANALYSIS
676
Figure 2. Complex modulus and loading number curve of each binder.
performance was obtained and when the regenerant content reached 10%, a poor anti-fatigue
performance was presented. So, as for the fatigue performance of regenerating asphalt, the
conclusion that there exists an optimum content and a maximum content of regenerant is
validated again.
5.1.4 Applicability and sensitive analysis of fatigue evaluation methods and indexes
From previous analysis, three kinds of fatigue evaluation methods and four kinds of fatigue eval-
uation indexes were presented, applicability and sensitive analysis was conducted between them.
677
Figure 4. DER curve and Nf1 index of 4% regenerant content of regenerating asphalt.
For the fatigue factor test—fatigue factor is the loss modulus of binders and it is mainly
affected by whether the binder is soft or not—softer binder meant a smaller fatigue factor, but
it can be easily found that for the same stress level, softer binder is much easier to undergo
678
deformation even though it has a better deformation capacity, so it’s hard to say that softer
binder displays a better anti-fatigue performance. As far as regenerating asphalt is concerned,
research results supported that fatigue factor test could not be recommended as a fatigue
evaluation method and fatigue factor index couldn’t be used to evaluate fatigue properties.
Time sweep test and LAS test showed almost the same tendency from the results analysis,
fatigue life was enhanced with regenerant dosage at 2% and 4% and was lower than the origi-
nal binder with regenerant dosage at 8% and 10%. The correlation analysis was investigated
between the three fatigue parameter-Nf1, Nf2 and Nf3 as shown in Figure 7. It can be seen
that a better correlation between Nf3 and Nf2 was obtained than the correlation between Nf1
and Nf2 which means that Nf3 and Nf2 showed almost the same tendency. But the difference
mechanism between the optimum regenerant content from the perspective of each index was
worth considering and a further study.
With the purpose of determining the sensibility between fatigue indexes, single-factor
analysis of variance was conducted taking Nf3 index as an example and presented in Table 3.
F value of ANOVA of Nf1 and Nf2 were 25.22 and 43.34 respectively. The conclusion can be
made that the three parameters were all sensitive to regenerant content particularly for Nf3
index, considering the definite meaning of Nf3, it can be recommended as a fatigue evaluation
index of regenerating asphalt which was recycled from matrix asphalt; however, the mixture
performance verification needs to be done in further study.
Figure 7. Correlation analysis results between the three fatigue parameter-Nf1, Nf2 and Nf3.
ANOVA Sum of Squares Degree of freedom Mean square F Value Significance level
679
that alkanes stretching vibration absorption peaks from 2923 cm−1 to 2924 cm−1and alkanes
bending vibration absorption peak at 1461 cm−1 and 1671 cm−1 decreased with regenerant
increased. Also the curves became flat which means that the aging degree of the asphalt is
reduced with increased regenerant, it means peak area at those range of wavelength influence
the fatigue life of recycled asphalt.
680
Figure 10. Complex modulus master curves of different binders.
ance, at this moment, asphalt and regenerant combined with each other very badly and part
of regenerant may be free among asphalt components.
As far as the phase angle curves results at intermediate frequency segment are concerned,
from Figure 11 it is evident that smaller phase angle means a better fatigue life by comparing
to fatigue indexes presented above.
Based on the testing and analysis presented herein, conclusions are summarized as follows:
1. Fatigue factor test couldn’t be recommended as the fatigue evaluation method for recycled
asphalt index-G*sin δ decreased with regenerant content increasing.
2. Considering its definite physical meaning and sensibility to regenerant content, Nf3 -fatigue
performance parameter index was recommended as the fatigue evaluation index of recy-
cled asphalt, but mixture performance verification needs to be done in future research.
3. Fatigue life could be used to determine the optimal and maximum regenerant content of
recycled asphalt. An optimal and a maximum regenerant content were found at 4% and
6% respectively in this paper. Recycled asphalt with a 10%-content of regenerant showed a
poor fatigue property.
681
4. FTIR test results showed that no chemical reaction happened between regenerant and
asphalt. The difference of fatigue property of recycled asphalt was influenced by alkanes,
stretching vibration absorption peak area from 2923 cm−1 to 2924 cm−1 and alkanes bend-
ing vibration absorption peak area at 1461 cm−1 and 1671 cm−1.
5. Complex modulus and phase angle significantly influenced the fatigue property of recycled
asphalt. A better fatigue life would appear when complex modulus was at an intermediate
range, smaller phase angle at intermediate frequency segment means a better fatigue life.
The paper discussed the fatigue property of regeneration of matrix aging asphalt extracted
from RAP. The applicability that has been discussed in the paper was suitable for modified
asphalt such as SBS modified asphalt and rubber asphalt or not need to be further verified.
Also, the mixture performance verification for fatigue properties of recycled asphalt needs to
be done in following research.
ACKNOWLEDGEMENTS
The authors are grateful to the financial support by National Science Fund for Distinguished
Young Scholars of China (51225803) and the Fundamental Research Funds for the Central
Universities (HIT.NSRIF.201662).
REFERENCES
Al-Qadi, Imad L. & Elseifi & Mostafa. 2013. Reclaimed Asphalt Pavement—A Literature Review [J].
Illinois Center for Transportation:, ISSN: 0197–9191.
A Shenoy. 2002. Fatigue Testing and Evaluation of Asphalt Binders using the Dynamic Shear Rheom-
eter. Journal of Testing and Evaluation., 30(4):303∼305.
Behzad Behnia & Eshan Dave & Sarfraz Ahmed. Effects of Recycled Asphalt Pavement Amounts on
Low-Temperature Cracking Performance of Asphalt Mixtures Using Acoustic Emissions[R]. Journal
of the Transportation Research Board: List of Issues, Volume 2208, DOI: 10.3141/2208–09.
Bonnetti K.S. & Kitae Nam & Bahia H.U. 2002. Measuring and Defining Fatigue Behavior of Asphalt
Binders. Transportation Research Record (No 1810).Transportation Research Borad: 33∼43.
Gonzalo V & Félix P.J & Rodrigo M & Adriana M & Ramón B. 2001. Experimental study of recycled
asphalt mixtures with high percentages of reclaimed asphalt pavement (RAP) [J]. Construction and
Building Materials: 25( 3): 1289–1297.
Bahia H.U., Zhai H. & Bonnetti 1999. Non-Linear Viscoelastic and Fatigue Properties of Asphalt Bind-
ers. The Journal of Association of Asphalt Paving Technologists, 68: 1∼34.
Jing-yun Chen. 2011. Experimental Study on Recyeling Methods of Asphalt Pavements[D]. DaLian:
Ph.D. Dissertation of Dalian University of Technology: 12–26.
Lee H.J. 1996. Uniaxial Constitutive Modeling of Asphalt Concrete Using Viscoelasticity and Con-
tinuum Damage Theory. Ph.D Dissertation of North Carolina State University:45∼56.
Maupin Jr & Diefenderfer G W & Stacey D. 2008. Evaluation of Using Higher Percentages of Recycled
Asphalt Pavement in Asphalt Mixes in Virginia [R]. Transportation Research Board: 8–22.
Planche J.P & Anderson D.A & Gauthier G. 2004. Evaluation of Fatigue Properties of Bituminous
Binders. Material and Structures. 37:356∼359.
Shen, J. & Amirkhanian, S. & Aune Miller, J. 2007. Effects of Rejuvenating Agents on Superpave Mixtures
Containing Reclaimed Asphalt Pavement [J]. Journal of Materials in Civil Engineering: 19(5): 376–384.
Shu X & Huang B.S & Dragon V. 2008. Laboratory evaluation of fatigue characteristics of recycled
asphalt mixture [J]. Construction and Building Materials: 22( 7): 1323–1330.
Tan Y.Q & Shan L.Y & Li X.M 2008. Fatigue Characteristic of Asphalt Pavements and Materials: Char-
acterization, Modeling, and Simulation. Geotechnical Special Publication, 182: 98∼107.
Thomas B, Raj D 2014. Backcalculation Method to Determine Effective Asphalt Binder Properties
of Recycled Asphalt Pavement Mixtures[R]. Journal of the Transportation Research Board: List of
Issues, Volume 2179, DOI: 10.3141/2179–09.
Y.R. Kim 1988. Evaluation of Healing and Constitutive Modeling of Asphalt Concrete by Means of
the Theory of Nonlinear Viscoelasticity and Damage Mechanics. Ph.D. Dissertation of Texas A&M
University: 80∼103.
Y.R. Kim 2003. Mechanistic Fatigue Characterization and Damage Modeling of Asphalt Mixture.
Ph.D. Dissertation of Texas A&M University: 19∼23.
682
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Bin Xu
Dalian University of Technology, Dalian, Liaoning, P.R. China
Research Institute of Highway, Ministry of Transport, Beijing, P.R. China
Chuanzhong Lu
Jiangsu Coast Expressway Management Co. Ltd., Yancheng, Jiangsu, P.R. China
1 INTRODUCTION
For porous asphalt pavement, large-porosity asphalt mixture is used as the surface layer. On
rainy days, the water can permeate into the pores of the surface and horizontally flow out of
the pavement. Such kind of pavement has the technical features such as high skid-resistance,
low noise, restraining water spray on rainy days, preventing “water-drifting” during high-
speed driving, and reducing glaring lights in the night (Cao et al. 2009). The porous asphalt
pavement, started in 1960s, was mainly applied to improve driving safety on the road surface,
and reduce the traffic noise in populous regions and those with dense road network. The
research and application of porous asphalt technology started in 1980s in China, and now
the high-viscosity modified asphalt is mainly adopted as the binder to improve the anti-
raveling performance. So far, porous asphalt pavement has been applied in more than ten
provinces, and the total length of porous asphalt section is more than 300 kilometers.
However, due to the large-porosity and open gradation characteristics, under the conditions
of repeated traffic loading and complex climate conditions, damages in porous asphalt pavement
are typically presented in the form of raveling, brittle cracking, pitting or a combination of them
(Zhao Su. 2012). Many scholars (Molenaar et al. 2006; Hagos. 2008; Akihiro et al. 2014) believe
that the most serious damage in porous asphalt is raveling (loss of the aggregate on the upper
surface), and this will significantly reduce the life expectancy of the pavement. Therefore only
683
with a good solution to raveling can a longer life cycle of porous asphalt pavement be expected.
In Texas, the United States (US), a research on preventive maintenance of porous asphalt with
rejuvenator was conducted, which found that, the rejuvenator can provide an additional unaged
asphalt film for the pavement and delay the raveling (Cindy et al. 2008). Zhang (2012) carried
out a research on the preventive maintenance technique of porous asphalt pavement by vir-
tue of the LVO (levensduur verlengend onderhoud)–ZOAB (zeer open asfalt beton) project in
Netherlands. Special rejuvenator was sprayed over a porous asphalt pavement. In this way, the
aggregate raveling is prevented and the service life of the pavement is extended.
This research is carried out based on the Enterprise Technical Innovation Project of Min-
istry of Transport of the People’s Republic of China. It aims at providing an applicable mate-
rial and method for preventive maintenance of porous asphalt with high-viscosity modified
asphalt as the binder. Based on the researches in this paper, an engineering application of
preventive maintenance technology for porous asphalt is conducted on Coast Expressway
(G15), the first porous asphalt section in southern rainy area in China.
A specially designed preventive maintenance material RC is selected for the research. It is con-
sidered to have the properties: 1) activate aged asphalt colloid and rejuvenate the activity and
elasticity of high-viscosity modified asphalt by refilling the polar compounds; 2) improve the
internal cohesive force and coherence of asphalt; 3) supplement the removed asphalt film on
aggregate surface. Another maintenance material used in this study is the Emulsified Asphalt
(EA), which is commonly used as preventive maintenance material in dense pavement. The
effects of these two types of materials on the property of porous asphalt are studied.
2.2 Adhesion
Adhesion of the preventive maintenance material refers to the physical and chemical adhesion
of the maintenance material with the aggregate in the mixture. The test is carried out under
humid condition. As in humid condition, the water can partly infiltrate into the contact sur-
face between the preventive maintenance material and the stones and remove the preventive
maintenance material from stones, the adhesion performance is evaluated by observing and
analyzing the removal degree of the material.
684
Table 1. Test results of rejuvenation of aged asphalt with maintenance materials.
Figure 1. The original core, wrapped core and the core after Cantabro Raveling Test.
The adhesion test on the preventive maintenance material and aggregate is conducted with
the water-boiling method of determining adhesion described in Standard Test Methods of
Bitumen and Bituminous Mixtures for Highway Engineering (JTG E20-2011). The aggregate
used for the test is Basalt. The specific test procedure is: 1) Place 500 g of 9.5-13.2 cleaned
aggregate in the oven at 105°C for 2 h; 2) Prepare the preventive maintenance materials, and
immerse 20 particles of aggregate into each of the preventive maintenance material; 3) place
the preventive maintenance material in the oven at 105°C for 15 min; 4) pick out the aggre-
gates and evenly put them in a plate in the constant temperature drier for certain period; 5)
take the aggregates out and place them under room temperature for another period of time;
6) place the aggregates in water bath of 80°C for 30 min and observe the adhesion situation.
As shown by the test results: The adhesion grade of RC is 5, which means RC can be well
attached to aggregate of the original pavement. In contrast, the adhesion grade of EA is 2.
The adhesion performance of RC with aggregate is obviously superior to that of EA.
Ai A1
A= × 100
A1 (1)
Bi B1
B= × 100
B1 (2)
C C1
C= i × 100
C1 (3)
686
Table 3. Results of test on application performance of preventive maintenance material RC with dif-
ferent amounts.
The influence coefficients of material amount on different properties of porous asphalt are
shown in Figure 2. With the increase of amount of RC, the performance influence coefficients
appear a rising trend expressed by an exponential (or logarithmic) function, and the correlation
coefficient reaches above 0.95. The raveling resistance performance significantly increases at the
initial stage with the increasing of material amount. However, the increasing trend slows down
and becomes stable with further addition of material. The permeability performance is only
slightly affected by the preventive maintenance material. The influence coefficients are all below
10% at the material amount observed. The influence coefficient on skid-resistance performance
slowly increases with the increasing amount when the amount is below 0.3 kg/m2. But with the
gradual addition of the material amount, the influence coefficient shows a sharply increase. As
shown in Figure 2, the increasing trend for each property changes around 0.3 kg/m2. Therefore,
0.3 kg/m2 is selected as the optimal amount of preventive maintenance material RC.
Brushing
Laboratory brushing and field brushing are both taken into account: laboratory brushing is
conducted with a small hairbrush, and the roller brush is used in the field brushing. Coat-
ing with the roller brush is shown in Figure 3. By observing the brushing effect, the material
amount was not evenly brushed on the road surface due to man-made factors. Therefore, the
brushing method is only suggested to be applied in laboratory study and repairing a small
sale road surface on field.
687
Figure 2. Influence of RC Amount on index performance.
688
Figure 5. Field spraying with intelligent spraying vehicle.
the intelligent spraying vehicle, it is recommended to adopt the intelligent spraying vehicle
to conduct the spraying of preventive maintenance material for the large-area preventive
maintenance engineering. The commonly used intelligent spraying vehicle in China is shown
in Figure 5.
Brief summary
By analyzing and studying the various spraying methods of maintenance materials, a sum-
mary for the spraying methods is given in Table 4.
4 ENGINEERING APPLICATION
In August 2015, engineering application of the preventive maintenance technology was con-
ducted for porous asphalt of Jiangsu Coast Expressway (G15) in China based on the achieve-
ments of this research. The special preventive maintenance material RC is adopted in the
project. Before preventive maintenance material spraying, the voids of porous asphalt were
first cleaned with the vacuum voids cleaning vehicle developed in China, to ensure the suffi-
cient contact reaction between the preventive maintenance material and porous asphalt mix-
ture. The permeability and friction coefficients were respectively measured before and after
the maintenance to investigate the functionality and security of porous asphalt pavement.
During the maintenance, the spraying amount was also measured, to ensure that the adopted
material was within the given range. Figure 6 is the picture of application of porous asphalt
preventive maintenance on field. Figure 7 shows the road surface treated by preventive main-
tenance technology. A further investigation is to be conducted in the future to evaluate the
effect of the preventive maintenance.
689
Figure 6. Process of preventive maintenance for porous asphalt pavement.
5 CONCLUSIONS
1. In this research, properties of preventive maintenance materials for porous asphalt are
investigated. It finds that the RC material shows promising effect on rejuvenating the
aged asphalt and repairing micro cracks of asphalt film; it also has good adhesion with
stone and can supplement asphalt film and improve adhesion performance. Thus it greatly
improves the raveling resistance performance of porous asphalt mixture.
2. The spraying amount of the preventive maintenance material RC shall meet the require-
ments of porous asphalt raveling resistance performance, and affects less on permeability
and friction coefficients of the original pavement. In this study, the spraying amount is
suggested to be 0.3 kg/m2 as it satisfies both the requirements.
3. Different technologies for applying the maintenance material on porous asphalt are dis-
cussed. The suitable application conditions for those methods are suggested.
4. Based on the materials and technology discussed in this study, a practical engineering
application of preventive maintenance for porous asphalt is carried out on an expressway
in China. A further investigation is to be conducted in the future to evaluate the effect of
the preventive maintenance.
690
ACKNOWLEDGEMENTS
REFERENCES
Akihiro M., Toshiro J., Takaaki N., Hiroshi I., Katsuya T., Akihiko K.. 2014. Construction and pave-
ment properties after seven years in porous asphalt with long life. Constr. Build. Mater, 50: 401–413.
Cao Dongwei, Liu Qingquan, Tang Guoqi. 2009. Porous Asphalt Pavement. Beijing: China Communi-
cations Press.
Cindy K. Estakhri, Allex E. Alvarez, and Amy Epps Martin. 2008. Guidelines on Construction and
Maintenance of Porous Friction Courses in Texas. FHWA/TX-08/0-5262-2. Texas Department of
Transportation, Austin, U.S.
Hagos, E.T. 2008. The Effect of Aging on Binder Properties of Porous Asphalt Concrete. Delft Univer-
sity of Technology, Delft, The Netherlands.
Molenaar, A.A.A., Meekerk, A.J.J., et al. 2006. Performance of Porous Asphalt. Journal of the Associa-
tion of Asphalt Paving Technologists, CD.
Research Institute of Highway Ministry of Communications. 2011. Standard Test Methods of Bitumen
and Bituminous Mixtures for Highway Engineering. Beijing: China Communications Press.
Yuan Zhang. 2012. Increasing the Service Life of Porous Asphalt with Rejuvenators, Sustainable Con-
struction Materials. 316–328.
Zhao Su. 2012. A Sustainable Maintenance Solution for Porous Asphalt Pavements via Rejuvenation
Technology, Sustainable Construction Materials. 387–403.
691
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Sustainability is an issue that today must be seriously considered in all struc-
tures. Sustainability may be described as a quality that reflects the balance of three primary
components such as economic, environmental and social impacts. A sustainable development
is the development that meets the needs of the present without compromising the ability of
future generations to meet their own needs. In the context of pavements the term sustainable
refers to a system characteristics that encompasses a pavement’s ability to: a) achieve the
engineering goals for which it was constructed, b) preserve and (ideally) restore surrounding
ecosystems, c) use financial, human, and environmental resources economically, and d) meet
basic human needs such as health, safety, equity, employment, comfort, and happiness. Long
life pavements or perpetual pavements affect positively the sustainability issue. This paper
describes in details the items considered in constructing a sustainable flexible pavement;
materials, design, construction, use-phase considerations, maintenance and preservation
treatments, and pavement end-of-life considerations. It also outlines the benefits of con-
structing a long life flexible pavement, also known as perpetual flexible pavement.
1 INTRODUCTION
Road infrastructure and in particular pavement infrastructure is one of the key factors
contributing to the quality of life and prosperity of the society. Any pavement structure by
the time is given to traffic usage starts to deteriorate over time and soon or later pavement
preservation or maintenance is necessary to be carried out so to continuously provide a high-
performing, safe, as well as, cost effective pavement network.
The current economic situation, world wide, the limited resources and finding and the
deterioration of the environment makes it more necessary than ever that authorities and
agencies seek ways to use the available resources and findings so to maximize the benefits and
protect the environment. This is to build sustainable and long lasting pavements.
Sustainability may be seen as a quality that reflects the balance of three primary compo-
nents: economic, environmental, and social impact.
2 DEFINITIONS
This paragraph provides some definitions used to the benefit of the reader.
Sustainable development: Sustainable development is the development that meets the
needs of the present without compromising the ability of future generations to meet their
own needs (Bruntdland Commission, 1987).
The term ‘Sustainable’ in the context of pavements: Refers to a system characteristics that
encompass a pavement’s ability to:
a achieve the engineering goals for which it was constructed,
b preserve and (ideally) restore surrounding ecosystems,
c use financial, human, and environmental resources economically, and
d meet basic human needs such as health, safety, equity, employment, comfort, and happi-
ness (FHWA-HIF-15-002, 2015).
693
Sustainable pavement in the context of environment (only): A pavement that minimizes
environmental impacts through the reduction of energy consumption, natural resources and
associated emissions while meeting all performances conditions and standards (Miller &
Bahia, 2009).
The durability of a pavement is related to its sustainability. Beginning of 2000’s two terms
related to durability have immerged: Long life pavement (European term) and Perpetual
pavement (USA term).
Long Life Pavement (LLP): A long-life pavement is a type of pavement where no signifi-
cant deterioration will develop in the foundations or the road base layers provided that cor-
rect surface maintenance is carried out (FEHRL, 2004).
Perpetual Pavements (PP): An asphalt pavement designed and built to last longer than 50
years without requiring major structural rehabilitation or reconstruction, and needing only
periodic surface renewal in response to distress confined to the top of the pavement (New-
comb et al., 2010).
The benefits of LLP or PP are:
a Low life-cycle cost, since deep pavement repairs or reconstruction during pavement’s life
time is not necessary.
b Lower user delay cost, since maintenance/rehabilitation works require lesser time hence
shorter period of traffic disruption.
c Lower environmental impact, since lesser material is required over pavement’s life time
maintenance/rehabilitation works and recycling may be applied.
Pavements started to be constructed thousands years ago. The type of construction each
period of time was to satisfy the current needs. Originally and for many years to come was
just to ensure that the road was passable throughout the twelve months of the year. Later
on, end of 19th and beginning of 20th century AC with the development of powered cars
and increase of traveling speed provision of comfort and safety was the predominant factors;
hence the development of flexible pavement. Figure 1 shows a graphical representation of the
historical evolution of pavements from 5th century BC to today.
As it can be seen, Figure 1, as early as the 5th century BC the pavements constructed by
the Greeks and later on by the Romans with the use of polygonal natural slabs exposed to the
user. The layers underneath were in sequence from bottom to top of rubble stones, smaller
694
Figure 2. Dion, Northern Greece, 424 BC Figure 3. Via Appia, Italy, 312 BC (3 m wide in
(4 m wide urban road). rural & 5 m wide in urban areas).
Figure 4. Via Egnatia, Northern Greece, 130 BC (3 m wide in rural & 5 m wide in urban areas).
stones, and gravel and sand layer. It was only until the end of 19th-beginning of 20th century
AC that asphalt mix started to be used in the top layer. With the tremendous increase of traf-
fic volume after 1950s the thickness of the asphalt layer also increased, as well as the total
thickness of the pavement.
Figures 2, 3 and 4 show examples of very old Greek and Roman roads.
All the above mentioned pavements achieved the engineering goals for which they were
constructed for, to preserve and restore surrounding ecosystems, using financial, human,
environmental resources in the most economically way and meeting the basic social needs.
It can therefore be concluded and even those days, nearly 2500 years ago, they use to build
sustainable pavements!
A typical flexible pavement structure consists of asphalt layers and unbound layers. The
distinct layers of a flexible layer are: surface course, binder course, asphalt base course, base
course, and sub-base course, which is sited over the subgrade (natural soil material) or over a
capping layer (imported good quality soil material).
695
When part of the asphalt base course and/or of the unbound layers are replaced with
hydraulically bound layers the pavement is known as semi-flexible or composite-flexible
pavement. Finally, when all unbound layers are replaced with asphalt layers the pavement is
known as full depth pavement.
5.1 Materials
5.1.1 Aggregate materials
The aggregate materials for asphalts, unbound and hydraulically bound mixtures can be:
Natural aggregates, crushed aggregates, slags, mine waste, demolition materials, or recycled
(pulverized) aggregates.
Good quality and long lasting aggregate materials is vital for sustainable construction.
Good quality means materials satisfying the engineering properties, such as: Geometrical
properties & cleanliness (particle size distribution, particle shape, percentage of crushed &
broken phases, cleanliness and quality of fines, etc.), Physical properties (resistance to frag-
mentation/crushing, resistance to polishing/abrasion/wear/attrition, particle density & water
absorption, durability against freeze/thaw, weathering, thermal shock, stiffening properties
of filler, etc), and Chemical properties (chemical composition, determination of unsoundness
of blast-furnace and steel slags, water solubility of filler aggregates, water susceptibility of
fillers for bituminous mixtures, etc.).
The requirements for each of the above mentioned properties can be found in appropriate
national standards. If European standards are employed these are CEN EN 13043/AC (2004)
and CEN EN 13242+A1 (2007).
Tips for improving sustainability from aggregate materials:
a Use crushed aggregates instead of uncrushed aggregates for unbound or hydraulically
bound mixes.
b Use only crushed, clean, and resistant to fragmentation aggregates for asphalt mixes.
c For surface course asphalt mixes use only crushed, clean and resistant to polishing/abra-
sion aggregates (limestones are not suitable).
d Minimize transportation cost.
e Reduce the amount of virgin aggregates used, and
f Built pavement on a subgrade with adequate bearing capacity (CBR >10 or even >20);
if required, provide capping layer or use stabilization techniques (cement or lime
stabilization).
5.1.3 Asphalt
The types of hot asphalts, in order of preference, used in Europe are: Asphalt Concrete (AC)
(dense), Asphalt concrete for thin (30–40 cm) (AC-TL), or for very thin layers (20–30 mm)
(AC-VTL), Stone Mastic Asphalt (SMA), Porous Asphalt (PA) or Open-Graded Asphalt
(OGA), Double Layered Porous Asphalt (2 L-PA), High-Modulus Asphalt (HiMA), Hot
Rolled Asphalt (HRA), Ultra thin layer asphalt concrete (10–20 mm) (UTLAC), Mastic
Asphalt (MA), and Soft Asphalt (SA).
The Asphalt Concrete (AC) (dense) is the most commonly used type of asphalt worldwide
for the construction of all asphalt layers. The aggregate particles are continuously graded and
the air voids, are normally between 3% - 6%.
Asphalt concrete for thin (30–40 cm) (AC-TL), or for very thin layers (20–30 mm) (AC-
VTL) is an asphalt mix developed in Europe (originally in France) in late ’80 s and over the
last 10 years is extensively used in many countries of Europe for surface layers. The aggregate
particles are gap-graded, and these asphalts provide very good noise reduction, have good
697
drainage ability (air voids within 8% to 15%) and due to lesser amount of material required
per unit area, they are cheaper then AC surface layer.
Stone Mastic Asphalt (SMA) is asphalt developed in Germany primarily for long lasting
surface layer, but also for binder course. The aggregate particles are gap-graded and the air
voids of the asphalt mix are within 3% to 4%. Due to high percentage of binder incorporated
it normally requires the addition of fibers. SMA is perhaps the most expensive material of
all asphalts.
Porous Asphalt (PA) or Open-graded asphalt (OGA) is exclusively for surface layer.
Almost single size aggregate particles are used, and PA (or OGA) provides the highest possi-
ble noise reduction, excellent surface drainage ability, and minimization/elimination of spray
(air voids > 18%). PA is more expensive than conventional AC since normally requires the
addition of fibers. They are normally laid at a thickness of 50 mm.
Double Layered Porous Asphalt (2 L-PA) is again for surface layer. The bottom layer is
with course aggregate (usually 11/16 mm), and the top layer with finer aggregate (usually
4/8 mm); total thickness approx. 70 mm. 2 L-PA seem to provide slightly greater service life
than PA but they are more expensive than PA.
High-Modulus Asphalt (HiMA) is a dense material with high stiffness modulus; it pro-
vides high resistance to rutting, has good spreading ability and good durability. It is used
only for base or binder courses and is more attractive to be used in heavily trafficked area.
Hot Rolled Asphalt (HRA) (limited use, UK mainly) is a gap-graded mix in which the
mortar of fine aggregate, filler and high viscosity binder are the major contributors to its
performance. Nowadays is used as binder/base course material and to a much lesser extend
as surfacing material. HRA is quit expensive in comparison to AC due to high percentage of
binder incorporated and the air void content is within 2% to 3%.
Ultra thin layer Asphalt concrete (10–20 mm) (UTLAC) is a propriety, gap-graded
open surface texture mix. It is exclusively used for surface layer and requires special paving
machinery.
Mastic Asphalt (MA), for surface layers (not to be confused with MA for waterproofing
bridge decks) is used in some countries of Europe. It is a ‘voidless’ mix with high filler con-
tent, high binder content, and requires the use of hard bitumen. It requires the addition of
chippings after laying, is of high initial cost but has long life duration.
Soft Asphalt (SA) (for Nordic countries only, and for secondary roads) is a dense mix for
all layers, it requires the use of soft binder and is of limited use.
The durability of those asphalts used as surfacing material varies which in turn affects the
construction sustainability. Figure 5 show the expected durability of all surfacing materials
when used in motorways or other secondary roads.
Apart from hot asphalts there are also warm and half-warm mix asphalts.
Warm Mix Asphalts (WMA) and Half-Warm Asphalts (HWA) are asphalts produced,
handled and compacted at temperatures lower than those of conventional hot asphalts (70°C
to 130°C). This is achieved by lowering the viscosity of the binder using additives (organic or
chemical), applying foaming process, or bitumen emulsion-based processes, or modification
of binder/aggregate mixing processes (proprietary processes), all known as WMA systems.
The aggregate mix gradation is the same as for the hot mix variety. Producing and using
WMA or HWA fuel consumption, environmental and paving benefits are resulted.
Finally, cold mix asphalts are also used. Cold mix asphalts are asphalts produced, handled,
and compacted at ambient temperatures (7°C to 45°C).This is achieved by using bitumen
emulsion as binder material. The aggregate mix gradation is the same as for the hot mix vari-
ety. However, only AC gradations are mostly used.
Exclusively for surface restoration, finer continuous aggregate mix gradations are used
and the cold mix is known as Slurry sealing or Micro-surfacing.
Producing and using cold mix asphalts further fuel consumption, environmental and pav-
ing benefits are resulted.
Table 1 provides information to the energy consumed and the greenhouse gases emit-
ted during the production and laying of 1 t of most commonly paving materials used, in
general.
698
Figure 5. Asphalts used in surface layers and their life expectancy (EAPA, 2007).
Table 1. Energy consumed and greenhouse gases emitted for the production of 1 t of road paving
material (Nikolaides, 2014).
g Always run Life Cycle Assessment (LCA) to quantify environmental impacts over the
entire life cycle of pavement systems.
h As for rehabilitation always choose long lasting solutions.
702
e As for the reflectivity, heat capacity and conductivity, very little can be done at the moment.
The future use of bio-products as a replacement of bitumen binder is promising.
REFERENCES
704
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The growing needs for sustainability demand that all asphalt concrete layers
in an asphalt structure can be produced with high quality, low production temperature and
using high percentages of reclaimed material. By using LEAB bitumen foaming technology,
binder/base layer asphalt mixtures containing up to 60% reclaimed material can be produced
at about 110°C with good performance. The lowered production temperature results in 25%
reduction of CO2 emission and 40% reduction in energy. With the support of the European
LIFE+ program it is shown that a prous asphalt layer can also be recycled horizontally. The
reclaimed porous asphalt is first decomposed into reclaimed stone and reclaimed mortar.
By rejuvenating and foaming of the reclaimed mortar, and mixing with the reclaimed stone,
porous asphalt is designed containing more than 90% recycled material and produced at
around 110°C. This porous asphalt is proven to be identical as the hot mix in terms of mortar
and mixture performances.
1 INTRODUCTION
With the growing concerns of government bodies on climate change and, the historic Paris
Agreement on Climate Change has been signed sustainability on 12 December 2015 by 195
nations. The agreement aims to keep the global temperature rise in this century below 2°C
and even striking to 1.5°C. As a result, sustainable technologies enabling the high quality
recycling of asphalt pavements are more and more popular in the Netherlands and all over
the world. The Netherlands is one of the most active countries in asphalt recycling. In the
Netherlands, a typical asphalt pavement structure on the Dutch motorway system consists a
porous asphalt (PA) surface layer placed over a dense binder and base layer. During renovation
activities, a large amount of reclaimed asphalt pavement is obtained and reused. According
to the European Asphalt Pavement Association, the Netherlands produces 9.3 million tons
of asphalt and 4.5 million tons of Reclaimed Asphalt (RA) every year (EAPA 2013). And
about 80% of the available RA is re-used in new asphalt production. However, most of the
asphalt recycling is downgrade recycling. For example, the reclaimed porous asphalt RA is
commonly not reused in constructing new asphalt mixtures, although research indicates that
a maximum of 30% of porous asphalt RA may be used in the production of these layers.
In binder and base layer asphalt mixtures, the RA can be used as an ingredient. Practically,
60% RA is used with success without extra addition of rejuvenators. Attempts with higher
percentage recycling up to 100% together with rejuvenators give also promising results.
From the point of view of social and economic needs, recycling horizontally is necessary.
Horizontal recycling means that the recycled surface layers such as porous asphalt, SMA
etc. to be reused as newly produced porous asphalt and SMA. And the recycled binder/base
layers to be reused as newly produced binder/base layers. And no asphalt is to be recycled as
base or subbase.
On the other hand, the application of the warm mix technology is also getting more atten-
tion due to its economic and social-environmental benefits (D’Angelo 2008). The principle of
the using foamed bitumen was developed in the 1950s (Csanyi 1957). When hot bitumen and
water meet each other, a large expansion in volume can be expected at a temperature around
705
100–110°C. This allows the binders to effectively coat the warm mineral aggregates and to
produce a half-warm/warm asphalt mixture at a temperature around 90–110°C (D’Angelo et
al 2008; Jenkins 2010).
As a result, this paper discusses the possibilities of high-quality high-percentage recycling of
a motorway asphalt structure by the use of foaming technologies. The total asphalt structure
will be horizontally reconstructed at a temperature of 110°C. The percentage of recycling
for binder/base layer is about 60% and the percentage of recycling for porous asphalt layer
is above 90%!
2.1 Introduction
As shown in Figure 1, the basic principle of the LEAB technology is to produce asphalt
mixtures using with the foam technology. The LEAB (Dutch acronym for Low Energy
Asphalt Concrete) technology, which was developed by BAM, has been successfully applied
in practices for more than 10 years (Jacobs et al. 2010). Until now, more than 400000 ton
LEAB asphalt mixtures have been produced in about 200 road construction projects includ-
ing motorways, provincial roads and streets. LEAB production facilities are installed in 4
asphalt plants in the Netherlands. The LEAB mixtures produced at 90–110°C have the same
composition and performance characteristics as their hot mix equivalents produced around
150–170°C. In this paper, the LEAB technology is used to produce a high quality bin/base
asphalt layer containing 60% RA.
2.2 Materials
Table 1 gives an overview of the materials used in the tests. There are three types of binder/
base asphalt mixtures described namely, LEAB 22 bin/base with 60% RA, AC bin/base with
60% RA and AC 22 bin/base without RA, respectively. It is observed that the LEAB 22 with
60% RA has identical compositions as its hot variant AC 22 bin/base with 60% RA. The
reclaimed asphalt 0/20 fraction has a bitumen content of 5.0% and a bitumen penetration
of 22 (0.1mm). In order to balance the total penetration of the mixtures, a 70/100 bitumen
with a penetration value of 75 (0.1 mm) was used for mixtures containing 60% RA. The
mixtures AC 22 bin/base with 60% RA and AC 22 bin/base without RA were produced at
around 165°C, and the mixtures LEAB 22 bin/base with 60% RA were produced at around
110 °C.
Figure 1. Principle of bitumen foaming and LEAB foaming installations in the asphalt plant (Jacobs
et al. 2010).
706
Table 1. Overview of materials and receipts used for binder/base layer asphalt mixtures.
708
3 SUSTAINABLE RECYCLING OF POROUS ASPHALT
3.1 Introduction
Porous asphalt is being applied as a surface layer on the Dutch motorway system since the
early 1980’s. Application of porous asphalt concrete on the primary road network is manda-
tory and as a result approximately 90% of this network has a porous asphalt surface layer.
This type of asphalt, which has an air void percentage around 20%, has great advantages with
respect to noise reduction and reducing splash-spray, etc. However, due to its high void ratio
porous asphalt can only perform when it is of highest quality. This implies that porous asphalt
can only be produced when there is full control over mix composition and the quality of used
ingredients. Because of this conventional recycling of this mixture is not feasible. It is thus of
importance to develop a new horizontal recycling approach for porous asphalt surface layers.
In 2013, under a grant of the European Life+ program, BAM proposed an innovative way
of horizontal recycling of porous asphalt. Under the project named “Low Emission2 Asphalt
Pavement (LE2AP)”, a 1 km porous asphalt layer will be constructed containing more than
80% of recycling material and produced as a warm mix (Huurman 2015). As a result, this
project will realize three goals; lowering CO2-emissions, increase the rate of recycling and
reducing noise.
In the LE2AP approach, the reclaimed porous asphalt passes through a rotary decomposi-
tion device and is separated into reclaimed aggregates with a small amount of bitumen and a
bitumen-rich mortar fraction. The rotary decomposition device is a machine that uses the glassy
behavior of asphalt mixtures at high frequencies. In the decomposition process, the milled off
asphalt is thrown against a steel wall with certain speed, which results in extremely high load
frequencies. At these frequencies, the bitumen and mortar show glass-like behavior at impact
and brittle failure occurs. This results in the peeling off of the mortar film that surrounds
individual aggregates. The reclaimed aggregates, called PA-stone, can be reused as high quality
aggregates without further treatment. The reclaimed mortar is first treated and homogenized
before it is reused as mortar for a new mixture. With a special foaming technology, the renewed
mortar is foamed and mixed with the well-fractionized PA-stone. As a result, high quality warm
produced porous asphalt containing a high percentage of reclaimed material can be produced.
3.2 Materials
Figure 4 and Table 2 illustrate the materials obtained after the rotary decomposition proc-
ess. The bitumen-rich mortar contains about 10.5% bitumen and the reclaimed aggregates
contain about 1% bitumen.
Table 3 gives the recipe of the LE2AP mortar and its fresh equivalent. The mortar contains
sand, filler and bitumen, which is the binding agent of the porous asphalt. For this reason it
is of great importance that the LE2AP mortar, produced on basis of reclaimed mortar, has
the same characteristics as a freshly produced mortar. The LE2PA mortar is designed such
that the bitumen content and the bitumen penetration are similar to that of the fresh mortar.
Figure 4. Materials obtained after decomposition: reclaimed mortar sand and PA-stone.
709
Table 2. Descriptions of materials obtained after decompositions of reclaimed porous asphalt.
Reclaimed mortar sand [%] 100 100 100 97.4 24.0 10.5
PA-stone 5/8 [%] 100 81.5 9.4 4.2 2.6 1.1
PA-stone 8/16 [%] 99.5 10.2 3.0 2.6 1.2 1.0
The designed LE2AP mortar contains 82.9% reclaimed material, new bitumen and rejuvena-
tor. Their rate of application were controlled by the log-pen blending law of bitumen.
Table 4 presents the compositions of involved mixtures. Two types of LE2AP porous
asphalt were produced at 110°C. One was produced by mixing hot LE2AP mortar at 170°C
with PA-stone preheated to 120°C. The other was produced by mixing foamed LE2AP mor-
tar with the reclaimed aggregates preheated to 110°C. In the mix design, the bitumen on the
PA-stone is taken into account. It is assumed that 25% of the bitumen in or on the PA-stone
is lost as active bitumen. This bitumen acts as black rock and cannot be reactivated. The
other 75% of the bitumen on/in the PA-stone is active in the new mixture. This bitumen may
either be reactivated or has penetrated the stone as pre-coating which binds the stone and the
LE2AP mortar into the stone. In either case this bitumen becomes of value in the new mix.
Due to the influence of the bitumen on the PA stone, the total bitumen penetration of the
mixture decreases from 89 to 71 pen.
710
3.3 Mortar design
In order to verify the properties of the LE2AP mortar, complex modulus and phase angle
master curves were determined by using the Dynamic Shear Rheometer (DSR). In the DSR,
a special mortar column setup with a height of 20 mm and a diameter of 6 mm is hereto used
(Huurman et al 2010). During the measurement, the specimens were subjected to a sinusoidal
loading signal at a range of frequencies and temperatures. DSR response tests were executed
on both fresh and aged LE2AP mortars. Aging was achieved by placing 2 mm thick mortar
slabs in a stove at 135°C for 44 hours. It is expected that the aging hardening that is obtained
in this manner is equivalent to 10 years of aging of porous asphalt in the Netherlands (Jemere
2010). Figure 5 presents the complex modulus and phase angle master curves of the mortars
before and after aging. The presented master curves are made at a reference temperature
of 20°C. Figure 5 clearly indicates two clusters of lines. One cluster contains the response
behaviour of virgin mortars whereas the second cluster indicates the response behaviour after
aging. The results clearly indicate that LE2AP mortar produced based on reclaimed mortar
has aging properties that are similar to the aging behavior of a freshly produced mortar.
Since the behaviour at low frequencies is important with respect to raveling (Huurman et al
2010), the result indicates that the LE2AP porous asphalt will have the same field perform-
ance as a freshly produced PA.
711
the produced LE2AP mortar at a temperature of about 170°C. From the mixer the mortar
may be pumped to a foaming nozzle with water injection. In order to improve the quality of
the obtained mortar foam, 0.4% foaming additive was used.
Figure 7 gives the results of some mortar foaming tests. The expansion factor of the
obtained foam may reach a value of 10 and the half-life is above 200s. The half-time of mortar
foam thus is even longer than the half-life of the bitumen foam. During the foaming experi-
ment, it is also noticed that the temperature of the foam remained at about 110°C, which is
slightly higher than the temperature of bitumen foam due to the presence of the hot sand
and filler fraction. In particular, no segregation of the sand and filler was observed during
foaming. As a result, it is expected that with this mortar foaming technology, a good quality
warm porous asphalt containing a high percentage of recycled materials can be produced.
712
Figure 8. Illustration of the ARTe and results of the ARTe tests.
In order to evaluate the ravelling resistance of the produced porous asphalt slabs the
Aachener Rafelings Tester, ARTe, was used according to Annex A of prCEN-TS 12697-
50:2014. The ARTe is specifically designed to determine the raveling resistance of surface
layers, and more specific of porous asphalt. The ARTe itself is an accelerated loading test.
One of the important indications from the ARTe is the stone loss after loading, which is an
indicator for ravelling susceptibility and thus the durability of the surfacing system.
As shown in Figure 8, the slab is fixed in a slab fixation box and is moving forwards and
backwards. During this movement, a set of two wheels rotates about a vertical axis while
applying a vertical load of 2500N to the test slab. The combined movements of the wheels and
the fixation box result in the application of large shear stresses to the slab surface acting to eat
away individual surface stones, i.e. raveling. The speed of the loading table is 0.3 ± 0.03 m/s
during the time that the wheels travelling over the slab. The tyre pressure of the rotating wheels
is about 200 ± 10 kPa during the test and the rotation speed is about 47 ± 1 rpm. The test is
conducted at a controlled temperature of 20°C. According to the test norm each slab is sub-
jected to a total of 600 cycles. After 300 cycles the slabs are rotated over 180° for another 300
cycles. During the period, the stone loss in mass is recorded as ravelling susceptibility.
Figure 8 gives the results of the ARTe tests for the two porous asphalt slabs. Unfortunately,
the results of the porous asphalt mixtures produced by mortar foaming were not fully avail-
able at the moment of the paper was written. As a result, only the results from the other
two types of mixtures are shown in the mixture performance section. Slight stone loss over
time can be observed. Considering the variation of the ARTe experiment, the two plates are
considered to be identical with respect to raveling susceptibility. It can be concluded that the
new recycling technology enables the production of mixtures with 93% reclaimed material
that s behave similar to freshly produced porous asphalt mixtures. Although results for the
foamed variant are not available, it is believed that warm porous asphalt produced by foaming
reclaimed mortar will have raveling properties that equal those of a freshly produced hot mix-
ture. This is because it was observed that wetting of the reclaimed stone and the workability of
the obtained mixture are both better than that of the warm mixture involved in these tests.
ACKNOWLEDGEMENTS
The financial support from the European Commission LIFE+ program is gratefully
acknowledged.
REFERENCES
Csanyi, L.H. 1957. Foamed asphalt in bituminous paving mixes. Highway Research Board Bulletin
10(160): 108–122.
D’Angelo J., Harm E., Bartoszek J., Baumgardner G., Corrigan M., Cowsert J., Harman T., Jamshidi M.,
Jones W., Newcomb Dave, Prowell B., Sines R. & Yeaton B. 2008. Warm—Mix Asphalt: European Prac-
tice, Report FHWA-PL-08-007.
EAPA, 2013. EAPA Asphalt in Figures 2013.
Huurman M., Mo L.T. & Woldekidan M.F. 2010. Unravelling porous asphalt concrete, towards a mech-
anistic material design tool. Road Materials and Pavement Design 11 (3):583–612.
Huurman M. 2015. Low Emission2 Asphalt Pavement, LE2AP. 6th Eurasphalt & Eurobitume Congress
2016: Prague, Czec. In press.
Jacobs M.M.J., van den Beemt C.M.A., Sluer B.W. 2010. Successful Dutch experiences with low energy
asphalt concrete. 11th International Conference on Asphalt Pavements: ISAP Nagoya, Japan.
Jemere Y. 2010. Development of a laboratory ageing method for bitumen in porous asphalt. MSc thesis,
Delft University of Technology, The Netherlands.
Jenkins K.J. 2000. Mix design considerations for cold and half-warm bituminous mixes with emphasis
on foamed bitumen. PhD Dissertation, University of Stellenbosch, South Africa.
Van de Ven M.F.C., Jenkins K.J., Voskuilen J.L.M. & van den Beemt R. 2007. Development of (half-)
warm foamed bitumen mixes: state of the art. Special Issue: foamed asphalt for Pavements, Interna-
tional Journal of Pavement Engineering, ISSN 1029-8436, 2007.
714
Pavement Structures
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
717
to its swelling and water loss shrinking (Yitagesu et al., 2012). The common semi-rigid base
asphalt structures are not well adapted to such differential deformations. Therefore, it is
meaningful to study alternative ones.. In this paper, two pavement structures with crush rock
and thick asphalt mixtures were presented. The ABAQUS codes were used to numerically
evaluate the additional stresses.
2.1 Theory
2.1.1 Consolidation theory
Using the transient analysis function of ABAQUS, the flow of water through both saturated
and partially saturated soils follows Darcy’s law which is rewritten as:
Q
= = − kgradH = kJ (1)
A
snv kgradH = kJ
k (2)
ε ij′ = αδ
αδ ij Δw (5)
where ε ij′ = the potential linear swelling strain is related to change in moisture content;
α = linear expansion coefficient; δ ij = Kronecker delta; Δw = change in compaction moisture
content. The stress related to change in moisture content will cause the mechanical strain in
the soils when the soils cannot freely swell, so the total strain can be rewritten as:
where ε ij = the observed (or mechanical) strain; ε ij″ = the strain was induced by expansion of
the soil. Hooke's law is workable between the strain induced by expansion of the soil and the
718
stress induced by expansion of the soil, so the observed (or mechanical) strain can be also
rewritten as:
1+ μ μ
ε ij = σ ij − δ ijσ + αδδ ij Δw (7)
E E
where σ ij = the mechanical stress: σ = the total volume stress;E = the elastic modulus; μ = the
Poisson’s ratio. The equation (Eq.7) can also be rewritten as:
E μE Eα
σ ij = ε ij + δ ij e − δ ij Δw (8)
1+ μ (1 + μ )(1 2 μ ) 1 − 2μ
Eα
σ ij" σ ij − δ ij Δw (9)
1 − 2μ
qs qr
q( h ) = qr + m (10)
⎡1 + α h n ⎤
⎣ ⎦
{ }
2
−m
1 − ahh
n 1
⎡1 + ahh n ⎤
⎣ ⎦
K ( h) = KS (11)
n m/2
⎡1 + ah ⎤
⎣ ⎦
θr θs a(1•m-1) n KS(m/s-1)
during swell/shrinkage tests follow a linear correlation. The relationship can be expressed in
the following form:
720
Figure 2. Variation of hydraulic conductivity with matric suction for black cotton soil.
Figure 3. Correlations of volumetric swell/shrinkage strains versus soil moisture content change for
black cotton soils.
the x-direction, but they were free to move in the y-direction. The evapotranspiration and
infiltration were used as an inlet or outlet discharge on each surface element.
2.2.3 Evapotranspiration/infiltration
A 200-day period of observation was considered from July 2014 to January 2015 in Nairobi.
The daily meteorological data collected from the meteorological station was used to calculate
the reference evapotranspiration. The FAO 56 Penman–Monteith method was used to calcu-
late reference evapotranspiration in this study (Zhang, 2004).
900
0.408 (R n − G) + u (e − e )
T0 =
ET T + 273 2 s a (13)
Δ + γ (1 0.34u2 )
where ET T0 = reference evapotranspiration;RRn = net radiation at the crop surface;G = soil heat
flux density; T = air temperature at 2 m height; u2 = wind speed at 2 m height; es = saturation
vapor pressure; ea = actual vapor pressure; Δ = slope vapor pressure curve; γ = psychrometric
constant. As the magnitude of the daytime soil heat flux beneath the grass reference surface
is relatively small, it was ignored. The daily rainfall and daily actual evapotranspiration from
July 2014 to January 2015 are shown in Figure 5.
721
Figure 4. Typical 2-D views of embankment section.
Figure 5. Daily evapotranspiration and rainfall at Nairobi during the period July 2014 to Jan 2015.
duration of the rainfall. While the highway embankment had the maximum deformation
underneath the edge of the pavement, the deformation decreased gradually as one moves
farther away from the edge of the pavement.
3 PAVEMENT STRUCTURE
K1q K2 (14)
where E = elasticity modulus of granular materials;θ = the first stress invariable; K1, K2
= regression constant related to the material and test (He et al., 1998a). For dense-graded
aggregate base, the value of K1 is between 19939 and 31925. The value of K2 is between 0.40
and 0.53 (He et al., 1998b). The average values of K1 and K2 were used in this study.
722
Figure 6. Vertical displacement on the top surface of embankment during the period July 2014–Jan
2015.
Total thickness 69 cm 76 cm 76 cm
Surface course 4 cm SMA-13 4 cm SMA-13 4 cm SMA-13
6 cm Superpave20 6 cm Superpave20 6 cm Superpave20
8 cm Superpave25
Base 24;cm ATB 12 cm ATB 38 cm CTB
15 cm CR 19 cm CTB
15 cm CR
Sub-base 20 cm ATB 20 cm ATB 20 cm LS
723
Figure 7. Schematic diagram of displacement loading.
724
Figure 9. Variation of maximum tension stress on the top surface of pavement structure with time.
Figure 10. Variation of location of the maximum tension stress with time.
Figure 11. Variation of maximum horizontal tension stress with elasticity modulus of sub-base.
pavement constructed on black cotton soil embankment at Nairobi should consider the pave-
ment temperature field in the future. The locations of the maximum tension stress on the top
surface of pavement structures are shown in Figure 10. It demonstrates that the location of
the maximum tension stress was not fixed; it fluctuated between 7.5 m and 10.5 m away from
the center of the pavement.
Figure 11 shows the maximum tension stress following the increasing elasticity modu-
lus of sub-base. The maximum horizontal tensile stress of all the three pavement structures
725
increased following the increase of the elasticity modulus of sub-base. The semi-rigid base
asphalt structure had higher sensitivity to the change of the elasticity modulus of sub-base
than the other two pavement structures.
5 CONCLUSIONS
The results of the numerical analyses indicate that there is considerable effect of differential
deformation within an expansive soil embankment. The following specific conclusions were
derived:
1. The significant settlement within expansive soil embankment due to soil drying was pro-
duced in the dry season and this settlement seems to be recovered during the following
rainy season. There were significant differential deformations on the top surface of expan-
sive soil embankment during the period of rainfall and evapotranspiration.
2. The differential deformations within the embankment greatly influenced the distribution
of horizontal stress in pavement structures. There was significant horizontal tensile stress
on the upper part of pavement, but the differential deformations had not caused tensile
stress on the lower part of pavement structure.
3. With increased differential deformations within the embankment, the value and loca-
tion of maximum horizontal tensile stress changed constantly. The tension stresses were
greater in the semi-rigid base asphalt structure than the pavement structures with thick
asphalt mixtures and crush rock.
4. The elasticity modulus of the sub-base had significant influence on the distribution of
horizontal tensile stress in pavement. The semi-rigid base asphalt structure had higher
sensitivity to the elasticity modulus of the sub-base than the other two pavement
structures.
REFERENCES
Anh-Minh, T., Yu-Jun, C., Viet-Nam, T., Szerman, Y., & Marchadier, G. 2009. Analysis of the railway
heave induced by soil swelling at a site in southern France. Engineering Geology 106: 68–77.
Cao, D.T. & Hu, C.S. 2001. Analysis for additional stress in pavement due to thawing settlement of per-
mafrost subgrade. JOURNAL OF CHONGOI NGJI AOTONG UNIVERSITY 20(3): 57–61.
Estabragh, A.R., Pereshkafti, M.R.S., Parsaei, B., & Javadi, A.A. 2013. Stabilised expansive soil behav-
iour during wetting and drying. International Journal of Pavement Engineering 14(4): 418–427.
He, Z.Y., Huang, W. & Dong, X.J. 1998a. Comparative Research Between Dynamic and Static Modulus
of Elasticity of Granular Materials. China Journal Of Highway and Transport 11(1): 15–20.
He, Z.Y., Shen, Y., Huang, W. & Dong, X.J. 1998b. Study on semi-rigid pavement structure with granu-
lar base. Chinese Journal of Geotechnical Engineering 20(2): 93–96.
Hong, G.T. 2008. EARTH PRESSURES AND DEFORMATIONS IN CIVIL INFRASTRUCTURE
IN EXPANSIVE SOILS (Dissertation for doctor degree). Texas A&M University, Texas (in USA).
Hong, G.T., Bulut, R., Aubeny, C.P., Jayatilaka, R. & Lytton, R.T. 2006. Prediction of Roughness of
Pavements on Expansive Soils. Unsaturated Soils: 415–426.
Hung, V.Q. 2002. UNCOUPLED AND COUPLED SOLUTIONS OF VOLUME CHANGE PROB-
LEMS IN EXPANSIVE SOILS (Dissertation for doctor degree). University of Saskatchewan, Sas-
katchewan (in Canada).
Jahangir, E., Deck, O., & Masrouri, F. 2013. An analytical model of soil–structure interaction with
swelling soils during droughts. Computers and Geotechnics 54: 16–32.
Ji, Y.B.B., Ma, B. & Wang, B.G. 2006. Influence of Subgrade Settlement Deformation in Frozen Soil
Zone on Asphalt Pavement Structure. China Journal Of Highway and Transport 19(5): 1–5.
Liao, G.Y., Huang, X.M. & Yang, Q.G. 2007. Adaptability of Different Asphalt Pavement Structures
on Soft Ground with Differential Settlements. Journal of Highway and Transportation Research and
Development 24(4): 34–46.
Miao, X.X., Yang, C.Y., & Chen, Z.D. 1993. Humidity Stress Field Theory in Swelling Rock Mass. Rock
and Soil Mechanics 14(4): 49–55.
726
Nyangaga, F.N. 1996. DEVELOPING A MODEL TO PREDICT PAVEMENT ROUGHNESS
DEVELOPMENT ON EXPANSIVE SOILS (Dissertation for doctor degree). Texas A&M Univer-
sity, Texas (in USA).
Puppala, A.J., Manosuthikij, T., & Chittoori, B.C.S. 2014. Swell and shrinkage strain prediction models
for expansive clays. Engineering Geology 168: 1–8.
Rajeev, P., & Kodikara, J. 2011. Numerical analysis of an experimental pipe buried in swelling soil.
Computers and Geotechnics 38: 897–904.
Van Genuchten, M.T.h. 1980. A closed form equation for predicting the hydraulic conductivity of
unsaturated soils. Soil Sci Soc Am J 44: 892–898.
Yang, G.L. & Ding, J.M. 2006. Model Test on Expansion and Shrinkage Deformation in Expansive Soil
Roadbed. China Journal of Highway and Transport 19(4): 23–29.
Yang, H.P., Zhang, R., & Zheng, J.L. 2006. Variation of deformation and strength of expansive soil
during cyclic wetting and drying under loading condition. Chinese Journal of Geotechnical Engineer-
ing 28(11): 1936–1941.
Yang, Q.G., Huang, X.M., Liao, G.Y. & Gu, C.C. 2007. Analysis of the Effect of Soft Ground with
Differential Settlements on Sandwich Pavement Structure. Journal of Highway and Transportation
Research and Development 24(1): 1–5.
Yitagesu, F.A., Werff, H.V.D., Meer, F.V.D., & Hecker, C. 2012. On the relationship between plasticity
and spectral characteristics of swelling soils: The 3–5 μm wavelength region. Applied Clay Science
69: 67–78.
Yu-Jun, C., Ta, A.N., Hemmati, S., Tang, A.M., & Gatmiri, B. 2013. Experimental and numerical inves-
tigation of soil-atmosphere interaction. Engineering Geology 165: 20–28.
Zhang, X. 2004. CONSOLIDATION THEORIES FOR SATURATED-UNSATURATED SOILS
AND NUMERICAL SIMULATION OF RESIDENTIAL BUILDINGS ON EXPANSIVE SOILS
(Dissertation for doctor degree). Texas A&M University, Texas (in USA).
727
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: In current asphalt pavement design methods, the wheel load is characterized
by a cylindrical load. However, field measurements of the contact pressure of tyres show that
the real spatial distribution of the pressure under heavy load could be better described with
the hyperbolic paraboloid function. Therefore, based on the general solution of the asym-
metric elastic layered theory, the general solution for an elastic half-space body subjected to
an asymmetric vertical load was determined. Then considering the hyperbolic paraboloid
loading boundary conditions for an elastic half-space body, the Hankel integral transform of
load function was solved and the analytical solution presented. Finally, the relevant compu-
ter program was compiled and the analytical formula of pavement deflection was obtained.
Compared with the mechanical response of an elastic half-space body subjected to a circular
uniformly distributed load, significant differences in the field distributions of both stresses
and displacements were noted. The first principal stress, first principal strain, and maximum
of shear stress and deflection throughout the pavement structure clearly increased. It was
also found that an asymmetric vertical load would aggravate the fracture and deformation
failure potential of asphalt pavement.
1 INTRODUCTION
In the design and distress evaluation of asphalt pavement, the accurate determination of
structural responses is of great importance. Current design methods are based on the pri-
mary assumptions for the materials- homogeneous, isotropic, and linearly elastic and sub-
jected to a circular uniform distributed load. The half space theory has been applied to road
engineering since Boussinesq (1885) derived analytical solutions for stresses and displace-
ments, using three stress functions. Subsequently, by employing the equations of elasticity
theory for three-dimensional problems in cylindrical coordinates (Love 1927), which derived
from the displacement function, , Burmister (1943,1945) derived a solution based on a
circular uniformly distributed load by the Hankel transform. Later, Perloff and Moaven-
zadeh (1967) derived a vertical displacement equation of viscoelastic half space based on a
uniform, cylindrical load. Following their work, the derivation of viscoelastic system was
gradually extended to two layers (Chou 1969), three layers (Elliott 1971, Ashton 1966), four
layers (Barksdale 1967), and any number of layers (Huang 1973). However, solving the elas-
tic sstem problem is cumbersome for these methods, which must first determine the specific
relationship between the stress or displacement function with components of stress or dis-
placement. Fortunately, by using the Hankel Laplace transform, Dazhi Guo (2001) derived
a general solution for layered elastic systems and then extended it to elastic and viscoelastic
multilayers on the basis of Lamé equations. The finite element method (FEM) also has been
extensively employed in the analysis of pavement design. Al-Qadi (2008) analysed the causes
of fatigue cracking, rutting, and top-down cracking by combining the 3D FEM with acceler-
ated pavement testing. Using the ABAQUS analysis tool, M.Ameri (2014) analysed the static
mechanical response with a Prony series representing the viscoelastic indices of an asphalt
729
mix. Since the recognition that the presence of water within the pavement structure would
contribute to excessive deflection, cracking, ravelling (Choubane et al. 1998), many studies
employed different theories, including a porous medium theory (Zejiao Dong 2008) and a
comprehensive energy-based framework (Niki Kringos 2008). The aforementioned study all
focused on the circular uniform distributed load or a rectangular uniform load. However, the
actual spatial distribution of trye ground pressure typically shows asymmetric characteristics
(M. De Beer 1996, Smail Hamlat et al. 2014, Gabriel Anghelachea 2012). Figure 1 shows the
tyre ground pressure distribution under two kinds of axle load (Xiaodi Hu 2005). Hence, it is
necessary to extend the elastic system solution to an asymmetry system.
The ground pressure distribution was an approximate hyperbolic paraboloid in Figure1.
Therefore, this study intended to derive an analytical solution for a half-space subjected to
a hyperbolic paraboloid vertical load. The soundness of output from the solution was then
confirmed by comparing it to the circular uniformly distributed load. It is expected that the
derivation for an asymmetric vertical load can provide an effective tool for pavement design.
In this paper, the surface load was applied in the form of a hyperbolic paraboloid vertical
load, which was mathematically expressed in rectangular coordinates as the following:
⎧ x 2 y2
⎪ p0 + 2 − 2 0< r ≤δ
p(r,θ ) = ⎨ ah bh (1)
⎪0 r >δ
⎩
⎧ r 2 cos 2 θ r 2 sin
siin2 θ
⎪ p0 + − 0< r ≤δ
p(r,θ ) = ⎨ ah2 bh2 (2)
⎪0 r >δ
⎩
where δ is the radius of the load area, ah bh is the parameter of the hyperbolic paraboloid
vertical load function, p0 is the load, and r,θ are the distance from the centre of the load
circle and the polar angle respectively.
Its series can be expressed as:
730
Figure 2. Model of hyperbolic paraboloid vertical load.
The general solution for the asymmetric condition (Dazhi Guo) that satisfies the equations of
equilibrium and compatibility for the theory of elasticity is:
∞
⎡∞ ⎛ k + 1 as k − 1 as ⎞ ⎤
σr ∑ ∫ξF
k =0 ⎣ 0
as
1,k
k J k (ξ
ξ −⎜
⎝ 2r
U k +1,k +
2r
U k ,k ⎟ ⎥ cos kθ
⎠⎦
∞
∞
⎡ ⎛ k + 1 as k −1 ⎞⎤
σ θas
a
= ∑ ⎢2 μ ∫ ξ F2as,kk J k (ξ r dξ − ⎜ U k +1,k + U k −11,k ⎟ ⎥ kθ
k =0 ⎣ 0
⎝ 2 r 2 r ⎠⎦
∞ ∞
σ zas
a
= ∑ ∫ ξ F3,k J k (ξ r )cos kθ dξ
k =0 0
∞
∞
⎡ ⎛ k + 1 as k − 1 as ⎞ ⎤
θ =∑
τ ras ∫ξF J k (ξ ))ddξ + ⎜ − U k −1,k ⎟ ⎥ sin kθ
a
U
⎣0k =0
4 ,k
⎝ 2r k +1,k 2r ⎠⎦
1 ∞
τ = ∑ ( k 1,
aas
θz 1,k )sin kθ (4)
2 k =0
1 ∞
τ zras
a
∑ ( H k ,k H kas ,k ) kθ
2 k =0
1+ μ ∞
u as = − ∑ (U k +11,k −U kas 1,k )cos kθ
2E k =0
1+ μ ∞
v as = − ∑ (U k +11,k + U kas 1,k )sin kθ
2E k =0
∞
1+ μ ∞
w as = − ∑ ∫
E k =0 0
F9,k J k (ξ )cos kθ dξ
∞ ∞
where H k ,k ∫ξF
0
as
5 ,k J k (ξ r )dξ H kas−1,k,k = ∫ ξ F6,k J k −1 (ξ r)
0
r dξ ,
∞ ∞
Uk ,k ∫ F ask Jk
0
r d U kas− k = ∫ F8,k J k −1 (ξ r)
0
r dξ
731
F5,k [ Akas (2 μ ξ z )B
)Bkas E k ]e ξz
[Ck (2 μ ξ zz)D
)Dk Fkas ]eξ z
F6,k [ Akas (2 μ ξ z )B
)Bkas E k ]e ξz
[Ck (2 μ ξ zz)D
)Dk Fkas ]eξ z
F7,k [ Akas (1 ξ z )Bkas 2 E k ]e −ξ z − [C
[Ck (1 ξ z )Dk 2 Fkas ]eξ z
F8,k [ Akas (1 ξ z )Bkas 2 E k ]e −ξ z − [C
[Ck (1 ξ z )Dk 2 Fkas ]eξ z
F9,k [ Akas (2 − 4 μ + ξ z )Bkas ]e ξ
[Ckas (2 4 μ − ξ z )Dkas ]eξ z
In the above equations, σ ras , σ θas , and σ zas are the radial stress, tangential stress and vertical
stress respectively.τ rasθ τ θasz and τ rzas are the shear stresses in the three directions respectively.
u as , v as and ω as are the displacements in the three directions respectively.E and μ are the
elastic modulus and Poisson ratio respectively. r is the distance from center to the calculated
point and z is the distance from surface to calculated point.
With equations (3) and (4), it can be determined that parameter k is equal to 0 and 2.
Hence, the load can be divided into two parts as follows:
Evidently, the mechanical equations were derived from the inversion of the Hankel trans-
form. Therefore, it was necessary to convert the surface boundary condition to the Hankel
transformed domain by employing the first Sonine finite integral formula as:
π
2v (v 1)
∫0
2
J μ ( x sinθ )si μ 1
θ cos 2v +1 θ dθ =
x v +1
J μ +v +1 ( x )
where Γ( ) is the gamma function and J m ( x ) is the Bessel function with order m.
Equation (4) taking Hankel transforms was:
(0) p0δ P0hδ 3 2 P0hδ 2
p0 (ξ ,θ ) J1 (ξδ ) J 2 (ξδ )
ξ ξ2
(2) P2hδ 3
p2 (ξ ,θ ) J3 (ξδ ))cos 2θ
ξ
pδ P hδ 3 2 P0hδ 2 P hδ 3
p(ξ ,θ ) J1 (ξδ ) J 2 (ξδ ) 2 J3 (ξδ ))cos 2θ (6)
ξ ξ 2
ξ
Meanwhile, with the condition of infinity that stress and displacement were zero, the value
of the parameters were determined to be:
Ck 0 Dk = 0, Fk = 0
(7)
5 GOVERNING EQUATIONS
Combining the boundary condition at the ground surface and infinity and substituting
x = ξδ into equation (4), the displacement and stress components of the elastic half-space
subjected to a hyperbolic paraboloid vertical load were derived as follows:
732
∞
⎛ z ⎞ −zx ⎛r ⎞
σ rh = − p0 ∫ ⎜1 − x e δ J1 ( x )J 0 ⎜ x ddx
0
⎝ δ ⎠ ⎝δ ⎠
∞
⎛ z ⎞ −zx ⎛r ⎞
− P0hδ 2 ∫ 1 − x ⎟ e J1 ( x )J 0 ⎜ x ⎟ dx
0
⎝ δ ⎠ ⎝δ ⎠
∞
1 ⎛ z ⎞ − δz x ⎛r ⎞
+2 P0hδ 2 ∫ 1 − x ⎟ e J 2 ( x )JJ 0 ⎜ x dx
d
0
x⎝ δ ⎠ ⎝δ ⎠
∞
p0δ 1 ⎛ z ⎞ −zx ⎛r ⎞
+ ∫
r 0 x⎝
1 − 2 μ − x ⎟ e δ J1 ( x )J1 ⎜ x ⎟ dx
δ ⎠ ⎝δ ⎠
∞
P0hδ 3 1 ⎛ z ⎞ − δz x ⎛r ⎞
+
r 0x ∫ ⎝
1 − 2 μ − x ⎟ e J1 ( x )J1 ⎜ x ⎟ dx
δ ⎠ ⎝δ ⎠
∞
2 h
δ3 1 ⎛ z ⎞ −zx ⎛r ⎞
− 0
r ∫x
0
2 ⎝
1 − 2 μ − x ⎟ e δ J 2 ( x )J1 ⎜ x ⎟ dx
δ ⎠ ⎝δ ⎠
∞
⎛ z ⎞ −zx ⎛r ⎞
− P δ cos
h
o 2θ ∫ 1 − x ⎟ e δ J3 ( )J 2 ⎜ x ⎟ dx
2
2
0
⎝ δ ⎠ ⎝δ ⎠ (8)
∞
3 δ cos 2θ 1 ⎛
h 3
z ⎞ − x
z
⎛r ⎞
+ 2
2r ∫0 x ⎜⎝1 2 μ − δ x⎟⎠ e δ J3 ( x )J3 ⎜⎝ δ x⎟⎠ dx
∞
P hδ 3 2θ 1 ⎛ z ⎞ −zx ⎛r ⎞
+ 2
2r ∫0
x ⎝
1 − 2 μ − x ⎟ e δ J3 ( x )J1 ⎜ x ⎟ dx
δ ⎠ ⎝δ ⎠
∞ z
− x r
σ θh 2 μ p0 ∫ e δ
J1 ( x )J 0 ( x )ddx
0
δ
∞ z
− x r
−2 μ 0hδ 2 ∫ e δ J1 ( x )J 0 ( x )ddx (9)
0
δ
∞ z
1 − x r
+4 μ 0 δ ∫ J 2 ( x )e δ J 0 ( x )ddx
h 2
0
x δ
∞
p0δ ⎛ z ⎞ 1 −zx ⎛r ⎞
− ∫
r 0⎝
1 − 2 μ − x ⎟ e J1 ( x )J1 ⎜ x ⎟ dx
δ ⎠x ⎝δ ⎠
∞
P0hδ 3 ⎛ z ⎞ 1 −zx ⎛r ⎞
−
r 0⎝∫ 1 − 2 μ − x ⎟ e J1 ( x )J1 ⎜ x ⎟ dx
δ ⎠x ⎝δ ⎠
∞
2 P0hδ 3 ⎛ z ⎞ 1 −zx ⎛r ⎞
+
r 0⎝ ∫ 1 − 2 μ − x ⎟ e J 2 ( x )J1 ⎜ x ⎟ dx
δ ⎠x ⎝δ ⎠
∞ z
− x ⎛r ⎞
−2 μ h
δ2 s 2θ ∫ J3 ( x )e δ J 2 ⎜ x ⎟ dx
2
0
⎝δ ⎠
∞
3 h
δ3 2θ ⎛ z ⎞ 1 −zx ⎛r ⎞
− 2
2r ∫0 ⎝ 1 − 2 μ −
δ
x ⎟ e J3 ( x )J3 ⎜ x ⎟ dx
⎠ x ⎝δ ⎠
∞
P2hδ 3 2θ ⎛ z ⎞1 z
− x ⎛r ⎞
−
2r ∫
0
⎝
1 − 2 μ − x ⎟ J3 ( x )e J1 ⎜ x ⎟ dx
δ ⎠x ⎝δ ⎠
733
∞
⎛ z ⎞ −zx ⎛r ⎞
σ zh = − p0 ∫ ⎜1 + x e δ J1 ( x )J 0 ⎜ x ddx
0
⎝ δ ⎠ ⎝δ ⎠
∞
⎛ z ⎞ −zx ⎛r ⎞
− P0hδ 2 ∫ 1 + x ⎟ e J1 ( x )J 0 ⎜ x ⎟ dx
0
⎝ δ ⎠ ⎝δ ⎠
∞
⎛ z ⎞ 1 −zx ⎛r ⎞ (10)
+2P0hδ 2 ∫ 1 + x ⎟ e J 2 ( x )JJ 0 ⎜ x ddx
0
⎝ δ ⎠ x ⎝δ ⎠
∞
⎛ z ⎞ −zx ⎛r ⎞
P2hδ 2 coss θ ∫
−P x e J3 ( x )J 2 ⎜ x ⎟ dx
0
⎝ δ ⎠ ⎝δ ⎠
∞
3P2hδ θ ⎛ z ⎞ 1 −zx ⎛r ⎞
τ rhθ =
2r ∫0 ⎝1 − 2 μ − δ x⎟⎠ e δ J3 ( x )J3 ⎜⎝ δ x⎟⎠ dx
d
∞
P2hδ 3 2θ ⎛ z ⎞ 1 −zx ⎛r ⎞ (11)
−
2r ∫
0
⎝
1 − 2 μ −
δ
x ⎟ e J3 ( x )J1 ⎜ x ddx
⎠ x ⎝δ ⎠
∞ ∞
P2hδ θ⎡ z
− x ⎛r ⎞ z
− x ⎛r ⎞ ⎤
τ θhz = − ⎢ ∫ xe δ J3 ( x )J3 ⎝ x ⎠ dx
d ∫ xe
xe δ J3 ( x )J1 ⎜ x ⎟ dx ⎥
⎝δ ⎠ ⎦ (12)
2 ⎣0 δ 0
∞
p0 z z
− x ⎛r ⎞
τ zrh = −
δ 0 ∫ xe δ J1 ( x )J1 ⎜ x ddx
⎝δ ⎠
∞ z
− x ⎛r ⎞
− P0hδ z ∫ xe δ J1 ( x )J1 ⎜ x ddx
0
⎝δ ⎠
∞ z
− x ⎛r ⎞
+2 P0hδ z ∫ e J 2 ( x )J1 ⎜ x ddx (13)
0
⎝δ ⎠
∞
P2hδ z 2θ z
− x ⎛r ⎞
−
2 ∫ xe
0
x δ J3 ( x )J3 ⎜ x ddx
⎝δ ⎠
∞
P2hδ z θ z
− x ⎛r ⎞
+
2 ∫ xe
0
δ J3 ( x )J1 ⎜ x dx
⎝δ ⎠
∞
1+ μ ⎛ z ⎞ −zx 1 ⎛r ⎞
uh = − p0δ ∫ ⎜1 − 2 μ − x e δ J1 ( x )J1 ⎜ x ddx
E 0
⎝ δ ⎠ x ⎝δ ⎠
∞
1+ μ h 3 ⎛ z ⎞ − x1
z
⎛r ⎞
− P0 δ ∫ 1 2 μ − x e δ J1 ( x )J1 ⎜ x ddx
E 0
⎝ δ ⎠ x ⎝δ ⎠
∞
1+ μ h 3 ⎛ z ⎞ 1 −zx ⎛r ⎞ (14)
+ 2 P0 δ ∫ ⎜1 2 μ − x ⎟ 2 e δ J 2 ( x )J1 ⎜ x ⎟ dx
E 0
⎝ δ ⎠ ⎝δ ⎠
∞
1+ μ h 3 ⎛ z ⎞1 z
− x ⎛r ⎞
− P2 δ cos ∫ μ 3 ( x )e
δ J
3⎜ x dx
2E 0
⎝ δ ⎠ ⎝ δ ⎟⎠
∞
1+ μ h 3 ⎛ z ⎞ 1 −zx ⎛r ⎞
+ o 2θ ∫ 1 2 μ − x ⎟ e δ J3 ( )J1 ⎜ x ⎟ dx
P2 δ cos
2E 0
⎝ δ ⎠ ⎝δ ⎠
∞
1+ μ h 3 ⎛ z ⎞1 z
− x ⎛r ⎞
vh = − P2 δ n 2θ ∫ ⎜1 − 2 μ − x ⎟ J3 ( x )e J3 ⎜ x ⎟ dx
2E 0
⎝ δ ⎠ x ⎝δ ⎠
∞
1+ μ h 3 ⎛ z ⎞1 z
− x ⎛r ⎞ (15)
− P2 δ n 2θ ∫ ⎜1 − 2 μ − x ⎟ J3 ( x )e J1 ⎜ x ⎟ dx
2E 0
⎝ δ ⎠ x ⎝δ ⎠
734
∞
1+ μ ⎛ z ⎞ 1 −zx ⎛r ⎞
wh = p0δ ∫ ⎜ 2 − 2 μ + x ⎟ e J1 ( x )J 0 ⎜ x ddx
E 0
⎝ δ ⎠ x ⎝δ ⎠
∞
1+ μ h 3 ⎛ z ⎞1 z
− x ⎛r ⎞
+ P0 δ ∫ 2 2 μ + x ⎟ J1 ( x )e δ J 0 ⎜ x ⎟ dx
E 0
⎝ δ ⎠ ⎝δ ⎠
∞
1+ μ h 3 ⎛ z ⎞ 1 −zx ⎛r ⎞ (16)
− 2 P0 δ ∫ ⎜ 2 − 2 μ + x ⎟ 2 e δ J 2 ( x )J 0 ⎜ x ⎟ dx
E 0
⎝ δ ⎠ ⎝δ ⎠
∞
1+ μ h 3 ⎛ z ⎞ 1 −
z
⎛r ⎞
P2 δ cos 2θ ∫ 2 2 μ + x ⎟ e δ J3 ( x )J 2 ⎜ x ⎟ dx
x
+
E 0
⎝ δ ⎠ ⎝δ ⎠
6 ANALYSIS RESULT
To visualise the influence of an asymmetric load on the mechanical response, numerical inte-
gration was carried out on equations (8) through (16) and the result was compared the result
with that of a circular uniformly distributed load. The elastic half-space was set with an
elastic modulus of 30 MPa and a Passion ratio of 0.25. A circular uniformly distributed load
was selected with a stress of 0.7 MPa and a contact radius (δ ) of 10.65 cm. To obtain an
equivalent load, the parameters of a hyperbolic paraboloid vertical load were calculated and
expressed as: ah bh = 15, δ = 10.65
65cm,, p0 = 0.542MPa.
Figure 8. τ max at different depths (location: cylindrical load r = 0.9 δ , the oxz plane under the parab-
oloid load r = δ , the oyz plane under Paraboloid load: r = 0 ).
7 CONCLUSION
To investigate the influence of an asymmetric load on pavement structural response, the ana-
lytical solution of an elastic half-space subjected to a hyperbolic paraboloid vertical load was
derived from a general solution at an asymmetric condition. A numerical integration on the
737
components of stress and displacement was then conducted. The conclusions of this study
are as follows:
1. The form of the analytical solution under the hyperbolic paraboloid vertical load was typi-
cally complex, including 42 infinite integrals containing Bessel functions of order 0, order
1, order 2 and order 3.
2. Compared to that of a circular uniform load, the impact of an asymmetric load on the
mechanical response of a pavement structure did not propagate to deeper areas but there
were remarkable differences in the distribution of stress and displacement at depths of
0 cm to 2 δ .
3. It was found that the components of ε1, τ max magnitude subjected to a hyperbolic parabo-
loid vertical load were much greater than those under a cylindrical load on the upper
surface, which may contribute to more serious damage. It implies that the asymmetry
of vertical or horizontal shear load should be carefully taken into account in pavement
design.
4. The elastic half-space solution under a hyperbolic paraboloid vertical load should be
extended to elastic multilayered systems in later studies.
ACKNOWLEDGEMENTS
REFERENCES
Al-Qadi I.L. Wang H. Yoo P.J. and Dessouky S.H. 2008. Dynamic analysis and in situ validation of
perpetual pavement response to vehicular loading[J]. Transportation Research Record 2087.
Ashton J E, Moavenzadeh F M. 1966. Viscoelastic Analysis of Three-layer Systems[J].
Barksdale R D, Leonards G A. 1967. Predicting performance of bituminous surfaced pavements[C]. Intl
Conf Struct Design Asphalt Pvmts.
Boussinesq, J., 1885. Application des Potentiels a l’etude de l’equilibre et du Mouvement des Solids
Elastiques; Gauthier-Villars, Paris.
Chou Y T. 1969. Stresses and Displacements in Viscoelastic Pavement Systems under a Moving
Load[R].
Choubane, B., Page, G.C. and Musselman, J.A. (1998). “Investigation of water permeability of coarse
graded superpave pavements.” Proceedings of Association of Asphalt Paving Technologists, Vol. 67:
1–13.
Dazhi Guo, Decheng Feng.(2001). The mechanics of multilayered elastic system[M]. Harbin Institute
of Technology Press.
De Beer, M. 1996. Measurement of tyre/pavement interface stresses under moving wheel loads. Heavy
Vehicle Systems, Special Series, International Journal of Vehicle Design, Vol. 3, No’s 1–4, pp.
97–115
D.M.Burmister. 1945. The General Theory of Stresses and Displacements in Layered Systems. I[J].
Journal of Applied Physics.
D.M.Burmister. 1945. The General Theory of Stresses and Displacements in Layered Soil Systems.
III[J]. Journal of Applied Physics.
Dazhi Guo, Decheng Feng.(2001). The mechanics of multilayered elastic system[M]. Harbin Institute
of Technology Press.
Elliott J F, Moavenzadeh F. 1971. Analysis of Stresses and Displacements In Three-Layer Viscoelastic
Systems[J]. Highway Research Record.
Gabriel Anghelachea, Raluca Moisescua. 2012. Measurement of stress distributions in truck tyre con-
tact patch in real rolling conditions[J]. Vehicle system dynamics.
738
Huang Y.H.1973. Stresses and Strains in Viscoelastic Multilayer Systems Subjected to Moving Loads[J].
Highway Research Record 457.
Love, A.E.H. (1927). Treatise on the mathematical theory of elasticity,Dover Publications, Inc., Mine-
ola, NY.
M.Ameri, M.Malakouti, P.malekzadeh. 2014. Quasi-static analysis of multilayered domains with vis-
coelastic layer using incremental-layerwise finite element method[J]. Mech Time-Depend Mater.
Niki Kringos, Tom Scarpas, Cor Kasbergen, 2008. Patrick Selvadurai. Modelling of combined
physical–mechanical moisture-induced damage in asphaltic mixes, Part1: governing processes and
formulations[J]. International Journal of Pavement Engineering.
N. Kringos, A. Scarpas, A. Copeland, J. Youtcheff. 2008. Modelling of combined physical–mechanical
moisture-induced damage in asphaltic mixes Part 2: moisture susceptibility parameters[J]. Interna-
tional Journal of Pavement Engineering.
Smail Hamlat, Ferhat Hammoum, Jean-Pierre Kerzreho. 2014. Evaluation of the distribution of local
pressures and the real contact area between the tyre and the road surface[J]. International Journal of
Pavement Engineering.
Perloff W H, Moavenzadeh F. 1967. Deflection of Viscoelastic Medium Due to a Moving Load[C]. Intl
Conf Struct Design Asphalt Pvmts.
Xiaodi Hu, Lijun Sun(2005). Measurement of tire pressure distribution of heavy truck[J]. Journal of
tongji university (natural science edition).
Zejiao Dong,Yiqiu T, Liping Cao. 2008. The Effect of Water on Pavement Response Based on 3D FEM
Simulation and Experiment Evaluation[J]. Pavements and Materials.
739
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Yared H. Dinegdae
Royal Institute of Technology KTH, Stockholm, Sweden
Björn Birgisson
Aston University, Birmingham, UK
1 INTRODUCTION
Many of the input parameters used in pavement design exhibit inherent variability (PMS
Object 2008, Shell 1978). The combined effect of the variances associated with these input
parameters, and other factors have a significant influence on predicted pavement perform-
ance (Timm et al. 1998, Darter et al. 1972, Noureldin et al. 1994). It is important for a
pavement design procedure to deliver designs that perform consistently at the desired level
of reliability while taking into account input parameters variabilities effect on perform-
ance. There are several studies which have been undertaken since the 1970s that focus on
how to address and incorporate reliability analysis into pavement design process (Darter
et al. 1973, Chua et al. 1992, Timm et al. 1999, Retherford & McDonald 2010). These
analyses used a variety of techniques, ranging from empirically derived safety factors to
Monte Carlo simulations and analytically based reliability analysis methods. A codified
design procedure using factor of safety (FOS) was one of the outcomes of these studies.
A typical example is the American Association of State Highway and Transportation
Officials (AASHTO) design procedure for flexible pavements (AASHTO 1993). In this
procedure, reliability factor is introduced as a positive spacing parameter between the
allowable and expected traffic repetitions. However, representing the various variabilities
involved with an empirically derived factor might not ensure designs of uniform reliability.
This problem can be addressed by developing a set of partial safety factors that directly
considers input parameters variabilities through a probabilistic method of uncertainty
propagation. Load and resistance factor design (LRFD) procedure is a typical example of
this kind of design method.
741
Load and resistance factor design (LRFD) procedure uses partial safety factors that are
derived through reliability analysis to account for the variabilities and uncertainties involved
in the design process. The magnitude of the partial safety factors depends on the level of safety
required, degree of influence and associated variabilities of input parameters (AASHTO
1997a). The LRFD method has been widely implemented in structural design specifications,
and there are limited studies regarding its applicability for pavement design purposes (ACI
1995, AISC 1994, Kim & Buch 2003, Dinegdae & Birgisson 2015). The partial safety factors
in the LRFD method are developed by modelling input parameters variabilities with a cer-
tain level of coefficient of variation (COV) and probability density function (pdf). The design
procedure is expected to deliver uniform performance for a given target reliability for all the
likely combination of different pavement design features. This may be difficult to achieve
considering the fact that input parameters variabilities exhibit a wide range in variation from
one project site to another. In the case when there is a wide deviation from the conditions
upon which the LRFD procedure was developed, the impact on estimated target reliability
will be higher and this can have a serious impact on the estimated total life cycle cost and
planned maintenance schedule.
This paper investigates input parameters variabilities influence on estimated target reli-
ability. In addition, the variabilities of the pavement performance measuring parameters
were also characterized through simulation. Various levels of input parameters COV in
conjunction with probability distribution functions were studied so as to establish the level
of influence of each characterization. For this purpose, a field pavement section that has
high quality laboratory and field data was designed for a given target reliability using the
mechanics-based LRFD procedure. The actual reliability of the optimized pavement sec-
tion for the various combinations of COV levels and probability density functions was esti-
mated using a two-component reliability analysis methodology. Monte Carlo simulation
was used to characterize the pavement performance measuring parameters. The significance
of each variation was established by comparing the estimated actual reliabilities with the
target reliability.
The main sources of uncertainty in pavement design as outlined by Huang (2004) are input
parameter inherent variability, model uncertainty and statistical uncertainty. The inherent
variabilities of input parameters are mainly related to the spatial variability in material and
cross-sectional properties and improper measurements. Model bias is introduced into the
design process due to the fact that the models that are used for response computation and
performance prediction are a simplified version of the much complex material behaviour. Sta-
tistical uncertainty arises due to lack of data, which signifies incomplete information regard-
ing the uncertainties involved. Appropriate statistical characterization of input parameters
variabilities is a prerequisite for any reliability analysis. These variabilities can be modelled
using the mean and variance of the random variables and if required with a full probability
density function (pdf).
In the development of the LRFD procedure for the mechanics-based analysis frame-
work, Dinegdae & Birgisson (2015) identified parameters such as asphalt layer thickness
(HAC), base modulus (EB), fracture resistance (DCSElim) and expected traffic (n) to have a
significant influence on predicted performance, and as such modelled these design inputs
as random variables. Dinegdae & Birgisson (2015) used average COV levels and repre-
sentative distribution functions to model the variabilities of these parameters. A literature
review regarding the variabilities of these parameters is presented in Table 1. As can be
seen in Table 1, the variabilities of these input parameters exhibit a wide range of variation.
Asphalt thickness variability can be modelled with either normal or log-normal probability
distribution functions with a COV range of 3%–25%. The variability of the base modulus
is also observed to follow both normal and log-normal distributions but with a relatively
wide COV range of 5%- 60%. The variability in traffic can be expressed in addition to
742
Table 1. Design inputs variabilities.
Variability
normal and log-normal distributions with an extreme value type-1 distribution function.
As there was limited information on the variability of DCSElim, which is a fracture resist-
ance parameter, Dinegdae & Birgisson (2015) used a log-normal distribution with a COV
level of 35%.
3 EVALUATION METHODOLOGY
The influence of input parameters variabilities on estimated target reliability were evalu-
ated through a sensitivity analysis that combine the variations of these input parameters
systematically. A total of 10 cases were established by combining various levels of COV with
distribution functions. These 10 cases were developed based on literature and can be grouped
into two broad categories. In the first category influence of COV levels was analyzed while
keeping the distribution functions the same. In this study only normal and log-normal proba-
bility density functions (pdf) were used to model input parameters variabilities. In the second
category, various combination of distribution functions influence was studied for the same
level of COV. The COV values used in the second category were the same values used for the
development of the mechanics-based LRFD procedure.
The analyses were undertaken by designing a field pavement section that has high qual-
ity field and laboratory data for a given target reliability using the mechanics-based LRFD
procedure and estimating the actual reliability of the optimized section through analytical
based reliability analysis method. The variabilities of the damage accumulation and pave-
ment life were determined for all the cases through Monte Carlo simulation. A total of
10,000 cycles were used to generate the distribution functions of the two parameters. The
following sections provide detailed information regarding the evaluation approach adopted
for this study.
743
Table 2. Design inputs variabilities.
COV (%)
HAC Normal 5 10 15
EB Normal 15 30 45
N Normal 30 40 50
DCSElim Normal 25 35 45
COV (%)
HAC Log-normal 5 10 15
EB Log-normal 15 30 45
n Log-normal 30 40 50
DCSElim Log-normal 25 35 45
744
Table 4. Design inputs variabilities.
HAC Normal 10
EB Log-normal 30
N Log-normal 40
DCSElim Log-normal 35
γ global DCSE
C acum f φAC H AAC ,φEB E B ,γ n n ) (2)
For this study, a pavement life function (D) that relates DCSElim with DCSEacum was defined
as follows:
DCSE
C Elim
D= (3)
DCSE
C Eacum
A section of the Interstate highway, I-75, located in the county of Charlotte, Florida State
was used for establishing the influence of input parameters variabilities on estimated target
reliability. A target reliability of 90% was selected, and the pavement section was designed
accordingly using the mechanics-based LRFD procedure. Table 6 presents the mean val-
ues of the input parameters of the optimized section. Information regarding the remaining
inputs such as gradation, volumetric, binder properties and cross-sectional information can
be obtained from Dinegdae et al. (2015).
HAC 16.8 cm
EB 377.8 MPa
N / year 573 (103 ESALs)
DCSElim 1.03 kJ/m3
Figure 1. Influence of COV levels on target reliability: (a) normal pdf, (b) log-normal pdf.
746
Figure 2. Distribution type influence on estimated target reliability.
comparable. This shows, regardless of the type of distribution function used in modelling
input parameters variabilities, the estimated target reliability is governed by COV level.
747
Figure 3. Monte Carlo generated damage accumulation fitted with log-normal distribution.
Figure 4. Monte Carlo generated pavement life distribution fitted with log-normal distribution.
The distribution of the pavement life as generated by the Monte Carlo simulation for Case
7 is shown in Figure 4. A further analysis was performed for all the 10 cases so as to establish
the distribution type and the COV range of the pavement life variability. Accordingly, the
chi-square goodness of fit test have shown that out of the 10 cases in four this variability can
be modelled with a log-normal distribution, in three cases with a normal distribution and in
the remaining three cases with neither. A wide COV range of 46.02% - 243.8% was obtained,
the minimum value was for Case 4 and the maximum was for Case 3. For Case 7 a COV value
of 77.43% was obtained. For cases that have the same level of COV but different distribution
functions, not that much difference in the COV levels of the pavement life parameter was
observed, which is similar with the results of the damage accumulation.
748
5 CONCLUSIONS
This paper studied the impact of input parameters variabilities on estimated target reliability.
In addition, the variabilities of damage accumulation and pavement life were also investi-
gated. A total of 10 cases which combine COV levels and probability density functions sys-
tematically were used to model input parameters variabilities. These variabilities are modelled
using normal and log-normal distribution functions and three levels of COV, that represent
low, average and high variability conditions. It was clear from the results that the number of
cases studied and the combination of variabilities selected were successful in demonstrat-
ing the influence of input parameters variabilities on the target reliability and performance
measuring parameters.
The analyses have shown that the COV values used in modelling the variabilities of input
parameters have a considerable influence on estimated target reliability irrespective of the
probability density functions used. It was also observed from the analyses that the type of dis-
tribution function used in modelling input parameters variabilities had a negligible influence
on estimated target reliability in the case when the COV level was constant. The variability of
pavement life can be modelled best with both normal and log-normal distribution functions
with a wide range in COV values. In the case of accumulated damage, the variability was
observed to follow a relatively narrow COV range and neither normal nor log-normal distribu-
tion functions succeeded in fitting the Monte Carlo generated frequency spectrum. Neverthe-
less, for practical purposes this variability can be assumed to follow a log-normal distribution
function. The mechanics-based LRFD procedure was able to deliver designs of uniform target
reliability for cases in which the same level of COV was used as the design procedure.
ACKNOWLEDGMENT
The authors would like to thank the Swedish transport administration, Trafikverket and the
Swedish construction industry organization for research and development, SBUF for provid-
ing the financial support for the project under which this work was performed.
REFERENCES
AASHTO, 1985. Proposed AASHTO guide for design of pavement structures (NCHRP Project 20-7/24,
Vol. 2). Washington, DC: American Association of State Highway and Transportation Officials.
AASHTO, 1993. Guide for design of pavement structures. Washington, DC: American Association of
State Highway and Transportation Officials
AASHTO, 1997a. LRFD Highway Bridge design specifications, SI units. Washington, DC: American
Association of State Highway and Transportation Officials. First Edition (1994 with 1996 and 1997
Interims).
Bush, D. 2004. Incorporation of reliability into mechanistic-empirical pavement design in Washington
and California. Retrieved from http://courses.washington.edu/kckapur/526/526projects/Dbush.pdf.
ACI. 1995. Building code requirements for reinforced concrete. Detroit, MI: American Concrete
Institute.
AISC 1994. Load and resistance factor design specification for structural steel buildings (2nd ed.). Chi-
cago, IL: American Institute of Steel Construction.
Chua, K.H., Kiureghian, A.D., & Monismith, C.L., 1992. Stochastic model for pavement design, Jour-
nal of Transportation Engineering, 118(6), 769–786.
Darter, M.I., McCullough, B.F., & Brown, J.L., 1972. Reliability concepts applied to the Texas flexible
pavement system (Record No. 407, pp. 146–161). Washington DC: Highway Research Board.
Darter, M.I., Hudson, W.R., & Brown, J.L., 1973. Statistical variation of flexible pavement properties
and their consideration in design. Proceedings of the Association of Asphalt Paving Technologists, 42,
589–615.
Dinegdae, Y.H., & Birgisson, B., 2015 Reliability-based calibration for a mechanics-based fatigue crack-
ing design procedure. Journal of Road Materials and Pavement De-sign.doi:10.1080/14680629.2015.
1055335.
749
Dinegdae, Y.H., Onifade, I., Jelagin, D., & Birgisson, B., 2015. Mechanics-based top-down fatigue
cracking initiation prediction framework for asphalt pavements. Journal of Road Materials and Pave-
ment Design. doi:10.1080/14680629.2015.1055335
Ditlevsen, O. 1981. Principle of normal tail approximation. Journal of Engineering Mechanics
Divsion,107(6), 1191–1209.
Hasofer, A.M., & Lind, N.C., 1974. An exact and invariant second-moment code format., Journal of
Engineering Mechanics Division, 100(1), 111–121.
Huang, Y. 2004. Pavement analysis and design (2nd ed.). Upper Saddle River, NJ: Prentice Hall, p. 505.
Kim, H.B., & Buch, N., 2003. Reliability-based pavement design model accounting for inherent variabil-
ity of design parameters. Transportation Research Board, 82nd annual meeting, Washington, DC.
Maji, A., & Das, A. 2008. Reliability considerations of bituminous pavement design by mechanistic
empirical approach. International Journal of Pavement Engineering, 9(1), 19–31.
Montgomery, D.C. 2005. Design and analysis of experiments: Response surface method and designs.
Hoboken, NJ: John Wiley and Sons.
Noureldin, S.A., Sharaf, E., Arafah, A., & Al-Sugair, F. 1994. Estimation of standard deviation of pre-
dicted performance of flexible pavements using AASHTO model (Transportation Research Record,
No. 1449, pp. 46–56). Washington, DC: TRB.
PMS Objekt, Vägverket., 2008. VVTK VAG, Borlänge: Vägverket.
Rackwitz, R., & Fiessler, B., 1978. Structural reliability under combined random load sequences. Com-
puters and Structures, 9(5), 484–494.
Retherford, J.Q., & McDonald, M., 2010. Reliability method applicable to mechanistic-empirical pave-
ment design method, Transportation Research Record, No 2154, pp. 130–137). Washington DC:
TRB.
Schueller, G.I., Bucher, C.G., Bourgund, U., & Ouypornprasert, W. 1987. On efficient computational
schemes to calculate structural failure probabilities. In Y.K. Lin & G.I. Schueller (Eds.), Lecture notes
in engineering, Vol. 31: Stochastic structural mechanics (pp. 388–410). New York: Springer-Verlag.
SHELL International Petroleum Company (1978). Shell pavement design manual–Asphalt pavement
and overlays for road traffic. London: Author.
Sun, L., & Hudson, W.R. 2005. Probabilistic approaches for pavement fatigue cracking predictions
based on cumulative damage using Miner’s law. Journal of Engineering Mechanics, 131(5), 546–549.
Timm, D.H., Birgisson, B., & Newcomb, D.E. 1998. Variability of mechanistic–empirical flexible pave-
ment design parameters. Proceedings of the 5th International Conference on the Bearing Capacity of
Roads and Airfields, Vol. 1, Norway 1998.
Timm, D.H., Newcomb, D.E., Birgisson, B., & Galambos, T.V. 1999. Incorporation of reliability into
the Minnesota mechanistic-empirical pavement design method (Final Report Prepared to Min-
nesota Department of Transportation). Minneapolis: Minnesota University, Department of Civil
Engineering.
750
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Xiaowei Wang, Xingyu Gu, Mingzhu Yang, Huixu Ma & Heng Shen
School of Transportation, Southeast University, Nanjing, Jiangsu Province, P.R. China
ABSTRACT: This study aims to evaluate the permanent deformation of asphalt mixtures under
field vehicle loads at urban intersections. In order to obtain the information of signal timing, rut-
ting and traffic, a field investigation was conducted at urban intersections in Nanjing (Jiangsu,
China). Five loading models were designed to simulate the load induced by buses crossing inter-
section at normal speed, deceleration, low speed, braking and jam. A Partial Triaxial Creep Test
(PTCT) and 3-stage model was employed to characterize the permanent deformation of asphalt
mixtures and to analyze the effects of loading models, loading levels, and additives on permanent
deformation. Test results showed that the vehicle behaviors ranked by pavement damage is jam >
braking > low speed > deceleration > normal speed. It is also shown that the pavement service life
can be promoted by adding fiber or an Anti-rutting Agent (ATA) into asphalt mixtures.
1 INTRODUCTION
It has been observed that permanent deformation can be a major failure mode of asphalt
pavement and makes serious effects to road driving safety and its service life (Walubita et al.
2012). Permanent deformation generally consists of rutting and shoving. A rutting failure
alone occurs under traffic loads moving at high speed away from the intersection, while both
rutting and shoving failures may occur under traffic loads at intersections (Hajj 2005). It has
been indicated that permanent deformation at intersections is much more severe than other
road sections with the same pavement structure (Li et al. 2013). In other words, the perma-
nent deformation resistance of pavement at intersections is weaker than the pavement away
from intersections (Tannoury 2007). The deformation resistance of mixtures is closely related
to the vehicle behaviors which controlled by intersection signal.
Theoretical methods and finite element methods have been widely used to analysis the
permanent deformation of asphalt pavement at intersections (Wu et al. 2011). For example,
a computer code 3D-Moving load analysis (3D-Move) was used to estimate the response of
an asphalt pavement to the truck during braking period (Hajj et al. 2006), and the results
showed that rutting was mainly developed under the steering single tire while shoving was
mainly caused by the dual tandems driving tire. Tangential and vertical forces were assessed
by establishing finite element models to evaluate pavement permanent deformation (Li et al.
2013), but the field vehicle behaviors at intersections especially braking and jam behaviors
were not considered. Some researchers evaluated the rutting resistance of asphalt mixtures
by developing a testing method which can simulate the working conditions of filed asphalt
pavements (Li et al. 2014, Gu et al. 2015). In those studies, characteristics of vehicle load at
intersections have been studied by many scholars, however, the evaluation of the pavements
permanent deformation resistance under actual vehicle running states have not been take
enough consideration. Therefore, it is necessary to employ a reasonable experiment method
to evaluate the deformation resistance of asphalt mixture at intersection under field vehicle
behaviors.
Many test methods having been developed to evaluate the deformation properties of asphalt
mixtures, such as wheel tracking test, uniaxial static creep test, triaxial creep test, bending creep
751
test and so on (Gu et al. 2015), but all of them have not been accepted to have a good cor-
relation with the field performance of asphalt pavements (Li et al. 2014). With the fact that
pavement structure suffered from not only vertical stresses but also lateral confinement stresses
which varied with depth and temperature, it is more appropriate to use a partial triaxial creep
test (PTCT) to simulate the field conditions of pavement structure (Xu et al. 2014).
This paper aims to design loading models simulating field vehicle behaviors at intersections
and to evaluate the permanent deformation resistance of asphalt mixtures under different vehi-
cle behaviors. A field investigation about signal timing, traffic information, and rutting was con-
ducted at signalized intersections in Nanjing (Jiangsu, China) to design the appropriate loading
models with the vehicle behaviors of normal speed, deceleration, low speed, braking and jam.
3-stage model was developed to evaluate permanent deformation resistance and to analyze the
influence of loading models and loading levels on permanent deformation of asphalt mixtures.
2 FIELD INVESTIGATION
In order to obtain the parameters of loading models, a field investigation about signal timing
and rutting was carried out at four signalized intersections in Nanjing.
* AC13 is a kind of asphalt concrete whose maximum nominal aggregate size is 13mm;
752
Table 2. Conversion coefficients for different vehicles.
Lane type Left turn Through Right turn Left turn Through Right turn
East Road and Longpan Road with Ruijin Road as the selected intersections and choosing
car as the standard vehicle. Convert the traffic volume of medium trucks, large buses and
buses into the volume of cars according to conversion coefficients (Table 2) from Code for
Design of Urban Road Engineering (Beijing general municipal engineering design & research
institute company 2012) and the results are presented in Table 3.
Parking index can characterize the passing capacity of a lane during a signal cycle. The
higher parking index means a lower passing capacity that vehicle needs cost a longer time
passing intersection. Parking proportion has the same meaning and all of them are control-
led by signals. As Table 3 shows, the parking index and parking proportion of left turn lanes
are the largest. The result indicated that the waiting time demanded for turning left vehicles
was the longest and straight vehicles were the second longest. The volume of through lanes
was the largest. So the rutting depth of left turn lane and through lane is bigger than right
turn lane, which is consistent to the investigation results.
According to the investigation results, the composition of traffic vehicle including 78%
cars, 14% medium truck, 1.6% large bus and 5.0% bus. Because the loading level of bus is
the largest, bus is selected as the testing vehicle, and other loading levels are also discussed
in this paper. The signal timings of through lanes were selected as the testing parameters
because rutting of through lanes is the most severe and traffic volume of through lanes is
the most.
753
Table 4. Characteristics of AC-20C.
3 LABORATORY TESTING
3.1 Materials
The binder course contributes most to the whole deformation of pavement structure (Li et
al. 2014), so only the binder course was analyzed in this paper. AC-20C which is a typical
mixture of the binder course was selected to be investigated within this research. SBS modifi-
cation asphalt, anti-rutting agent (ATA) and fiber were employed to prepare the binder. The
mixture composition is detailed in Table 4. The cylindrical specimen of 150mm in diameter
and 80mm in height was compacted using a gyration compactor. The target void of AC-20C
is 4%.
754
Figure 1. Finite element models.
4 LOADING MODELS
Loading period and dwell period were determined by vehicle speed, but vehicle behaviors
are unpredictable and fluctuant. As a result, loading period and dwell period are difficult to
ascertain. A calculation method has been widely used which was first proposed by Huang.
Results showed that the extreme distance where vehicles are able to make difference is six
times the radius of tire contact area (Huang 1998). Loading period can be calculated accord-
ing to the following formula:
12r
t= (1)
V
Where r is the radius of equivalent circular and value is 0.151m when vehicle load is the
standard vehicle (Deng 2005). Dwell period can be calculated according to the space head-
way and speed. The loading mode of permanent deformation test of asphalt mixture usually
adopts a single pattern of 0.1s-0.9s (loading period of 0.1s and dwell period of 0.9s) (Gu et
al. 2015, Huang & Zhang 2010). However, the vehicle behaviors are various at intersections.
It is unsuitable to use a single immobilized loading model to simulate the loading progress.
Studies have verified that haversine load was more proper to simulate the load induced by
continuous vehicle movement (Kim et al. 2004). As for intermittent movement of braking
and jam, a square wave was more suitable for loading mode because standing time of the
vehicles was significantly longer than running time.
Vehicle behaviors can be classified into five types: normal speed, deceleration, low speed,
braking and jam according to the traffic investigation as introduced before. Characteristics
of each behavior are presented in Table 6. Based on the characteristics of each behavior, five
loading models were designed and shown in Table 6 and Figure 2. The specimens were placed
in the oven at 60 °C for 4 hours. The tests were conducted using a UTM-25 system with five
loading models. The value of load is 0.83MPa (bus standard load), so the load applied by
loading head is 3.66kN.
Figure 3(a) illustrates the rutting depth (RD) of asphalt mixtures with the time of loading
repetitions in the PRCT at 60 °C.Loading models under the behavior of braking and jam are
755
Table 6. Characteristics of each vehicle behavior and loading parameter.
Jam
Loading mode Normal speed Deceleration Low speed Braking Red Green
Speed (km/h) 64 40 20 0 0 5
Space headway (m) 16 10 4 – – 2
Loading period (s) 0.1 0.16 0.32 95 95 1.5
Dwell period (s) 0.9 0.9 0.72 55 0 1.5
vibration waves, so they produced the rutting depth curves with a vibration upward trend.
Figure 3(b) shows the development of rutting depth in the first 150s. For braking, dwell
period began at 95s to simulate the actual working conditions, while a continuous vibra-
tion loading with a cycle of 3s was used to simulate the jam condition. It can be seen from
756
Figure 3. The rutting depth development under different loading models.
Figure 3(b), there is a rapid decrease of RD at 95s in braking, while the rutting depth under
jam condition has a continuous vibration increase at 95s.
From Figure 3, it can be seen that the entire deformation process of mixture under the
loading models of normal speed, deceleration and low speed were accorded with a 3-stage
model (Zhou et al. 2004), however the deformation under the behaviors of braking and jam
inconsistent with it, because the loading models of those two behaviors are vibrative. But if
separate the deformation curve of braking and jam showed in Figure 3 into two parts: load-
ing-deformation curve A and unloading-deformation curve B presented in Figure 4. Curve A
and B are apparently accorded with the following 3-stage model (Zhou et al. 2004).
Table 8. Results of 3-stage model under different loading levels (60 °C).
loading levels, PTCT was performed with four loading levels: 0.24Mpa (loading of car),
0.5Mpa, 0.7Mpa, and0.83Mpa (loading of bus). Braking has the highest proportion in all
of vehicle behaviors at intersections. Therefore, the loading mode was adopted correspond-
ing to braking as the selected loading mode. The test was conducted at 60 °C and results are
presented in Figure 5 and Table 8.
It is found from Figure 5 and Table 8 that with the increase of loading level, the deforma-
tion of mixtures increased. There is no failure and rutting depth is almost the same under the
loading level of 0.24Mpa and 0.50Mpa. When loading level is 0.70MPa and 0.83MPa, the
permanent deformation of asphalt mixtures accumulated quickly and led to failure at 3750s
and 1425s respectively. Pavement service life under the loading level of 0.83MPa is shorter
2.6 times than the loading level of 0.7MPa. It is also can be conclude that overload is a vital
factor causing severe permanent deformation on asphalt pavement.
In order to prevent intersection pavements from failure, various additives have been incorpo-
rated with asphalt mixture at intersection to improve the permanent deformation resistance.
This paper used anti-rutting agent (ATA) and basalt fibers as selected additives. The com-
759
Figure 6. RD development under different additives.
positions of mixture are shown in Table 4. Loading level is 0.83Mpa and test temperature is
60°C. The test result of mixture under the loading mode of braking is presented in Figure 6
and Table 9.
Table 9 shows that t1st and t2nd of the mixtures with ATA and basalt fibers are increased
greatly than ordinary mixture. It is also found that 2nd decreased greatly while t2nd increased
about 4 times. Rutting depth decreased significantly as well as the permanent deformation
resistance was greatly improved.
7 CONCLUSIONS
Based on the field vehicle behaviors at intersections, Partial triaxial creep tests (PTCT) with
five loading models were designed in this study to evaluate the permanent deformation of
asphalt mixtures. A field investigation was conducted at a signalized intersection in order to
obtain the parameter of loading models. 3-stage model was employed to characterize the
permanent deformation of mixtures. Critical points of this model were developed to evaluate
the permanent deformation resistance of mixtures. Several conclusions can be summarized
as followed:
1. According to the critical points of 3-stage model and pavement damage, vehicle behaviors
rank as: jam > braking > low speed > deceleration > normal speed.
2. Intersection vehicle behaviors means different loading period which increased permanent
deformation of asphalt mixture. Longer loading period and shorter dwell period makes
jam and braking have a more serious damage on the service life of the pavement.
3. When loading level is higher than 0.70MPa and is integrate with the disadvantageous
vehicle behaviors, the permanent deformation of asphalt mixture will rapidly increase and
lead to pavement failure.
4. The permanent deformation resistance can be greatly improved by adding ATA and
basalt fibers. This would be an efficient method to prevent pavements from failure at
intersections.
760
8 DISCUSSION
From what discussed above, vehicle behaviors at intersections play an important role in per-
manent deformation of asphalt mixtures. The highlights of this paper are the use of a well-
designed test method and loading models to evaluate the permanent deformation of asphalt
mixtures under field intersection conditions. A new testing method to simulate horizontal
force induced by braking to evaluate the permanent deformation of asphalt mixtures needs
to be further studied.
REFERENCES
Beijing general municipal engineering design & research institute company 2012. Code for design of
urban road engineering. Beijing: China Architecture & Building Press.
Deng, X. 2005. Subgrade and pavement engineering. Beijing: China Communications Press.
Gu, X., Dong, Q., Yuan, Q. 2015. Development of an innovative uniaxial compression test to evaluate
permanent deformation of asphalt mixtures. Journal of Materials in Civil Engineering 27(1): 1–9.
Hajj, E.Y. 2005. Hot mix asphalt for Nevada’s intersections. University of Nevada. Ph. D dissertation,
Reno: University of Nevada.
Hajj, E.Y., Sebaaly, P.E., Siddharthan, R.V. 2006. Response of an asphalt pavement mixture under a
slow moving truck. Asphalt Concrete: 134–146.
Huang, X. & Zhang, Y. 2010. A new creep test method for asphalt mixtures. Road Materials and Pave-
ment Design 11(4):969–991.
Huang, Y. 1998. Analysis and design of pavement. Beijing: China Communications Press.
Kim, S.H., Little, D.N., Masad, E., Lytton, R.L. 2004. Prediction of anisotropic resilient responses
for unbound granular layer considering aggregate physical properties and moving wheel load. Pro-
ceedings of aggregates: asphalt concrete, portland cement concrete, bases and fines. Twelfth annual
symposium, Denver. 24p.
Li, L., Huang, X., Wang, L., Li, C. 2013. Integrated experimental and numerical study on permanent
deformation of asphalt pavement at intersections. Journal of Materials in Civil Engineering 25(7):
907–912.
Li, Q., Ni, F., Gao, L., Yuan, Q., Xiao, Y. 2014. Evaluating the rutting resistance of asphalt mixtures
using an advanced repeated load permanent deformation test under field conditions. Construction
and Building Materials 61: 241–251.
Research institute of highway ministry of transport. 2008. Field test methods of subgrade and pave-
ment for highway engineering. Beijing: China Communications Press.
Tannoury, G.Y. 2007. Laboratory evaluation of hot mix asphalt mixtures for Nevada’s intersectons—
Phase II. Master thesis, Reno: University of Nevada.
Walubita, L.F., Zhang, J., Das, G., Hu, X. et al. 2012. Hot-mix asphalt permanent deformation evalu-
ated by Hamburg wheel tracking, dynamic modulus, and repeated load tests. Transportation Research
Record 2296: 46–56.
Wu, Z., Chen, X., Yang, X., Zhang, Z. 2011. Finite element model for rutting prediction of flexible pave-
ment with cementitiously stabilized base-subbase. Transportation Research Record 2226: 104–110.
Xu, T., Wang, H., Li, Z., Zhao, Y. 2014. Evaluation of permanent deformation of asphalt mixtures
using different laboratory performance tests. Construction and Building Materials 53: 561–567.
Yuan, Q. 2012. Research on experiment method of asphalt pavement permanent deformation at high
temperature. Master thesis, Nanjing: Southeast University.
Zhou, F., Scullion, T., Sun, L. 2004. Verification and modeling of three-stage permanent deformation
behavior of asphalt mixes. Journal of Transportation Engineering 130(4): 486–94.
761
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: A long service life of asphalt pavements can be achieved through a good and
durable interlayer bond. A new apparatus for Cyclic Testing of the Interlayer Bond (CTIB) in
double-layered asphalt specimens has been developed to determine the shear stiffness at the
interface between asphalt layers. In order to examine the interactions of repeated traffic load-
ing, acceleration and braking processes as well as temperature, an extensive testing procedure
has been created taking into account all the parameters that influence the interlayer bond.
Experimental tests have been carried out under sinusoidal repeated shear loading conditions
at varying frequencies and temperatures under different normal stresses. The dependence of
the combined effect of temperature, shearing frequency and normal pressure on shear stiff-
ness has been approximated with sigmoid master functions.
1 INTRODUCTION
2 BACKGROUND
The interlayer bond is achieved through the adhesion between the asphalt binder of the two
layers and the applied tack coat, the interlocking of the aggregate particles at the interface
763
and the friction between the surfaces of the two asphalt layers. To ensure a good interlayer
bond, the top surface of the underlying layer should be given a good clean before spraying it
with a bitumen emulsion and before placing the next layer.
The pavement is loaded permanently in both a vertical direction by wheel loads and a
horizontal direction by braking and acceleration processes, which may fully change the three-
dimensional stress state in the entire pavement structure and may lead to horizontal displace-
ments of the layers if a poor interlayer bond is present. Temperature variation and moisture
also cause additional stresses, which can compromise the functionality of the pavement. This
may result in a reduction or a loss of load capacity, premature material fatigue, pothole for-
mation, separation of the layers, delamination of wearing course, slippage cracking, distor-
tion and significantly shorter service life of the entire asphalt construction.
A full bond between the asphalt layers in pavements is expected to exist if there is a full-
surface and rigid connection at the interfaces. During load transfer the asphalt layers act as
a unit and relative displacements between the asphalt layers do not exist. When there is a
poor or no bond between the layers, large displacements may appear. A good interlayer bond
contributes to a long service life, driving comfort and road safety.
In Germany the assessment of the interlayer bond in asphalt pavements is currently car-
ried out with a static test according to Leutner by determining the shearing-off force as a cri-
terion for the evaluation of the achieved bond strength, which is regulated in FGSV (2007).
In many countries there are some similar methods with different test conditions (Raab et al.
1999, Sholar et al. 2004, Leng et al. 2008, Tashman et al. 2006). Some testing procedures and
test devices are improved so that normal stresses can additionally be applied (West et al. 2005,
Uzan et al. 1978, Canestrari & Santagata 2005, Mohammad et al. 2005, Mohammad et al.
2012). The disadvantage of the static tests is that they cannot simulate the real loading state
in the pavement due to repeated loads, and therefore they are inapplicable for the accurate
description of the bond behavior.
Several devices have been developed over the years, which make the initiation of shear
stresses in the specimens possible via cyclic loads with or without interaction of normal
loads (Wellner & Ascher 2007, Romanoschi & Metcalf 2003, de Bondt 1999, Crispino et al.
1999, Monismith et al. 1994, Sanders et al. 1999). All testing devices differ in terms of test-
ing procedure, type of load (force-controlled or position-controlled testing procedure), test-
ing temperature and normal stresses. At TU Dresden, Wellner & Ascher (2007) have used a
shear apparatus for cyclic shear tests with static normal force. The detailed testing procedure
includes temperatures ranging from –10°C to 50°C, frequencies in the range of 1 Hz to 15 Hz
and normal stresses ranging from 0 to 1,1 MPa. The purpose of the project has been to find
a “bonding factor” for good and poor bond.
Within the ongoing project “Cyclic Shear Stiffness and Shear Fatigue Test for Evaluation
and Optimization of the Interlayer Bond in Asphalt Pavements”, a new improved apparatus
for CTIB has been designed at the Institute of Urban and Pavement Engineering at TU
Dresden. Its purpose is to simulate as accurately as possible the real load conditions which
exist on in-service asphalt pavements. The present work introduces this apparatus which is
an improved version of the testing apparatus used by Wellner & Ascher (2007). An extensive
testing procedure has been created to take into account all the parameters that influence the
interlayer bond, thereby facilitating the examination of the interactions of repeated traffic
loading, acceleration and braking processes as well as weather-related effects. The examined
shear stiffness at the interface between the two layers (Equation 3) has been defined as shear
stress amplitude divided by the ratio of shear displacement and gap between the two steel
adapters holding the double-layered asphalt specimen (Fig. 1). It has been calculated at dif-
ferent systematically varied shearing frequencies, normal stresses, and temperatures. In order
to analytically determine the shear stiffness at any desirable parameter combination a sig-
moid function has been developed and the results have been compared with the experimen-
tally determined ones. This function can further be implemented in the German pavement
design (FGSV 2009) to assess the effect of the interlayer bond on the whole asphalt pavement
and predict its service life.
764
Figure 1. Specimen preparation for CTIB.
3 EXPERIMENTAL PROGRAM
765
Figure 2. Test apparatus mounted in the temperature chamber of a servo-hydraulic testing machine
(left) and schematic representation of the apparatus for CTIB (right).
which was fastened to jowl B. The vertical shear displacement was measured using two sen-
sors, which were attached to jowl B. Another two sensors were mounted horizontally on
jowl A to measure its horizontal motion. The mechanical model of the test apparatus is also
shown in Figure 2.
4 EXPERIMENTAL RESULTS
The measured and the calculated values obtained in the course of the experiment were
recorded in log files. In order to exclude the initial deviations at the beginning of each shear-
ing frequency, only the data from the last five cycles at its end were used for the calculation of
the shear stiffness. It was not reasonable to show the variation of the shear stiffness with time
because of the long tempering duration between the subtests. Therefore, a sequential num-
bering was chosen on the x-axis showing only values during the subtests. The shear strain is
766
Figure 3. Testing procedure.
τs
tan γ s = (1)
Gs
For small shear angles tanγs ≈ γs. The shear strain can be calculated as the ratio of the
shear displacement sw to the gap between the steel adapters ds, which was 1.0 mm for all tests
(Fig. 1).
sw
tan γ s γs = (2)
ds
The shear stiffness was calculated as follows:
τs τ
Gs = = s (3)
γ s ⎛ sw ⎞
⎜⎝ d ⎟⎠
s
with
Fs
τs = (4)
A
where Fs = shear force amplitude [N]; γs = shear strain [-]; τs = shear stress [MPa]; A = cross
section at the interface [mm2]; sw = shear displacement amplitude [mm]; ds = gap between steel
adapters [mm]; Gs = shear stiffness [MPa/mm].
A comparison between the average shear stiffness values at 200 g/m2, 300 g/m2 and 400 g/m2
of C60BP1-S bitumen emulsion applied on clean surfaces as well as the temperature, normal
stress and frequency gradients from the tests are shown in Figure 4. The layer combination
was binder course on base course and three asphalt specimens were tested for each amount of
bitumen emulsion. It was observed that the normal stress and the shearing frequency influ-
767
Figure 4. Gradients of the average shear stiffness for 200 g/m2, 300 g/m2 and 400 g/m2 bitumen emul-
sion C60BP1-S applied on clean surface.
ences the shear stiffness to various degrees at different temperatures. The shear stiffness of
the interlayer bond decreased rapidly with increasing temperature and grew with increasing
normal stress, as it occurs under traffic load. The impact of the normal stress on the shear
stiffness was considerably smaller at low temperatures compared to the higher temperatures
irrespective of the tack coat amount used. As can be seen in the diagram, the shearing fre-
quency has a significant impact on the shear stiffness. The latter increased with increasing
frequency, whereby the smallest increment was observed at the temperature extremes. The
highest shear stiffness was observed for 300 g/m2 while the lowest was measured for 200 g/m2
tack coat at all temperatures. The shear stiffness values of the interlayer bond with 400 g/m2
bitumen emulsion was always between the stiffness values of the other two amounts. It is
assumed that because of the high roughness of the base course’s surface the smaller amount
of 200 g/m2 penetrates into the pores and therefore strong adhesion cannot be achieved.
Obviously 300 g/m2 bitumen emulsion is the optimal amount to produce a stronger bond
for this particular asphalt layer combination. The interlayer bond does not get better when
increasing the amount of tack coat above 300 g/m2, because the bitumen emulsion starts
acting as a “lubricating film” between the layers, thereby reducing the interlocking at the
interface.
Since there is always an aggregate interlocking on rough interfaces, it is important to
separate the adhesion from the friction. It was assumed that when no normal pressure was
induced, basically the effects of adhesion between the asphalt binder of the two layers and the
applied bitumen emulsion would take place. The temperature dependence is a characteristic
feature of the bituminous binder, which makes the adhesion temperature dependent too. At
the lowest temperature the adhesion at the interface was strong. The friction was insignificant
because the displacements were smaller than 30 μm and therefore it could be neglected. The
effects of adhesion, aggregate interlocking and friction took place concurrently at tempera-
tures of 10°C and 30°C. Adhesion strength decreased with increasing temperature, which in
turn led to continuous reduction in shear stiffness indicating a successive deterioration of the
interlayer bond. At 50°C there was no adhesion at all, and the shear stiffness at the interface
768
was achieved only through the aggregate interlocking and the friction between the surfaces
of the two asphalt layers. The results show that at 50°C all three shear stiffness gradients are
nearly the same at any normal stress and frequency. The shear stiffness values are approxi-
mately zero MPa/mm for all three tack coat amounts at the lowest frequency when no normal
stress is applied. This means that there is a complete loss of friction at the interface indicating
an inadequate interlayer bond. Once normal pressures and higher shearing frequencies were
applied, the friction was activated again and the shear stiffness increased.
In order to examine only the effect of adhesion, without having aggregate interlocking and
friction at the interface, both layer’s surfaces were given a fine polish before applying the tack
coat. It was observed that the shear stiffness gradients for 200 g/m2 and 300 g/m2 are superim-
posed at all temperatures (Fig. 5). The average values for all three tack coat amounts were almost
the same. Slightly better adhesion was found to exist for 400 g/m2 at the highest normal pressure
and at higher frequencies, more pronounced at -10°C and 10°C. The adhesion at 400 g/m2 was
weaker than the adhesion at the other two bitumen emulsion amounts when no normal stress
was induced, which was evidence that the interlayer bond produced with the highest amount
was more prone to the combined influence of all three parameters. At a temperature of 50°C all
shear stiffness gradients were approximately the same and there was a complete loss of adhesion
and friction at a shearing frequency of 0.1 Hz when no normal pressure was applied.
Due to the additional effects of aggregate interlocking at -10°C and of the combination of
both aggregate interlocking and friction at 10°C and 30°C for normally produced interlayer
bond (Fig. 4) the shear stiffness values for all three tack coat amounts are generally higher
than those shown in Figure 5. At 50°C all shear stiffness values depreciate identically, indicat-
ing high deterioration of the interlayer bond.
For the analytical evaluation of the shear stiffness, it was necessary to develop a master
function which takes into account the combined influence of temperature, normal pressure
and frequency. First, the dependence between the shear stiffness and the temperature-fre-
quency equivalence was established at the four normal stresses. The regression used for the
master curves which approximated the experimental values most accurately was the sigmoid
Figure 5. Gradients of the average shear stiffness for 200 g/m2, 300 g/m2 and 400 g/m2 bitumen emul-
sion C60BP1-S between finely polished course surfaces.
769
function. This function approaches asymptotically the minimum and the maximum values of
the shear stiffness. The temperature-frequency equivalence was computed as follows:
⎛ ⎞
⎜ 1 ⎟
m f ⋅ α T ) = log f ⋅ E ⎛ 1 (5)
⎜ − a ⋅⎜ −
1 ⎞⎟
⎟
⎝ e R ⎝ T + 273.15 TR + 273.15 ⎠ ⎠
Gs Gs ,min +
(G s, a Gs ,min ) (6)
1+ e ( m ⋅a + b )
a c1 ⋅ ln σ N + c2
(7)
b d1 ⋅ σ N + d2
(8)
Figure 6. 3D plot of the dependence between shear stiffness calculated with Equation 9, normal pres-
sure, temperature and shearing frequency.
770
After substituting the regression parameters a and b in Equation 6 the equation becomes:
Gs Gs ,min +
(G s, a Gs ,min ) (9)
1 + e[m c ⋅ N +c
+c d N +d ]
The 3D plot in Figure 6 shows the dependence between the shear stiffness determined with
Equation 9, the normal pressure, the temperature and the shearing frequency. Comparison
between the experimentally obtained shear stiffness values and the calculated values with the
regression showed very good agreement.
5 CONCLUSIONS
To determine the shear stiffness at the interface at different temperatures and normal stresses
under sinusoidal repeated shear loading conditions at varying frequencies, a new test appa-
ratus and extended test procedure have been developed. The experimental results show that
the shear stiffness of the interlayer bond decreases rapidly with increasing temperature. It has
been observed that higher normal stresses and shearing frequencies have a positive effect on
the interlayer bond shear stiffness increasing it significantly. Notwithstanding the tack coat
amount, the impact of the normal stress on the shear stiffness was considerably smaller at lower
temperatures compared with that at higher temperatures. The results from the tests of the bond
produced with C40BF1-S bitumen emulsion show that at all temperatures the highest shear
stiffness is achieved with 300 g/m2 tack coat followed by the amount of 400 g/m2. The lowest
shear stiffness has been found to result from the use of 200 g/m2 tack coat at all temperatures.
In order to exclude the effects of aggregate interlocking and friction at the interface and to
test solely the effect of the adhesion, asphalt specimens of finely polished layer surfaces have
been used. It has been observed that the shear stiffness gradients for 200 g/m2 and 300 g/m2
are equal at all temperatures. The interlayer bond produced with tack coat amount of 400 g/
m2 has been found to be most prone to the combined influence of temperature, shearing
frequency and normal stress.
In both the normally produced bond and the bond between polished surfaces a complete
loss of adhesion and friction occurs at a temperature of 50°C at the lowest frequency when
no normal pressure is applied indicating an inadequate interlayer bond.
A master function for the shear stiffness Gs has been developed. The regression which
approximates most accurately the experimentally determined shear stiffness values is the sig-
moid function. This function takes into account the combined influence of temperature,
normal pressure and shearing frequency and the calculated shear stiffness agrees very well
with the experimentally assessed one.
ACKNOWLEDGMENTS
The authors would like to thank the German Federation of Industrial Research Associations
“Otto von Guericke” (AiF) and DFG under grant KA 1163/31 (FOR 2089) for financial
support.
REFERENCES
Bondt de, A.H. 1999. Anti-reflective Cracking Design of (Reinforced) Asphaltic Overlays. Ph.D. Dis-
sertation, Delft University of Technology.
Canestrari, F. & Santagata, E. 2005. Temperature effects on the shear behaviour of tack coat emul-
sions used in flexible pavements. The International Journal of Pavement Engineering, Vol. 6, No. 1,
pp 39–46.
771
Crispino, M., Festa, B., Giannattasio, P. & Nicolosi, V. 1997. Evaluation of the interaction between the
asphalt concrete layers by a new dynamic test. 8th International Conference on the Structural Design
of Asphalt Pavements, Washington State University, Seattle, pp 741–754.
Forschungsgesellschaft für Straßen- und Verkehrswesen (FGSV) 2007. Technische Prüfvorschriften für
Asphalt TP Asphalt-StB, Teil 80 – Abscherversuch, Köln.
Forschungsgesellschaft für Straßen- und Verkehrswesen (FGSV) 2009. Richtlinien für die rechnerische
Dimensionierung des Oberbaus von Verkehrsflächen mit Asphaltdeckschicht RDO Asphalt 09,
Köln.
Leng, Z., Ozer, H., Al-Qadi, I.L. & Carpenter, S.H. 2008. Interface Bonding Between Hot-Mix Asphalt
and Various Portland Cement Concrete Surfaces: Laboratory Assessment. Transportation Research
Record: Journal of the Transportation Research Board, No. 2057, Transportation Research Board of the
National Academies, Washington, D.C., pp 46–53.
Mohammad, L.N., Elseifi, M.A., Bae, A., Patel, N., Button, J. & Scherocman, J.A. 2012. Optimiza-
tion of Tack Coat for HMA Placement. In Transportation Research Board, NCHRP REPORT 712,
Washington, D.C.
Mohammad, L.N., Wu, Z. & Raqib, A. 2005. Investigation of the Behavior of Asphalt Tack Coat Inter-
face Layer. Louisiana Transportation Research Center.
Monismith, C.L., Hicks, R.G., Finn, F.N., Sousa, J., Harvey, J., Weissman, S., Deacon, J., Coplantz,
J. & Paulsen, G. 1994. Permanent Deformation Response of Asphalt Aggregate Mixes. SHRP Strategic
Highway Research Program SHRP-A-415, Washington, D.C.
Raab, C. & Partl, M.N. 1999. Methoden zur Beurteilung des Schichtenverbunds von Asphaltbelägen.
ASTRA-Project FA 12/94.
Romanoschi, S.A. & Metcalf, J.B. 2003. Errors in Pavement Layer Moduli Backcalculation due to
Improper Modeling of the Layer Interface Condition. In Transportation Research Board Annual
Meeting CD-ROM.
Sanders, P.J., Brown, S.F. & Thom N.H. 1999. Reinforced Asphalt Overlays for Pavements. Ph.D. Dis-
sertation, University of Nottingham.
Sholar, G.A., Page G.C., Musselman, J.A., Upshaw, P.B. & Moseley, H.L. 2004. Preliminary Investiga-
tion of a Test Method to Evaluate Bond Strength of Bituminous Tack Coats. Journal of the Associa-
tion of Asphalt Paving Technologists, Vol. 73, pp 771–806.
Tashman, L., Nam, K. & Papagiannakis, T. 2006. Evaluation of the Influence of Tack Coat Construction
Factors on the Bond Strength Between Pavement Layers. Washington Center for Asphalt Technology,
Washington State University.
Uzan, J., Livneh, M. & Eshed, Y. 1978. Investigation of Adhesion Properties between Asphalt Concrete
Layers. Asphalt Paving Technology, Vol. 47, Proceedings of the Association of Asphalt Paving Tech-
nologists, Technical Sessions.
Wellner, F. & Ascher, D. 2007. Untersuchungen zur Wirksamkeit des Haftverbundes und dessen
Auswirkung auf die Lebensdauer von Asphaltbefestigungen. Schlussbericht zum AiF-Projekt Nr.
13589 BR/1, Dresden.
West, R.C., Zhang, J. & Moore, J. 2005. Evaluation of Bond Strength between Pavement Layers. NCAT
Report 05–08, The Alabama Department of Transportation.
772
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The material of stress absorption interlayer has the superiority in delaying
or restraining the reflection cracks of old concrete road surface, preventing water from sur-
face, therefore it can prolong the service life of pavements. In order to improve the service
performance of stress absorb layers, in this study methods including using bitumen with
high viscosity and elastic recovery, mix proportion optimum design and adding fiber into
asphalt mixture were explored. The performances of stress absorption interlayer, including
shearing in high temperature, cracking in low temperature, stability in water, were tested.
The tests indicate that fiber-reinforced stress absorption interlayer has excellent service
performance.
1 INTRODUCTION
Reflective cracking is a common problem found in both semi-rigid base of asphalt pave-
ment and old cement concrete pavement. The formation of reflection cracks causes the
diffusion of surface water into the pavement and soil bases, which accelerates the destruc-
tion of pavement and seriously affects the performance of asphalt pavement. However,
to include the stress absorption interlayer in asphalt pavement can better ease the stress
concentrated in the cracking areas of semi-rigid base or cement concrete plate joints. The
interlayer can also lower the load and thermal stress of the based layer of asphalt overlay,
so as to effectively delay the formation of reflection cracking, and prevent the infiltration
of surface water to extend the service life of the pavement (Chen et al, 2005; Li et al, 2007;
Leonard, 2005).
The asphalt mixture of stress absorption interlayer is made up of a large proportion
of fine aggregate, mineral fillers and modified asphalt binders with exceptional high con-
centration of content and elastomeric polymer. The maximum nominal grain size of the
asphalt mixture is 4.75 mm, with a mineral aggregate gradation of 0∼2.36 mm, which
accounts for more than 70% of the total aggregates. Moreover, the special modified
asphalt content is between 8%∼10%. As the asphalt is modified to have high viscosity and
high elasticity, the material can effectively prevent reflection cracking because it is not
only denser, but also with good deformation resistance, high elastic recovery ability and
good resistance against fatigue. In other words, this new structure of asphalt pavement
material has different technical performances compared to other types of asphalt mixtures
(Donald, 1990).
Therefore, by integrating the current stress absorption interlayer technology, adjusting the
composition and material optimization, and including fibers, this paper improves the per-
formance of stress absorption interlayer of asphalt mixture, and discusses the stability of
such mixture under different water and temperature conditions.
773
2 MATERIAL COMPOSITION
Table 1. The result of asphalt bitumen with high viscosity and elastic recovery.
Penetration 0.1 mm 87 65
Ductility(5°C) Cm 55 41
Softening Point(R&B) °C 97 79
Rotational viscosity(135°C) Pa⋅s 2.9 1.7
Dynamic viscosity(70°C) Pa⋅s 28680 4057
Dynamic viscosity(80°C) Pa⋅s 9087 1869
Isolation, Softening point difference °C 3.1 1.8
Elastic recovery(25°C) 3 min % 99 68
1h % 100 98
Weight loss % 0.005 −0.012
Penetration ratio % 84 77
Ductility(5°C) cm 54 28
Elastic recovery(25°C) 3 min % 95 67
1.5 h % 100 91
774
mixture so that not only the performance of mixture is improved, but also the construction
workability. The volume index for gyratory compaction of fiber-enhanced asphalt mixture is
designed to be 100 times, and respective results are shown in Table 2.
When gyratory compaction times G = 10, void ratio requirement is greater than 4%, which
ensures the ability of asphalt mixture to resist permanent deformation to a certain degree;
when gyratory compaction times G = 25, void ratio requirement is 1.5∼2.5%, and the require-
ment for voids of mineral aggregate is greater than 17%, which ensures the construction
workability of stress absorption interlayer of asphalt mixture and its overall road perform-
ance, especially the waterproof function.
For general stress absorption interlayer, when gyratory compaction times G = 50, void
ratio requirement is 0.5∼1.5%, voids of mineral aggregate requirement is greater than 17%,
and bitumen saturation requirement is greater than 90%. For enhanced stress absorption
interlayer, when gyratory compaction times G = 100, void ratio requirement is 0.5∼1.5%,
voids of mineral aggregate requirement is greater than 17%, and bitumen saturation require-
ment is greater than 90%, which ensures the overall performance of heavy loaded asphalt
mixture under high temperature.
10 >4 >4 — — — —
25 1.5∼2.5 1.5∼2.5 >17 >17 — —
50 0.5∼1.5 — >17 — >90 —
100 — 0.5∼1.5 — >17 — >90
Mesh size(mm) 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075
775
3 HIGH TEMPERATURE PERFORMANCE EVALUATION
The stress absorption interlayer is located in parts of the joints or cracks within the structure,
where shear stress is greater than other parts(Yang et al, 2005). Hence, the ability to withstand
shear stress is one of the evaluation index to determine the stability performance of the stress
absorption interlayer under high temperature. To complete the evaluation for the shearing
properties of the stress absorption interlayer, the single penetration test is introduced, which
acts on the specimen through the pressed steel indenter. The diameter of the indenter head
is small enough, so that the force experienced can be more consistent with that acting on
the actual road. The temperature designed for single penetration test is 60°C, loading rate is
1 mm/min, and the diameter of cylindrical specimens of asphalt mixture samples is 100 mm.
Formula (1) provides the calculation method for determining the shear strength.
τ0 f × fh × σ p
(1)
F
σp =
A (2)
Where τ0 = Shear strength; σ p = Vertical stress value; F = Load limit; A = Steel indenter
cross-sectional area; f = Shear stress coefficient. For mixture with maximum nominal size
less than or equal to 16 mm, f = −0.0018 h + 0.357; for mixture with maximum nominal
size greater than 16 mm, f = 0.35 (h > 6 cm); f = −0.0127 h + 0.427 (h = 4∼6 cm), in which h
is the specimen height in cm. fh = Non-standard specimen height correction factor, where
specimen number = 1. For mixture with non-standard height maximum nominal size less
than or equal to 16 mm, fh = 0.0416 h + 0.5834, in which h is the height of the specimen
in cm.
According to the above test method, single penetration tests are performed to both mix-
tures with general stress absorption interlayer and enhanced stress absorption interlayer, and
results are shown in Table 5.
Table 5. Comparison of single penetration test results of two types of asphalt mixture.
776
Under high temperature, the single penetration test actually represents the shear deforma-
tion capacity of asphalt mixtures. The shear deformation capacity of common hot asphalt
mixture mainly depends on the interlocking dependencies of materials, while the binder
properties of asphalt mixture are the key determinants for shear deformation capacity of
mixture with stress absorption interlayer, which has a suspension structure. From the above
test results, improvements can be observed among loading limit, vertical stress value and
shear strength for the asphalt mixture with enhanced stress absorption interlayer. Also, a
more substantial increase is observed in shear strength of about 20%. This means that the
addition of fiber provides a better shear deformation capacity for the stress absorption inter-
layer of asphalt mixtures.
Table 6. Comparison of low-temperature bending test results of two types of asphalt mixture.
The maximum
damage Flexural The maximum Stiffness
The maximum deflection strength bending tensile modulus
Category failure load(N) (mm) (MPa) strain (εB) (MPa)
777
Table 8. Comparison of freeze thaw and splitting test results of two types of asphalt mixture.
stress absorption interlayer mixture are basically the same, i.e. 2.291 MPa and 2.392 MPa
respectively, which indicating a similar tensile strength. However, the maximum displacement
values under tensile stress are 0.91 mm and 1.70 mm respectively, which means that by add-
ing fiber, the toughness of mixture has doubled, and the mixture is able to withstand greater
displacement with a better anti-cracking performance at low temperature.
The freeze thaw splitting test of water stability of stress absorption interlayer mixture is
designed to test the freeze thaw splitting strength ratio of both common and enhanced mix-
ture types. Test results are shown in Table 8.
The above test results for both the common and enhanced stress absorption interlayer
mixtures are able to meet the standard requirements, this is because of the small porosity in
the stress absorption interlayer. Comparing to the common type, the enhanced interlayer has
improved in failure test load and tear strength. Also, an increase in value can be observed in
freeze thaw splitting intensity ratio. Therefore, due to the reinforced fibers in stress absorp-
tion interlayer under increased compaction work, the water stability performance is improved
to a certain extent.
6 CONCLUSION
This paper has discussed and analyzed the performance of asphalt mixture by applying of
high viscosity and elastic recovery modified asphalt, optimizing design methods, and adding
fibers to improve the performance of stress absorption interlayer of asphalt mixture. The
conclusions are as follows
1. High viscosity and elastic recovery asphalt mixtures should be used for stress absorption
interlayer. Such asphalt has characteristics of high viscosity and high elasticity, where
its dynamic viscosity is much larger than that of SBS modified asphalt, and has a better
recovery ability.
2. Gyratory compaction method provides volume indicators for different frequency of gyra-
tory compaction, and the designed compact frequencies for common and enhanced stress
absorption interlayer of asphalt mixture are 50 and 100 times respectively. The optimum
asphalt content is determined by equating the middle value of a range of porosity values
from compaction to the target porosity value, while other indexes, for example, the volume
indicators of spinning 10 and 25 times, should meet the technical requirements.
3. Adding fibers could significantly improve the performance of stress absorption interlayer
of the asphalt mixture. The shear resistance capacity can be improved by 20% for high
temperature, and the displacement value under maximum tensile stress has increased for
nearly 1 time for low temperature. In addition, although the testing values under the bend-
ing test at low temperature and the freezing and thaw splitting test have increased, there is
little difference. In other words, the evaluation results of material differences under these
two types of test methods are not obvious.
778
REFERENCES
Chen Shuanfa, Zheng Mu-lian, Yang Bin. 2005. Thermal stress influence factors of asphalt overlay
on cement concrete pavement cracking slab. Journal of Traffic and Transportation Engineering
5(3):25–30.
Donald K C. 1990. Designing asphaltic concrete mixes which are more readily compacted. Asphalt Pav-
ing Technologists 59(1):341–359.
Leonard M. 2005. Experiences with Reflective Crack Relief Projects. Transportation Research Record
Li Zuzhong, Chen Shuanfa, Zhang Dengliang. 2007. Fatigue Test of Material Tension and Tension
Compression in Stress Absorbing Layer. Highway 10:190–194.
Liao Wei-dong, Chen Shuan-fa,Liu Yun-quan. 2003. Research on anti-fatigue performance of STRATA
stress absorbing layers. Journal of Wuhan University of Technology 25(12):1–4.
Liao Weidong, Liu honghai, Zhang Changbo. 2005. Gradation Feature and Construction Control
Technology of STRATA Stress Absorbing Layer.Highway5:13–17.
Liao Weidong, Chen Shuan-fa, Liu Shao-yu. 2009. Experimental Research on Service of Stress Absorp-
tion Interlayer. Journal of Highway and Transportation Research and Development 26(3):11–16.
Phil B. 2004. Interlayer and Design Considerations to Retard Reflective Cracking. Transportation
Research Record No.1896
Yang Bin, Chen Shuan-fa,Liao Wei-dong,et al. 2005. STRATA stress-absorbing layer’s impact on load
stress and thermal stress in overlay. Journal of Highway and Transportation Research and Develop-
ment 22(9):27–30.
Zhang Zhong-qi, Wu Jianglong, Di Jin. 2003. Elementary exploration on the construction technology
of STRATA stress absorbing layers. Journal of Wuhan University of Technology 25(12):38–40.
Zhang Yong-qing, Jia Shuangying. 2005. Evaluation method for asphalt pavement performance of free-
way. Journal of Chang’an University (Natural Science Edition)25(2): 11–15.
Zhang Peng. 2005. Defence methods for reflective crack sin existed cement concrete black pavement.
Journal of Chang’an University (Natural Science Edition) 25(3):16–18.
779
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
D. Ren
Delft University of Technology, Delft, The Netherlands
Southwest Jiaotong University, Chengdu, China
L.J.M. Houben
Delft University of Technology, Delft, The Netherlands
Wilfred Nijssen
National Highway Agency, Maastricht, The Netherlands
Berwich Sluer
Boskalis Nederland, Nieuw-Vennep, The Netherlands
1 INTRODUCTION
Continuously reinforced concrete pavement (CRCP) with porous asphalt (PA) wearing
course have been built on several major roads in the Netherlands since 1990: A76, in the
very south-east of the country, 1990, A12, south of Utrecht, 1998; A5, near Amsterdam
Airport Schiphol, 1999–2003; A50, Eindhoven, 2004–2005; and A73, between Venlo and
Echt-Susteren, 2007. This type of AC/CRCP composite pavement provides low noise and
good friction, as well as reduced splash and spray. The expected lifetime for the PA wearing
course is 8 to 10 years but the overall PA/CRCP system is assumed to be a ‘perpetual’ pave-
ment, where only the surface layer needs to be replaced over time when needed. In general,
those AC/CRCP pavement sections are in good condition without severe distresses, rutting,
or ravelling in the PA wearing course after more than 8 years in service. However, there are
few minor to medium transverse reflective cracks in the PA wearing course in several sections,
and some wide reflective cracks require maintenance during its service period. There are
781
several reflective crack models available for the asphalt concrete layer over Portland cement
concrete (PCC) or CRCP slab (Jayawickrama & Lytton 1987; Owusu-Antwi et al. 1999; Lyt-
ton et al. 2010). Reflective cracking in composite AC overlay of PCC pavement is considered
as the result of the horizontal and vertical movements at the joints in the underlying PCC
pavement caused by temperature cycles and traffic loading. However, those models are lim-
ited to an asphalt overlay over an old concrete pavement, and none of them is suitable for
the Netherlands conditions, where PA is placed over the newly constructed CRCP slab as the
development of the crack pattern in CRCP is influenced by the thermal insulated effect of
the PA wearing course (Khazanovich et al. 2013). As most of the PA wearing courses in those
PA/CRCP motorways in the Netherlands are approaching their design life, it provides a good
opportunity to study the mechanism of reflective crack development in the PA/CRCP pave-
ment type. This paper presents the results of the crack pattern of the underlying CRCP slab
and its influences on the occurrence of reflective cracking in the PA wearing course though
field investigations on three PA/CRCP motorways, A5, A50 and A73 in the Netherlands. The
risk of punchout development and horizontal cracking in the CRCP slab, and the effect of
polymer modified PA mixtures to reduce reflective cracking are also discussed.
2 TEST SECTIONS
782
Table 1. Construction and investigation data of surveyed PA/CRCP sections in The Netherlands.
A5 A50 A73
783
2.3 Reflective crack in porous asphalt wearing course
Reflective cracking is defined as a crack that is initiated by a discontinuity such as a joint or
crack in the underlying PCC or CRCP layer, which propagates through the asphalt surfacing
due to vertical and/or horizontal movement of the crack caused by temperature variation and
load transfer. Contraction of the PCC or CRCP layer due to decreasing temperature results
in the opening of the joints or cracks that induce horizontal stresses in the PA overlay. Traffic
loadings over the discontinuities in the PCC or CRCP cause differential vertical deflection
that result in shearing and bending stresses in the overlay (Trevino et al. 2004). Figure 2
shows some reflective cracking in the PAC/CRCP composite pavement on A73.
Table 2. Crack spacing statistics of CRCP slab in PA/CRCP motorways in the Netherlands.
Age (years) 8 13 13 8 8
Reinforcement 0.67% 0.60% 0.60% 0.70% 0.70%
Number of cracks 52 120 68 404 803
Mean spacing (m) 1.92 2.50 4.41 2.63 1.88
SD spacing (m) 1.45 1.29 2.02 1.66 1.25
Max. spacing (m) 6.3 5.5 9.4 11.1 9.8
Min. spacing (m) 0.15 0.20 0.70 0.10 0.20
Number of reflective cracks 15 -- 39 27 15
Number of reflective cracks/100 m 15.0 -- 13.0 2.6 1.0
Percentage of reflective cracks 28.8% -- 57.4% 6.68% 1.87%
784
the longitudinal reinforcement is, the lower the mean crack spacing is in CRCP. However, an
extremely large mean crack spacing of 4.41 m is found for the truck lane of A5, which may
be attributed to the low percentage of longitudinal reinforcement. Besides, in contrast to
conventional CRCP roads, the PA wearing course on the CRCP slab could reduce the mag-
nitude of the thermal stress in CRCP, more especially in the first winter after construction.
The PA wearing course may reduce both the absolute temperature drop and the temperature
gradient along the CRCP slab, which then reduces the number of cracks occurring during
this critical period. The smaller number of transverse cracks, in other words the larger mean
crack spacing, for the PA/CRCP composite pavement structure likely is the result of the
asphalt layer insulating the CRC layer from extreme temperature variations. Accounting for
the pavement surface albedo effect, which is the effect of color on the degree of absorption
of solar radiation, one would expect that the surface of AC over PCC pavement would have
higher maximum and daytime temperatures than that of a single layer PCC pavement. This is
confirmed with field measured temperature data in the AC/PCC composite pavement (Khaz-
anovich et al. 2013). However, larger thermal gradients in the composite pavement system do
not necessarily create a larger thermal gradient through the concrete slab itself. The presence
of an AC overlay may create an insulating effect, wherein the gradient in the PCC slab in the
composite pavement is less severe than its exposed CRCP counterpart.
As shown in Table 2, another feature of the crack pattern in the CRCP slab for the three
PA/CRCP motorways in the Netherlands is the large maximum transverse crack spacing,
for instance, the maximum crack spacing as large as 11.1 m is found in the left carriageway
of A73. The crack pattern has shown that nearly all the reflective cracks in the PA wearing
course are located on long crack segments of CRCP, which is due to the larger crack move-
ment on those locations, as shown in Figure 3.
As compared to the very few transverse reflective cracks in the PA wearing courses on A73,
larger numbers of reflective cracks are observed in the two other PA/CRCP motorways, A5
and A50, in the Netherlands. On average there are 13 and 15 reflective cracks per 100 m long
section for the A5 truck lane and the A50 emergency lane, respectively, while they are only
as few as 1.0 to 2.6 cracks per 100 m long section in the right and left carriageway of A73,
respectively, after 8 years in service. Besides, only 1.87% and 6.68% of the total number of
transverse cracks in the underlying CRCP induced a reflective crack in the above PA wearing
course in the right and left carriageway of A73, respectively. However, in the surveyed sec-
tions on A5 57.4% of the transverse cracks in the underneath CRCP slab reflected through
the PA wearing course. In addition to the influences of the wider crack spacing, having larger
horizontal crack movements of the CRCP slab, the excellent performance with negligible
reflective cracking on A73 is believed to be attributed to the use of polymer-modified bitu-
men for both the PA layers. Several studies have demonstrated that the use of polymer modi-
fied asphalt mixtures and asphalt rubber hot mixtures are effective to reduce the occurrence
of reflective cracking (Sousa et al. 2003; Trevino et al. 2004; Rao et al. 2013).
Figure 3. Example of observed transverse cracks in CRCP slab and reflective cracks in PA wearing
course on A5.
785
3.2 Crack spacing distribution
The crack spacing histograms for the CRCP slab for both carriageways of A73, as well as the
results for three other PA/CRCP sections on A5, A50 and A12 (Braam 2003), are presented
in Figure 4. In addition, the results obtained by the authors of the crack spacing distributions
of a conventional CRCP section on E17 (near City of Ghent, Belgium) and on E313 (near
city of Herentals, Belgium, with an active crack control method) are also presented here (Ren
et al. 2013 and 2014).
The crack spacing range between 0.8 m and 3.0 m is generally considered as a favorable
crack pattern (AASHTO Design Guide 1993). A crack spacing larger than 3.0 m will lead to
larger crack movements that will significantly increase the risk of reflective cracks in the PA
wearing course, whereas a short crack segment has the high risk of punchout development.
As shown in Figure 4, an ideal crack pattern is observed for the right carriageway (HRR) of
A73, A12 and E313, as both the large crack spacings and the short crack spacings are less
than 15%, and more than 65% of the crack spacings falls into the ideal crack spacing range,
0.8 to 3.0 m. This favorable crack pattern would reduce both the risk of punchout and reflec-
tive cracks. There are as much as 66.2%, 17.7%, and 31.9% of the crack spacings exceeding
3.0 m for the truck lane of A5, A50 and the left carriageway (HRL) of A73, respectively.
The extremely large number of large slab segments in the truck lane of A5 appears to be the
primary cause of the reflective cracks in the PA wearing course. On the other hand, very low
Figure 4. Crack spacing histograms in CRCP slab in four PA/CRCP motorways in The Netherlands
and two CRCP motorways in Belgium.
786
percentages of short crack spacings are found in the surveyed sections in the Netherlands,
except the emergency lane of A50, and it would decrease the risk of punchout development
that is more severe than reflective cracks in the PA wearing course.
Figure 5 shows the comparison of the cumulative crack spacing distribution between the
PA/CRCP composite pavements on A73, A5 and A50 in the Netherlands and conventional
CRCPs (without PA) on E17 and E313 in Belgium. The high percentages of large crack seg-
ments on A5 and the left carriageway of A73 are clearly demonstrated.
It is generally believed that the greater the crack spacing, the larger the crack width. In
addition, the slab length at both sides of a crack in CRCP contributes to the crack width.
Figure 6 presents the percentage of the reflective cracks as a function of the sum of the slab
length at both sides of a crack in the CRCP slab. The result of A5 illustrates a tendency that
the larger crack spacings do increase the potential of the initiation of a reflective crack in the
PA wearing course. However, Figure 6 shows that the slab length of CRCP between cracks
is not the dominant effect in inducing reflective cracks in the PA wearing course on A73 and
Figure 5. Comparison of cumulative crack spacing distribution in the underlying CRCP slab.
Figure 6. The percentage of reflective cracks in the PA wearing course in A73 and A5 in relation to all
the observed cracks in CRCP slab as a function of the crack spacing categories.
787
A50. It may be interpreted that the concrete volume changes contributing to the crack width
movement of the underlying CRCP pavement due to temperature variations is limited to the
bond stress transfer length instead of the whole slab length (Braam 2003; Won 2009). The so-
called transfer length is defined as the distance from a crack where the undisturbed situation
is again reached, i.e. the bond stress is zero. With respect to common CRCP structures, this
stress transfer length is thought to vary from 0.3 m to 0.6 m (Braam 2003; Won 2009).
Crack width is considered one of the most important factors determining the performance
of CRCP. The basic principle of CRCP is that the longitudinal reinforcement keeps cracks
tight. Tight cracks will maintain a good load transfer and prevent water to get into the crack
that would cause reinforcement corrosion, especially in regions where de-icing salts are used.
According to EN 1992 and EN 206, a crack width of 0.4 mm for CRCP is allowed accord-
ing to most environment classes that the pavement is subjected to. The first impressions of
the observed reflective cracks are strongly correlated with the crack width of the underneath
CRCP slab. It is found that the reflective cracks are mainly initiated at the locations where a
wider crack is present in the CRCP slab, as shown in Figure 7(a). On the other hand, the fine
cracks in the CRCP slab are less prone to induce a reflective crack in the PA wearing course,
as shown in Figure 7(b).
Figure 8 shows two pictures illustrating the crack width measurements by Digital micro-
scope on the surface of the CRCP slab in the truck lane of A5. Table 3 summarizes the crack
widths for two groups, with and without a reflective crack in the PA wearing course in A50
andA5. For instance, the crack width measurements were performed on 37 cracks in the
truck lane of A5, and 24 of them are with reflective cracks. It shows that the cracks with a
reflective crack in the PA wearing course have a higher mean crack width, 0.45 mm, as com-
pared to that of 0.38 mm for the cracks without a reflective crack. 16 out of the 24 cracks
Figure 7. Crack width on the surface of CRCP slab in A5, (a) a wide crack with reflective crack in PA
wearing course; (b) a fine crack in CRCP slab without reflective crack.
Figure 8. Crack width on the surface of CRCP slab in A5 by Digital microscope, (a) a wide crack in
CRCP slab with reflective crack in PA wearing course; (b) a fine crack in CRCP slab without reflective
crack in PA wearing course.
788
Table 3. Crack width statistics of CRCP slab surface in A50 and A5.
A50 A5
With Without With Without
reflective reflective reflective Reflective
Sections cracks cracks cracks cracks
Number of cracks 2 6 24 13
Mean crack width (mm) 0.45 0.37 0.45 0.38
Standard deviation of crack width (mm) 0.06 0.06 0.14 0.10
Maximum crack width (mm) 0.49 0.47 0.91 0.59
Minimum crack width (mm) 0.41 0.30 0.27 0.23
Figure 9. Crack width and the sum of slab length at both sides of a crack for A50, the crack widths
were measured by microscope at different depth on drilled cores.
(66.7%) with a reflective crack exceed the crack width limitation of 0.4 mm according to EN
1992 and EN 206. On the other hand, only 5 out of the 13 cracks (38.5%) without a reflective
crack exceed the allowable crack width of 0.4 mm. A similar tendency is also observed based
on the limited number of crack width measurements on A50.
It is generally believed that the greater the crack spacing, the larger the crack width. In
addition, the slab length of both sides of a crack in CRCP are contributing to the crack
width. The crack width and the sum of slab length at both sides of a crack for A50 is plot-
ted in Figure 9. It clearly shows a strong relationship between the reflective cracks in the PA
wearing course and the crack pattern in the underlying CRCP. Nearly all the reflective cracks
in PA occurred when the crack width of the CRCP layer exceeds 0.25 mm. In another way,
the reflective cracks in PA also are located at the larger sum of slab length at both sides of a
crack in CRCP. The overall result indicates the significant influence of the crack pattern in
CRCP on the reflective crack in PAC in the PAC/CRCP composite pavement.
The crack width and the sum of slab length at both sides of a crack for A5 are plotted in Fig-
ure 10. However, no clear tendency can be observed, which may be attributed to the unreliability
and the divergence of the crack width measurements on the surface of a CRCP slab (Ren 2015).
Horizontal cracks at the location of reinforcement were observed in the CRCP layer for
nearly one third of the drilled cores in A50 and A5 (Ren & Houben 2013 and 2015a). It may
be related to the type of coarse aggregate and larger crack movement. No corrosion of lon-
gitudinal reinforcement in the emergency lane has been observed (Ren & Houben 2013 and
2015a). It indicates the good condition of CRCP in the PAC/CRCP composite pavement.
5 SUMMARY OF FINDINGS
Based on the results of the surveyed crack pattern in the PA/CRCP composite pavement of
three motorways in the Netherlands, the following findings are observed:
789
Figure 10. Crack width and the sum of slab length at both sides of a crack for A5, the crack widths
were measured by digital microscope on the CRCP slab surface.
After more than 8 years in service (some section even elder than 13 years, such as A5), the
surveyed PA/CRCP composite pavement sections (on A5, A50, and A73) in the Netherlands
are in quite good condition, without severe distresses, rutting, or ravelling in the PA wearing
course. There are only a few reflective cracks, of low severity to medium severity, in the PA
wearing course. One could consider filling the most severe reflective cracks but there is no
need for replacement of the PA wearing course in the near future.
The results of observed crack pattern show a tendency that the larger crack spacings in the
CRCP slab will increase the potential of the initiation of a reflective crack in the PA layer.
Besides, there is a weak relationship between the crack width and the crack spacing: the
smaller the crack spacing, the smaller the crack width. A similar tendency was also observed
for the influence of the crack width in the CRCP slab on the initiation of a reflective crack
in the PA wearing course. The larger the crack width on the CRCP slab surface, the higher
the risk of initiation of a reflective crack into the PA layer. Considering the above-mentioned
features of crack pattern of CRCP slab in A5, the stress absorbing membrane interlayer is
recommended to apply to prevent the development of the reflective cracks in the PA wearing
courses.
The transverse crack pattern of the underlying CRCP slab on A73 is characterized as a
rather ideal crack pattern: the short crack spacings (<0.4 m) are less than 15%, and more than
65% of the crack spacings falls into the ideal crack spacing range, 0.8 to 3.0 m. This favour-
able crack pattern has a low risk of punchout occurrence and a lower percentage of reflective
cracks in the PA wearing course.
As compared to two other PA/CRCP motorways, A5 and A50, the excellent performance
of A73 showing negligible reflective cracks in the PA wearing course is believed to be largely
attributed to the use of polymer-modified bitumen for both layers of the twin-layer PA wear-
ing course. The use of the polymer modified asphalt mixtures is effective to reduce the occur-
rence of reflective cracking.
Several horizontal cracks at the location of the reinforcement are observed in the CRCP
layer. It may be related to the applied type of coarse aggregate and the environmental loading
during the construction. No corrosion of longitudinal reinforcement in the emergency lane
has been observed. It indicates the good condition of CRCP in the PAC/CRCP composite
pavement.
ACKNOWLEDGEMENT
The assistance and financial support received from Rijkswaterstaat and Boskalis Nederland
are greatly appreciated.
790
REFERENCES
AASHTO Guide for Design of Pavement Structures 1993. American Association of State Highway
Transportation Officials, Washington, D.C., United States.
Braam, C.R. and E.M. Horeweg. 2003. Cracking behaviour of continuously reinforced concrete pave-
ments: A Comparison between Measurements and Theory (in Dutch). Report 25.5–03–20, Delft
University of Technology, the Netherlands.
CROW. 2005. Cement Concrete Pavement Manual-Basic Structures. Report Publication 220, CROW,
Ede, the Netherlands.
Houben, L. J. M. 2006. Structural Design of Pavements Part IV: Design of Concrete Pavements, Delft
University of Technology, Delft, the Netherlands.
Jayawickrama, P.W., and R.L. Lytton. 1987. Methodology for Predicting Asphalt Concrete Overlay
Life against Reflection Cracking. Proceedings 6th International Conference on Structural Design of
Asphalt Pavements. University of Michigan, United States.
Khazanovich, L., J.T. Balbo, L. Johanneck, R. Lederle, M. Marasteanu, P. Saxena, D. Tompkins, M.
Vancura, M. Watson, J. Harvey, N.J. Santero, and J. Signore. 2013. Design and Construction Guide-
lines for Thermally Insulated Concrete Pavements. Report MN/RC 2013–2, Minnesota Department
of Transportation, United States.
Lytton, R.L., R.L. Tsai, S-I, Lee, R. Luo, S. Hu, and F. Zhou. 2010. Models for Predicting Reflec-
tion Cracking of Hot-Mix Asphalt Overlays. NCHRP Report 669, National Cooperative Highway
Research Program, Transportation Research Board, Washington, D.C., United States.
Owusu-Antwi, E.B., L. Khazanovich, and L. Titus-Glover. 1999. Mechanistic-Based Method for Pre-
dicting Reflective Cracking in Asphalt Concrete-Overlaid Pavements. In Transportation Research
Record: Journal of the Transportation Research Board, No. 1629, TRB, National Research Council,
Washington, D.C., pp. 234–241.
Rao, S., M.I. Darter, D. Tompkins, M. Vancura, L. Khazanovich, J. Signore, E. Corleri, R. Wu, J. Har-
vey, and J. Vandenbossche. 2013. Composite Pavement Systems, Volume I: HMA/PCC Composite
Pavements, Report S2-R21-RR2, SHAP2 Renewal Research, Transportation Research Board, Wash-
ington, D.C., United States.
Ren, D., L.J.M. Houben, and L. Rens. 2013. Cracking Behavior of Continuously Reinforced Concrete
Pavements in Belgium: Characterization of Current Design Concept. In Transportation Research
Record: Journal of the Transportation Research Board, No. 2367, TRB, National Research Council,
Washington, D.C., pp. 97–106.
Ren, D. and L.J.M. Houben. 2013. Crack pattern of CRCP and reflective cracking in PA/CRCP com-
posite pavement: field survey of A50, Report, Delft University of Technology, the Netherlands.
Ren, D., L.J.M. Houben, L. Rens, and A. Beeldens. 2014. Active Crack Control for Continuously Rein-
forced Concrete Pavements in Belgium through Partial Surface Notches. In Transportation Research
Record: Journal of the Transportation Research Board, No. 2456, TRB, National Research Council,
Washington, D.C., pp. 33–41.
Ren, D. and L.J.M. Houben. 2015a. Investigation of Crack Pattern of CRCP of Motorways A5, Report,
Delft University of Technology, the Netherlands.
Ren, D. and L.J.M. Houben. 2015b. Investigation of Crack Pattern of CRCP of Motorways A73,
Report, Delft University of Technology, the Netherlands.
Sousa, J.B., J.C., Pais, R. Saim, G.B., Way, and R.N., Stubstad. 2003. Mechanistic-Empirical Over-
lay Design Method for Reflective Cracking. In Transportation Research Record: Journal of the
Transportation Research Board, No. 1809, TRB, National Research Council, Washington, D.C.,
pp. 209–217.
Trevino, M., S-M, Kim, A. Smit, B.F. McCullough, and Y. Yildirim. 2004. Asphalt Concrete Over-
lays on Continuously Reinforced Concrete Pavements: Decision Criteria, Tack Coat Evaluation, and
Asphalt Concrete Evaluation. Report FHWA/TX-05/0–4398–3, Centre for Transportation Research,
University of Texas, Austin, United States.
van Breugel, K., van der Veen, C., Walraven, J. C., and Braam, C. R. 1998. Concrete Structures Sub-
jected to Temperature and Shrinkage Deformations-Theory and Practice (Betonconstructies Onder
Temperatuur-En Krimpvervormingen), [In Ducth], BetonPrima, Delft, the Netherlands.
Won, M. 2009. Evaluation of MEPDG with TxDOT Rigid Pavement Database. Report FHWA/TX-09
/0–5445–3, Centre for Transportation Research, University of Texas, Austin, United States.
791
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
C.F. Wang
School of Material Science and Engineering, Chang’an University, Xi’an, P.R. China
A.M. Sha
Key Laboratory of Ministry of Education for Special Area Highway Engineering, Chang’an University,
Xi’an, P.R. China
Z.Y. Sun
School of Information Engineering, Chang’an University, Xi’an, P.R. China
ABSTRACT: Proper filler bitumen ratio of the asphalt mixture has important impacts on
the performance of the constructed pavement. Both the asphalt and mineral filler have their
own appearance in color, thus, the asphalt mortars mixed with asphalt and mineral filler by
different filler bitumen ratios have their own color. The paper provides a new method of the
filler bitumen ratio measurement by using the pattern matching algorithm. The cross sec-
tion images of the asphalt pavement were acquired and the mortar parts were segmented by
using gray scale transformation. Then the measurement could be accomplished by searching
the types of the asphalt and filler and matching their normalized histogram patterns. The
histogram comparison method used were the histogram correlation combined with Bhatta-
charyya coefficient. The asphalt pavement with the AC-13 and AC-20 gradations were tested,
and the results showed that the pattern matching method is worth trying.
1 INTRODUCTION
Mineral filler is a kind of limestone powder with the diameter less than 0.075 mm, proper
mineral filler bitumen ratios of the asphalt mixture has important impacts on the perform-
ance of the constructed pavement. The skeletons formed by coarse aggregates are filled by
the mineral filler to make the asphalt mixture have good high temperature performance and
strength. The cohesive of the asphalt mortar could be enhanced by the mineral filler with
proper content (Liu, L. et al. 2007, Alvarez, A.E. et al. 2012), the aging performance of the
asphalt mixture is also affected by the mineral filler and the durability, water stability (Wang,
B.G. et al. 2007, Kong, Z.Z. et al. 2014) and the high and low temperature performance (Tan,
Y.Q. et al. 2010, Zhou, Z.X. 2013) of the asphalt mixture could also be improved by the min-
eral filler. During the construction of the asphalt pavement, the usage of the proper mineral
filler with proper content and fineness is critical to the quality and durability of the asphalt
pavement (Hu, H.M. et al. 2004, Remisova, E. 2015, Ji, L. et al. 2014). Thus, the measure-
ment of the mineral filler content of the asphalt mixture is of importance.
The measurement of mineral contents in asphalt pavement mainly involves the measure-
ment of mineral filler in the extraction liquid and the mineral filler in the asphalt mixture
after extraction. Conventional testing method of the mineral filler content include the Burn-
ing-Sieving method and centrifugal elutriation method (Su, Z.G. 2005). The Burning-Sieving
method is empirical and time-consuming and the centrifugal elutriation method needs addi-
tional centrifuge systems. Thus, this paper aims to provide a new method of the filler bitumen
ratio measurement by using the pattern matching method.
793
2 PRINCIPLE OF THE FILLER BITUMEN RATIO MEASUREMENT
794
Figure 3. Image of the asphalt pavement and its normalized histogram.
∑ (H )(H (4i) − Η )
L 1
i=0 A i)
i) B Β
r(A,B)= (1)
∑ ( ) ∑ (H (4i) − Η )
L 1 2 L 1 2
i=0
HA ( i)
i Α i=0 B Β
L −1
∑ i=0 ∑ i=0 HA ( ∑ i=0 ( i))
L 1 L 1
r(A,B)= 1 i − ΗΒ(4 i)
HA ( i) i i)
i 0 (2)
In equation (1) and (2), HA(4i) is the normalized histogram value of the image A and the
HB(4i) is the normalized histogram value of the image B. The 4i is used to sample the histo-
gram value and improve the efficiency of the comparison algorithm.
In the histogram correlation, the bigger the result is, the higher the matching degree. But in
the Bhattacharyya method, the smaller result shows a higher matching degree. The histogram
correlation method combined with Bhattacharyya method was used and the flow chart is in
Figure 4.
795
The principle of the filler bitumen ratio measurement of asphalt pavement based on histo-
gram matching is used with the assumption that the mineral filler is uniformly in the mortar.
Where T is the threshold value of the Otsu. The color of the aggregate particles is brighter
than the color of the mortar, thus, the mortar could be extracted by the thresholding value of
the voids and the Otsu thresholding value T by using the equation (3).
796
Figure 6. Original image and the image after segmentation of the asphalt pavement.
If the normalized histogram of mortar parts image in the asphalt pavement are obtained,
select related mortar image with the same bitumen and filler type, and compare their normal-
ized histogram by using the small application, the Histogram correlation value and the Bhat-
tacharyya coefficient could be calculated to decide if the filler bitumen ratio of the asphalt
pavement is the same as the mortar samples.
Sieve size /mm 16 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075 filler
Sieve size /mm 26.5 19 16 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075
Figure 10. Sample image 1, the mortar parts image and its normalized histogram.
Figure 11. Sample image 2, the mortar parts image and its normalized histogram.
798
Figure 12. Sample image 3, the mortar parts image and its normalized histogram.
Figure 13. Sample image 4, the mortar parts image and its normalized histogram.
Figure 14. Sample image of AC-20 asphalt pavement (left: No. 5, middle: No. 6, right: No. 7).
The bitumen mortar color palette will be enriched when the mortar mixed with new types
of bitumen and mineral filler.
5 CONCLUSIONS
Proper mineral filler bitumen ratio of the asphalt mixture has important impact on the per-
formance of the constructed pavement. Conventional testing method of the mineral filler
bitumen ratio is empirical and time-consuming. The paper provides a new method for filler
bitumen ratio measurement based on histogram pattern matching. For both of the bitumen
and filler have their own color, by means of machine vision, the color distribution charac-
teristics of asphalt mortar could be extracted as the histogram information in different color
799
space. The filler bitumen ratio of the constructed asphalt pavement can be decided by using
the histogram correlation combined with Bhattacharyya coefficient. The matching condition
is the correlation between the mortar palette and the mortar parts image of the asphalt pave-
ment is bigger than 0.9 and the Bhattacharyya coefficient is smaller than 0.3.
Although the color palette provided by the bitumen mortar will be enriched and through
the measurement most of the filler bitumen ratio could be decided if the type of the bitumen
and the mineral filler is the same, sometimes this method probability got completely wrong
results for some mineral fillers have almost the same color and the machine vision tool could
not distinguish.
For the histogram matching method for filler bitumen measurement is based on color dis-
tribution characteristics, the image acquisition should control the acquisition environment
by using uniform lighting to improve the accuracy and reduce measure error.
If the color palette is enriched to a certain, the filler bitumen ratio measurement of asphalt
pavement is more effective than Conventional testing methods. The color palette and the
histogram matching could be achieved by using the image database and the content-based
image retrieval techniques.
ACKNOWLEDGEMENTS
The work is supported by the Fundamental Research Funds for the Central Universities (No.
310831161014, No. 310831163201 and No. 310821161016) and the open research fund of
Key Laboratory of Highway Engineering of Sichuan Province, Southwest Jiaotong Univer-
sity (No. LHTE003201103).
REFERENCES
Alvarez, A.E. & Ovalles, E. & Caro, S. 2012. Assessment of the effect of mineral filler on asphalt-ag-
gregate interfaces based on thermodynamic properties. Construction and Building Materials, 28(1):
599–606.
Elbayoumi H. & Suheir M. & Isa, N.A.M. & Salamah, S. A. 2015. Improved image magnification algo-
rithm based on Otsu thresholding. Computers and Electrical Engineering, v46: 338–355.
Gary, B. & Adrian, K. Learning OpenCV: Computer Vision with the OpenCV Library. O’Reilly Media,
Inc., 2008.
Hu, H.M. & Sha, A.M. 2004. Effects of mineral filler fineness and content to the performance of
asphalt mixture. Anhui Architecture, (2):108–109.
Ji, L. & Li, Y.L. & Sun, W.G. & Yuan, P.Y. 2014. Controlling Dosage, Fineness and Density Index of
Mineral Filler in Asphalt Mixture. Journal of Highway and Transportation Research and Develop-
ment, 31(8): 25–29+52.
Kong, Z.Z. & Zhang, Y.M. & Zhang, A.Q. 2014. Influence of granite mineral filler on the water stability
of asphalt concrete. Journal of China & Foreign Highway, 34 (5): 287–290.
Liu, L. & Hao, P.W. & Xiao, Q.Y. & Wang, H.N. 2007. High temperature properties and evaluation
method of asphalt mortar. Journal of Chang’an University (Natural Science Edition), 27(5): 30–34.
Remisova, E. 2015. Study of mineral filler effect on asphalt mixtures properties. Bituminous Mixtures
and Pavements VI—Proceedings of the 6th International Conference on Bituminous Mixtures and
Pavements, ICONFBMP, p 49–54.
Su, Z.G. 2005. Test research for determining asphalt content of asphalt admixture with ignition method
and acentric method. Fujian construction science & technology. 1: 46, 58.
Sun, Z.Y. & Sha, A.M. & Yao, Q.L. & Zhang, H.L. 2005. Realization of threshold segmentation algo-
rithm in asphalt mixture. Journal of Chang’an University (Natural Science Edition), 25(6): 34–38.
Tan, Y.Q. & Li, Z.H. & Zhang, X.Y & Dong, Z.J. 2010. Research on high—and low-temperature prop-
erties of asphalt-mineral filler mastic. Journal of Materials in Civil Engineering, 22(8): 811–819.
Wang, B.G & Li, P. & Zhang, Z.Q. & Sun, H.W. 2007. Influence of mineral powder on aging properties
of asphalt mortar. Journal of Chang’an University (Natural Science Edition), 27(4): 6–9.
Zhao, D.H. 2015. Study of the mineral filler for permeable asphalt mixture. Journal of China & Foreign
Highway, 35(03): 285–289.
Zhou, Z.X. 2013, Research on relationship of filler and the high and low temperature performance of
asphalt mortar and performance prediction of asphalt mortar. Hunan University.
800
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Hongyan Ma
Fuzhou Univeristy, Fuzhou, China
Feng Zhang
Harbin Institute of Technology, Harbin, China
Ruxin Jing
Delft University of Technology, Delft, The Netherlands
Decheng Feng
Harbin Institute of Technology, Harbin, China
1 INTRODUCTION
Frost heave disease of Seasonal frozen soil has serious effect on infrastructure in cold region.
The development of frost heave is an extremely complex process. It is determined by the dis-
tribution of temperature, moisture and stress field, in the meantime by the coupling of these
factors. It is important to predict the deformation of frost heave efficiently and accurately,
which is the basis for anti-freezing design and evaluation of frost disease.
Currently, domestic and foreign scholars made a wealth of research results on the predic-
tion of frost heave. R. R. Gilpin developed a model of the frost heave phenomenon in soil.
This model predicts ice lensing and heave rates as a function of the basic properties (thermal
conductivities and particle size) and the externally applied boundary conditions (surface tem-
peratures and overburden pressure) (Gilpin, 1980). Konrad and Morgenstern revealed that
a freezing soil can be characterized by two parameters, the segregation-freezing temperature
and the overall permeability of frozen fringe. They presented a model that indicated how
the overall permeability of frozen fringe can be calculated without detail measurements at
the scale of the frozen fringe (Konrad and Morgenstern, 1980). Kevin O’Neill and Robert
D. Miller explored a numerical model which simulated frost heave in saturated, granular,
air-free, solute-free soil. This model is based on equations developed from fundamental ther-
momechanical considerations and previous laboratory investigations (O’Neill and Miller,
1985). Nixon J. F. extended and modified an approximate analytical technique of Gilpin and
accounted for the effect of distributed phase change within the freezing fringe in both the
head and mass transfer component of the formulation (Nixon, 1991). Bronfenbrener, Leonid
801
and Bronfenbrener, Regina presented a generalized model for secondary frost heave in freez-
ing fine-grained soils. This model assumes that the non-instantaneous crystallization process
takes place in the kinetic zone, and that the rate of crystallization is a function of supercool-
ing. The model presented predicts the frost heave and freezing processes in porous media
with reasonable accuracy and satisfactorily reflects observed phenomena (Bronfenbrener
and Bronfenbrener, 2010b, Bronfenbrener and Bronfenbrener, 2010a). Azmatch Tezera F.
et al. proposed a new fundamental approach to determine the ice lens initiation condition
using the Soil Freezing Characteristics Curve (SFCC). It is demonstrated that an ice lens
initiates close to the so-called ice-entry value defined using the SFCC. Ice lens initiation con-
ditions for different boundary conditions were determined in a laboratory using the SFCC
and were then compared with the ice lens initiation conditions from a one-dimensional open
system frost heave tests. The results using the SFCC showed good agreement with the values
determined experimentally (Azmatch et al., 2012). Groenevelt Pieter H. and Grant Cameron
D. presented a unifying theory for the process of heave in freezing soils. They pointed the
value of the heave index is completely dominated by the unsaturated hydraulic conductivity
function of both the unfrozen soil below the frozen fringe and the soil layer at sub-zero tem-
perature (Groenevelt and Grant, 2013).
In China, Wenshan Bing established a simplify predict formula of frost heave based on the
thermal conduction theory (Bing, 1984, Bing and zhou, 1995). Huimin Dai and Xionglong
Wang established a 5 grade classification and criteria for the subsoil frost heave of highway
bridge based on in-situ frost heave experiments on both clay and sand (Dai and Wang, 1989,
Dai and Wang, 1993). Tiexing Wang built a two dimensional moisture migration model for
frozen soil subgrade, which is proposed that the hydraulic head should be made up of gravi-
tational head, martin suction head, temperature head and phase change head (Wang, 2005,
Wang and Hu, 2001, Wang et al., 2000). Xuesong Mao established a heat-moisture-stress
field coupling model based on the non-stationary temperature field control equation, the
finite element control equation of the moisture movement and the two-dimensional numeri-
cal calculation model of deformation in the subgrade (Mao, 2004). Jining Yu proposed a
theoretical calculation method of frozen heave and phase change temperature is a supple-
mentary condition of formation for ice lens based on solid phase increment theory and ther-
moelasticity theory (Yu et al., 2006). Anping Zhao studied seasonal frozen soil in Chanchun
to research micro structure change in freeze-thaw circle, applied with grey relation analysis
method and rough set theory to estimate the soil micro structure effect on its frost heave, dis-
cuss correlations between micro structure parameter and frozen-heave factor (Zhao, 2008).
Zhiguo Chen and Zheren Wang modified formulas of calculation of earth frost depth and
highway frost depth. In addition, according to stress state of pavement subjected to frost
action in subgrade soil, the calculation formula and design method for pavement anti-frost
damage were brought forward (Zhiguo and Zheren, 2009). For the development of frost
heave during 1-D freezing process of saturated oil, Yang Zhou established a separate ice frost
heave model based on the ice lens growth of coupled moisture and heat transfer (Zhou, 2009,
Zhou et al., 2010). Jiankun Liu, Huya Tian and Tianliang Wang et al. studied the effects of
load type on frost heave of fine grained soil, and pointed out that frost heave are inhibited
both under static and dynamic load (Tian et al., 2011, Wang and Yue, 2013, Yan et al., 2013).
Hui Bing and Ping He investigated the variation of frost heave amount, redistribution con-
dition of water and salt and change law of dry density (Bing and Hui, 2011). Based on the
separated ice model, considering overburden pressure, deformation of soil porosity and criti-
cal separation pressure, Kun Hu formulated governing equation coupled heat and moisture
transfer and built up frost heave model which could explain the evolving rules of segregation
ice during soil freezing (Hu, 2011). Yuanming Lai, Shimin Zhang and Wenbing Yu designed
and investigated a new kind of embankment structure, which consists of geotextile, crushed-
rock layer and geomembrane, both in the laboratory and at the field of Budongquan located
at Qinghai–Tibet Plateau (Lai et al., 2012). Dongqing Li, Jiazuo Zhou and Kun Zhang built
a Thermal-Hydro-Mechanical Coupling mathematical model based on the relationship of
matric potential, pressure potential and temperature (Fang et al., 2013, Li et al., 2012). Dai-
chao Sheng, Sheng Zhang, Zhiwu Yu and Jiasheng Zhang presented a simple frost heave
802
model which assumed frost heave is caused by the formation of ice lenses in a freezing soil.
The formation of ice lenses is governed by the Claperon equation of thermodynamics and
relies on the existence of a frozen fringe between the frozen and unfrozen zones (Sheng et al.,
2013). Jiazuo Zhou et al. proposed the concept of separating void ratio as a judge criterion
for the formation of ice lenses, adjusted the hydraulic conductivity to describe the unfrozen
water gathering at the front of ice lenses and growth of ice lens, and established a mathemati-
cal model of coupled water, heat and stress (Zhou and Li, 2012, Zhou et al., 2014).
As the research of frost heave from the simple thermodynamic analysis to multi-phys-
ics coupling simulation, the predict results of frost heave is more and more close to real-
ity. Through a review of freezing process, a thermal-hydro-mechanical coupling model can
perfect describe the development process of frost heave. Although there is a few research
on multi-physics coupling simulation, the predict deformation of frost heave is not good
enough. Therefore, in order to further understand the distribution of the temperature field,
moisture field and stress field, and the development of frost heave deformation, first a multi-
physics coupling control equation was established, based on energy conservation law, mass
conservation law and Darcy’s law. Then based on VG model and the relationship between
saturation and hydraulic head, a numerical model which is combined with the apparent heat
capacity method was established. Finally the model was verified by compare the results from
lab test. This model was used to predict the development of the frost heave deformation and
evolution law of each parameter for seasonal frozen soil. And it can provide a theoretical
basis for anti-freezing design and evaluation of frost disease.
803
∂ ∂∂T
T ∂ ∂T ∂ ∂∂T
T ∂S ∂T
( λT ) + ( λT ) + ( λT ) + Lnρi i =Cv (1)
∂x ∂x ∂y ∂y ∂z ∂z ∂t ∂t
1 ∂e ∂Si e 1 ∂e ∂Si e
ρw [ (1 Si ) ] + ρi [ S ]
1 + e ∂t ∂ 1+ e 1 e ∂t ∂ 1+ e
k k L
− ρ w [ ∇( ∇ w ) + ∇( ∇ ) + ∇( w∇ )]=0
γw γ w T0
(2)
where ρw is density of water, e is void ratio, kw is permeability coefficient, γ w is bulk density
of water, uw is pore water pressure, T0 is freezing point and z is head position.
In above equation, permeability coefficient should be the function of temperature as below
(Gilpin, 1980):
kw
= [ − (T T0 )]β (3)
kw0
where kw0 is the permeability coefficient of soil at unfrozen stage, β is pore structure
factor.
Base on Claperyron function, the relationship between pore water pressure and pore ice
pressure can be written as follow:
uw ui T
− =L (4)
ρw ρi T0
∂[d( '
+ δ ij )]
+d
ij w
0 (5)
∂y j
i
d σ ij' = dε σ '
ij
where D is standard elastic matrix, ε σ ' is strain increments which caused by effective
ij
stress.
ε σ ' =ε e εT
ij
804
where, ε e is elastic strain, ε T is temperature strain.
n n0
ε e=
1− n
ε β (T T0 )
∂ n − n0 ∂
{D[ − β (T T0 )]} + δ iij uw dbi = 0
∂y j 1− n ∂z
(6)
⎧ θs θr
⎪θ + h<0
θ ( ) = ⎨ r [ (α h ) ]m (7)
⎪⎩θ s h≥0
where α is the reciprocal value of the intake pressure, n is the pore distribution in soil particle,
m is composition constant factor and m = 1–1/n,θ is the residual volumetric water content
of soil, θs is volumetric water content of saturation soil, and θ is the actual volumetric water
content of soil.
The relationship between saturation and head can be written as follow:
⎧ 1
⎪− h<0
Se ( h ) = ⎨ [ (α h )n ]m (8)
⎪⎩1 h≥0
805
t
Δh ∫ [SPP e
0
0
aP
Pe
× grandT ]dt (10)
where Δh is the deformation of frost heave, SP0 is the initial segregation potential which is
equal to the ratio of flow velocity and temperature gradient, Pe is the stress at freezing fronts
and a is the sensitive parameters of pressure.
Formula (10) is a time integral function. It is use flow velocity, temperature gradient and
stress distribution at each time to calculate the deformation of frost heave.
3 MODEL VALIDATION
806
3.2 Numerical model
A two dimension model which width is 10 cm and height is 5 cm was built. The thermal
parameters of silty clay are shown in Table 2.
According to the governing equation of frost heave, the input parameters (boundary con-
dition and initial value of each physical field) are shown in Table 3. The computing step is
30 mins and computing time is 50h.
Figure 3 is the temperature of 1# sample during test under one direction freezing. From
Figure 3, it can be found, a. the temperature of sample was decreased with increasing of
time; b, for the top part of sample, the temperature decreased faster; c. part of sample was
still at unfrozen stage from 0.5 to 7.5 hours; d. the sample was full frozen after 7.5 hours; e,
temperature of sample was –15°C after 42 hours which is the same temperature as top plate.
807
Figure 3. The temperature with test time of 1# sample.
No. 1# 3# 5#
Δh1 . n ⋅ Se h0 (11)
Where Δh1 is the deformation of frost heave, n is the soil porosity and h0 is the initial
height of sample.
In equation (11), the deformation of frost heave is caused by moisture phase change, and
in equation (10), the deformation of frost heave is caused by moisture migration.
From Table 4, it can be found test results are smaller than calculate results. This is because
the internal pores of unsaturated soil is not completely filled with water, some of the pores
are filled with air. In freezing process, for unsaturated soil, the increased volume of moisture
will fill the empty pores of sample and the total volume of sample is not changed. Therefore,
the frost heave deformation of unsaturated soil can be ignored in field construction.
808
From Figure 4, the test results increased before 10 hours and decreased after that. This
is because cooling system was not working well at 10 hours, then the sample is thawing at
that moment. Later, after cooling system returned well, the deformation of frost was not
increased any more.
From Figure 5, the frost heave deformation of all three sample is larger than 1 mm. For
the 6# sample, the saturation is 100% and the frost heave deformation is 1.0 mm which is
the smallest of these three sample. For the 4# sample, the saturation is 95.5% and the frost
heave deformation is 3.0 mm which is the largest of these three sample. For the 2# sample, the
saturation is 100% and the frost heave deformation is 1.2 mm. From 2# and 4# sample, which
saturation are both 100%, it can be found the deformation of frost heave has inverse relation-
ship with dry density. It is same as the conclusions from Lizhou Wu et al (Wu et al., 2011).
From frost heave curve, it can be found there are three stages in freezing process, they are
frost shrink, frost heave and stable stage. For 2#, the frost shrink stage is from 0 to 3 hours,
the frost heave stage is from 3 to 12 hours and the stable stage is after 12 hours.
Compare Figure 4 with Figure 5, it can be found, the frost heave deformation of saturated
soil has significantly larger than that of unsaturated soil. It is because there is no potential
809
suction in unsaturated soil that unable to provide a driving force for the water supply. The
other reason is no water filled pore has negative effect of volumetric change in freezing proc-
ess. Therefore, the moisture content of soil has a direct effect on the deformation of frost
heave. In other words, the more moisture content in soil, the more deformation of frost
heave.
Figure 6 shows the test and predict results of frost heave of soil.
Figure 4 to Figure 6 show, for maximum deformation of frost heave, the predict results
is almost same as the test results. However, for the growth stage of frost heave deformation,
the predict results is not accurate enough. This is because, a. the increase rate of frost heave
deformation is quite high at growth stage; b. the growth stage of frost heave is quite short.
Therefore, the predict model which was built in this paper is just used for the prediction of
maximum deformation of frost heave and it need to be developed to obtain more accurate
results in further research.
4 CONCLUSION
1. Base on Van Genuchten model and apparent heat capacity method, a multi-physics
coupling numerical model of seasonal frozen soil was established under thermal-hydro-
mechanical.
2. The multi-physics coupling numerical model was validated by the frost heave test of silty
clay. It shows this model can accurate predict the maximum deformation of heave frost.
3. The relationship between dry density, temperature, saturation and frost heave was found
based on the frost heave test of silty clay.
4. The predict model which was built in this paper is not accurate to predict the frost heave
deformation at growth stage and it need to be developed in further research.
REFERENCES
Azmatch, T.F., Sego, D.C., Arenson, L.U. & Biggar, K.W. 2012. New ice lens initiation condition for
frost heave in fine-grained soils. Cold Regions Science and Technology, 82, 8–13.
Bing, H. & Hui, P. 2011. Experimental study of water and salt redistributions of saline soil with different
freezing modes. Rock and Soil Mechanics, 32, 2307–2312.
Bing, W. 1984. Design theory of highway pavement anti-frost layer. Journal of Glaciology and Geocryol-
ogy, 6, 69–76.
Bing, W. & Zhou, J. 1995. Study of calculation methods for frozen depth of road. Journal of Harbin
University of Architecture and Engineering, 28, 11–16.
Bronfenbrener, L. & Bronfenbrener, R. 2010a. Frost heave and phase front instability in freezing soils.
Cold Regions Science and Technology, 64, 19–38.
Bronfenbrener, L. & Bronfenbrener, R. 2010b. Modeling frost heave in freezing soils. Cold Regions Sci-
ence and Technology, 61, 43–64.
Dai, H. & Wang, X. 1989. Frost heave susceptbility of highway bridge foundation soil in seaonal frost
region. China Journal of Highway and Transport, 4, 18–25.
Dai, H. & Wang, X. 1993. Frost heave susceptbility of highway bridge foundation soil in seaonal frost
region. Journal of Glaciology and Geocryology, 15, 377–382.
Fang, J., Li, D., Zhou, J. & Zhang, K. 2013. Technical study on anti-frost heaving in seasonally frozen
soil subgrade. China Journal of Highway and Transport, 26, 9–14.
Gilpin, R.R. 1980. A model for the prediction of ice lensing and frost heave in soils. Water Resources
Research, 16, 918–930.
Groenevelt, P.H. & Grant, C.D. 2013. Heave and Heaving Pressure in Freezing Soils: A Unifying The-
ory. Vadose Zone Journal, 12.
Hu, K. 2011. Development of separated ice model coupled heat and moisture transfer in freezing soils.
Xuzhou: China university of mining & technology.
Konrad, J.M. & MorgensterN, N.R. 1980. A mechanistic theory of ice lens formation in fine-grained
soils. Canadian Geotechnical Journal, 17, 473–486.
810
Konrad, J.M. & Morgenstern, N.R. 1981. The segregation potential of a freezing soil. Canadian Geo-
technical Journal, 18, 482–491.
Konrad, J.M. & Morgenstern, N.R. 1984. Frost heave prediction of chilled pipelines buried in unfrozen
soils. Canadian Geotechnical Journal, 21, 100–115.
Lai, Y., Zhang, S. & YU, W. 2012. A new structure to control frost boiling and frost heave of embank-
ments in cold regions. Cold Regions Science and Technology, 79–80, 53–66.
Li, D., Zhou, J., Zhang, K. & Chang, F. 2012. Modelling and numerical analysis of moisture heat and
stress in seasonal frozen soil. China Journal of Highway and Transport, 25, 1–7.
Mao, X. 2004. Study on coupling model of the moisture-heat-stress fields in the permafrost subgrade.
xi`an: Chang’an University.
Nixon, J.F. 1991. Discrete ice lens theory for frost heave in soils. Canadian Geotechnical Journal, 28,
843–859.
O’Neill, K. & Miller, R.D. 1985. Exploration of a Rigid Ice Model of Frost Heave. Water Resources
Research, 21, 281–296.
Sheng, D., Zhang, S., YU, Z. & Zhang, J. 2013. Assessing frost susceptibility of soils using PCHeave.
Cold Regions Science and Technology, 95, 27–38.
Thomas, H.R., Cleall, P., Li, Y.C., Harris, C. & Kern-Luetschg, M. 2009. Modelling of cryogenic proc-
esses in permafrost and seasonally frozen soils. Geotechnique, 59, 173–184.
Tian, Y., Liu, J. & Peng, L. 2011. Experimental study on frost action of fine-grained soils under dynamic
and static loads. Chinese Journal of Geotechnical Engineering, 32, 1882–1887.
Van Genuchten, M.T. 1980. A closed-form equation for predicting the hydraulic conductivity of unsatu-
rated soils. Soil science society of America journal, 44, 892–898.
Wang, T. 2005. Analysis of frost heave on subgrade in permafrost regions. China Journal of Highway
and Transport, 18, 1–5.
Wang, T. & Hu, C. 2001. A numerical model of moisture migration for frozen soil subgrade. China
Journal of Highway and Transport,, 14, 5–8.
Wang, T., Hu, C., Wang, B. & Hou, J. 2000. A finite element method for thermal field analysis of frozen
soil subgrade on the consideration of all field-factors. China Journal of Highway and Transport, 13,
8–11.
Wang, T. & Yue, Z. 2013. Influence of fines content on frost heaving properties of coarse grained soi.
Rock and Soil Mechanics, 34, 359–364.
Wu, Z., Xu, Q. & Huang, R. 2011. Analysis of freezing-thawing test process of unsaturated clay Rock
and Soil Mechanics, 32, 1025–1028.
Yan, H., Wang, T. & Liu, J. 2013. Experimental study of repeated frost heave and thaw settlement prop-
erties of silty sand. Rock and Soil Mechanics, 34.
Yu, J., Tan, F. & Fu, W. 2006. Analysis of frozen heave deformation in process of frozen soil. Rock and
Soil Mechanics, 1, 203–206.
Zhao, A. 2008. A study on the mechanism of microstructure of frost heaving in subgrade soil in sea-
sonal frost zone. Changchun: Jilin University.
Zhiguo, C. & Zheren, W. 2009. Anti-frost design moethod for highway pavement in seasonal frost region.
China Journal of Highway and Transport, 22, 34–40.
Zhou, J. & Li, D. 2012. Numerical analysis of coupled water, heat and stress in saturated freezing soil.
Cold Regions Science and Technology, 72, 43–49.
Zhou, J., Wei, C., Wei, H. & Tan, L. 2014. Experimental and theoretical characterization of frost heave
and ice lenses. Cold Regions Science and Technology, 104–105, 76–87.
Zhou, Y. 2009. Study on frost heave model and frost heave control of frozen soils. Xuzhou: China uni-
versity of mining & technology.
Zhou, Y., Zhou, G. & Wang, Y. 2010. Separate ice frost heave model for coupled moisture and heat
transfer in saturated soils. Chinese Journal of Geotechnical Engineering, 32, 1746–1751.
811
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Peiwen Hao
Key Laboratory of Highway Engineering in Special Region of Ministry of Education,
Chang’ an University, Xi’an, China
Mundia Muya
School of Engineering, University of Zambia (UNZA), Lusaka, Zambia
Haiwei Zhang
Highway College, Chang’an University, Xi’an, Shaanxi Province, Xi’an, P.R. China
ABSTRACT: There is a growing interest of many African countries to revamp their neglected
and unsustainable railways to attract a transportation modal shift from the roads. In order to
promote regional trade and transportation integration with border concessions, investors are
faced with problems of railway track gauge conversions to match the needs of the rolling stock
technology of the 21st century at the same time maintain business operations with the old roll-
ing stock on the same track structure. The objective of the work documented in this paper was
to numerically evaluate the impact of track gauge conversions on traffic load induced Permanent
Deformation (PD) of low embankment on soft sub-grade. A method to predict the traffic load
induced settlement of low embankment on soft sub-grade is proposed. Using the user-defined
material subroutines (UMAT) in ABAQUS, a 2-D Finite Element (FE) model was formulated.
These models are converted into a numerical formulation for implementation in FE analysis and
the traffic load induced dynamic stress in the sub grade are calculated by using the multi-layer
elastic theory. Then the plastic vertical strain in the sub-grade is calculated by an empirical equa-
tion, whose constants are related to the physical and mechanical properties of the sub-grade soil.
The method was applied to analyze a 700 m long section of a low embankment on the soft black
cotton soil of Nakuru plains in Kenya using single and dual track gauge respectively. The corre-
sponding results showed that the application of traffic loads on alternate rail tracks due to gauge
conversions have a significant effect on the permanent deformation of the sub grade soil. The
depth significantly influenced by traffic loading was found to be close to 6 m below the base of
the embankment. The analysis also shows that increasing the thickness and stiffness of the sub
grade is a very effective way of reducing the traffic load induced permanent deformation of soft
sub grade soil. The proposed method can be used for settlement analysis on low embankments
as well as a useful tool for making decisions on railway track gauge conversions.
1 INTRODUCTION
For low railroad embankments (less than 3.0 m high) constructed on soft subsoil, the per-
manent deformation of subsoil due to traffic load controls the design life as well as the
maintenance costs of the track structure. To select a cost effective design, it is desirable to
predict the traffic load induced time dependent settlement. Several factors affect the traf-
fic loading induced deformation, such as, the strength and deformation characteristics of
813
the subsoil, properties of the pavement and sub-grade of railroad, and the magnitude and
number of traffic load repetitions. Any useful settlement prediction method should consider
these factors directly or indirectly. The existing methods can be divided into 3 groups as
numerical, equivalent static loading and empirical methods. Theoretically, explicit simulation
of the response of subsoil under repeated load is preferable, although it is difficult to as the
response of subsoil under traffic load is a three-dimensional (3D) problem and the number of
load applications is extremely large. Furthermore, a considerable number of soil parameters
are needed in the modeling. With a time consuming calculation, the accuracy of the results
may not be guaranteed. Equivalent static load methods involve a static load to represent the
traffic load, and the one-dimensional consolidation theory is used to predict the settlement.
Although this method is simple, both the 3D traffic loading transfer mechanism and the
mechanism of repeated load induced settlement are not taken into consideration. A number
of empirical equations have been proposed to predict the permanent deformation of cohe-
sive soil under repeated loading whose constants in the empirical equation are determined
experimentally and the strength of the subsoil is directly included in the equation with its
physical state considered indirectly. The values of the constants are normally applicable to
remolded (compacted) cohesive soil but not suitable for natural clay deposit and the effect
of initial static deviator stress is not considered. A method to predict the traffic load induced
settlement of low embankment on soft sub-grade is proposed. Application of the proposed
method to a test section with 4 subsections at the Nakuru plain of the northern railway cor-
ridor of Kenya is described. Discussions are presented on the applicability of the proposed
method to design low embankments supporting a dual gauge track structure.
It is commonly accepted that when the dynamic deviator stress exceeds a critical level
(dynamic strength), the plastic strain increases rapidly with repeated loading. Also, under
repeated load, the response of the normally consolidated soil is different from that of over
consolidated soil. Assuming that the shear stress acting in the soil is below dynamic strength
of soil (not a failure problem), and the subsoil is in a normal to slightly over consolidated
state. Then, considering the effect of initial static deviator stress, a new empirical equation for
calculating the cumulative plastic strain of soft subsoil under traffic loading is proposed as:
m n
⎛q ⎞ ⎛ q ⎞
ε p = a ⎜ d ⎟ ⎜1 + is ⎟ N b (1)
⎝ qf ⎠ ⎝ qf ⎠
Where ε p = Cumulative plastic strain (%), N = the number of repeated load applications.
qis = Initial static deviator stress, and a, b, m, and n are constants related to the plasticity
and compression indices of the subsoil.
qd = Traffic load induced dynamic deviator stress
qf = Static failure deviator stress of soil
Dynamic deviator stress qd . To estimate the value of qd, the 3D traffic load transfer mecha-
nism and the characteristics of multi-layer foundation using the Burmister's multi-layer elas-
tic solution (Burmister, 1945) is considered suitable for this purpose.
Constants b and m. The parameter b controls the increment rate of plastic strain with the
number of repeated load applications. The constant b is not sensitive to the magnitude of
dynamic deviator stress it is mainly influenced by soil type2. The parameter m influences both
the magnitude and distribution of plastic strain with depth.
Constant a. The parameter a controls the magnitude of plastic strain with the traffic load
induced deformation comprising of two parts, dynamic consolidation and shear deforma-
tion. The amount of dynamic consolidation deformation is directly related to the compres-
sion index (Cc) of the soil as well as affecting the magnitude of shear deformation. Therefore,
it is rational to relate a with the compression index (Cc) of subsoil.
814
Table 1. Adopted constants for the empirical equation.
a σ Cc (2)
For the case under study, the value of Cc is about 0.8 to 1.0, with α value of 8.0.
Constant n. The constant n controls the effect of initial static deviator stress on traffic
load induced plastic deformation determined from laboratory triaxial tests. The adopted
constants for specimen sections are shown in Table 1.
The average embankment height on the existing meter gauge railway line is 0.75 m whose test
section was divided into 4 subsections and several subgrade treatment methods were examined.
The methods include (a) surface lime treatment, (b) surface cement treatment, (c) light-weighted
fill material, (d) geogrid reinforcement, and (e) cement column improvement of subsoil. A typi-
cal subsoil profile at the site is shown in Fig. 1 together with physical properties. In the figure,
Ac means alluvial clay, as means alluvial sand, and Hc means High non-marine clay deposit.
The soft deposit is about 20 m thick with 3 clay layers sandwiching 2 thin sand layers. About
1.0 m thick surface crust is in an over consolidated state with an Over Consolidation Ratio
815
Figure 2. Typical soil profile at test sections.
(OCR) of 2.0 to 4.0 (mainly due to weathering and aging). The deposit below 1.0 m depth is in
a normal to slightly over consolidated state. The compression index of the clay layers is about
0.8 to 1.0 for upper layer (Ac1) and 1.0 to 1.5 for lower soft layers (Ac2 and Ac3).
The settlements of the railroad on the exisiting meter gauge railway were monitored using
strain gauages at selected locations. Since the upper subsoil was in a lightly over consolidated
state, field measurement together with finite element analysis confirmed that the settlement
due to 0.75 m embankment loading (about 70 mm) was almost finished before monitored set-
tlements because the rail line has been operational for more than 30 years. In this regard, it is
considered that the measured settlement was caused by rail traffic load only. The total traffic
816
intensity (up and down direction) was 2 train pairs/day (10–22 freight wagons 45 kN each
depending on availability of goods to transport) with tractive force of 2700 tons, 18 kN axle
loada. The adopted Young’s modulus (E) and thickness (H) of each layer are summarized in
Table 2. The modulus for cement treated layers was estimated as 100 times the unconfined
compression strength (qu). The modulus for natural subsoil was assumed to be 200 times
the undrained shear strength (Su). The calculated dynamic deviator stress was 3 to 5 kPa
just below the subgrade, and reduced to about 1.0 kPa at 6 m depth. Table 3 gives calculated
settlement results for a CM subsection (see Fig. 2) as an example. The static undrained shear
strength (Su) of the subsoil is calculated by the following empirical equation3.
sσ v′0 (OCR
C )
m
Su (3)
Where σ’v0 = effective vertical stress. Values of S = 0.35 and OCR = 1 (m = 0.8) were used.
3.1 Variation of traffic load induced settlement with time on dual gauge set up
A numerical model is formulated using the user defined material properties (subroutines) in
Abaqus 2-D model. The traffic load is modelled over a time step evaluated as elapsed time
after dual guage track is open to traffic as shown in Figure 4 and 5. The dual gauge has
three rail tracks with one outer track supporting both meter and standard gauge (1435 mm)
dynamic loads. These models are converted into a numerical formulation for implementation
in FE analysis and the traffic load induced dynamic stress in the sub grade are calculated by
using the multi-layer elastic theory. Then the plastic vertical strain in the sub-grade is calcu-
lated by an empirical equation developed in this study (eq.1), whose constants are related to
the physical and mechanical properties of the sub-grade soil such as plasticity index and com-
pression index.The FE and Empirical equation calculated traffic load induced settlements
are shown in Figs. 3 and 4 for each subgrade soil treatment. The calculation predicted the
settlements well, which indicates that the proposed method is a useful tool for designing the
low road embankment on soft subsoil. In Figure 3, CM and CS subsections are compared,
which had different depth of improvement. For CS subsection, soil-cement columns (36%
replacement ratio by area) improved the upper subsoil. The dynamic deviator stress below
the columns was small and the traffic load induced settlement was much smaller than CM
subsection. In Figure 4, ECM and CMG subsections are compared. For these two subsec-
tions, the strength and stiffness of the cement treated layer was different, and CMG subsec-
tion had a lower strength and stiffness. The settlement of CMG subsection was more than
twice of ECM subsection. This case history indicates that an increase in the thickness and
stiffness of subgrade is efficient for reducing the traffic load induced permanent settlement.
817
Figure 4. Settlement of ECM and CMG subsections.
about 6 m below the base of the embankment. The plastic strain reduced very quickly within
the upper 2 m. It also indicates that the weaker the subgrade (CMG subsection), the larger
the plastic strain in upper subsoil. This observed results is indicative of the impact of dual
gauge track on settlement pattern of the sub grade soil.
4 CONCLUSIONS
The Government of Kenya plans to upgrade the existing meter gauge rail track to dual gauge
standard (1435 mm) on the same formation. The Impact of Dual Gauge Railway Tracks on Traf-
fic Load Induced Permanent Deformation of Low Embankments is demonstrated by proposing
an empirical method for predicting the traffic load induced permanent deformation of low road
embankment on soft subsoil. The method considers the 3D traffic load transfer mechanism, the
magnitude and the number of traffic load applications, multi-layer ground conditions, and the
strength and compression characteristics of soft subsoil. A numerical model is then formulated
using the user defined material properties (subroutines) in Abaqus 2-D model. These models are
converted into a numerical formulation for implementation in FE analysis and the traffic load
induced dynamic stress in the sub grade are calculated by using the multi-layer elastic theory.
Then the plastic vertical strain in the sub-grade is calculated by an empirical equation developed
in this study, whose constants are related to the physical and mechanical properties of the sub-
grade soil of the test section such as plasticity index and compression index. The method was
applied to analyze a 700 m long test section at the Nakuru wetlands of Kenya’s northern cor-
ridor meter gauge railroad. The impact of having dual gauge rail tracks is demonstrated by com-
paring the settlement effects on sub-grade with four different soil treatment methods. From the
results it shows that the proposed method can be used for calibrating the design of low railroad
embankment on soft subsoil. The analysis also revealed the following:
818
a. For the case investigated, the significant influence depth of traffic load is about 6 m below
the base of the embankment as compared to 2.5 m for single track gauges.
b. Increasing the thickness and stiffness of subgrade is an effective way of reducing the traf-
fic load induced permanent deformation on soft subsoil under dual gauge rail tracks.
REFERENCES
Burmister, D.M. (1945). “The general theory of stresses and displacements in layered system I”. Journal
of Applied Physics, Vol. 16, pp. 89–94.
Carr, Gary A. Dynamic Response of Railroad Track Induced by High Speed Trains and Vertical Stiff-
ness Transitions with Proposed Method of Measurement, Masters Thesis, Department of Mechani-
cal Engineering, Tufts University, September 1999. pp. 1–17
Civil Engineering Department, Saga Prefecture (1996). “A study on predicting the differential settlement
and rational design method for road on soft subsoil”. Research Report, Saga Prefecture, Japan, p 77
D. Bowness, A.C. Lock, W. Powrie, J.A. Priest and D.J.Richards, Monitoring the dynamic displace-
ments of railway track, Proc. IMechE, J. Rail and Rapid Transit 221 (2007) 13–22
D. Li, T.R. Sussman and E.T. Selig, Procedure for railway track granular layer thicMkness
determination,Report No. R-898, Association of American Railroads,Transportation Technology
Center, Pueblo, Colorado,USA, 1996, Transportation Research Board, Washington, D.C., pp. 1–17.
Ladd, C.C. (1991). “Stability evaluation during stage construction”. Journal of Geotechnical Engineer-
ing, ASCE, Vol.117, No. 4, pp. 541–615.
Li, D. And Selig, E.T. (1996). “Cumulative plastic deformation for fine-grained subgrade soils”. Journal
of Geotechnical Engineering, ASCE, Vol. 122, No. 12, pp. 1006–1013.
Samang, L. (1997). Settlement of soft cohesive deposit induced by cyclic loading. Dr. of Engineering
Dissertation, Saga University, p. 240.
Thompson, Randy and Li, Dingqing. Automated Vertical Track Strength Testing Using TTCI’s Track
Loading Vehicle. Technology Digest. February 2002
Zarembski, Allan M. and Choros, John. On the Measurement and Calculation of Vertical Track Modu-
lus. Proceedings American Railway Engineering Association. 1980, Vol. 81, pp. 156–173.
819
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The influence of principle stress distribution on semi-rigid base asphalt pave-
ment damage are analyzed. Three-dimensional finite element model under double-round uni-
form load was established for multi-layer asphalt pavement with semi rigid base and stress
components of each node were analyzed based on Drucker-Prager (D-P) criterion. A new
index, Asphalt Pavement Potential Damage Index (APPDI3D), was proposed. The results
show that: The different components of the principle stress are corresponding to different
pavement early failure under different temperature. Combined with the Mohr circle and the
location in the pavement structure, APPDI3D can accurately predict the damage form of
asphalt pavement, which provides a theoretical basis for the interpretation of pavement fail-
ure mechanism and optimization of pavement structure and material design.
1 INTRODUCTION
821
potential damage index), was proposed and used to explain the pavement distress of Ji-Qing
expressway.
2 OBJECTIVES
The study aims to interpret the rutting and Top-Down cracking damage by the tension-
compression composite shear theory under the normal and high temperatures, and achieve
the integrated design of structure and material.
3 THEORETICAL BACKGROUND
rπ ≤ rσ (1)
I1 = 1 + σ 2 + σ3 (2)
J2 =
1
6
( 1 − 2
2
+ 2 − 3
2
+ 3 − 1
2
) (3)
where I1 is the stress tensor of first invariant; J2 is the stress deviator of second invariant; σ1
is the first principal stress; σ2 is the second principal stress; σ3 is the third principal stress. The
compression stress is positive and the tensile stress is negative.
The vector radius of an arbitrary point on π plane inner the yield curve is expressed as:
rπ 2J2 (4)
And the vector radius of an arbitrary point on π plane on the yield curve is shown as:
rσ 2 J2′ (5)
where
J2′ − ⋅ I1 = k (6)
1 sin ϕ 3c cos ϕ
α= k= (7)
3 3 + sin ϕ 2 3 + sin2 ϕ
where c is the value of cohesion and ϕ is the value of internal friction angle.
In the yield surface figure of Drucker–Prager (D–P) criterion (Figure 1).the π plane posi-
tion and the yield curve size of each point are different because the I1 and J2 of an arbitrary
node are different in the pavement structure (Zheng.2005).
Asphalt Pavement Potential Damage Index (APPDI3D) was defined to represent the fatigue
state of each point. The formula of the calculation of APPDI is defined as:
rπ 2J2
APPDI3D = = (8)
rσ 2 J2′
822
Figure 1. Yield curve and yield surface.
While:
1) When APPDI3D < 1, it is indicated that the stress vector radius of this point is less than
that of the yield curve of the point, the Mohr circle and the envelope are not intersect, that
will generate elastic deformation, and which restores after unloading.
2) When APPDI3D > 1, that will produce a plastic deformation under the external loading.
And the Mohr circle will intersect with the envelope. The plastic deformation would not
restore after unloading. The early damage would be caused by plastic deformation cumu-
lative under cyclic loading or fatigue loading.
823
Table 1. Standard axle calculation parameters.
Parameters BZZ-100
Ei/MP a c/MP a
Figure 2. Boundary conditions and structure Figure 3. 3-D Finite element model for
layers. pavement structure.
Figure 5. The distribution of maximum value and the composition of principal stresses under high
temperature.
Figure 5 shows that the regions where the value of APPDI3D is greater than 1 are mainly
distributed around loading circle area in the up-layer. The number of nodes which the value
of APPDI3D is greater than 1 is 30, and it can be divided into three categories. 1) The 1st state
is tension-compression composite shear stress state in which the absolute value of tensile
stress is greater than the compression stress and the third principle stress is less than zero (
|σ3|>|σ1| and σ3 < 0 ). The nodes’ number of this type is nine. The maximum value of APPDI3D
is 1.2624, in which the σ1 equals 0.0291 MPa and σ3 equals −0.0822 MPa.The tension com-
pression stress ratio is 1:0.354, and the ratio of tensile stress and tensile strength is greater
than 0.5. It is located in 1.4 cm of the tire loading on the surface and the failure mode is
deduced as Top-Down cracking. 2) The second state is tension-compression composite shear
825
stress state in which the absolute value of tensile stress is less than the compression stress and
the third principle stress is less than zero (|σ3| > |σ1| and σ3 < 0 ). This category has 12 nodes.
The maximum value of APPDI3D is 1.1922, in which the σ1 equals 0.1013 MPa and σ3 equals
−0.0730 MPa. The tension compression stress ratio is 1:0.509, and the ratio of tensile stress
and tensile strength is greater than 0.5. It is located in 0.7 cm of the tire loading on the surface
and the failure mode is deduced as rutting. 3) The third state is pure compression shear stress
state in which the third principle stress is greater than zero (0 < σ3 < σ1). There are 9 nodes
belong to this situation. The maximum value of APPDI3D is 1.0768, in which the σ1 equals
0.3407 MPa and σ3 equals 0.0095 MPa. It is located in 0.5 cm of the tire loading with depth
of 1 cm under the pavement surface and the failure mode is deduced as rutting.
Figure 6 shows the π plane projection of different stress states based on D-P criterion.
The contour map of APPDI3D of Y = 0 profile and Y = −r/4 (r = 0.1065 m) profile in the
asphalt surface layer are shown in Figure 7. The lines AA' and BB' represent the inter-layer
segmentation line respectively among the surface layer, middle layer and bottom layer.
Figure 7 shows the value of APPDI3D is greater than 1 exists in two sections, they are
mainly distributed around the edge of the loading circle area 1∼2 cm of pavement surface.
The potential damage developed from edge of loading circle to the underneath of loading
826
area along with a small angle. The tensile stress will appear at the edge of the tire loading
and is greater than the tensile strength of material, so it is possible to generate the cracking.
Generally, it won’t produce cracking due to the healing ability of asphalt mixture. However,
cracks will be appeared if the temperature changes largely, such as the raining and diurnal
temperature difference. Meanwhile, the situation that tensile stress is less than the tensile
strength and 3D is greater than 1 will also appear in this region. It won’t produce tensile dam-
age, but it is easier to close to the material yield curve to produce shear failure to form rutting.
Therefore, the cracks and rutting coexistence phenomenon may be appeared in the asphalt
pavement structure under the high temperature at the same place.
Surface layer 4 1200 350 0.28 0.08 1.30 0.35 830 600 0.19 0.13 0.86 0.60
Middle layer 5 1000 500 0.22 0.10 1.10 0.53 710 400 0.15 0.08 0.76 0.40
Bottom layer 6 700 0.12 0.63 430 400 0.08 0.07 0.48 0.43
827
5 CONCLUSIONS
REFERENCES
Baek C. 2010. Investigation of Top-Down cracking mechanisms using the viscoelastic continuum dam-
age finite element program [D]. Raleigh, North Carolina: North Carolina State University.
Freitas D et al. 2005. Effect of construction quality,temperature,and rutting on initiation of Top-Down
cracking [J]. Transportation Research Record, v1929: p. 174–182.
Hu Chunhua. 2009. Three dimensional finite element analysis of Top-Down cracking for asphalt pave-
ments [C]. ICTE, p. 3248–3253.
Industry standard of the People’s Republic of China (JTG D50–2006). High Asphalt Pavement Design
Specification. Beijing: China Communications Press.
Taeyoung Yun & Y. Richard Kim (2013). Viscoelastoplastic modeling of the behavior of hot mix asphalt
in compression[J]. KSCE Journal of Civil Engineering, v17, p. 1323–1332.
Ulhdtz, P. 2002. “Analytical tools for design of flexible pavements”, Keynote Address delivered at the
8th ISAP Conference held in Copenhagen.
Sun Li-jun. 2005. Structural behavior study for asphalt pavements [M]. Beijing: China Communications
Press, 304–332.
Wang Gang et al. 2012. Research on anti-deformation properties of high modulus asphalt concrete [J].
Journal of Tongji University (Natural Science), v02, p. 217–222.
Wang H. & I.L. Al-Qadi (2010). Near-surface pavement failure under multiaxial stress state in thick
asphalt pavement [R]. Washington D.C. 89th Transportation Research Board Annual Meeting,
Washington D.C.
Wang Yajun. 2011. Research on disease types and treatment technology for asphalt pavement of Ji-Qing
expressway [D]. Changchun, Jili: Jilin University.
Yao Zhanyong et al. 2012. Analysis of rutting change rule on asphalt pavement overlay [J]. Advanced
Materials Research, v368–373, p. 3131–3136.
Yan Cuixiang. 2010. Study on rutting development rule of Ji-Qing expressway asphalt pavement after
maintenance [D]. Jinan, Shanong: Shandong University.
Zheng Ying-Ren & Zhang Liang (2005). Geotechnical plastic mechanics [M]. Beijing: China Building
Industry press.
828
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Nan Zhang
Institute of Material Science and Engineering, Chang’an University, Xi’an, China
ABSTRACT: A study was conducted on the factors affecting the binding performance
of ultra-thin surface for the purpose of preventing ultra-thin surface between layers shear
failure and improving the integrity and durability. In the study, cored two-layer road tray
samples with different inter-layer treatments were tested on shear strength testing apparatus.
The binder material, binder dosage, existing road surface characteristics, and temperature
were analyzed for their effect on inter-layer binding performance. The results suggest: The
pavement when milled helps increase inter-layer binding performance, but the contribution
of milling varies depending on the type of binder, and so the choice shall be made as appro-
priate to the actual situation. Binding course shear strength drops appreciably with a rising
temperature, indicating inter-layer failure is more likely in hot seasons.
1 INTRODUCTION
With the common growth of traffic volume and the influence of the environment of high tem-
perature, more and more asphalt pavement need for maintenance. Ultra-thin surface as a new
kind of curing technology, more and more widely used on the asphalt pavement. Ultra-thin
skin coat with 2–2.5 cm is easy to damage[1], especially the shear failure occurs between ultrathin
overly and old asphalt pavement more easily, In order to improve performance between layers,
it need more detailed study of the influence factors, according to a lot of studies have found
that the influence factors of bonding layer mainly including ultra-thin surface materials, inter-
face form of old pavement [2], bonding layer material, dosage and temperature. Reference both
at home and abroad of the study found that direct shear test is used to study the performance
of bonding layers is feasibility, this paper based on the direct shear test to study the different
influence factors on the importance of the influence of the bonding performance[3∼5].
In order to keep the 2 cm thin layer mask surface with three layers structure, the paper
choose the sieve of 7.1 mm to control the gradation of asphalt mixture. the asphalt mixture
named UTO - 7. The asphalt mixture (UTO-7, AC-16) Marshall test results as shown in the
Table 1.
830
Figure 4. Before and after shear test.
F
τf = (1.1)
100S
Above formula:
τf——Shear strength (MPa);
F——Horizontal shear (N);
S——interface area of specimen (cm2)
3 TEST RESULTS
831
Figure 5. Test results of shear strength.
From the analysis of above, The shear strength for milled surface is general higher than the
shear strength for non-milled surface, It can included that milled surface on the bottom of
old pavement can improve the combination between the ultra-thin overly and old pavement,
increase the shear strength. With the increase of temperature, shear strength with four differ-
ent bonding layer materials is reduced, under high temperature, thin layer surface is easy to
generate shear failure.
Compared the biggest shear strength of four bonding layers under 60°C, It can conclude
that shear strength of the different bonding layers were improved after milled. And the per-
formance of high viscosity modified asphalt binder is best among all of bonding layers after
been milled, however the biggest shear strength of bonding layers is Nova Bond when it is
non-milled. Milled and non-milled on the old pavement are not linear relation to improve the
832
shear strength of bonding layers. It mainly because of the texture depth of the bottom plate
is different on different pavements after milled.
Above model:
τf——Shear strength (MPa);
T——Temperature (N);
S——Surface Texture Depth (mm);
D——Dosage of asphalt(kg/m2);
Untreated Milled
833
Table 6. Model summary.
Table 7. ANOVA.
Table 8. Coefficient.
Standard Coefficient
Bootstrap(1000) estimate
Influencing factors Beta of standard error df F Sig.
As the principle of probability theory and statistical known, It can conclude that the influ-
encing factor is an important factor if the Significant (sig) is less than 0.05,. From the Table 8,
Significant (sig) of temperature and surface texture depth are less than 0.05, temperature and
surface texture depth are important influencing factors of shear strength.
With the same analysis method can analyze Nova bond bonding layer, It can get the fol-
lowing model:
However, From the Table 11, Significant (sig) of temperature and dosage of high viscosity
modified emulsified asphalt are less than 0.05, and temperature and dosage of high viscosity
modified emulsified asphalt are important influencing factors of shear strength.
From above models, whatever bonding layer materials, The influence of temperature on
the shear strength is the largest, but dosage of asphalt and surface texture depth has different
influence on shear strength for different bonding materials. It can conclude that milled on
834
Table 11. Coefficient.
Standard Coefficient
Bootstrap(1000) estimate
Influencing factors Beta of standard error df F Sig.
the bottom of old pavement can obviously enhance the surface texture depth, bonding layer
shear capacity was improved obviously. However the significance can be used to determine
the importance influence factors. If sig of its surface structure depth is less than 0.05, it sig-
nificant to adopt milled on the old pavement to enhance shear strength.
4 CONCLUSIONS
With temperature rising, shear strength of bonding layer significantly decrease, in high tem-
perature season interlayer are prone to damage.
When the bottom of pavement was milled, the surface texture depth obviously to be
enhanced. However the significance(sig) can be used to determine the importance influence
factors. If sig of its surface structure depth is less than 0.05, it significant to adopt milled on
the old pavement to enhance shear strength. It needs to choose proper dosage of bonding
layers materials and the form of processing interface according to actual situation.
REFERENCES
Wang Xiang-heng. Research on interfacial shearing failure and function layer for asphalt pavement[D].
Xi’’n:Chang’an University, 2009 (in Chinese).
Chen Ming-xing. Research on test method and evaluation indexes about interlayer bonding of asphalt
pavement[D]. Xi’’n:Chang’an University, 2009 (in Chinese).
Zhou Ze-hong, Zheng Nan-xiang, Ji Xiao-ping. Design of synchronous under-seal based on anti-
shear strength between interlayers[J]. Journal of Chang’an University (Natural Science Editon),
2011,31(2):21–24 (in Chinese).
Su Xin-guo, Yan He, LU Sheng-di, Huang Xue-wen, Chen Hua-xin. Influencing factors on bond-
ing effect between asphalt layers[J]. Journal of Chang’an University (Natural Science Editon),
2013,33(3):21–26 (in Chinese).
Laith T. Evaluation of the influence of tack coat construction factors on the bond strength between
pavement of Transportation, 2006.
Lang Yang. Research on multi-functional shear apparatus of asphalt mixture[D]. Xi’’n:Chang’an
University, 2009 (in Chinese).
835
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: In this paper the feasibility of using Phosphorous Slag (PS) which replaces
limestone chips as the aggregate of base course and replaces part of cement is studied. Exper-
iment shows that when the rate of PS instead of limestone chips is 25% or PS powder instead
of cement ratio is 40%, the 90th day age strength increased by 22% and 21% respectively,
when the stress is larger, add PS micro powder mixture at the grass-roots level of fatigue
performance is better than that of ordinary cement stabilized mixture. Compared with the
ordinary cement stabilized macadam, mixed with PS material can decrease the mixture of
dry shrinkage strain and temperature shrinkage strain, can to a certain extent, improve the
crack resistance of cement stabilized gravel mixture
1 INTRODUCTION
Yellow phosphorus production in China rank number one of the world, the consequent yel-
low phosphorus slag have brought severe influences to the environment (Shiyong and Jinkai,
1997, Yang et al., 2005). Therefore, how to fully use these Phosphorous Slag (PS) in a reason-
able way has become an urgent task nowadays. The use of recycling waste PS in cement con-
crete industry have been becoming a trend, which can make full use of the PS, reduce its harm
to the environment and the cost of material, even is expected to improve materials properties
and brings great economic value and environmental benefits (Chen et al., 2009, HU et al.,
2007, Jian-xin and Lin, 2007, LIU and YANG, 2006, Ding-yian et al., 2000). Cement added
with PS as raw material or admixtures have lots of advantages such as high cement strength,
good wear-resisting property and fracture resistance property, strong ability to resist ero-
sion and low hydration heat (Liu et al., 2007, Yanling and Chenggui, 2005, Dong-xu et al.,
2002, Dongxu et al., 2001, Li et al., 2000, Li and Wu, 1995, Zhenhao, 1990). But this kind of
cement has some defects, such as low early strength, longer condensation time. The Strength
and durability of concrete and masonry mortar added PS as reactive aggregates are superior
to those mixed by Inactive aggregates (LIU et al., 2008, Yunxing, 1998).
According to the thought of adding PS materials into the cement concrete, it is also fea-
sible that applying PS to cement stabilized macadam base mixture. It has been studied that
using PS to replace limestone chips in different ratio and consisting of cement stabilized
composition aggregate of PS and limestone, proving cement stabilized PS and stone material
is good pavement base material (Jin et al., 2004, Bao-yu et al., 2007, Chong-jun et al., 2007,
Ming-kai et al., 2007). But there are no further researches in the investigation of influences
for cement stabilized stone mixtures’ property, especially the late strength, anti-creaking
property and anti-fatigue performance, which caused by mixing of PS in cement stabilized
stone. In addition, there still lack the research in which the effects on cement stabilized stone
after replacing part of cement with PS powder. Hence, the research team undertook the task
of performing some laboratory tests on cement stabilizing gravel added PS in two different
ways. One was replacement of limestone chips as the aggregate of the base course. The other
was replacement of cement. By laboratory tests it was investigated whether the mixing of PS
837
have effects on cement stabilizing gravel mixtures’ properties, such as later strength, crack
resistance and fatigue property.
2 METHOD
838
Figure 1. Composite gradation for the PS and aggregates.
The long-term strength of the six retained mixes was determined after 90 days of age using
the unconfined compression test. The cylindrical specimens were made 15 cm in diameter
and 15 cm long using compaction energy equivalent to that used for the Modified Proctor
test. There were nine parallel tests in specimens labelled F1 through F6.
The flexural strength of these mixes was determined after 90 days of age. Beams
10 × 10 × 40 cm3 were tested in a three-point bending setup. There were four parallel tests in
specimens labelled F1 through F6.
The fatigue test of these mixes was also determined after 90 days of age. Beams
10 × 10 × 40 cm3 were tested in a three-point bending setup with compressive fatigue mode,
at a frequency of 10 Hz, stress ratio from 0.5 to 0.8.
Cement-treated materials undergo thermal shrinkage (temperature variation) and drying
shrinkage (moisture loss). Drying shrinkage of these mixes was determined at room temperature
after 7 days of age and specimens were also same as beams in flexural strength test. Anti-drying
shrinkage property of these mixes was assessed by two indicators, cumulative water loss rate
and shrinkage strain. Drying shrinkage strain mainly occurs in earlier days, so it was supposed
to increase the indicator reading in the initial stage of test. The indicators were measured every
four hours in the first three days and subsequently measured two times a day and. They were
measured at one time every other day after seven days later and measured one time every seven
days after fifteen days later. There were three parallel tests in specimens labelled F1 through F6.
Thermal shrinkage of these mixes was determined at a step-down temperature varied from
50 to 0 °C after 7 days of age using strain gauge method. The temperature dropped at every
10 °C and was kept constantly for three hours when it meeting the desire degree. Specimens
were also same as beams in flexural strength test. There were three parallel tests in specimens
labelled F1 through F6.
839
Figure 2. Compressive strength of all retained mixes. Figure 3. Flexural strength of all retained mixes.
This finding is in agreement with the results found in the literature. This could be explained
by the fact that PS consist mainly of vitreous of silicate and aluminate and part of crystal of
C3S and C2S. It has the potential hydraulicity property which would be activated by activa-
tor calcium hydroxide produced in cement hydration process, making PS gradually hydrate
and form cementing property. Therefore the strength of cement treated mixes were enhanced
and the structure of their more compacted.
840
Figure 4. Strain-N curve of mixes F1, F2, F6.
mixes F4, F5, F6 were decreased 9%, 11% and 25% compared with mixes F1. It was indicated
that replacing part of cement with PS fine powder has an effect on water loss rate and shrinkage
strain of mixes, but shrinkage coefficient is not related to the addition of PS fine powder.
The influence of PS on drying shrinkage property of cement treated mixes explained by the
reasons below. To begin with, replacing cement with PS powder lead to decrease of cement
content, which can help to the result of less water loss ratio and shrinkage strain of mixes.
841
Figure 7. Thermal shrinkage strain of all retained mixes in different temperature interval.
What’s more, PS fine powder have retarding effect, making it function as filling and water-re-
ducing agent, which can balance binding material volumetric uneven change in the hydration
process and help to decrease the shrinkage strain of mixes. In addition, chemical substances
of PS powder react with calcium hydroxide produced in cement hydration process and form to
C-S-H cementitious materials, hence, the amount of calcium hydroxide and pore are much less
in the mixes, leading to drying shrinkage property enhanced with addition of PS powder.
REFERENCES
Bao-Yu, C., Yi-Xia, L., Yan-Hua, W. & Ying, C. 2007. Research on the Modified Quicklime Stabilized
Phosphorous Slag-crushed Stone Composite Aggregate. Journal of Wuhan University of Technology,
29, 119–121.
Chen, X., Zeng, L. & Fang, K. 2009. Anti-crack performance of phosphorus slag concrete. Wuhan
University Journal of Natural Sciences, 14, 80–86.
Chong-Jun, Y., Jin, Z. & Wei, C. 2007. Experimental Study of Phosphorous Slag Pavement Course
Material. Coal Ash China, 27–28+35.
Ding-Yian, W., Li, Z., Kun-He, F. & Yong-He, L. 2000. Application Study of Phosphotous Slag Admix-
ture in Concrete [J]. Yunnan Water Power, 2, 014.
Dong-Xu, L., Lin, C., Zhong-Zi, X. & Zhi-Min, L. 2002. A blended cement containing blast furnace
slag and phosphorous slag. Journal of Wuhan University of Technology-Mater. Sci. Ed., 17, 62–65.
Dongxu, L., Jinlin, S., Lin, C. & Xuequan, W. 2001. The influence of fast-setting/early-strength agent
on high phosphorous slag content cement. Cement and concrete research, 31, 19–24.
Hu, P.-G., Xu, D.-L., Song, Q. & Li, H. 2007. Study of effect of phosphorous slag on cement concrete
performance and mechanism researching [J]. Concrete, 5, 015.
Jian-Xin, L.Q.-M.C. & Lin, Y. 2007. Research of the influences of phosphorus slag on concrete physical
property [J]. Shanxi Architecture, 21, 106.
843
Jin, Z., Ming-Kai, Z. & Wei-Guo, S. 2004. A Study on Basecourse of Cement Stabilizing Phosphorous
Slag and Gravel. Highway, 186–189.
Li, D., Shen, J., Mao, L. & Wu, X. 2000. The influence of admixtures on the properties of phosphorous
slag cement. Cement And Concrete Research, 30, 1169–1173.
Li, D. & Wu, X. 1995. The study of character of blended cement of blast furnace slag and phosphorous
slag. J. Cem. Lime, 5, 2–5.
Liu, D., Fang, K. & Shi, Y. 2007. Effects of phosphorus slag on hydration properties and pore structure
of cement paste. Journal-Chinese Ceramic Society, 35, 109.
Liu, G.-P. & Yang, N. 2006. Study on Some Basic Performances of Concrete with Phosphorus Slag
Powder [J]. Journal of Wuhan University of Science and Engineering, 8, 021.
Liu, X.-W., Cao, J.-X., Chen, Q.-L. & Yang, L. 2008. Study on the application of phosphorus slag pow-
der in the ready-mixed commercial mortar [J]. Concrete, 6, 033.
Ming-Kai, Z., Jin, Z. & Wei-Guo, S. 2007. Design, Preparation and Property of Phosphorous Slag Road
Base Materials. Acta Scientiarum Naturalium Universitatis Sunyatseni, 46, 159–160.
Shiyong, L. & Jinkai, X. 1997. The compositional characteristics of guizhou yellow phosphorus slags
[J]. Acta Mineralogica Sinica, 3, 014.
Yang, J., Xiao, B. & Wang, X. 2005. Progress and Prospects for Comprehensive Utilization of Yellow
Phosphorus Slag. Multipurp. Util. Miner. Resour, 5, 37–41.
Yanling, C.L.S.G.P. & Chenggui, Z. 2005. Effect of Retardation Mechanism of Phosphorous Slag on
Portland Cement [J]. Bulletin of the Chinese Ceramic Society, 4, 012.
Yunxing, F.N.D.J.S. 1998. The flowability and strength as well as durability of superfine phorsphorus
slag concrete [J]. Low Temperature Architecture Technology, 1.
Zhenhao, W.S.Z. 1990. Hydration features of new phosphorous slag cement [J]. Journal of The Chinese
Ceramic Society, 4, 015.
844
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: In order to evaluate the fatigue damage of asphalt mixtures under repeated
loads, the Digital Image Processing (DIP) was used to investigate the microstructural charac-
teristics of air-voids. A field test track was constructed to reflect the internal states of asphalt
pavement, and several cores were drilled to slice into specimens with different air-void ratios.
Test temperatures were −10°C, 0°C and 10°C, the internal structures were scanned by X-ray
Computed Tomography (X-CT) device before and after fatigue damage, and the influence
of air-voids on fatigue damage was investigated. The results indicate that air-void shape and
distribution can be measured by X-CT images and reconstructed by DIP effectively; fractal
dimension is a quantitative index to represent the complexity of air-void morphology. The
effects of temperature and initial air-void ratio on fatigue performance are significantly, and
change of air-void ratio presents a linear relation with respect to the complexity of air-void
morphology.
1 INTRODUCTION
Asphalt mixture is a composite material that includes aggregates, air-voids and asphalt mastic,
each microstructure has obvious effect on the performance. Air-voids are one of the critical
structures in asphalt pavement, some functions such as drainage and noise reduction relate
with air-voids significantly. However, air-voids aggravate the fatigue damage under vehicle
loads and reduce the strength of asphalt mixture as well, causing macro cracks and other
distresses in asphalt pavement. Therefore, air-voids can be regarded as defects in asphalt
mixture.
In recently years, Digital Image Processing (DIP) based on X-ray Computed Tomography
(X-CT) has been widely used to analyze the origin and influencing factors of distresses in
asphalt mixture, the development of pavement distress can be presented through the micro-
structures reconstructed by X-CT images (Apeagyei 2014, Presti 2014, Mohajeri 2014, Arega
2013, Braz 2011, Moriyoshi 2013). As a main structure in asphalt pavement, air-voids have
significant effect on the mechanical performance and durability of asphalt mixture (Castillo
2015), investigation about the influence of air-voids on damage state of asphalt mixture is
an important research in pavement engineering, and the mechanisms that air-voids affect
the mechanical behavior of asphalt pavement are presented according to the X-CT scan-
ning technology (Aliha 2015, Yan 2015, Khosravi 2013, Kodippily 2014). Coleri used X-CT
apparatus to scan the internal structures of asphalt pavement before and after permanent
deformation, then the change of air-voids was detected and the effect of air-voids on rutting
distress was evaluated (Coleri 2012, Coleri 2013), Sefidmazgi also investigate the influence
of internal microstructures on anti-deformation performance of asphalt mixture using X-CT
845
images and DIP (Sefidmazgi 2012). Khan developed a 3D numerical microstructural model
according to X-CT data, indicating air-voids can cause stress concentration, and the mois-
ture damage of asphalt mixture was investigated as well (Khan 2013, Khan 2015). Besides,
the air-voids are extended to form cracks commonly due to stress concentration, and the
conclusion of presented research show that air-voids have adverse influence on mechanical
performance of asphalt mixture (Zhang 2014).
The air-void distribution is uneven because of the mineral aggregates gradation. The gra-
dation parameters of asphalt mixture, especially the content of fine aggregates, determine
the air-void distribution obviously (Tashman 2012). Many researches show that the air-void
characteristics, such as uniform state, discrete level of morphology and distribution law, can
be determined through X-CT images and DIP (Presti 2014, Rose 2015, Hassan 2015). How-
ever, most of them study the asphalt mixture molded in laboratory, it is hard to reflect the
real internal state of air-void distribution and characteristics, and the investigations about the
relationship between air-voids and fatigue damage are less.
This paper measures the morphology and distribution of air-voids in test specimens with
different compaction levels, the effect of air-voids on fatigue performance of asphalt mixture
is evaluated using the DIP and dynamic fatigue test. The research findings determine the
relationship between compaction and air-voids, and also present the influence of air-voids
on fatigue damage.
846
Table 1. Material parameters of SMA-11S.
Figure 2. Asphalt mixture specimens form ISAC test track. (a) Paving. (b) Compacting. (c) Core sam-
ples. (d) Test specimens.
Figure 5. X-CT images at different depths in pavement. (a) 30 mm. (b) 90 mm. (c) 150 mm. (d) 210 mm.
(e) 270 mm.
It can be obversed from Figure 5 that the compaction level of asphalt mixture increases
with the increasing depth in pavement. The air-void ratio nearby the pavement surface has the
maximum value, as shown in Figure 5(a), it is decreased with the increase of depth, and there
are almost no air-voids in the asphalt mixture illustrated in Figure 5(d) and Figure 5(e). There-
fore, the asphalt mixtures in different depths of pavement have dfferent internal structures.
The gray distribution of original X-CT images of asphalt mixture shows that gray value
does not evenly distribute but gradually increases from image center to bonder due to energy
attenuation, scatter and noise of X-ray, as shown in Figure 6, therefore, the greater the radius,
the greater the brightness from a visual point of view is.
The main functions of DIP applied to original X-CT image of asphalt mixture are extract-
ing and analyzing microstructures, this processing can distinguish air-voids from aggregates
848
Figure 6. Gray distribution of original X-CT image.
Figure 7. X-CT gray image and microstructure. (a) Original gray image. (b) Binary image of
Air-voids.
and asphalt mastic accurately. Artificial Neural Network (ANN) has been developed to extract
microstructures (Zelelew 2011, Kutay 2010). Futhermore, the gray image can be converted
into binary image to represent the different microstructures, and a method named Otsu has
been proved to be an effective binary operation for gray image (Xu 2011, Mizushima 2013).
The basic principle of Otsu is dividing the image into target and background, and the least
squares method calculates the optimal segmentation threshold. The maximum variance of
two groups means minimum probability.
Assuming X is an image with L gray level, the amount of pixels which has i gray level is Ni
and the total number of pixels in the whole image is represented by N, therefore, the prob-
ability density function is shown in equation (1):
Pi Ni / N (1)
Selecting a threshold k, the pixels of image can be divided: C0 is a group of pixels with a
gray level of [0, 1, …, k−1], and C1 is a group of pixels with a gray level of [k, k−1, …, L−1].
The definition of the maximum between-class variance δ2(k) is represented by equation(2):
δ 2 ( ) ω 0 ( μ μ0 )2 + ω1( μ μ1 )2 (2)
where, μ is average gray level of whole image, ω0, μ0, ω1 and μ1 represent pixel proportion and
average gray level of C0 and C1 respectively:
L −1 k −1 k −1 L −1
μ ∑ ω = ∑ Pi , μ0 ∑ ω μ ∑ iPi / ω1 ω1 1 ω0 (3)
i =0 i =0 i =0 i =k
k varies from 0 to L-1, then, the corresponding δ2(k) is calculated and the maximum one is
the optimal segmentation threshold.
However, the original Otsu cannot solve the problem that uneven gray distribution of
X-CT image. In this research, an improved Otsu method was used to extract the binary image
849
of air-voids and other microstructures (Hu 2015), and then, the microstructure characteris-
tics were analyzed by Matlab software, as shown in Figure 7.
where, lmmd represents the maximum distance of cracks or air-voids in length direction, and
lmnd represents the minimum distance of cracks or air-voids in width direction.
850
Figure 9. The changes of air-void ratio.
Figure 10. Microstructure characteristics of air-voids before and after fatigue damage. (a) Tempera-
ture −10°C. (b) Temperature 0°C. (c) Temperature 10°C.
The SI value and air-void area can be compared to determine the microstructures change
after fatigue damage. Due to the pixel of gray image was 80 μm, microstructures which width
is greater than 80 μm can be identified. In this research, microstructures, which SI was greater
than 10 and air-void area was than 5 mm2, were assumed as cracks. Selecting five cross-sec-
tions from each test specimen, the statistical results are listed in Figure 10.
851
The average SI and air-void area of individual microstructure are increased after fatigue
damage at different temperatures. Some data points which SI is greater than 10 and air-void
area is less than 5 mm2 show that cracks are commonly appeared after fatigue damage. It can
be observed from Figure 10 that the new cracks in asphalt mixture at temperature 0°C is more
than that at temperature −10°C and 10°C, the phenomenon indicates that the morphology or
shape of air-voids affects the damage state of asphalt mixture during fatigue test besides the
temperature, the air-voids with a great initial SI cause cracks more easily. Furthermore, some
large air-voids are compressed at temperature 0°C and 10°C due to the deformation.
Figure 11. Relationship between fractal dimension and change of air-void ratio. (a) Temperature
−10°C. (b) Temperature 0°C. (c) Temperature 10°C.
852
The above analysis reflects that increasing compaction level and reducing the complexity
of air-voids can improve the fatigue resistance of asphalt mixture. Besides the construction
technology, the aggregate particles also effects the morphology of air-voids, and the relevent
research will be launched in next step.
5 CONCLUSION
This paper investigates the air-void change in asphalt mixture before and after fatigue dam-
age, X-CT device was used to obtain the gray image of internal structures, the relationship
between air-void characteristics and fatigue damage was evaluated by digital image process-
ing. Results of this study can be summarized as follows.
1. The air-void ratio decreases with the increase of depth, especially at the range from 90 mm
to 150 mm, the asphalt mixture nearby the pavement surface presents a adverce fatigue
resistance because of high porosity.
2. The air-void ratio of asphalt mixtures are increased after fatigue damage at different test
temperatures, and research shows that increasing the temperature aggravate fatigue dam-
age during the low temperature range from −10°C to 10°C.
3. Utilizing the shape index can identify the cracks and determine air-void change before and
after fatigue damage, air-voids with great shape index causes cracks distress easily.
4. The damage of asphalt mixture can be evaluated by the fractal dimension of air-voids,
the relationship among them is linear, the result indicates that decreases the complexity of
air-voids can effectively reduce the fatigue damage of asphalt mixture.
ACKNOWLEDGMENTS
The work underlying this project was carried out under the research grant number FOR
2089, on behalf of the grant sponsor, the German Research Foundation (DFG).
REFERENCES
Arasan, S.; Yener, E.; Hattatoglu, F.; Akbulut, S.; Hinislioglu, S. 2010. The relationship between the
fractal dimension and mechanical properties of asphalt concrete. International journal of civil and
structural engineering 1(2):165–70.
Arega, Z.A.; Bhasin, A.; Kesel, T.D. 2013. Influence of extended aging on the properties of asphalt com-
posites produced using hot and warm mix methods. Construction and Building Materials 44: 168–174.
Apeagyei, A.K.; Grenfell, J.R.A.; Airey, G.D. 2014. Observation of reversible moisture damage in
asphalt mixtures. Construction and Building Materials 60: 73–80.
Aliha, M.R.M.; Fazaeli, H.; Aghajani, S.; Nejad, F.M. 2015. Effect of temperature and air void on mixed
mode fracture toughness of modified asphalt mixtures. Construction and Building Materials 95: 545–555.
Baer, J.U.; Kent, T.F.; Anderson, S.H. 2009. Image analysis and fractal geometry to characterize soil
desiccation cracks. Geoderma 154: 153–163.
Braz, D.; Barroso, R.C.; Lopes, R.T.; Motta, L.M.G. 2011. Crack detection in asphaltic mixtures by
computed tomography. NDT&E International 44: 195–201.
Coleri, E.; Harvey, J.T.; Yang, K.; Boone, J.M. 2012. A micromechanical approach to investigate asphalt
concrete rutting mechanisms. Construction and Building Materials 30: 36–49.
Coleri, E.; Harvey, J.T.; Yang, K.; Boone, J.M. 2013. Micromechanical investigation of open-graded
asphalt friction courses’rutting mechanisms. Construction and Building Materials 44: 25–34.
Castillo, D.; Caro, S.; Darabi, M.; Masad, E. 2015. Studying the effect of microstructural properties on
the mechanical degradation of asphalt mixtures. Construction and Building Materials 93: 70–83.
Farhidzadeh, A.; Dehghan-Niri, E.; Moustafa, A.; Salamone, S.; Whittaker, A. 2013. Damage Assess-
ment of Reinforced Concrete Structures Using Fractal Analysis of Residual Crack Patterns. Experi-
mental Mechanics 53: 1607–1619.
Hu, J.; Qian, Z.D.; Wang, D.W.; Oeser, M. 2015. Influence of aggregate particles on mastic and air-voids
in asphalt concrete. Construction and Building Materials 93: 1–9.
853
Hassan, N.A.; Airey, G.D.; Yusoff, N.I.M.; Hainin, M.R. et al. 2015. Microstructural characterisation
of dry mixed rubberised asphalt mixtures. Construction and Building Materials 82: 173–183.
Kutay, M.E.; Arambula, E.; Gibson, N.; Youtcheff, J. 2010. Three-dimensional image processing meth-
ods to identify and characterise aggregates in compacted asphalt mixtures. International Journal of
Pavement Engineering 11(6): 511–528.
Khosravi, H.; Abtahi, S.M.; Koosha, B.; Manian, M. 2013. An analytical–empirical investigation of the
bleeding mechanism of asphalt mixes. Construction and Building Materials 45: 138–144.
Kodippily, S.; Henning, T.F.P.; Ingham, J.M.; Holleran, G. 2014. Computed Tomography Scanning for
Quantifying Chipseal Material Volumetrics. Journal of Computing in Civil Engineering 28(3): 1–10.
Khan, R.; Grenfell, J.; Collop, A.; Airey, G.; Gregory, H. 2013. Moisture damage in asphalt mixtures
using the modified SATS test and image analysis. Construction and Building Materials 43: 165–173.
Khan, R., Grenfell, J.; Collop, A. 2015. Micromechanical modelling of typical Dense Bitumen Mac-
adam (DBM) from laboratory testing and X-ray Computed Tomography (CT). Construction and
Building Materials 77: 1–6.
Mizushima, A.; Lu, R. 2013. An image segmentation method for apple sorting and grading using sup-
port vector machine and Otsu’s method. Computers and Electronics in Agriculture 94: 29–37.
Moriyoshi, A.; Takahashi, N.; Ikeda, O.; Kawashima, M.; Akabane, T. 2013. Strain distribution in
asphalt mixtures during the wheel tracking test at high temperatures. Construction and Building
Materials 40: 1128–1135.
Mohajeri, M.; Molenaar, A.A.A.; Van de Ven, M.F.C. 2014. Experimental study into the fundamental
under standing of blending between reclaimed asphalt binder and virgin bitumen using nanoindenta-
tion and nano-computed tomography. Road Materials and Pavement Design 15(2): 372–384.
Presti, D.L.; Khan, R.; Hassan, N.A.; Airey, G.; Collop, A. 2014. Laboratory Mix Design of Asphalt
Mixture Containing Reclaimed Material. Advances in Materials Science and Engineering 2014: 1–11.
Presti, D.L.; Hassan, N.A.; Khan, R.; Airey, G. 2014. Reclaimed Asphalt Test Specimen Preparation
Assisted by Image Analysis. Journal of Materials in Civil Engineering 27(8): C4014005 1–5.
Rose, A.A.; Arambula, E.; Liu, G.; Glover, C.J. 2015. Use of a Total Air Void Diffusion Depth to
Improve Pavement Oxidation Modeling With Field Validation. Petroleum Science and Technology
33: 1198–1207.
Sefidmazgi, N.R.; Tashman, L.; Bahia, H. 2012. Internal structure characterization of asphalt mixtures
for rutting performance using imaging analysis. Road Materials and Pavement Design 13(S1): 21–37.
Tashman, L.; Pearson, B. 2012. Characterisation of stone matrix asphalt mixtures. International Journal
of Pavement Engineering 13(4): 297–309.
Xu, X.Y.; Xu, S.Z.; Jin, L.H,; Song, E. 2011. Characteristic analysis of Otsu threshold and its applica-
tions. Pattern Recognition Letters 32: 956–961
Yan, K.; Ge, D.D.; You, L.Y.; Wang, X.L. 2015. Laboratory investigation of the characteristics of SMA
mixtures under freeze–thaw cycles. Cold Regions Science and Technology 119: 68–74.
Zelelew, H.M.; Papagiannakis, A.T. 2011. A volumetrics thresholding algorithm for processing asphalt
concrete X-ray CT images. International Journal of Pavement Engineering 12(6): 543–551.
Zhang, Y.Q.; Luo, X.; Luo, R.; Lytton, R.L. 2014. Crack initiation in asphalt mixtures under external
compressive loads. Construction and Building Materials 72: 94–103.
854
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The mechanistic design of asphalt pavements is based upon the scien-
tific knowledge in engineering mechanics and material technologies of both, asphalt and
unbound structural layers. The deformation behaviour of the Unbound Granular Materials
(UGM) used in structural pavement layers is very complex due to their inherit nonlinear
stress dependency and the effect of bulk material factors, such as water content, density,
grain size distribution, etc. Thermal effects plays an important role on the mechanical defor-
mation of UGM when the following two conditions are met: (1) water is present and (2) the
temperature fluctuates in the saturation range (i.e. changes from positive to negative °C, or
vice versa). Due to extreme climatic conditions, these freeze and thaw cycles are very com-
mon in most of the granular base layers of flexible pavements in central European countries.
In this context, this paper presents selected results of an investigation about the effect of the
temperature on the mechanical deformation of UGM under cyclic triaxial conditions using
the results of Repeated Load Triaxial (RLT) tests with Variable Confining Pressure (VCP).
1 INTRODUCTION
The deformation behaviour of Unbound Granular Materials (UGM) under cyclic load-
ing consist of an elastic and a plastic part, and the magnitude of the stress/strain response
depends mainly on the type and magnitude of the applied load ([1], [2]). Furthermore, there
are several material related factors that influence the mechanical response of UGM notice-
able and making it very complex. Some of these factors are the material density, moisture
content, grain size distribution and particle shape ([3], [4]).
Thermal effects play an important role on the mechanical deformation of UGM when the
following two conditions are met: (1) water is present and (2) the temperature fluctuates in
the saturation range (i.e. changes from positive to negative °C, or vice versa). Due to extreme
climatic conditions, these freeze and thaw cycles are very common in most of the granular
base layers of flexible pavements in central European countries.
In the field, water can be found in capillary form or as absorbed water within the granular
grains of UGM. The water binding mechanism and binding strength differs between each
different water form and, therefore, it is not possible to define a uniform melting point. This
leads to a change in aggregate conditions at different temperatures. The freezing tempera-
ture of water is dependent on the free enthalpy, also known as “Gibbs free energy”, which
decreases with an increasing binding of water molecules and ions concentrations. There-
fore, water within UGM freezes first in pore spaces [5], where there is a minor contribution
from binding forces. Only with a continuous and prolonged temperature fall is there a phase
change in the bounded water. This complex process involves diverse interactions at the ther-
mo-hydraulic and chemico-physical level [6].
The properties between frozen and unfrozen soils under different loading conditions
differ significantly. The driving factor for this difference is the presence of both, ice and
liquid water. Freezing generally results in an increase in strength and stiffness of the soil
855
([7], [8]). Simonsen reported a large temperature dependency in all soils investigated but dif-
ferent stress dependency [9].
To the knowledge of the authors, there are no previous investigations about the mechani-
cal deformation behaviour of natural unbound granular materials considering the effect of
temperature (frost-thaw cycles) under small strains. Therefore, this paper presents the results
of an experimental investigation to determine the effect of the temperature on the mechanical
deformation of UGM under cyclic triaxial conditions.
2 LABORATORY TESTS
The thermal stress of UGM (oven-dry state) within the service temperature range is classi-
fied for the elastic material behaviour as insignificant effect size. This is no longer valid when
water is present in the material and temperature changes occur from positive to negative or
negative to positive °C Temperature range (or opposite) occur. The climatic conditions in
Germany cause both positive and negative °C temperatures in UGM, which can change fre-
quently. Therefore, a focus of the present thesis is to study the elastic deformation behaviour
of UGM under triaxial conditions and cyclic loading with additional thermal influence on
the material. Of particular importance is the study of the mechanical deformation behaviour
during the physical state change of existing water in UGM from fluid to crystalline state dur-
ing the cooling of the material or from crystalline to the fluid state form during the heating
of the material. This paper reports selected results of a phd thesis [10] conducted at the Chair
of Pavement Engineering at Technische Universität Dresden, which aimed to investigate the
thermal influence on the deformation behaviour of UGM in pavement structures.
3 EXPERIMENTAL PROGRAM
The RLT tests were performed with the RLT test apparatus of the Institute of Urban and
Pavement Engineering at TU Dresden. The RLT apparatus consists of a test frame, a hydrau-
lic unit, a test cylinder and a climatic chamber. In axial direction, tension and compressive
dynamic loads up to 5 MPa can be applied. The cell pressure can be both static and dynamic
and it ranges between 0.05 MPa and 1.5 MPa. Dynamic loads can be up to 10 Hz. The cli-
matic chamber supports temperatures between −20°C and +70°C. For measuring axial defor-
mations, two inductive displacement measuring systems (external) and a magnetic measuring
system (internal) are available. In the radial direction, the measuring system consists of nine
inductive displacement transducers.
856
Figure 1. Grain size distribution of the Sandy Gravel 0/45.
and 103% and three moisture contents MC = 2.0 M.-%, 4.0 M.-% und 6.0 M.-%. The speci-
men identification is presented as follow (Material_DoC_MC).
857
(Dummys), to control the specimen core temperature.listing facts use either the style tag List
summary signs or the style tag List number signs.
Figure 3 presents the results of a typical temperature run in terms of measured tempera-
ture in the chamber, cell, and specimen over a period of about 29 h for a complete freezing
and thawing cycle. The period A covers the specimen pre-tempering, period B the cooling or
freezing cycle and period C the warming or thawing.
The duration of the freezing and thawing period in the triaxial cell and in the specimen are
shown. The discontinuous shape of the temperature curve at nearly -1°C can be explained by
the phase transition period. During this phase transition period, the aggregate state of the
existing water changes from liquid to solid (frozen) and back to liquid; the water as a liquid
and the ice crystals coexist. The temperature of the whole system remains nearly equal to
the freezing/melting point due to slow removal of heat when being in contact with air. It can
be observed that the freezing process is greatly slowed and the temperature will not drop for
about 30 minutes once the freezing starts but continue dropping once the freezing process has
been completed. Figure 4 shows the measured values and the approximation of the sample
core temperature throughout the cooling phase and in the temperature range between −3.0°C
and + 0.5°C (Figure 5). For the approximation of the core temperature dependent of water
contend, degree of compaction and temperature difference between core and lateral surface a
spline function was used. Particularly cubic splines with the following approach were used:
Figure 3. Time dependent temperature in chamber, cell and specimen core; right: measured (grey
points) and approximated values (red line). On the example of an specimen with MC = 4 M.−%.
DoC = 97% and a temperature change of 40 K.
858
Figure 5. Cooling phase, temperature range of -3.0°C and -0,5°C (Detail A), schematic representation
of the test points (TP-F).
Ti (t ) = a b11 ⋅ (t ti ) + c11 (t − ti ) d ⋅ (t ti )
2 3
(1)
Ti(t) is the calculated temperature depended on time, a11 to d11 are coefficients and (t0-di)
is an defined time step.
Based on this temperature distribution within the specimen dependent on time, the meas-
uring points for the tests under repeated loading were defined. In total 10, measuring points
(see Table 1 and Table 2) during the freezing (TP-F) and thawing period (TP-T) were chosen
to measure the change of stiffness during this period. The measurement was stopped after
the freezing process had been completed at a specimen temperature of -10°C. In the positive
temperature range, the measurements were completed after the thawing process has been
finished at a temperature in the centre of the specimen of + 3°C.
3.5 Tables
For the characterization of the mechanical deformation behaviour the elastic axial strain
ε11_el can be described in dependence of the average stress p and the deviatoric stress q.
The deviatoric stress q is calculated using the second invariant of the deviatoric stress
tensor:
⎡1
q 3 J2 = ⎢ ⋅ [ − + − + − ]⎤⎥ (3)
⎣2 ⎦
1 1.00 9.890 A
2 117.75 −0.219 TP-F1
3 127.75 −0.696 TP-F2
4 137.75 −1.025 TP-F3
5 147.75 −1.225 TP-F4
6 157.75 −1.308 TP-F5
7 167.75 −1.313 TP-F6
8 177.75 −1.317 TP-F7
9 187.75 −1.325 TP-F8
10 197.75 −1.329 TP-F9
11 207.75 −1.511 TP-F10
12 — — —
1 1045 −10.00 A
2 1109 −2.59 TP-F1
3 1119 −2.04 TP-F2
4 1129 −1.72 TP-F3
5 1139 −1.50 TP-F4
6 1149 −1.37 TP-F5
7 1159 −1.29 TP-F6
8 1169 −1.23 TP-F7
9 1179 −1.19 TP-F8
10 1189 −1.13 TP-F9
11 1199 −1.06 TP-F10
12 1265 3.73 C
q = σ1 − σ 3 (4)
The average stress p can be calculated using the first invariant of the stress tensor:
1 1
σm I1 (σ 1 + σ 2 + σ 3 ) (5)
3 3
860
Figure 6. Applied stress paths.
Under the assumption of isotropy (independency of direction) the material behaviour for
any stress state can be expressed by two relevant material parameters (Young’s modulus and
Poisson’s ratio) using Hook’s law. However, in case of non-linear material behaviour, as it is
known for unbound granular materials, the Hook’s law can only be applied for infinitesimal
stress and strain changes. Following, the required calculations for analyzing the laboratory
test results must be carried out incrementally. For rotational symmetry and isotropy follows:
σ xx σ yy = σ xx / yy /σ = σ σ / 33
(6)
σ xx σ yy = σ xx / yy /σ = σ σ / 33 (7)
861
1
ε11
E
(σ11 − 2 μ σ 22 33 )
(8)
1
ε 22 /33
E
(σ 22 /33 − μ (σ 22 /33 σ11 ))
(9)
The Young’s modulus can now be calculated by:
( − )⋅( + ⋅ / 33 )
E= 11 22 /
(10)
Δε11 ⋅ ( 11 + 22 / 33 ) − 2 ⋅ Δσ 22 / ⋅ Δ σ / 33
Δε 22 / 33 Δσ 11 − Δε11 Δσ 22 / 33
μ= (11)
2 ⋅ Δε 22 / 33 Δσ 22 / 33 − Δε11 ( Δσ 11 + Δσ 22
2 / 33 )
Only the vertical deformation of the specimens during loading and discharge are accu-
rately measured and evaluated. Following, exclusively the stress point is in the focus of con-
sideration. From
and by differentiation with respect to the desired independent variable (with the same direc-
tion as the dependent variable) the calculation of the Young’s modulus as a tangent modulus
follows:
d f (σ 11; σ 22 / 33 ) 1
11
= (13)
dσ 11 E
4 RESULTS
The following chapter contains selected results focusing on the material characteristics such
as Young’s modulus and Poisson’s ratio.
Figure 7. Young´s modulus at different degrees of compaction (DPr) and constant water content (w);
quasi-completely frozen.
Figure 8. Young´s modulus at different water content (w) and constant degrees of compaction (DPr);
quasi-completely frozen.
863
Figure 9. Poisson’s ratio at different degrees of compaction (DPr) and constant water content (w);
quasi-completely frozen.
Figure 10. Poisson’s ratio at different water content (w) and constant degrees of compaction (DPr);
quasi-completely frozen.
strain relationship is influenced while the temperatures change from positive to negative and
consequently the aggregate state of the water changes; regardless of the material condition.
After evaluating all 10 test points of the freezing phase (during the phase-transfer from fluid
to ice), it could be found, that the decreasing amount of fluid water does not only lead to a
reduction of the elastic axial strains, but also to an elastic material response, which can be
described as “solid-state-like”, whereby the latter results in an increasing linear elastic behav-
iour. Opposite behaviour can be found during the thawing process. When the temperature
rises the freezing line retreats farther and farther to the inside of the specimen and the elastic
strains increase. Although the strains already increase, linear elastic material behaviour can
still be assumed during this process. Firstly, with the start of measuring point “TP T3”, the
expected nonlinearities could be found again. However, the sample with the lowest water
content of 2 M.-% is once again an exception: At the first measuring point (TP-F1) a linear
elastic response was observed, as the crystallization of water was far advanced.
Figures 11 and 12 show the temperature-dependent increase of the Young’s modulus with
respect to the different degrees of compaction and water content. Due to a very rapid decrease
of the elastic strains as a result of the freezing progresses, the Young’s modulus also rises
relatively quickly and significantly. At the end of the transition phase of the water (test point
TP-F10), the Young’s modulus reaches nearly identical values as the quasi-complete freezing.
Significant differences regarding the influence of the degree of compaction can not be
determined within the framework of the conducted studies. It is very clear, that even within
the period when the main part of the phase-transfer takes place, the water content affects the
value of the Young’s modulus considerably. A difference in the water content in the mixture
864
Figure 11. Young’s modulus as a function of sample core temperature and different degrees of
compaction.
Figure 12. Young’s modulus as a function of sample core temperature and different water contents.
Figure 13. Young’s modulus and Poisson’s ratio for different degrees of compaction and for different
water content.
*(Note: The calculated values apply for test times where linear elastic material behaviour has been
assumed TP-F3, F7 TP, TP-F10).
of 2 to 3 M.-% causes a difference in the specific example of the calculated Young’s modulus
between factor 3 and 4. Regarding samples with 4 M.-% and 5 M.-% water content as well
as a core temperature allowing to assume linear elasticity, the calculated Poisson’s ratio was
about 0.3. The only exception to this is again the specimen with the lowest water content of
2 M.-%. Its calculated Poisson’s ratios are lower due to higher compressibility with values
within a range of 0.1 and 0.2 (Figure 13).
865
5 CONCLUSIONS
Unbound granular materials show, in addition to the well known temperature independence
in a positive temperature range, a temperature independence when the water existing in the
unbound mixture is quasi-completely crystallized and this state remains unchanged (or a fur-
ther cooling is going on). Consequently, the mass fractions of water in fluid and crystalline
form can be considered as constant. Temperatures between the two aforementioned situa-
tions (independent of whether a cooling or heating process exists), influence the deformation
behaviour of the unbound granular material. Of particular importance is the period of phase
transition of water, due to the fact that the ongoing change of the state of aggregation (fluid-
crystalline) has a significant influence on the material structure and performance.
The degree of compression is of great importance for the deformation behaviour of
unbound granular materials given that no thermal influence is present. However, within the
research presented here, the degree of compression could not be defined as a relevant influ-
encing factor on the material behaviour—both during the phase of transition as well as the
frost penetration. Variations in the material properties and the small specimen deformations
due to thermal influence are possible reasons for this. It could be deduced, that the water
content seems to be the most influencing factor, not only in case of no thermal influence, but
also under thermal influence.
A rising water content leads by using thermally uninfluenced materials to an increase of
the elastic material response (drained conditions) when applying the same stress state. The
use of thermally influenced materials shows the contrary phenomenon. The elastic strain
decreases with an increasing water content (and when applying the same stress state) signifi-
cantly. This can be explained by the rising crystalline water content in the unbound granular
material.
The stress dependent performance of unbound granular materials, that is well known for
a thermal uninfluenced state could only be detected—for a thermal influenced state—at the
beginning of the cooling phase or at the end of the warming phase. Even at a relatively
low cooling time, the unbound granular materials showed a linear elastic material behav-
iour under the defined test procedure and test conditions, including the defined stress levels.
Therefore, it can be concluded that the characteristic material properties, Young’s modulus
and Poisson’s ratio, can simply be constant, stress-independent variables.
REFERENCES
Andersland, O.B./Ladanyi, B. Frozen Ground Engineering. Wiley & Sons Verlag, 2003.
Barksdale, R.D./Itani, S.Y. Influence of aggregate shape on base behaviour. Transportation Research
Record, 1989, No. 1227, pp. 173–182
Farouki, O.T. Thermal Properties of Soils. Series of Soils. Series of Soil Mechanics, 1986, Vol. 11.
Hass, H. North American Tunneling. Proceeding. Society for Mining, Metallurgy, and Exploration.
Inc, 2012.
Lekarp, F. et al. STATE OF THE ART.I: Resilient Response Of Unbound Aggregates. Journal of Trans-
portation Engineering, 2000.
Lekarp, F. et al. STATE OF THE ART.II: Permanent Strain Response of Unbound Aggregates. Journal
of Transprtation Engineering, 2000
Mancuso, C. et al. Unsaturated Soils: Research and Applications, 2012, 2. Aufl. Springer Verlag.
Patzak, J. Mechanical deformation behavior of a base course material with particular effect of temperature—
Experimental analysis using the example of a gravel base course. Dissertation, TU Dresden, 2015.
Simonsen, E. et al. Resilient Properties of Unbound Road Materials during Seasonal Frost Conditions.
Journal of Cold Regions Engineering, 2002, Volume 16, Issue 1.
TL SoB-StB Technische Lieferbedingungen für Baustoffgemische und Böden zur Herstellung von
Schichten ohne Bindemittel im Straßenbau. Ausgabe 2004/Fassung 2007.
Van Niekerk et al. Estimation of mechanical behaviour of unbound road building materials from physi-
cal material properties. Proceedings, 5th International Conference on the bearing Capacity of roads
and airfields, 1998, R.S. NORDAL and G. RAFSDAL, eds., VOL. 3, pp. 1221–1233
866
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Yiluo Zhang
Chang’an University, Xi’an, Shaanxi, China
ABSTRACT: In order to reduce the use of crushed stone and cement in base course,
Composite Soil Stabilizer Stabilized Gravel Soil (CSSSGS) has been gradually used in
road construction. This paper presents details of a study that deals with determination of
mechanical properties (unconfined compressive strength, splitting strength, compressive
modulus of resilience and splitting modulus of resilience) and durability (water stability,
freezing resistance, and shrinkage) of CSSSGS compared with Cement Stabilized Macadam
(CSM) based on the target unconfined compressive strength (5 MPa). The study reveals
that CSSSGS has the better mechanical properties (unconfined compressive and splitting
strengths) than CSM, and the durability is close to that of CSM. The potential of CSSSGS
for cracking resistance is good within 28 days of curing time. Overall, CSSSGS can be a
alternative material for CSM, and this research provides a solid foundation for applying
CSSSGS to entity projects.
1 INTRODUCTION
Cement stabilized macadam as base course material for highway is commonly used in China.
However, the production processes of crushed stone and cement can pose many problems for
society, such as environmental pollution, breaking ecological balance and increasing energy
consumption. So new alternative materials should be found to reduce the use of crushed
stone and cement.
Soil-stone mixture is a common geological body, which can be seen everywhere. Natural
gravel soil is widely distributed in China, and regional difference was very obvious. Natu-
ral gravel soil can be seen more easily in the southwest area of China, such as Guangxi,
Sichuan and Yunnan provinces, due to their special terrain, ground feature and climate
characteristics (Liu et al., 2013). For highway engineering in these areas, natural gravel soil
can be a potential road-building material.
Cement and lime are usually utilized to improve the mechanical properties and durabil-
ity of soils that exhibit significantly changes in behaviour under fluctuating environmental
conditions. Unfortunately, the production processes of cement and lime can release a large
number of greenhouse gases, dust and poisonous gases. Besides, the main raw material used
for production of cement and lime is stone, which can bring environmental and ecological
problems. Therefore, supplementary cement-materials, such as fly ash, slag, and cement kiln
dust, has been investigated to stabilize soils. Fly ash, a ndustrial by-product, can reduce the
amount of clay size particles, plasticity index and the swell potential of expansive soils and
these functions can be explained by cation exchange capacity values, which also can indicate
the changes in the mineralogy of the fly ash stabilized soils (Nalbantoğlu, 2004). The combi-
nation of fly ash, cement and lime is commonly used to reinforce soils. Obtaining the same
stabilizing effect of soils, the addition of fly ash to cement stabilized soil can reduce the
867
dosage of cement, and 10% replacement ratio is seem as the most effective mixing propor-
tion (Zhang et al., 2011; Horpibulsuk et al., 2009; Jongpradist et al., 2009). Furthermore,
the engineering properties of soils and crushed stone can also be significantly improved by
cement-lime-flyash (Gu, 2008; Sivaguru and Chakrabarti, 1984). On the other hand, polymer
has been studied to enhance road materials. The existing researches show that polymer can
promote application potential of base course and subgrade by increasing their compressive
strength, stiffness or water stability (Rodway, 2001; Iyengar et al., 2012).
The use of fly ash and polymer to replace a certain amount of cement and lime for stabiliz-
ing natural gravel soil as road material can lead to low-cost construction especially for the local
population and can provide an environmentally friendly means of reducing the use of CSM.
However, it is necessary to investigate the engineering properties of natural gravel soil stabilised
with composite soil stabilizer, which contains cement, quick lime, fly ash and polymer. So the
objective of the present study is to investigate the mechanical and durability characteristics of
CSSSGS by laboratory performance tests. The laboratory tests included unconfined compres-
sive strength, splitting strength, compressive modulus of resilience, splitting modulus of resil-
ience, water stability, freezing resistance, and shrinkage. In order to analyze the suitability of
CSSSGS as base course or subbase course, the laboratory tests for CSM are also conducted, and
comparative results are presented to examine the relationship between CSSSGS and CSM.
2 EXPERIMENTAL PROGRAM
868
Table 2. Testing results of technical indexes of natural gravel soil.
Crushing value (%) Liquid limit (%) Plastic limit (%) Plasticity index Classification
Sieve size (mm) Median Gradation range Sieve size (mm) Median Gradation range
ured according to Chinese standard JTG E42-2005 and JTG E40-2007, and the results are
showed in Table 2.
In this research, 32.5# Portland cement was used, which was produced in the city of Tong-
chuan, Shaanxi Province in China. The basic technical properties of cement tested on the
basis of the Chinese standard (GB 175-2007) are shown in Table 3.
The crushed stone was composed of four kinds of aggregates with different size ranges,
namely, XG3, G11, G8, G3. The aggregate has a maximum size of 31.5 mm. The composite
gradation of the aggregate referred to the median of the gradation range (C-C-2), which was
prescribed for CSM as base course material of low-grade highway (JTG/T F20-2015). The
gradation of the crushed stone adopted in this study is shown in Table 4.
CSSSGS CSM
Dosage (%) OMC (%) MDD (g/cm3) Dosage (%) OMC (%) MDD (g/cm3)
and composite soil stabilizer can be got through the relationship between unconfined com-
pressive strengths and binder contents. The optimum dosages of cement and composite soil
stabilizer were used for all performance tests in this study. The technology roadmap of this
study is shown in Figure 2.
Rc . Rc (1
(1 Cv ) (1)
Ri . Ri (1
(1 Cv ) (2)
where Rc 0.90 is the representative value of compressive strength with 90% guarantee rate; Rc
is the average value of compressive strength; Cv is the coefficient of variation; Ri 0.90 is the
representative value of splitting strength with 90% guarantee rate; Ri is the average value of
splitting strength. The representative value was adopted as the test value. The compressive
modulus of resilience can be calculated by
ph
Ec = (3)
l
Rt
K= × 100 (4)
R0
where K is the water stability coefficient (%); Rt is the representative value of compressive
strengths of the specimens soaked in water for t days (MPa); the guarantee rate is 90%.
Freezing and thawing tests were carried out according to T 0858-2009 (JTG E51-2009).
The cylindrical specimens were cured for 28 days in the curing room. At the end of curing
time, the specimens were unwrapped from plastic bags and soaked in water for one day.
Then, some specimens were used for testing the unconfined compressive strength and other
specimens were subjected to 5 cycles of freezing and thawing. Frost resistance coefficient can
be calculated by
RDC
R= × 100 (5)
Rc
where BDR is frost resistance coefficient (%); RDC is the representative value of compressive
strengths of the specimens subjected to 5 cycles of freezing and thawing (MPa); Rc is the
representative value of compressive strengths of the control specimens (MPa); the guarantee
rate is 90%.
The beam specimens were used to determine the drying shrinkage coefficient on the basis
of T 0854-2009 (JTG E51-2009). The specimens were cured under standard curing condition
for 6 days plus 1 day soaked in water. Then, these specimens were used to measure drying
shrinkage deformation and water loss. At last, the mass of oven-dry specimen was measured.
The drying shrinkage coefficient can be calculated by
εd
αd = (6)
Δω
871
ΔL
εd = (7)
L
where α d is the drying shrinkage coefficient (με/%); ε d is the drying shrinkage strain (με);
Δω is the rate of water loss (%); ΔL is shrinkage deformation of the specimen for Δω (mm);
L is the length of the specimen (mm). In this test, the average value was adopted as the test
value.
Splitting modulus of resilience tests were determined on a hydraulic testing machine under
strain-control at a loading speed of 1 mm/min, according to T 0852-2009 (JTG E51-2009).
The cylindrical specimens were cured for 7, 28 and 90 days in the curing room. At the end of
curing time, the specimens were unwrapped from plastic bags and soaked in water for one
day. Then, some specimens were used for testing splitting modulus of resilience. Splitting
modulus of resilience of specimen can be calculated by
p p0
Ei = (0.27 + 1.0 μ ) (8)
dllX
lY × ( .135 + 0.. μ )
lX = (9)
1.794 − 0.0314 μ
where Ei is the splitting modulus of resilience (MPa); p is the level of load (N); p0 is the initial
load (N); d is the diameter of the specimen (mm); lX is rebound deformation of horizontal
direction (mm); lX is rebound deformation of vertical direction (mm); μ is Poisson’s ratio.In
this test, the average value was adopted as the test value.
CSSSGS CSM
872
Figure 3. Influence of age on unconfined compressive strength.
tensile strengths when the hydration of cement clinker approximately completes (especially,
curing time exceeds 90 days). After about 50 days of curing, the unconfined compressive
and splitting tensile strengths of CSSSGS are higher than that of CSM at the same curing
times, and the 180-day unconfined compressive and splitting tensile strengths of CSSSGS
are 20.7% and 19.3% higher than those of CSM, respectively. For CSSSGS, the ratio of 150-
day splitting tensile strength to 180-day splitting tensile strength is as high as 96%, which
indicates that the 150-day splitting tensile strength of CSSSGS can be used for pavement
structure design.
Figure 5 shows the contrast relationship between the compressive modulus of resilience
of CSSSGS and CSM at various curing times. For CSSSGS and CSM, the compressive
873
modulus of resilience increases with the increase of age. The compressive modulus of resil-
ience of CSSSGS increases rapidly up to 90 days of curing and then it levels off at 150 days
of curing. For CSM, significant modulus increase occurred between 7 and 28 days of cur-
ing, and the increasing rate of the compressive modulus of resilience tends to be zero after
150 days of curing. The maximum modulus of CSSSGS is 17.3% less than that of CSM.
This is probably because natural gravel soil contains about 27% of fines. The compressive
modulus of resilience of base course is mainly controlled by the modulus of raw materials
and reaction products and structure type for base course (Gu, 2008). The modulus of the
fines is lower than that of crushed stone, which leads to CSSSGS having lower modulus
compared to CSM.
CSSSGS CSM
874
Table 8. Results of frost resistance tests.
Materials Average value Representative value Average value Representative value BDR (%)
CSSSGS CSM
Rc0.90 (MPa) Ri0.90 (MPa) Ec (MPa) K (%) BDC (%) α d (με/%) ε (με)
X=
∑ n
i =1 Xi
(8)
n
1 n
( )
2
S* = ∑ Xi
n − 1 i =1
X (9)
S*
Cv = (10)
X
where X is average value of one-property indicator of one set of specimens; X i is each origi-
nal test result of one-property indicator; n is the number of one set of specimens; S * is stand-
ard deviation; Cv is coefficient of variation. In order to save space, the statistical analysis
results are not presented, and the coefficient of variation for all tests in this study doesn’t
exceed 10%, meeting the requirements of Chinese standard JTG E51-2009.
4 CONCLUSIONS
In this study, the effect of composite soil stabilizer stabilization on natural gravel soil was
examined by using laboratory performance tests. The relationship between the performance
tests results of CSSSGS and CSM was also analyzed to evaluate the suitability of CSSSGS
as base course or subbase course, with the target unconfined compressive strength of 5 MPa.
Based on the test results, the following conclusions are made:
1. The comparison results of CSSSGS and CSM are summarized in Table 11.
2. The unconfined compressive and splitting tensile strengths and compressive modulus of
resilience of CSSSGS and CSM increase with the increase of age, but the increasing rate
decreases with time. Comparing 180-day unconfined compressive and splitting tensile
strengths, CSSSGS outperforms CSM. Besides, the compressive modulus of resilience of
CSSSGS is close to CSM. It can be concluded that from the sense of mechanics propertis
analysis, CSSSGS can be used to replace CSM as road base course or subbase course.
3. The water stability and freezing resistance of CSSSGS decrease with the increase of soak-
ing time and cycles of freezing and thawing, respectively, which is very close to that of
CSM. The drying shrinkage of CSSSGS is 54.9% more than that of CSM, so the curing
of CSSSGS base course should be strengthened.
4. From the results of failure strain of indirect tensile test, it can be concluded that CSSSGS
has the satisfactory potential of cracking resistance, compared with CSM with 28 days of
curing time.
5. Overall, CSSSGS is a potential material to replace CSM as road base course or subbase
course.
REFERENCES
876
Gu, Y., 2008. Study on the Material Composition of Cement-Lime-Flyash Stabilized Aggregates and
Pavement Performance. Thesis. Chang’an University.
Horpibulsuk, S., Rachan, R. and Raksachon, Y., 2009. Role of fly ash on strength and microstructure
development in blended cement stabilized silty clay. Soils and Foundations, 49(1): 85–98.
Iyengar, S.R., et al., 2012. Pavement subgrade stabilization using polymers: characterization and per-
formance. Journal of Materials in Civil Engineering, 25(4), 472–483.
Jongpradist, P., Jumlongrach, N., Youwai, S. and Chucheepsakul, S., 2009. Influence of fly ash on
unconfined compressive strength of cement-admixed clay at high water content. Journal of Materials
in Civil Engineering, 22(1): 49–58.
JTG/T F20-2015, 2015. Technical Guidelines for Construction of Highway Roadbases. Beijing: China
Communications Press.
JTG E42-2005, 2005. Test Methods of Aggregate for Highway Engineering. Beijing: China Communica-
tions Press.
JTG E40-2007, 2007. Test Methods of Soils for Highway Engineering. Beijing: China Communications
Press.
JTG E51-2009, 2009. Test Methods of Material Stabilized with Inorganic Binders for Highway Engi-
neering. Beijing: China Communications Press.
Liu, D., et al., 2013. Experimental Analysis of the Water Weaken Effect on Gravel Soil Roadbed Locat-
ing in the Reservoir Region with Different Compactedness. Journal of Civil Architectural & Environ-
mental Engineering, 35(4): 94–100.
Nalbantoğlu, Z., 2004. Effectiveness of class C fly ash as an expansive soil stabilizer. Construction and
Building Materials, 18(6): 377–381.
Rodway, B., 2001. Polymer stabilisation of clayey gravels, In Proceedings for 20 th ARRB Conference,
19–21 March 2001, Melbourne.
Sivaguru, N. and Chakrabarti, P.K., 1984. PAPER NUMBER 367. EXPERIMENTAL STUDY ON
PERFORMANCE OF LFA BASE COURSES FOR FLEXIBLE PAVEMENTS, Journal of the
Indian Roads Congress, 45(367): 567–634.
Zhang, X.D., Liang, Y. and Yue, Y., 2011. Experimental Study of Fly Ash Activator in the Application
of Aeolian Soil. Advanced Materials Research, 255–260: 3361–3365.
Zhang, D., 1990. Principles of reinforcing soils. Beijing: China Communications Press.
Zhang, D. and Zheng N., 1991. On the Anti-shrinkage Cracking Performance of Semi-rigid Base Course
Materials. China Journal of Highway and Transport, 4(1): 16–22.
Zhou, X., 2005. Research on Mix Design and Pavement Performance of Cement Stabilized Macadam.
Thesis. Chang’an University.
877
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The Cement Stabilized Material (CSM) has been accepted and utilized in
multiple areas of China as semi-rigid base. To predict the fatigue damage of asphalt pavement
with semi-rigid base and identify the effects of design parameters on fatigue life, simulating
models are established by ABAQUS based on a nonlinear fatigue damage model and modu-
lus attenuation law of CSM. The results show that the fatigue life of asphalt pavement with
semi-rigid base increases exponentially as the thickness of structural layers increases. It is
efficient to extend the service life of asphalt pavement by increasing the thickness of struc-
tural layers, especially that of semi-rigid base. The modulus of semi-rigid base and sub-base
is not the higher the better, it should be determined rationally to meet the requirements of
specification. It is concluded that the proposed numerical model can serve as an efficient tool
to predict the fatigue life of asphalt pavements.
1 INTRODUCTION
Asphalt pavement with semi-rigid base has already been universally applied in China.
Fatigue cracking of Cement Stabilized Material (CSM) is the main distress that governs
the service life of semi-rigid asphalt pavement. Fatigue damage increases with applied
loading cycles in a cumulative manner. Therefore, the calculation of cumulative fatigue
damage plays an important role in life prediction of pavement structure (Kim, 2002; Lee,
2000). The tensile stresses in the semi-rigid base caused by repeated traffic loading initi-
ate microcracks and make them propagate and coalesce to form macrocracks. As a result,
dynamic modulus of semi-rigid base would attenuate under cyclic loading. Fixed values
used in current Chinese specifications do not consider the modulus attenuation pattern of
CSM (Yang, 2006). The consistent use of initial modulus cannot reflect the accumulative
damage state and variable mechanical responses in pavement structure during the fatigue
process. However, in countries such as Australia and South Africa, the two-stage design
method has been adopted to simulate the variations in the behavior of CSM, in which
the effective fatigue life phase and equivalent granular phase of semi-rigid materials are
employed to calculate the service life of semi-rigid base (Sha, 2009). AASHTO (2002)
Pavement Structure Design System takes the modulus attenuation characteristics of CSM
into consideration by converting the accumulated fatigue damage into residual modulus
(Jia, 2009).
In this research, the fatigue damage evolution model was utilized based on secondary
development of ABAQUS finite element software. In order to estimate the fatigue life of
semi-rigid asphalt pavement, the modulus attenuation model of semi-rigid base materials
was also included in the fatigue damage analysis. By calculating the fatigue life of semi-rigid
asphalt pavement under different structural design parameters, the influence of thickness and
modulus on the fatigue life of asphalt pavement was analyzed.
879
2 CALCULATION THEORY
where:
D is the damage level in decimal form (1 ≥ D ≥ 0);
N is cumulative traffic loading (fatigue life);
a* p, q are the damage characteristic parameters;
σ is the tensile stress.
Plane strain model is applied in this research, the constitutive equation is shown in
Equation 2:
⎡ μ ⎤
⎢ 1 1− μ
0 ⎥
⎧σ x ⎫ ⎢ ⎥ ⎧ε x ⎫
⎪ ⎪ E ( μ ))(( D) ⎢ μ ⎥⎪ ⎪
⎨σ y ⎬ = ⎢ 1 0 ⎥ ⎨ε y ⎬ (2)
⎪⎩τ xz ⎪⎭ ( + μ )( μ ) ⎢1 − μ ⎥ ⎪ε ⎪
⎢ 1− μ ⎥⎩ z ⎭
⎢ 0 0 ⎥
⎣ 2(( μ ) ⎦
In the finite element calculation of fatigue damage, the model is implemented in ABAQUS
software through a user-defined subroutine of material (UMAT). UMAT is invoked after
each loading cycle to recalculate the element stiffness matrix, reflecting the influence of inner
fatigue damage on stiffness of material. The fatigue damage increment ΔD is calculated by
the Chaboche nonlinear damage evolution model in Equation 1 and then added to damage
degree D to regenerate the element stiffness matrix.
*Summarized in Table 4.
881
Table 2. Pavement structural parameters with different thickness of semi-rigid base.
(Sha, 2009). In this study, when analyzing the effect on fatigue life caused by the thickness of
structural layers, the initial modulus of semi-rigid base and sub-base were set as 10000 MPa
and 3000 MPa respectively. According to the AASHTO 2002 design guide for pavement, the
typical modulus values of semi-rigid base and sub-base in equivalent granular phase were
689 MPa and 172 MPa. To simplify the finite element model, the influence of temperature
and frequency on the dynamic modulus of asphalt concrete was not taken into account. The
modulus values of asphalt concrete were recommended by Chinese specification for design
of highway asphalt pavement at 10 Hz and 20°C.
a* 1 / [cK
KT ( p / 2) q ]
(3)
lg Ncr p lg σ n + lg c
(4)
where:
Ncr is fatigue life of crack formation;
a* p, q are the damage characteristic parameters.
882
Table 5. Pavement structural parameters with different modulus of asphalt layer.
Table 6. Pavement structural parameters with different initial modulus of semi-rigid base.
Material a* p
KT is the stress concentration factor. In this research, KT was set as 1, because it was
assumed that the road materials were homogeneous without regard for microscopic defect.
According to the fatigue equation and test data in the related research (Jia, 2009), the
value of p and c could be obtained, q was usually set as zero in the fatigue damage evolution
to confirm that the effective stress was proportionate to the logarithmic fatigue damage rate.
The value of a* could be obtained from Equation 3. By arranging the computing results, the
parameters of nonlinear damage evolution model are listed in Table 8.
Figure 3. The development of fatigue damage and tensile stress at the bottom of asphalt layer.
884
Figure 4. Relationship between thickness of asphalt layer and fatigue life of each structural layer.
equivalent granular phase, the fatigue failure occurs in asphalt layer first. After that point,
the modulus of semi-rigid base keeps reducing as the loading cycles continue to increase.
885
Figure 5. Relationship between thickness of semi-rigid base and fatigue life of each structural layer.
Figure 6. Relationship between thickness of sub-base and fatigue life of each structural layer.
Figure 7. Relationship between modulus of surface layer and fatigue life of each structural layer.
leads to a decline in its fatigue life, while the fatigue life of base and sub-base would have a
different range of growth. This is mainly because the increase of modulus leads to the rise of
tensile stress at the bottom of asphalt layer, reducing the fatigue life of surface layer under the
control stress loading mode. And because of the ascending load in the surface layer, the load-
sharing ratio of semi-rigid base and sub-base decreases, which prolongs their fatigue life.
According to the relationship between modulus of semi-rigid base and the fatigue lives of
structural layers presented in Figure 8, the enhancement in the modulus of semi-rigid base is
886
Figure 8. Relationship between modulus of semi-rigid base and fatigue life of each structural layer.
Figure 9. Relationship between modulus of semi-rigid sub-base and fatigue life of each structural layer.
also detrimental to its own fatigue life. This is chiefly caused by the increasing load-sharing
ratio of semi-rigid base to bear the vehicle loads. Meanwhile, the decrease of its fatigue life
also accelerates the attenuation rate of modulus in the base. Therefore, the support capacity
to the asphalt layer provided by the semi-rigid base is weakened, reducing the fatigue life of
the surface layer. However, the enhancement in semi-rigid base reduces the load share in sub-
base, prolonging the service life of the semi-rigid sub-base.
Figure 9 plots the influences on fatigue life of structural layers induced by the modulus of
sub-base. According to the calculation results, the increase in the modulus of sub-base brings
about the reduction in the fatigue lives of all layers, especially its own fatigue life. The reason
for this is that the sub-base shares more traffic loads as its initial modulus goes up, which
leads to the decrease of its fatigue life. The modulus of sub-base decreases rapidly, owing to
the shortening of its service life, which weakens the support to the upper layers. So the tensile
stress at the bottom of both semi-rigid base and surface layer rises up. This will accelerate the
damage accumulating rate in these two layers and shorten their service life.
On the basis of the above laws, the modulus of semi-rigid materials is not the higher the
better, it should be kept within reasonable bounds. As for the practical construction of semi-
rigid base and sub-base, the content of inorganic binder materials mixed in semi-rigid mate-
rials should be determined rationally to meet the strength requirement in accordance with
the specification. The modulus gradient of all structural layers should be considered when
determining the design parameters of pavement structure.
887
5 CONCLUSIONS
A fatigue damage evolution model integrated with the modulus attenuation law of semi-rigid
base materials was employed to investigate the fatigue performance of semi-rigid asphalt
pavement. The model was implemented in the ABAQUS software and the fatigue life of
pavement structure was evaluated, major conclusions can be drawn.
1. For asphalt layer, the fatigue damage mainly locates beneath the tire-road patch area. For
sub-base and base, the area that fatigue damage accumulates the highest is under the cen-
terline of dual wheels due to the dissemination of loading pressure by semi-rigid base.
2. The asphalt pavement with semi-rigid base goes through effective fatigue life phase and
equivalent granular phase if the asphalt surface is thin, and the fatigue failure develops
from bottom to top in the pavement structure under this circumstance. The surface asphalt
layer will suffer from fatigue failure first when it is relatively thick. The fatigue failure
develops from top to bottom in this situation. Fatigue life of asphalt pavements with
semi-rigid base increased exponentially as the thickness of pavement structure increased.
It is effective to prolong fatigue life of asphalt pavements by increasing the thickness of
semi-rigid base. There is little possibility of fatigue distress in semi-rigid base when thick
base is used.
3. Due to the sensitivity of semi-rigid materials to fatigue damage and significant decrease
in modulus during the fatigue process, the modulus of the semi-rigid base and sub-base
is not the higher the better. But the design of semi-rigid base should meet the strength
requirement in accordance with the specification.
REFERENCES
Jia, K. 2009. Research on the fatigue performance of semi-rigid base course material. Ph.D. Dissertation,
Xi’an: Chang’an University.
Kim, Y.R., Little, D.N. & Lytton, R.L. 2003. Fatigue and healing characterization of asphalt mixtures,
Journal of Materials in Civil Engineering 15(1): 75–83.
Liao, G. & Huang, X. 2008. The application of ABAQUS finite element software in road engineering.
Nanjing: Southeast University Press.
Lee, H.J., Daniel, J.S. & Kim, Y.R. 2000. Continuum damage mechanics-based fracture model of asphalt
concrete. Journal of Materials in Civil Engineering 12(2):105–113.
Sha, A., Jia, K. & Li, X. 2009. Fatigue performances of semi-rigid base course materials. Journal of
Traffic and Transportation Engineering 9(3):29–33.
Sun, Z. 2008. Research on the fatigue damage of asphalt pavement based on damage mechanics. Ph.D.
Dissertation, Nanjing: Southeast University.
Sun, Z. & Huang, X. 2012. Nonlinear fatigue damage characteristics analysis of asphalt pavement
considering interlayer contact. Journal of Central South University (Science and Technology) 43(6):
2385–2389.
Yang, M., Feng, D. & Sha, A. 2006. JTG D50-2006, Specifications for Design of Highway Asphalt
Pavement. Beijing: China communication press.
Yao, Z., Shang, Q. & Zhao, Z. 2007. The influence analysis of the semi-rigid asphalt pavement
configuration stress and distortion by interface conditions. Journal of Shandong University
(Engineering Science) 37(3):93–99.
Zhang, H. & Zhao, J. 1998. Application damage mechanics of hardware. Beijing: National Defense
Industry Press: 151–164.
888
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Steel reinforcements within the asphalt pavement layers have widely demon-
strated to be a successful technology to avoid the propagation of bottom-up fatigue cracks.
However, the tensile stresses, induced by heavy vehicles at the bottom of bound layers, induce
the crossbars unthreading compromising their proper functioning. In addition, the high tem-
perature during laying produces dilatations of the steel reinforcement.
The main topic of this paper is the study of the asphalt layer—steel reinforcement system,
analyzing the mechanical behavior and the construction materials optimization. Pull-out
tests were carried out to evaluate the effects of three different crossbars steel reinforcement in
terms of pull-out resistance. The reinforcement shape and the coating materials influence the
behavior of the steel wire mesh and its interaction with the asphalt mixture.
1 INTRODUCTION
Geosynthetics have been widely used as reinforcement in structures with unbound materials,
such as pavements, slopes, retaining walls and embankments. It has been shown that, in a
pavement system, the inclusion of geosynthetics can significantly improve the pavement per-
formance (Xiaochao et al. 2008, Tataranni et al. 2015). They influence, in fact, the aggregate
contact and interlocking, which control the load-bearing and load-transferring capability of
asphalt pavements (Dondi et al. 2012, 2014). A lot of studies have shown that the presence
of steel geosynthetics reinforcement at the bottom of asphalt layers should reduce the prin-
cipal stresses and the vertical deformations, thus increasing the bearing capacity and service
life. In particular the steel mesh provides a decrease of the tensile strain at the bottom of the
asphalt mixture, hindering the initiation and propagation of bottom-up cracking (Said et al.
2009, Namir et al. 2013). Vicari (2007) presents an overview of the main results obtained
from researches carried out by Universities around the world on pavements reinforced with
a steel mesh, in terms of improvement of the fatigue life. This study shows that in all cases
the pavement fatigue life increases more than 50%. Baek and Al-Qadi (2006) evaluated the
role of a single steel reinforcement wire in a two layered beam specimen on delaying crack
development by numerical analysis. The crack initiation time was delayed and the growth
rate decreased since the steel reinforcement held and redistributed concentrated stresses
around a crack tip. They mentioned that the role of the steel reinforcement was affected by
the interface conditions, the HMA material properties and the temperature. Bondt and Scar-
pas (2006) studied whether the shape and the bond of the reinforcement can influence the
behavior of these products inside the superstructures. Overseas experience has indicated that
the placement/construction procedures and the field conditions can strongly influence the
field performance of reinforcement products. Defects in a reinforced asphalt overlay due to
installation and construction problems (e.g. bonding between pavement layers, bulging of the
reinforcing material, etc.) may significantly reduce the effectiveness of a reinforced asphalt
overlay. The reinforcement product to be applied is much dependent on the location of the
reinforcing product relative to the crack zone and the surrounding medium (asphalt or con-
crete) affects the effectiveness of the reinforcement products (Binh Vuong et al. 2009). Most
889
commercial steel products have a much higher stiffness and tensile strength than asphalt, and
their thermal dimensional stability (dimensional variation because of temperature increase)
is fundamental considering the high temperature of the asphalt mixture during laying. The
steel dilatation could change significantly the geometry and physical properties of the rein-
forcement, compromising their role. In order to reduce the risk of steel dilatation, a slurry
seal layer was introduced between the reinforcement and the asphalt layer.
In this research context, the aim of the paper is to evaluate the effects of four steel rein-
forcements in terms of pull-out resistance. To this end pull-out tests were carried out in agree-
ment to prescriptions techniques PTV 867 (version 0.8, 15/4/2013). The study was carried out
in two different phases:
− phase 1: materials analysis and sample test construction using four different steel reinforce-
ment types.
− phase 2: pull-out tests and analysis of the results.
The effects of the different steel reinforcements within the asphalt layers will be studied,
analyzing the crossbars shape effects and their interaction with the asphalt layers.
2 METHODOLOGY
890
Figure 1. a) Mesh with the smooth bar. b) Mesh with one hump bar. c) Mesh with two humps bar. d)
Product with torsioned flat wire.
1 2319 6.23
2 2320 6.17
3 2318 6.25
Average 2319 6.22
gates weight, and 5.13% by mixture weight. The volumetric asphalt concrete characteristics
are reported in Table 2.
The amount of AC used is equal to 6.2 kg and was calculated according to EN 12697-33,
in order to obtain a bulk density of 2250 kg/m3 and a layer thickness of 40 mm after com-
paction. The emulsion type C55B1, at a rate of 0.2 kg/m2 of residual binder, was applied on
the top of the base layer and then the steel reinforcing netting was placed. The slurry seal
0/6.3 mm at a rate of 17 kg/m3 was laid in order to obtain 10 mm of thickness (Figure 2).
At the end the asphalt top layer (40 mm) was mixed and compacted using an asphalt
concrete AC 0/12. The aggregates bulk density was obtained according to UNI EN 1097-6,
and is equal to 2690 kg/m3. The mixture bitumen percentage is equal to 5.35% by aggregates
weight, and 5.07% by mixture weight. Also for the top layer, 6.2 kg of AC was weighted in
order to obtain a bulk density of 2250 kg/m3 and a layer thickness of 40 mm after compac-
tion. The compaction of both AC layers was performed with a static compaction at the
891
temperature of 160°C. The samples were prepared in pairs for each type and 14 samples were
made in compliance with EN 12697-33 (Figure 3).
Table 3 shows the numbering of the fourteen blocks and the type of steel reinforcing prod-
uct they contain.
892
The tests were performed on a tensile test device at room temperature in the displacement
controlled mode, with a displacement rate of 1 mm/min. A preload of 200 N was applied.
The tests were stopped at a displacement of 30 mm (Figure 5). The confinement pressure of
0.1 N/mm2 was distributed over the specimen faces. To obtain the required pressure a torque
wrench was used, considering a vertical force of 1750 N acting on each bolt. The applied
moment was determined with the following formula:
M = k d Fp,cd, (1)
where Fp,cd = 1750 N (vertical force on each bolt), k = 0.13 (factor depends on the type of
bolt), d = 520 mm (length of the torque wrench). The torque moment applied for each bolt
was equal to 120 Nm.
3.1 Criterion 1
Figure 6 shows the force versus displacement curves of the fourteen pull-out tests. Figure 8
shows the same data in the range up to 2 mm. From the force-displacement curves, the behav-
ior of the samples was analyzed related to the shape and diameter of the bar reinforcement.
The smooth bars force-displacement curves (RM A1, RM B1, RM C1 and RM D1) show the
maximum tensile force at rather small displacement (0.93 and 0.52 mm). In particular, RM
A1 and RM B1 reached a maximum force of 2346.1 N and 1917.2 N, respectively, and the
average value of the two tests (2140.63 N) complies with the first criterion.
893
Figure 6. Overview of all tests.
Table 4 shows the maximum force and the displacement at maximum force for each
specimen.
On the contrary, the RM C1 and the RM D1 maximum forces did not satisfy the first cri-
terion. This result is related to the sample curing which in this case was less than two weeks.
Therefore, the curing time of the slurry seal layer can affect the test results in terms of maxi-
mum force reached. Figure 6 shows that the samples with one hump bar (RM A2, RM B2,
RM C2 and RM D2) reached a maximum tensile strength value at a much higher displace-
894
ment than the samples with smooth bars. In particular, the force-displacement curves of RM
A2 and RM B2 reached an average maximum force of 2079.2 N at an average displacement
of 13 mm. The RM C2 and RM D2 bars exhibit a higher tensile strength than the bars hav-
ing the same shape but a smaller diameter. In this case, the average maximum force registered
was close to 3500 N at an average displacement of 18 mm. The results are strongly influ-
enced by the size of the bar, in particular, the 4.4 mm bar diameter experiences more friction
between the bar and the slurry seal and consequently has a higher pull-out resistance. The
one hump bars 4.4 mm and 3.9 mm both satisfy the first criterion. As shown in Figure 6 the
RM A3 force-displacement curve has a maximum force value of 4933.6 N. After the maxi-
mum, the force decreased but the level remained high throughout the test and was still above
4.5 kN at the end of the test. The bar shape with two humps has substantially influenced the
test results; the load was transferred on the wire mesh reinforcement that has dissipated the
energy transmitted to the tensile device. For the RM B3, the force-displacement starts from
zero and reaches the maximum value of 5332 N quickly. At this point the reinforcing bar
was broken and the test was interrupted. The same phenomenon occurred for the RM C3 at
7058.3 N. On these specimens, the hump placed on the outer boundary was exposed to the
steel box, generating a confinement that has broken the bars (Figure 7).
Out of the fourteen specimens studied, the RM D3 sample has reached the maximum
tensile force peak. In this case, the position of the two humps and the 4.4 mm bar diam-
eter have guaranteed a high pull-out resistance. For the two humps bar the first criterion is
always satisfied. Making a comparison between RM A3 and RM D3, it follows that the latter
reaches higher values since it has a larger diameter. For the double torsion bars, A and B,
the force–displacement curves have reached the mean maximum force value of 5.2 kN. This
value, similar to the value reached by the two humps bar of RM 3.9 mm, was lower than the
peak tensile force reached by the two humps bars RM 4.4 mm.
3.2 Criterion 2
Through the second criterion the cross reinforcement anchoring degree is rated by checking
the slope obtained from the pull-out curves between 1 mm and 3 mm. As described above, for
the criterion to be satisfied the average slope value must be greater than 200 N/mm.
In Table 5 the slope values of the 14 specimens are reported. For the samples with smooth
bar the mean slope values are negative and equal to −186.76 N/mm for RM A1 and RM B1,
and −66.72 N/mm for RM C1 and RM D1. The negative parameters depend on the smooth
bar tensile behavior. As seen in the previous paragraph, these specimens reached the maxi-
mum force at a displacement less than 1 mm and in the next log lost strength; therefore, the
curve trend in the 1–3 mm displacement range is decreasing.
895
Table 5. Slope values in 1–3 mm displacement range.
The one hump bars have satisfied the second criterion only for RM 4.4 mm (234.43 N/
mm). Also for this criterion the highest values were obtained in tests conducted on the bars
with two humps. In particular, for the 3.9 mm types (RM A3 and RM B3) a mean value of
the slope equal to 722.35 N/mm has been obtained (Figure 9); for the LBG types (RM C3,
RM D3) a mean value of 891.78 N/mm has been found (Figure 9). Therefore both types with
two humps have satisfied the second criterion. The curves extrapolated from tests on double
torsion bars have reached a mean value of 354 N/mm; also in this case the second criterion is
896
Figure 9. RM A3, RM B3, RM C3, RM D3 pull–out curve in 1–3 mm displacement range.
met, however, the values obtained are considerably lower compared to the bars with 2 humps
of the type RM A3, RM B3, RM C3 and RM D3.
4 CONCLUSIONS
The main objective of this research was to study how the different types of transversal bars
in steel reinforcements react to a pull-out test and therefore their influence on the pavement
structure behavior. Based upon this research, the following concluding remarks can be made:
• The good quality of the Slurry Seal has a significant role in terms of protection of the steel
geogrid at high temperatures and in terms of increasing the adhesion between the bars and
the pavement structure. Therefore, the friction between the slurry seal layer and the steel
reinforcement confers greater resistance to the system.
• The stress-strain response of the system is closely related to the different types of bars in
terms of both shape and diameter. An increase of the number of humps and the diameter
of the bars results in a higher maximum tensile force reached; the same factors also do
increase the anchor level.
• The curing time of the asphalt samples is relevant; bitumen oxidation, and consequently
its stiffening, supports the interaction between the steel bars and the asphalt concrete.
Based on the analysis of the results, the samples with two humps bars have exhibited the
best performance in terms of tensile resistance and anchorage. Finally it can be stated that
the reinforcement performance, especially for those with one and two humps, may depend
on their installation and position. The humps must not be installed on the outer boundary
of the road to avoid shear failure of the bar. At the end, a surface layer of minimum 40 mm
bituminous mixture must be laid in order to guarantee the superstructure performance.
ACKNOWLEDGEMENTS
The authors take this opportunity to thank Eng. Marco Vicari and the Maccaferri Group for
the considerable help they gave us in carrying out the tests.
897
REFERENCES
Said S.F., Carlsson H., Hakim H. 2009. Performance of flexible pavements reinforced with steel fabric.
Bearing Capacity of Roads, Railways and Airfields 1: 1133–1141.
Xiaochao T., Ghassan R.C. and Angelica Palomino A., 2008. Evaluation of geogrids for stabilising
weak pavement subgrade. International Journal of Pavement Engineering 9: 413–429.
Tataranni P., Sangiorgi C., Simone A., Vignali V., Viola P. and Dondi G. 2015. Effects on bonding of
anti-reflective cracking solutions at the top bituminous interface of a small airport pavement: a labo-
ratory and modeling study. 8th RILEM Symposium on testing and characterization of sustainable
and innovative bituminous materials, October 7–9, Ancona, Italy.
Dondi G., Simone A., Vignali V. and Manganelli G. 2012. Numerical and experimental study of granu-
lar mixes for asphalts. Powder Technology 232: 31–40.
Dondi G., Vignali V., Pettinari M., Mazzotta F., Simone A. and Sangiorgi C. 2014. Modeling the DSR
complex shear modulus of asphalt binder using 3D discrete element approach. Construction & Build-
ing Materials 54: 236–246.
Dondi G., Simone A., Vignali V. and Manganelli G. 2012. Discrete element modelling of influences
of grain shape and angularity on performance of granular mixes for asphalts. Procedia - Social &
Behavioral Sciences 53: 399–409.
Dondi G., Simone A., Vignali V. and Manganelli G. 2012. Discrete particle element analysis of aggre-
gate interaction in granular mixes for asphalt: combined DEM and experimental study. Proceedings
of 7th RILEM International Conference on Cracking in Pavements, June 20–22, Delft, The Nether-
lands, ISBN 978-94-007-4565-0.
Namir G.A., Saad F.I., Nabil H.J. 2013. Experimental Study On Surface Steel-Reinforcement For
Asphalt Pavements. Journal of Engineering and Development 17: 110–127.
Baek, J. and Al-Qadi I.L. 2006. Effectiveness of Steel Reinforcing Interlayer Systems on Delaying
Reflective Cracking, Proceedings of Airfield and Highway Pavement: Meeting Today’s Challenges
with Emerging Technologies, ASCE Atlanta, United States.
Vicari M. 2007,”Reinforcement with Double Twist Steel Wire Mesh: Modeling and Laboratory Experi-
ences to Evaluate the Design Life Improvement of Asphalt Pavements, 4th International SIIV Con-
ference, Palermo, Italy.
De Bondt, A.H. and Scarpas A. 2006. Design of reinforced overlays. 85th Annual Meeting TRB, Wash-
ington, D.C., United States.
Vuong B., Choi X., Hoque Z. 2009 Assessment of reinforced asphalt products for road applications
using finite element modelling. Proceedings of the Thirteenth International Flexible Pavements Con-
ference, Surfers Paradise, Queensland, Australia.
Prescriptions techniques, PTV version 08, 867. Treillis d’armature en acier, 2013.
898
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Xiaodi Hu
Transportation Research Center, Wuhan Institute of Technology, Wuhan, Hubei Province, P.R. China
1 INTRODUCTION
Each layer of pavement structure contributes to resistance to the PD. Several researchers
studied on the PD contribution of each AC layer. Hu, Zhang, and Huang (2011) found that
the PD contribution of the top AC layer, the intermediate AC layer, the bottom AC layer
was about 16%–26%, 42%–54% and 26%–40%, respectively. Hu, Zhong, and Lubinda (2015)
studied the PD contribution of each layer through both normal strain and shear strain, they
obtained the PD contribution from normal/shear strain, for normal strain, the PD contribu-
tion of the top AC layer, the intermediate AC layer, the bottom AC layer was about 15%–50%,
20%–50% and 10%–50%, respectively. For shear strain, the results were 15%–40%, 30%–54%
and 20%–40%, respectively.
Although previous studies shown important results, there are still some missing in the PD
contribution of each layer. Non-uniformity of the TPCP is widely accepted, and with longi-
tudinal TPCP, lateral TPCP as well. 3D-FE was utilized to model the real TPCP. Moreover,
modulus has a significant effect on the PD contribution, it was not settled by personal, in this
paper, and it would be calculated by the master curves under different temperature, modulus
would approach the real circumstance which would make the results more credible.
Besides, previous researchers focused on heavy trucks, Hu, Zhong, and Lubinda (2015)
analyzed only three types of light trucks in their study. In this study, more attention would be
focused on the overloaded light trucks, furthermore, more cases would be considered.
899
2 METHODOLOGICAL APPROACH
In this section, the methodological approach is discussed and includes the following aspects:
loading configuration, boundary conditions, TPCP, pavement structure, and modeling
parameters.
900
load of 17.88 kN with tire inflation pressure of 390 kPa, this three types of wheel load repre-
sent overloaded, which equal to 150% and 250% of the standard light trucks, respectively.
Ts = Ta (3)
Td(min) = Ts(min)+5.1*10−2d−6.3*10−5d2 (4)
Table 1. The highest and lowest temperature of each month (unit: °C).
Month
1 2 3 4 5 6 7 8 9 10 11 12
High temperature 16.3 20.7 27.6 30.3 33.7 35.9 37.2 37.6 35 29.8 24.1 17.2
Low temperature −5.8 −3.3 −0.2 5.38 11.9 17.2 21.8 19.6 13.9 7.77 0.77 −4.8
901
Table 2. The Modulus and Poisson ratio of AC layers for modeling.
Month 1 5085 3214 4465 0.45 0.35 0.35 39257 24639 50877 0.2 0.2 0.2
2 4191 2623 3359 0.45 0.45 0.35 34877 21596 41319 0.2 0.2 0.2
3 3090 1904 2145 0.45 0.45 0.35 30251 18430 32172 0.2 0.2 0.25
4 2741 1697 1799 0.45 0.45 0.45 23329 13797 20371 0.25 0.25 0.25
5 2355 1431 1438 0.48 0.45 0.45 17228 9842 11953 0.25 0.25 0.25
6 2133 1290 1244 0.48 0.45 0.45 13482 7490 7767 0.25 0.25 0.35
7 2014 1214 1143 0.48 0.48 0.45 10891 5904 5336 0.35 0.35 0.35
8 1977 1191 1112 0.48 0.48 0.45 12090 6633 6412 0.35 0.35 0.35
9 2220 1345 1319 0.48 0.45 0.45 15727 8891 10181 0.25 0.25 0.25
10 2803 1719 1859 0.45 0.45 0.45 20886 12197 16771 0.25 0.25 0.25
11 3595 2232 2680 0.45 0.45 0.35 28902 17516 29691 0.2 0.2 0.25
12 4890 3085 4215 0.45 0.35 0.35 37448 23378 46826 0.2 0.2 0.2
Table 3. The Modulus (unit: Mpa) and Poisson ratio of base, sub-base layer and subgrade at different
month.
Base 12000 12000 9000 9000 9000 9000 9000 9000 9000 9000 12000 12000 0.25
Sub-base 12000 12000 9000 9000 9000 9000 9000 9000 9000 9000 12000 12000 0.25
Subgrade 70 70 50 50 50 50 50 50 50 50 70 70 0.4
According to literature, the shear strain (γxz) and normal strain (εz) were selected for the
subsequent PD analysis (Hu, Zhong & Lubinda, 2015). In this study, normal/shear strain was
utilized to analyze its different contribution to the PD in each AC layer.
Since PD is related to the magnitude of the shear/normal strain and the thickness of each
AC layer, TS, IS and BS represent shear/normal strain of top AC layer, intermediate AC layer
and bottom AC layer, respectively. Besides, TA, IA and BA represent the PD contribution of
the top AC layer, intermediate AC layer and bottom AC layer, respectively. The relationship
between TS, IS, BS and TA, IA, BA is shown as following:
902
3.1. Normal strain contribution to PD of each AC layer
In order to analyze the PD contribution of each AC layer, normal strain contribution to PD
of each AC layer was shown in Figure 2, which list in two parts, (a), (b) and (c) shown the
High temperature group; (e), (f) and (g) shown the Low temperature group. (a), (e) and (b),
(f) and (c), (g) represented the TA, IA and BA, respectively.
Standard represent wheel load of 25 kN with tire inflation pressure of 600 kPa, 250–1073
represent wheel load of 10.73 kN with tire inflation pressure of 250 kPa, 390–715 represent
wheel load of 7.15 kN with tire inflation pressure of 390 kPa, 390–1073 represent wheel
load of 10.73 kN with tire inflation pressure of 390 kPa, 390–1788 represent wheel load of
17.88 kN with tire inflation pressure of 390 kPa.
Based on Figure 2, several conclusions could be inferred as following:
1. Normal strain contribution to PD of each AC layer changes due to different months,
namely, it varies with the change of temperature or modulus of each layer.
2. Each type of tire makes different contribution to PD at each AC layer.
For standard, TA, IA, BA due to normal strain is 14.88%–24.3%, 38.15%–50.27%,
25.43%–44.44%, respectively. Besides, TA is the smallest among these types of loading,
however, BA is the biggest. This finding is consistent with the literature results (Hu,
Zhong & Lubinda, 2015).
For 250–1073, TA, IA, BA is 23.50%–33.60%, 35.76%–44.68%, 21.72%–38.58%, respec-
tively. Besides, IA is the smallest among these types of loading, however, TA is the
biggest.
For 390–715, TA, IA, BA is 19.53%–29.42%, 44.75%–51.64%, 19.33%–34.97%, respec-
tively. Besides, BA is the smallest among these types of loading, however, IA is the
biggest.
For 390–1073, TA, IA, BA is 22.19%–32.96%, 40.78%–47.81%, 19.91%–36.23%,
respectively.
For 390–1788, TA, IA, BA is 20.80%–30.95%, 35.99%–45.76%, 23.29%–40.93%,
respectively.
3. Under each type of tire, the PD at each AC layer has the same trend at each month both
in highest temperature months and lowest temperature months.
904
Figure 3. Shear strain contribution to PD of each AC layer.
the PD contribution at each layer is much closer than 390–1073 and 390–1788. This may
implied that when the combination of wheel load and tire inflation pressure is reasonable,
overloading and less tire inflation would not do more damage to the pavement.
In this study, five kinds of different types of loading combination with 24 kinds of pave-
ment structures were analyzed by incorporation with 3D-FE modeling. On the basis of large
number of analysis in the study as described, the following conclusions can be drawn:
1. The PD contribution of each layer has the same variation tendency with the varied modu-
lus, which include all of the light trucks, overloading trucks.
905
2. The PD contribution of each layer is different, however, both normal strain and shear
strain contribute the biggest in the intermediate layer than the other two layers. This find-
ing emphasize the importance of the intermediate layer, moreover, designing a reasonable
intermediate layer has a significant influence on the pavement service life.
3. With the change of temperature, the modulus of each layer varied, consequently, the PD
of each layer changes.
4. With combination of wheel load and tire inflation pressure changes, the PD contribution
tendency of each layer keeps the same.
REFERENCES
Al-Qadi, I.L., & Yoo, P.J. 2007. Effect of surface tangential contact stress on flexible pavement response
(with discussion). Journal of the Association of Asphalt Paving Technologists, 76, 558–582.
De Beer, M., Fisher, C., & Jooste, F.J. 1996. Determination of pneumatic tyre/pavement interface con-
tact stresses under moving loads and some effects on pavements with thin asphalt surfacing layers.
Proceedings of the eighth international conference on asphalt pavements, Seattle, WA, 1, 179–227.
Hu, M., Zhang, J., & Huang, X. 2011. Analysis of rutting characteristics of semi-rigid base asphalt
pavement. Journal of Highway and Transportation Research and Development, 28(6), 15–18.
Hu, X., & Walubita, L.F. 2008. Modelling tensile strain response in asphalt pavements: bottom-up and/
or top-down fatigue crack initiation. Road Materials & Pavement Design, 10(1), 125–154.
Hu, X., 2003. Measuring distribution of tire ground pressure and stress response analysis of asphalt pave-
ment using those results (Dissertation for doctor degree). Tongji University, Shanghai (in Chinese).
Hu, X., Zhong, S., & Walubita, L.F. 2015. Three-dimensional modelling of multilayered asphalt con-
crete pavement structures: strain responses and permanent deformation. Road Materials & Pavement
Design, 16(3), 1–14.
Hu, X., Zhou, F., Scullion, T., & Walubita, L.F. 2009. Proposed loading wave forms and loading time
equations for m-e pavement design and analysis. Transportation Research Board, CD-ROM.
Huber, GA., et al. 1994. Weather Database for the Superpave Mix Design System, SHRP-A-648 A.
National Research Council. Washington DC.
Jia lu. 2008. The temperature prediction method and application of asphalt pavement (Dissertation for
doctor degree). Tongji University, Shanghai (in Chinese).
Kennedy, T.W., Huber, G.A., Harrigan, E.T., Cominsky, R.J., Hughes, C.S., & Quintus, H.V., et al.
1994. Superior performing asphalt pavements (superpave)’. the product of the shrp asphalt research
program. Binders.
Loulizi, A., Al-Qadi, I.L., & Elseifi, M. 2006. Difference between in situ flexible pavement measured
and calculated stresses and strains. Journal of Transportation Engineering, 132(7), 574–579.
Myers, L.A., Roque, R.B., & Ruth, E. 1999. Measurement of contact stresses for different truck tire
types to evaluate their influence on near-surface cracking and rutting. Journal of the Transportation
Research Board, 1655(23), 175–184.
Park, D.-W., Martin, A.E., & Masad, E. (2005). Effects of nonuniform tire contact stresses on pavement
response. Journal of Transportation Engineering, 131(11), 873–879.
Siddharthan, R.V., Krishnamenon, N., El-Mously, M., & Sebaaly, P.E. (2002). Investigation of tire con-
tact stress distributions on pavement response. Journal of Transportation Engineering, 28, 136–144.
Specification for design of highway asphalt pavement (JTG D50–2006). 2006. Beijing: People Traffic.
Wang yuefeng et al. 2012. Study on temperature modification of dynamic modulus of asphalt mixture.
Journal of China & Foreign Highway. 2012,32(2), 210–214(in Chinese).
Wang, H., & Al-Qadi, I.L. 2009. Combined effect of moving wheel loading and three-dimensional con-
tact stresses on perpetual pavement responses. Transportation Research Record: Journal of the Trans-
portation Research Board, 2095(1), 53–61.
Zopf, C., Wollny, I., & Kaliske, M. 2015. Numerical modelling of tyre–pavement-interaction phenom-
ena: constitutive description of asphalt behaviour based on triaxial material tests. Road Materials &
Pavement Design, 16(1), 133–153.
906
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: This paper reviews vibration data analysis and relates them to the deter-
mination of pavement failure modes as a roughness assessment method. Pavement rough-
ness directly influences ride comfort and surface distortion. All previous studies examined
roughness from both points of view and correlated it with acceleration data, gathered either
by conventional (accelerometers) and unconventional methods (smartphone applications,
mobile sensor networks). The paper compares the pros and cons of current methods used for
pavement condition assessments. Road tests using three different sizes of vehicle were con-
ducted to provide and supplement knowledge of the potential of using alternative methods.
While the research is ongoing, the paper provides current findings in a road test and recom-
mends future needs for improvements.
907
the users. Another definition of roughness is “the deviation of a surface from a true planar
surface with characteristic dimensions that affect vehicle dynamics and ride quality” [Fed-
eral Highway Administration, 2005]. Agencies and transportation specialists often convert
roughness into an index such as the Present Serviceability Index (PSI) or the International
Roughness Index (IRI). The paper compares the pros and cons of current methods used for
pavement condition assessments as well as analyses vibration data as a means of pavement
roughness evaluation.
909
Table 1. (continued)
Name & Authors Equipment
Lakusic et al. [2011] shows that the system encounters difficulties when registering irregularities
in the direction of the vehicle, such as longitudinal cracking, rutting on pavement.
Tests done with smartphone-based accelerometers also present errors [Islam et al., 2014a]
[Douanghphachanh et al., 2014]. Few studies exist that involve smartphone-based accelerom-
eters, hence the lack of numerous examples. The common method of determining the ampli-
tude values of vibration consists of doubly integrating the acceleration values. However, a
study [Arraigada et al., 2006] that compares lab tests and full scale tests shows differences
between the expected values of vibration (obtained through double integration) and the real
values of vibration. Differences need to be corrected. This leads to the conclusion that the
IRI indices obtained through double integration by the authors don’t correspond to the real
amplitude values.
2 PRESENT STUDY
This study comprises three parts: theoretical part, practical part, analysis part. The theoreti-
cal part consists of the car modeling as a mass-stiffness-damper system used to support the
road test and pavement condition assessment. The second part consists of the road test and
data collection. The third part consists of processing the test results, developing an algorithm
and drawing conclusions.
910
2.1 The theoretical part
The theoretical part consists in developing a car model, shown in Figure 1. The car is mod-
eled as a half-vehicle consisting of a spring-mass-damper system.
911
Figure 4. Toyota Corolla LX (Sedan). Figure 5. Smartphones placements.
this apps were stable and consistent to be used for analysis. The myVibrometer mobile app
exports a.csv file format, which is easy to import into GIS software. The application collects
data points at a rate of 50 points per second on a real time scale including date and time. The
vibration data were reported in units of gravity. The vibration data collected were in X, Y and
Z coordinates. The data collected from myVibrometer also directly provides the longitude
and latitude within the same.csv file. Each data set can be quickly imported into the mapping
software and would be easy to scale up the data sourcing and provides easy data management.
Based on output files, the myVibrometer app records and exports the following data: vibration
in X, Y and Z coordinates (g), heading (degrees), course (degrees), speed (m/s), altitude (m),
latitude (degrees), longitude (degrees), and date and time (Table 2). The comprehensive data
set makes this application a good candidate for use in this project (Ho et al. 2016).
Vibration data were recorded in a gravity of acceleration unit. These data were imported
from Excel to a programed Matlab using double integral coding to obtain the displacement
values. We exported in back to Excel for plotting and manipulation. The differences in plot-
ted lines lengths are influenced by time. We counted and considered 50 readings per second.
Because the first set of readings/second was not always complete (it would contain for exam-
ple 17 or 40 readings) it was eliminated for precision purposes.
912
Table 2. Example of vibration data input in a spreadsheet.
Figure 7. Displacement results for Toyota. Figure 8. Displacement results for Buick.
Figure 9. Displacement results for Ford. Note: all vertical values are expressed in meters while all
horizontal values are shown in second..
Figure 7 and Figure 9 show the displacement data for the left lane of the Beaver street in
Flagstaff Arizona where the smartphone oriented longitudinal to the road (as shown in pre-
vious figured, the smartphone attached to the armrest. Three tests were done with each car
on the left lane. The average values for each vehicle were taken and plotted against each other
in order to see how damping affects the results. All vertical altitude values shown in Figure 7
to 9 are calculated in meter along with time unit (in second). It is significant to observe how
a dumping system of individual vehicle influences the “pattern” of vibration responses along
on the same road. The dumping system of Corolla show consistent among three road tests.
The Buick’s dumping system has also displays close agreement among the three road tests.
913
The first two vibration altitudes collected by the Truck are close but are a bit away from the
third collection. This difference might be due to the aging problem of the vehicle (purchased
in 2001) that the dumping system is getting old and need to be repaired if it will continuously
be used for road test.
4 CONCLUSIONS
While this project is still ongoing, the paper is presented to provide updated information with
what methods have been implemented in the US. No conclusions can be drawn at this time, but
the following steps will be enunciated. The displacements will be calibrated and a formula for
the differences due to the car damping system will be found. The road profile will be plotted
in a special software. Data will be compared with DOT data about the same roads and will be
adjusted accordingly. Conclusions will be drawn about: the smartphone orientation influence,
the car damping system influence, the altitude influence on vehicle vibrational responses.
REFERENCES
B. Yu, X. Yu(2006, August) Vibration-Based System for Pavement Condition Evaluation [Online].
Available: https://www.researchgate.net/publication/268594808_Vibration-Based_System_for_Pave-
ment_ Condition_Evaluation
Buick website [Online]. Available: http://www.buick.com/regal-mid-size-luxury-sedan.html
Federal Highway Administration(2005, May). HPMS Field Manual[Online]. Available:
https://www.fhwa.dot.gov/ohim/hpmsmanl/pdf/appe.pdf
C.H. Ho, G. Slim, J. DeGeyter, S. Monahan.(2016) Application of Geographic Information Systems
and Vibration Mobile Apps in Road Condition Assessment of Bike Trails. The Annual Meeting of
the Transportation Research Board, Washington, D.C. January, 2016.
Ford website [Online]. Available: http://www.ford.com/trucks/superduty/trim/f250xl/?searchid = 217072
954%7C13433996794%7C2720235141&ef_id = VZYTjQAAAJX-SjXT:20151023031405:s
J. Eriksson et al.(2008, June). The Pothole Patrol: Using a Mobile Sensor Network forRoad Surface Moni-
toring [Online]. Available: http://www.ece.northwestern.edu/∼peters/references/PatholePatrol.pdf
L. Gregg, W. Foy( 1955, January) Triaxial acceleration analysis applied to the evaluation of pavement
riding qualities [Online]. Available: http://uknowledge.uky.edu/cgi/viewcontent.cgi?article = 2273&co
ntext = ktc_researchreports
M. Arraigada, M. Partl(2006, March). Calculation of displacements of measured accelerations, analysis
of two accelerometers and application in road engineering [Online]. Available: http://www.strc.ch/
conferences/2006/Arraigada_Pratl_STRC_2006.pdf
R. Barbosa(2011, September). Vehicle Dynamic Response Due to Pavement Rough-
ness Soc. Mech. Sci. & Eng. [Online]. vol.33 no.3 Available: http://www.scielo.br/scielo.
php?script = sci_arttext&pid = S1678–58782011000300005
S. Islam et al.(2014, May). Use of Cellphone Application to Measure Pavement Roughness [Online].
Available: https://www.researchgate.net/publication/269126333_Use_of_Cellphone_Application_to_
Measure_Pavement_Roughness
S. Islam, W. Buttlar et al.(2014, June). Measurement of Pavement Roughness Using Android-Based
Smartphone Application [Online]. Available: http://cedb.asce.org/cgi/WWWdisplay.cgi?319307
S.Lakusic, D. Brcic, V. Lakusic(2011, November). Analysis of vehicle vibrations—new approach
to rating pavement condition of urban roads [Online]. Available: https://www.researchgate.net/
publication/269786651_Analysis_of_Vehicle_Vibrations__New_Approach_to_Rating_Pavement_
Condition_of_Urban_Roads
Texas Department of Transportation(2011, January). Manual Pavement Design Guide.[Online]. Avail-
able: http://onlinemanuals.txdot.gov/txdotmanuals/pdm/introduction_ch1.htm
Toyota website [Online]. Available: http://www.toyota.com/corolla/
V. Douanghphachanh, H.Oneyama(2014, December). A Study on the Use of Smartphones
for Road Roughness Condition Estimation [Online]. Available: http://link.springer.com/
article/10.1186%2F1687–1499–2014–114
914
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Pekka Eskelinen
Department of Electrical Engineering and Automation, Aalto University, Finland
1 INTRODUCTION
Important road pavement characteristics include, but are not limited to, its load carrying
capacity and durability regarding climatic extremes and traffic. On the other hand, already
limited construction funding forces public authorities and private companies to some com-
promises. Standards and methods have been developed and adopted to maintain a reason-
able balance between these factors (PANK 2011). One typically used evaluation criterion is
the air void content, in other words the compaction degree, of the pavement which is often
judged by measuring the real part of permittivity by ground penetrating radar (GPR).
Several studies have been published in recent years related to pavement quality evaluation,
see for example Saarenketo & Scullion (2000) and Fauchard et al. (2015). We want to bring
new views to this topic, first of all by introducing comparisons of different methods also
against our microwave radar and, secondly, by measurements made on a real highway site,
contrary to often published results from dedicated test areas (Leng et al. 2011).
The aim of this paper is to estimate the suitability of GPR and microwave radar meas-
urements in density evaluation of thin pavement layer. This is achieved by comparing the
performance of the commercial 1 GHz GPR against the microwave radar in permittivity
estimation. The microwave radar results are also examined with respect to drill core sample
results including conventional laboratory density measurements and permittivity measure-
ments with vector network analyzer (VNA) scanner configuration.
Non-destructive testing utilized two different radars in a new highway section in southeast-
ern Finland. After the each lift of construction, 104 meters long lines were measured with
our new microwave radar which operates at 12–18 GHz. The frequency range was chosen
915
to obtain small pulse width in time domain. A detailed description of the microwave radar
system is given in Huuskonen-Snicker et al. (2015). The first three asphalt concrete (AC)
layers were constructed in 2014 and the last stone mastic asphalt (SMA) in 2015. The total
time span of the construction was roughly one year. After construction completion the road
section was measured with a commercial 1 GHz GPR system manufactured by GSSI® and
with our microwave radar. Five lines, aligned as shown in Figure 1a, were measured with
the microwave device, but only the line 1 with the GPR. The cross section of the pavement
structure is shown in Figure 1b.
The real part of permittivity from the GPR system data is calculated with modified version
of the method shown in Maser & Scullion (1991). The ground reflection signal is divided in
two halves and two permittivity values are calculated from each half against the metal refer-
ence. Schematic figure of the permittivity extraction is shown in Figure 2a. For the microwave
Figure 1. a) Location of test lines and core samples on the road. Only line 1 was measured with the
conventional GPR. 150 mm diameter drill core samples are marked with L-1, and L-2 and 100 mm
diameter drill core samples with S-1 to S−8. b) Cross-section describing the road structure. Drill core
subsample ids marked for ACbase1, ACbase2, and 2nd AC-layers.
Figure 2. a) Schematic chart describing the signal split into first and second halves and extraction of
peak to peak value with pulse GPR radar. b) Schematic chart describing extraction of the maximum
peak with the microwave radar.
916
Table 1. Test methods for drill core samples.
L-*-ABC X X
L-*-AB X X
L-*-A X X X X
L-*-B X X X X
L-*-C X X X X
S-**-A X X X
S-**-B X X X
S-**-C X X X
*1,2
**1–8.
radar the signal is transformed to time domain and the permittivity is calculated from the
ground reflection peak against a reference material. The schematic chart of the process is
shown in Figure 2b.
Two drill core samples (D = 150 mm) going through all AC pavement layers were taken at
selected spots of the lines for further analysis in the laboratory. The average sample thickness
was 185 mm, consisting of three successive asphalt layers. Both samples were studied with
our VNA-based microwave scanner, which covers 7–17 GHz frequency range (Pellinen et al.
2015). Each sample was scanned five times across a rectangle of 50 mm by 50 mm, first as
taken from the road, then by sawing the successive asphalt layers away one by one. In addi-
tion, another eight drill core samples (D = 100 mm) were taken through all AC pavement lay-
ers. Conventional laboratory measurements were performed to all drill core samples to define
their physical properties, e.g. bulk and maximum densities. From these measurements, the air
void content of each layer was calculated (SFS-EN 12697–5 2010 & SFS-EN 12697–6 2010).
Various tests and methods are summarized in Table 1.
3 RESULTS
Typical permittivity recordings with the two radars are shown in Figure 3a. The permittivity
distribution estimation of the same data obtained with kernel density estimation (Figure 3b).
The commercial GPR system with the permittivity calculated from the first half of the signal
has the average permittivity of 4.76 when the 12–18 GHz radar measurements have the aver-
age permittivity of 3.94. The standard deviations are 0.23 and 0.76, for GPR and microwave
radars respectively. Based on the second half of the GPR signal, the average permittivity is
5.42 with the standard deviation of 0.41. The average and the standard deviation are higher
compared to the result from the first half of the signal.
The general trend based on the microwave radar measurements shows that the ACbase1-
layer has the lowest permittivity values with average permittivity of 3.4, and the average of
the permittivity values increased to 4.5–4.8 for the rest of the AC-layers. The summary of all
radar measurements are shown in Table 2.
The measurements with VNA show that the ACbase1-layer has the highest average permit-
tivity of 5.53, and the average standard deviation 0.74. The permittivity of the ACbase2-layer
has an average permittivity value of 5.27 which is lower than for the first layer. The standard
deviation on average is 0.74 which is equal compared to the ACbase1-layer. The 2nd AC-layer
has the lowest average permittivity with the value of 5.23, and with the standard deviation
of 0.86.
The conventional laboratory measurements show that on the driving lane the ACbase1-layer
has on average the highest bulk density, 2.396 Mg/m3. The 2nd AC-layer has the lowest bulk
917
Figure 3. a) Permittivity of line 1 with SMA-layer measured with the commercial GPR and 12–18
GHz microwave systems, b) Kernel densities of permittivities measured with GPR and 12–18 GHz
microwave systems. GPR permittivity values are calculated from the first half of the signal.
918
Figure 4. The kernel density plots calculated for each pavement layers shows the general trend in per-
mittivity results obtained with the VNA scanner method.
density, 2.307 Mg/m3, and the ACbase2-layer has the bulk density between the ACbase1 and 2nd
AC-layer, 2.335 Mg/m3. On the passing lane, the highest bulk density is with the 2nd AC-
layer, on average 2.380 Mg/m3, and the ACbase2-layer has density on average 2.345 Mg/m3.
Figure 4 shows kernel density plots of the average real part of permittivity for each pave-
ment layer measured with the VNA scanner method. The general trend in VNA measure-
ments is that the ACbase1-layer has the highest permittivity values and the lowest permittivity
values are encountered with the 2nd AC-layer. A summary of all measured results for the drill
core samples are given in Table 3.
4 DISCUSSION
The permittivity values measured with GPR differ from the results measured with the micro-
wave radar. Both methods, the first and the second half of the signal, for the permittivity
estimation from the GPR data show higher results than the permittivity obtained with the
microwave radar. This effect is probably caused by the interference of two reflections, the
surface and the deeper reflections. This phenomenon may occur when the signal pulse width
is more than two times the layer thickness (Annan 2003). In microwave radar measurements,
the pulse width is small which means that the upper layer of the material is dominating in the
measurement results.
The permittivities measured with microwave radar and VNA-scanner show inverted rela-
tionship. The ACbase1-layer has the lowest average permittivity when results from the micro-
wave radar are used but on average have highest permittivities. The highest permittivities
from the microwave radar measurements are with the ACbase2-layer when the drill core sam-
ples indicate that the average permittivities are between the highest and lowest results.
Drill core samples show that on average the permittivity results from VNA measurements
and the densities obtained with conventional laboratory measurements are agreeing with
each other (Figure 5). The air void contents calculated based on the maximum and bulk
919
Table 3. Drill core results including VNA scans.
* Maximum density is calculated from subsamples as a weighted mean based on the relative volume.
** Air void contents are calculated based on the average densities.
densities on average are not following clear relationship with VNA permittivity measure-
ments. The results contain a lot of variability which affects the conclusion about the exact
relationship between the permittivity and the bulk and maximum densities.
The unknown degree of compaction impose another problem when analyzing different
measurements. This concerns especially results of ACbase1-layer. Successive pavement layers
were measured with microwave radar soon after construction, but the drilling through the
entire pavement was months after the ACbase1-layer was constructed. This may make mean-
ingful comparisons of radar-based results and conventional laboratory measurement results
impossible.
920
Figure 5. a) Relation between average bulk density measured with conventional laboratory method
(SSD) and permittivity measured with VNA method. b) Relation between average maximum density
measured with conventional laboratory method and permittivity measured with VNA method.
5 CONCLUSIONS
In this paper, we have shown permittivity results obtained with 1 GHz commercial GPR,
microwave radar and VNA scanner. In addition, conventional laboratory density measure-
ments have been presented. Based on the analysis, the pavement bulk density estimation based
on the permittivity is not reasonable. The commercial 1 GHz GPR is showing higher permit-
tivity estimates on thin layer pavements compared to the microwave radar. This is probably
due to inference of two signals from different structural layers. The variations between differ-
ent drill core samples in measured bulk and maximum densities lead to inaccurate estimates
in bulk densities and air void contents of different layers with small amount of samples. Due
to the results presented in this paper, a new statistically based approach is needed to assess
the quality of asphalt pavement layers with radar.
ACKNOWLEDGEMENTS
Authors would like to thank the Finnish Transport Administration for funding this research.
REFERENCES
Annan, A.P. 2003. Ground penetrating radar principles, procedures & applications. Sensors & Software
Inc.
Fauchard, C & Beaucamp B. & Laguerre L. 2015. Non-destructive assessment of hot mix asphalt com-
paction/density with a step-frequency radar: case study on a newly paved road. Near Surface Geo-
physics 13(3): (289–297).
Huuskonen-Snicker, E & Eskelinen P. & Pellinen T. & Olkkonen, M.-K. 2015. A New Microwave Asphalt
Radar Rover for Thin Surface Civil Engineering Applications. Frequenz 69(7–8): (377–381).
Maser, K.R. & Scullion, T. 1991. Automated Detection of Pavement Layer Thicknesses and Subsurface
Moisture Using Ground Penetrating Radar. TRB paper.
PANK. 2011. Päällystealan neuvottelukunta ry. Asfalttinormit. Finnish Pavement Technology Advisory
Council PANK. Finnish Asphalt specifications, PANK ry (in Finnish). Helsinki: Edita.
Pellinen, T. & Eskelinen, P. & Huuskonen-Snicker, E. & Hartikainen, A. 2015. Assessment of air void
content of asphalt using dielectric constant measurements by GPR and with VNA. Aalto University
publication series SCIENCE + TECHNOLOGY. http://urn.fi/URN:ISBN:978-952-60-6288-4 (refer-
enced 1-12-2015).
921
SFS-EN 12697-5. 2010. Bituminous mixtures. Test methods for hot mix asphalt. Part 5: Determination
of the maximum density. Helsinki: Finnish Standards Association SFS.
SFS-EN 12697-5. 2010. Bituminous mixtures. Test methods for hot mix asphalt. Part 6: Determination
of bulk density of bituminous specimens. Helsinki: Finnish Standards Association SFS.
Saarenketo, T. & Scullion, T. 2000. Road Evaluation with ground penetrating radar, Journal of Applied
Geophysics 43: (119–138).
Zhen Leng, Z. & Al-Qadi, I. & Lahouar, S. 2011. Development and validation for in situ asphalt mixture
density prediction models. NDT & E International 44(4): (369–375).
922
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: In this investigation, the axle load and annual average daily traffic data for
three National Highways (NH 5, NH 58 and NH 79) in India were collected. Analysis of the
data revealed that considerable amount of overloading exist in these highways. It was found
that while the gross vehicle weight were within the limits, the rear axles were overloaded. Sta-
tistical analysis was carried out on all the axle load data and it was seen that a mixed normal,
lognormal and Weibull distribution can fit the data. Using the moment statistics of the axle
load distribution, the load spectra factor was calculated for various axle types. The sensitiv-
ity of overloading in the calculation of equivalent single axle load was quantified using the
vehicle damage factor and truck factors. It was also observed that the efficiency and accuracy
of LSF was substantially higher to consider it as an alternative input for pavement design.
1 INTRODUCTION
In recent years, India has been making strides towards developing a robust mechanistic—
empirical pavement design procedure. While advances have been made in terms of material
characterization, alternate pavement cross sections and distress transfer functions for differ-
ent levels of reliability, analysis of traffic and axle load data is lagging and hence need greater
impetus. The concept of equivalent axle load factor along with vehicle damage factor is still
being used in India. Unlike USA and Europe, where considerable amount of traffic and
axle load data exists, the database related to axle load distribution is not available for pave-
ment design and evaluation purposes and such data are indeed collected only during a new
pavement construction project. As far as India is concerned, the standard practice related to
vehicle dimensions and weights is prescribed through IRC:3-1983 (1983) and IRC:37-2012
(2012) provides the guidelines related to bituminous pavement design.
It will be useful if one could use methodologies to handle the large amount of axle load
data while carrying out pavement design. In this paper, one such approach is outlined. The
work presented here is part of the ongoing investigation on developing warrants for use of
modified binders in bituminous pavement construction sponsored by the Department of Sci-
ence and Technology, Government of India.
2 LITERATURE REVIEW
The studies conducted by Salama et al.,(2006) stated that the magnitude of damage was
observed to be dependent on the axle type whereas Weissmann et al., (2013) stated that the
magnitude of damage depends on vehicle class as well. Haider et al., (2007) outlined a meth-
odology to estimate individual axle load spectra for various axle configurations using truck
weights and their proportions on the highway. In the study by Prozzi and Hong (2007), the
axle load distribution for the axle types such as tandem, single axle with dual wheels and
steering axles were fit using mixed log normal distribution. Timm et al., (2005) developed
923
a linear combination of lognormal and normal distributions to characterize the complex
axle load spectra. The studies conducted by Turochy et al., (2005) asserted the significant
difference in the observed axle load distribution between each month and the annual aver-
age, between directions at a given site and between each site and a combination of all sites.
Fuentes et al., (2012) have shown that different classes of vehicles have different effect of
overloading reflected in the truck factor. Also the truck factor calculated for a specific site
was considerably different from the historic data available for the same site.
Studies performed on overloading have considered the vehicles to be overloaded when at
least one of its axles is overloaded (Pais et al.,2013, Fuentes et al., 2012). Pais et al., (2013) in
their study have found the influence of overloading on specific axles on truck factor. Almedia
et al., (2015) developed a methodology to estimate the truck factor considering vehicle pave-
ment interaction.
3 DATA COLLECTION
The traffic and axle load data for three National Highways (NH) are selected for analysis
in this investigation. NH 5 connects Jharpokharia in Odisha and Chennai in Tamil Nadu
(Figure 1(a)). This highway is 1533 km long. The traffic and axle load data for NH 5 was
collected at Kallaparu toll plaza in the south bound direction. The 538 km long NH 58 high-
way runs from New Delhi in Uttar Pradesh to Mana Pas in Uttarakhand (Figure 1(b)). The
traffic and axle load data for NH 58 was collected at Shivaya toll plaza in the south bound
direction. The shortest among the three highways considered in this investigation is NH 79,
connecting Ajmar in Rajasthan and Indore in Madhya Pradesh which is of length 500 km
(Figure 1(c)). The traffic and axle load data for NH 79 was collected at 189th km in the south
bound direction. NH 5 is a part of golden quadrilateral, which is a highway network con-
necting the India’s major industrial, agricultural and cultural centers. The north side of NH
58 is a part of border road near Indo–Tibet border and Indian army is one of its major users.
On the south side of NH 58, we have the national capital city (New Delhi) and this portion
also connects the religious pilgrim centers. NH 79 is in the middle part of India and it covers
many industrial, commercial and tourist places. The axle load survey was carried by allowing
the randomly selected vehicles to pass over the pressure pads. The traffic and axle load survey
details for the three National Highways are shown in Table 1.
4 DATA ANALYSIS
924
Figure 1. National Highways (www.mapsofindia.com).
axles. It is also very clear that the overloading analysis shown in Table 3 does not reveal the
extent of overloading for each of the vehicle types. In such scenario, parameters such as
average overloading ratio (AOLR) and maximum overloading ratio (MOLR) were used to
compare the overloading nature of different highways (Wen et al., 2005). Here, AOLR is the
mean of the overloading ratio (ratio of actual weight of vehicle to the permissible weight of
vehicle (Table 2)) and MOLR is the maximum of the overloading ratio. Table 4 shows the
925
Table 2. Vehicle classification and their legal axle loads.
Axle analysis
Axle type NH 5 NH 58 NH 79
Steering axle (SS) 49 42 53
Single axle dual wheel (SD) 49 25 47
Tandem axle 80 84 93
Tridem axle 80 100 76
Vehicle analysis
Vehicle type NH 5 NH 58 NH 79
Class 1 73 20 56
Class 2 74 95 94
Class 3 59 100 62
Class 5 100 – –
Class 6 80 – –
Class 7 75 33 55
Class 8 90 – –
AOLR and MOLR comparison of different highways for four different vehicle types only.
This is because these are the only four vehicle types that are present in common in all the
three highways considered here. NH 79 was found to be less overloaded compared to NH 5
and NH 58. From Table 3, one can observe that the vehicle type ‘class 3’ is 59% overloaded
in NH 5,100% overloaded in NH 58 and 62% overloaded in NH 79. However, when the same
vehicle class 3 is considered in Table 4, one can see that the order of overloading for the high-
926
Table 4. Maximum and average overloading characteristics.
AOLR MOLR
NH
Vehicle type NH 5 NH 58 NH 79 NH 5 NH 58 79
ways considered here has changed. Thus it is important to notice that overloading analysis
based on mere axle and vehicle class alone cannot alone reveal the overloading nature on
these highways. It is now required to see whether such overloading can truly be captured by
the appropriate axle load analysis procedures currently followed.
VDF =
∑ EALL i
,
Total no. of
o vehicles (2)
where, EALF (SS) is the equivalent axle load factor for ‘SS’, ∑EALi represents the sum of the
EAL (Equivalent Axle Load) of all axle types and EAL represents the product of EALF and
the frequency corresponding to each load group. Table 5 shows the VDF values obtained for
different highways for with (actual axle loads) and without overloading (the overloaded axles
were limited to legal axle limits) condition. It was noticed that when the overloaded axles
were limited to legal axle limits, the reduction in VDF values were substantially higher for
NH 58 and followed by NH 5.
4.3.2 Calculation of TF
Unlike VDF, TF is calculated for each vehicle class based on their axle configuration. The
vehicles were grouped under the respective vehicle types as per Table 2. The TF for each
vehicle was calculated separately by summing up the EALF values of the axles in a vehicle.
The average of these TF values was considered as the TF for that particular vehicle type. In
this paper the TF for ‘With overloading’ and ‘Without overloading’ conditions were found
out for the three National Highways and is shown in Table 6. A significant reduction in the
truck factor value (Table 6) was observed for NH 58 (class 2) and NH 5 (class 6) when the
overloaded axles were reduced to legal axle limits.
927
Table 5. VDF values.
Loading condition NH 5 NH 58 NH 79
With overloading 18 21 4
Without overloading 3 2 2
Vehicle NH 5 NH 58 NH 79 NH 5 NH 58 NH 79
Class 1 15 5 3 4 1 2
Class 2 3 61 5 1 3 2
Class 3 14 15 3 1 2 2
Class 5 30 – – 4 – –
Class 6 50 – – 3 – –
Class 7 21 9 6 3 3 3
Class 8 7 – – 3 – –
Using these VDF and TF values, the ESAL values for three National Highways were calcu-
lated for an annual growth rate of 5%, design life of 10 years, a directional distribution factor
of 50% and a lane distribution factor of 75% using equation (3) (IRC:37-2012, 2012).
ESAL =
365 × (( + r) − )
n
×A D×F, (3)
r
where, ‘ESAL’ is the equivalent standard axle load, ‘r’ is the annual growth rate of commer-
cial vehicles in decimal, ‘n’ is the design life in years, ‘A’ is the Average Annual Daily Traffic
(AADT),’D’ is the lane distribution factor and ‘F’ is the VDF or TF or load spectra factor
(LSF, to be discussed in the section 4.3.4). Table 11 shows the ESAL values for ‘with over-
loading’ and ‘without overloading’ conditions by VDF and TF method for the three National
Highways. The ESAL values calculated using VDF and TF were found to be almost similar
for both the loading conditions.
nth moment
−μn+
σ 2 n2
⎛ n⎞
e 2 λ nΓ ⎜1+ ⎟ ,
⎝ k⎠
cannot reject the null hypothesis that the sample belong to fitted distribution at any signifi-
cance level (α) less than or equal to p.
Table 9 shows the parameters for mixed normal/log-normal distributions fitted for multi-
modal data. In Table 9 weight ‘w’ is explained as follows: In a mixture distribution, more than
one distributions are combined; the probabilities associated with each component is known
as weights/mixture weights.
929
Table 9. Multi-modal fit parameters.
Parameters
σ (Standard
Description Number of modes W (Weight) Μ (Mean) deviation)
∑
I
x q it
i =1 1
m
LSF= , (4)
m
L
s
where, ‘t’ is the bin width, ‘qi’ is the normalized frequency of the axle load in the ith bin of a
given axle load spectrum, xi is the load associated with the ith bin and Ls is the standard axle
load for a particular type of axle. It is convenient in a rigorous pavement design process to
directly calculate the damage using such distributions. For instance, the LSF can be deter-
mined using equation (5) (Prozzi and Hong, 2007)
m
⎛ x⎞ Mm
LSF=∫ ⎜ ⎟ f(x) dx= m , (5)
⎝ Ls ⎠ Ls
where, f(X) is axle load statistical function for a given axle load type ‘Ls’ and ‘Mm ‘ is the mth
order moment of the distribution. It is now shown easily that the two moment statistics can
be used to estimate the damage associated with the axle load spectra. The fact that one could
use non-integer case lends credibility to the analysis presented here. Table 10 shows the cal-
culated LSF values. Similarly, extending the concept discussed in equation (5) to multimodal
distributions, the LSF can be derived as given in equation (6).
m
⎛ x⎞ k
⎛ Mk m ⎞
LSF=∫ ⎜ ⎟ f(x) dx = ∑ ⎜ w k m ⎟ , (6)
⎝ Ls ⎠ k =1 ⎝ Ls ⎠
where, ‘wk’ is the weight of the kth mode; ‘k’ is the total number of modes and ‘Mkm’ is the
mth moment of the kth mode.
From Table 11, one can observe that the ESAL values calculated using VDF and LSF
method for NH 5 and NH 58 is found to be similar. However, the ESAL values calculated
using TF method for ‘with overloading’ condition of the same highways is found to be less
compared to that of other two methods. While for NH 79, which is less overloaded, the ESAL
values obtained are almost similar irrespective of the method of calculation. This is due to
the fact that the truck factor takes into account the gross vehicle weight unlike axle loads by
VDF and LSF and hence it does not exhibit the required sensitivity to overloading.
930
Table 10. LSF values.
NH 58 SS 2.2
SD 3.45
Tandem 47.5
NH 5 SS 2.11
SD 7.19
Tandem 12.82
NH 79 SS 0.73
SD 3.14
Tandem 4.44
ESAL (msa)
Method of
calculation Loading condition NH 5 NH 58 NH 79
5 SUMMARY
This paper focused on the analysis of axle load and AADT data collected from three
National Highways in India. It was found that a large number of vehicles with chassis
configuration not listed in IRC:3-1983(1983) are plying on these highways. Detailed axle
load analysis revealed that considerable amount of overloading exists. The ESAL values
were determined using vehicle damage factor, truck factor and load spectra factor. Influ-
ence of overloading was found to be more in the case of vehicle damage factor compared
to truck factor.
Statistical modeling of the axle load data was carried out. Using such analysis, the moments
were calculated and used to find the ESAL values. The load spectra factors calculated using
the moment statistics were found to be more sensitive to overloading than the vehicle dam-
age factors.
In the complex computation in the pavement design, it will be useful to have the axle load
data represented in the form of statistical distribution since the computation of the damage
can be performed easily. New highways can also be designed easily if one knows the axle load
distribution of a similar type of highway. The statistical parameters of the axle load data can
be used as inputs in the MEPDG design as well. This paper presented the first approach for
axle loads in Indian National Highways.
ACKNOWLEDGEMENTS
The authors thank Department of Science and Technology for funding this investigation.
The grant number is DST/TSG/STS/2011/46.
The authors thank M/s V.R. Techniche, Delhi and L&T IDPL, Chennai for sharing the
traffic and axle load data.
931
REFERENCES
Almeida, A.M.M. & Santos, L.G.P. 2015. Methodological framework for truck-factor estimation con-
sidering vehicle-pavement interaction. ASCE journal of transportation engineering. 141(2): 10.1061/
(ASCE)TE.1943–5436.0000726.
Fuentes, L. G., Macea, L.F., Vergara, A., Flintsch, G. W., Alvarez, A. E., & Reyes, O. J. 2012. Evaluation
of truck factors for pavement design in developing countries. Procedia-Social Behavioral Science. 53:
1140–1149.
Haider, W.S. & Harichandran, R.S. 2007. Relating axle load spectra to truck gross vehicle weights and
volumes. ASCE Journal of Transportation Engineering.133(12): 696–705.
Huang, Y.H. 2009. Pavement Analysis and Design. 2nd Edition, Pearson Prentice Hall. New Jersey.
IRC:3–1983.1983. Dimensions and weights of road design vehicles. Indian Road Congress. New Delhi,
India.
IRC:37-2012. 2012). Guidelines for the design of flexible pavements. Indian Roads Congress. New Delhi,
India.
Johnson, R.. 2005. Chapter 5: Probability Densities. Miller and freund’s probability and statistics for
Engineers, 7th Edition, Prentice-Hall of India, New Delhi. 166–175.
Pais, J. C., Amorim, S. I. R., & Minhoto, M. J. C. 2013. Impact of traffic overload on road pavement
performance. ASCE Journal of Transportation Engineering. 139(9):873–879.
Prozzi, J. A. & Hong, F. 2007. Optimum statistical characterization of axle load spectra based on load-
associated pavement damage. International Journal of Pavement Engineering. 8(9): 323–330.
R Development Core Team. 2011. R: A language and environment for statistical computing, R founda-
tion for statistical computing. URL: http://www.R-project.org (visited 1st Oct. 2015).
Salama, H.K., Chatti, K., & Lyles, R.W. 2006. The effect of heavy multiple axle trucks on flexible
pavement damage using in- service pavement performance data. ASCE Journal of Transportation
Engineering. 132(10): 763–770.
Timm, D.H., Tisdale, S.M., & Turochy, R.E. 2005. Axle load spectra characterization by mixed distribu-
tion modelling. ASCE Journal of Transportation Engineering. 131(2): 83–88.
TRB.2002. Regulation of weights, lengths, and widths of commercial motor vehicles, committee
for the study of the regulation of weights, lengths, and widths of commercial motor vehicles, special
report 267, USA.
Turochy, R.D., Baker, S.M., & Timm, D.H. 2005. Spatial and temporal variations in axle load spectra
and impacts on pavement design. ASCE Journal of Transportation Engineering. 131(10): 802–808.
Weissmann, A. J., Weissmann, J., Papagiannakis, A., & Kunisetty, J. L. 2013. Potential impacts of
longer and heavier vehicles on Texas pavements. ASCE Journal of Transportation Engineering.
139(1):75–80.
Wen, H., Peng, J., Xu-hong,L. & Jie, H. 2005. Site survey and analysis of highway trucks overloading
status Quo in Anhui. Journal of the Eastern Asia Society for Transportation Studies. 6: 1790–1803.
932
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
933
2 SCOPE AND OBJECTIVES
3 LITERATURE REVIEW
An attempt to develop the climatic data files for Egypt was conducted by Elshaeb et al.
2014. Average climatic data was collected from a weather data website and missing hourly
data was interpolated accordingly. Using the interpolated data, the study shows that pave-
ment performance was significantly affected by the change in the climatic data and the
most significant influence was on the predicted rutting of the asphalt layer. No signifi-
cant change in the predicted alligator fatigue cracking was found for the different climatic
locations.
A study conducted by Breakah et al. 2011 investigated the influence of using accurate
weather site specific data had on pavement performance in Iowa State, USA. County-specific
climatic data was collected and the performance of three pavement cross sections was pre-
dicted using MEPDG and results using different climate data sources were compared. The
study recommended the usage of more comprehensive climatic data for use in MEPDG.
A comparative study was performed by The University of Maryland, USA. Five sections
representing different US climatic conditions were designed by A93 design procedures using
each state’s local data for three levels of traffic. Sections were then evaluated using MEPDG
and pavement performance was analyzed. The study concluded that subgrade stiffness had no
impact on AC layer calculation using the A93 design procedures and AC layer is a function of
base layer properties only. Results showed that states located in warm regions showed worse
performance than those in mild to low temperature regions. The study finally concluded that
although the same serviceability was achieved for the different pavement sections using the
A93 pavement design procedure, different performance for these sections were predicted using
the MEPDG (Carvalho & Schwartz 2006).
934
4.2 Status
The MEPDG has an available database of more than 800 weather stations covering the
United States and Canada. For each weather station, required weather data is provided on
hourly basis for approximately 60 to 66 months. No such detailed data is available for weather
stations in Egypt for use in the MEPDG.
4.3 Approach
The goal is to acquire the hourly weather data for Egypt’s major cities for the preparation
of hourly climatic data (hcd) files. Acquired data is to be sorted, checked for accuracy,
formatted and validated to be accepted by MEPDG as an input. A user friendly computer
application (Weather ME©) is developed under this study especially for this process. Raw
weather data on hourly basis is obtained from the internet from www.wunderground.com.
www.wunderground.com provides detailed real-time weather information for locations all
over the world. The website was firstly originated 1991 at the University of Michigan and
is run by a group of academic and professional meteorologists. In 1995 the website was
launched on the internet and in 2012 it became part of The Weather Channel Companies.
(wunderground 2013).
Weather data available on www.wunderground.com are reported in Aviation Routine
Weather Report (METAR) format. METAR code is regulated by the World Meteorological
Organization (WMO) in consort with the International Civil Aviation Organization (ICAO).
It is predominantly used by pilots in fulfillment of a part of a pre-flight weather briefing, and
by meteorologists, who use aggregated METAR information to assist in weather forecasting.
METAR contains a report of wind, visibility, runway visual range, present weather, sky con-
dition, temperature, dew point, and altimeter setting (OFCM 2005).
935
4.6 Data validation
A comparative approach was adopted to validate the generated files. Weather ME© was used
to create an hcd file for a weather station that exists in the NCDC weather database accom-
panying the MEPDG. Comparison was then conducted between the generated file and the
file within the MEPDG weather database. Weather station 03144 at Imperial County Airport
in California was chosen due to the similarity in weather conditions to Egypt according to
the climatic zone classification (Kottek et al. 2006).
Linear regression analysis was then conducted on the two sets of data. Analysis results for
the five weather parameters range between 0.983 and 1.11 for the slope of the straight line,
0.181 to 3.29 for the intercept and 0.9 to 0.994 for R-squared. Minitab 15 statistical package
was used for the analysis. Values were found acceptable given the unusual large number of
data points for each parameter (approximately 49,000 data point).
5.1 Concept
The two design procedures have major differences in the way pavement sections are designed.
A93, adopted by the Egyptian Code of Practice, produces pavement layers thickness using
an empirical equation derived from the AASHO Road test. On the other hand, MEPDG uses
pavement layers thicknesses and other inputs to predict the pavement performance at the end
of its design life. The predicted performance is then checked against pre-defined performance
criteria. Several iterations of layers characteristics and systems would probably be required to
reach an optimized economic pavement section.
Methodology
Given these major differences, a direct comparison was then found to be unpractical and
instead, the methodology illustrated in Figure 1 was adopted:
– Typical three layered pavement sections were designed using A93 design procedures.
– Performance of each of the designed sections was predicted using MEDPG. Two perform-
ance indicators were observed, rutting and fatigue cracking
– The terminal Present Serviceability Index (PSI) for the designed sections was then calcu-
lated using the predicted IRI.
936
5.2.1 Traffic
Three different traffic conditions were selected for this study, which are low, medium and
high. The different traffic levels were chosen to represent traffic conditions of local, collector
and arterial roads according to Egyptian Code Part 3 functional classifications (ECP No.104
2008).
Values for typical Annual Average Daily Traffic (AADT) and percentage trucks were cho-
sen according to Egyptian Code—Part 6 table 2–3 (ECP No.104 2008).
For the consistency of the comparison, traffic data used for designing the pavement sec-
tions using A93 design procedure must be as close as possible to the traffic input into the
MEDPG software. Given its deficiencies, the Equivalent Single Axle Load (ESAL) concept
applied for traffic data input in A93 design procedure is replaced by Axle Load Spectra
(ALS) in MEPDG. ALS requires more detailed and comprehensive data input requirements.
ESAL to be used for design using A93 is calculated using inputs that would be used for traf-
fic data input in MEPDG. Traffic characteristics used and the calculated ESAL values are
shown in Table 2.
Traffic Levels
937
used in Egyptian Code. Material properties for both Granular Base and Subgrade used for
designing the pavement sections are shown below in Tables 4 and 5.
The MEPDG level 3 input allows for input of MR as a representative (design) value. The
above mentioned values were directly used as inputs for MEPDG.
5.2.3 Environment
The A93 design procedures allow seasonal variations in subgrade strength and layer coeffi-
cient. According to interviews done with pavement designers, contractors and governmental
authorities in Egypt, such consideration is usually not adopted. Accordingly, the subgrade
strength in this study was kept constant during the design life span of the pavement. Litera-
ture shows that the MEPDG was found to be very sensitive to environmental conditions. As
previously mentioned, part of this research is to create a weather database for Egypt for use in
MEPDG. Weather data files are created and used as inputs for the evaluation of pavements.
The weather data hcd file for each city was added to the MEPDG weather database and used
accordingly. As for depth to ground water table level, a typical value of 4 m (13 ft) depth was
used for all locations. Figure 6 illustrates the software input screen showing Egypt’s weather
stations added to the MEPDG software database.
CBR (%) 80
MR (psi) 51,800
Layer coefficient (a2) 0.133
Drainage Coefficient (m2) All cities = 1
Alexandria = 0.8
Poisson Ratio 0.35
Coefficient of Lateral Pressure (ko) 0.5
Type A-2–4
Reliability% R 95%
938
5.3 Sections designed using A93
Values for layers thicknesses, in inches, as designed using A93 are listed in Table 8:
R 5 e(− × IRI )
(1)
where, PSR = Present Serviceability Rating; and IRI = International Roughness Index (m/km)
PSI values are taken to be the same as PSR values due to the close correlation between
both indices (Huang, 2004). Figure 3 shows a summary of terminal PSI values for different
locations and traffic levels in Egypt.
Vehicle operational speed 30,50,60 mph (low, medium, high traffic levels)
IRI (in./mi.)
939
Figure 7 shows that terminal PSI values at the end of the design life for pavement sections
are inconsistent for different locations with the same traffic levels. According to A93, pave-
ment sections are designed to perform up to a design terminal PSI, 2.5 in this case, which
should be close to the actual terminal PSI values. Values for terminal PSI are higher than
the design PSI, which indicates an overestimation of layers’ thicknesses produced by A93. It
was also expected that an equal terminal PSI value would result with the variation of traffic
levels for the same location, where the increase in layer thickness should have counted for the
increased traffic loading on the section.
Results for predicted distresses performance for the designed sections are illustrated in
Figure 4.
Results show that rutting increased with the increase in traffic level and in cities of rela-
tively higher temperature (Aswan, Assiut, Sharm El Sheikh and Hurghada). Fatigue cracking
showed a different trend of lower values for sections designed under medium and high traffic
940
loading. This implies that sections designed under medium and high traffic loading, according
to A93, perform better than those designed under low traffic conditions with respect to fatigue
cracking, i.e. less fatigue cracking is present for high trafficked road sections. It is worth noting
that medium and high traffic loadings have traffic loading conditions relatively high to those
used during the AASHO road test. This may suggest here that A93 design procedure pro-
duces unstable and less reliable results for conditions different than that of the AASHO road
test. It is also worth noting that predicted performance is not the same for different locations
with different environmental conditions. This also implies that A93 predicts performance incor-
rectly, where the MEPDG predictions of distresses for pavements designed using the A93 guide
should also be consistent, regardless of location of the project or traffic conditions.
6.1 General
Environmental conditions are considered as an important input for pavement design and
have a considerable influence on pavement’s performance. Sensitivity analysis conducted
under this study was performed on One-At-A-Time (OAT) basis, where a reference pavement
section was chosen and predicted performance is evaluated under different environmental
conditions. Other design inputs were kept constant. MEPDG performance predictions were
then made for this variation with respect to alligator fatigue cracking and rutting. Reference
values for this study are chosen to be the pavement structure in Cairo under medium traffic
conditions. Sensitivity evaluation criteria adopted by Bayomy, et al. (2012) was used in this
study. The sensitivity of measures of effectiveness for each parameter is evaluated according
to the Distress Ratio (DS) between the highest and lowest value of the evaluated distress;
DS = Highest Distress Value/ Lowest Distress value, shown in Table 10.
Sharm
Aswan Alexandria Hurghada Cairo Assiut El Sheikh
DS Sensitivity
941
7 CONCLUSIONS
The following findings and conclusions were made from this study:
– Weather data hcd files for six major Egyptian cities representing Egypt’s climatic con-
ditions have been prepared. The preparation was done using a computer application
(Weather ME©) developed under this research.
– A93 produces over-designed pavement sections for pavement sections designed for the cit-
ies under investigation. Terminal PSI values for all sections exceeded the targeted design
PSI value of 2.5 used as an input value for designing these sections.
– For traffic conditions beyond those of the AASHO road test, A93 produces inconsistent
results for pavement performance. Terminal serviceability for different sections under the
same traffic loading is not the same or even close to each other, which is in contrary to
what the A93 depicts.
– Relatively low values for fatigue cracking that are predicted using MEPDG for sections under
medium and high traffic loading are believed to enhance the suggestion that A93 produces
over designed pavement sections for conditions beyond those of the AASHO Road test.
– Rutting and fatigue cracking are significantly sensitive to variations in environmental con-
ditions. Being two important performance indicators, this emphasizes the significance of
environmental conditions and its impact on pavement design and assures the necessity and
importance of using accurate input weather data for designing pavement sections.
REFERENCES
Al-Omari, B., & Darter, M. l. (1994). Relationships Between International Roughness Index and Present
Serviceability Rating. Transportation Research Record, 1435, 130–136.
Bayomy, F. M., & El-Badawy, S. M. (2011). Study of the Effectiveness of ITD Pavement Design Method.
Idaho: National Institute for Advanced Transportation Technology.
Bayomy, F., El-Badawy, S., & Awed, A. (2012). Implementation of the MEPDG for Flexible Pavements
in Idaho. Idaho: Idaho Transportation Department. Report No. FHWA-ID-12–193
Breakah, T. M., Khedr, S. A., & Bayoumy, F. M. (2009). Using Film Thickness to Substitute VMA
Criteria for Egyptian Hot Mix Asphalt. Turin, Italy: Sixth International Conference on Maintenance
and Rehabilitation of Pavements and Technological Control (MAIREPAV6).
Breakah, T. M., Williams, R. C., Herzmann, D. E., & Takle, E. S. (2011). Effects of Using Accurate
Climatic Conditions for Mechanistic-Empirical Pavement Design. ASCE Journal of Transportation
Engineering, 137 (1), 84–90.
Carvalho, R. L., & Schwartz, C. W. (2006). Comparisons of Flexible Pavement Designs—AASHTO
Empirical Versus NCHRP Project 1–37 A Mechanistic Empirical. Transportation Research Record:
Journal of the Transportation Research Board No, 1947, 167–174.
ECP No.104. (2008). Egyptian Code of Practice for Urban and Rural Roads—Part 3, Ministerial Decree
No. 369 - Year 2008 (2nd ed.). Cairo, Egypt: HBRC, Ministry of Housing and Development.
ECP No. 104. (2008). Egyptian code of Practice for Urban and Rural Roads—Part 6, Ministerial Decree
No. 369 - Year 2008 (2nd ed.). Cairo, Egypt: HBRC, Ministry of Housing and Development.
Elshaeb, M.A, El-Badawy,S.M, Shawaly E. A. 2014. Development and Impact of the Egyptian Climatic
Conditions on Flexible Pavement Performance. American Journal of Civil Engineering and Architec-
ture, vol. 2, no. 3: 115–121. doi: 10.12691/ajcea-2–3–4.
Huang, Y. H. (2004). Pavement Analysis and Design (2nd ed.). New Jersey: Pearson Prentice Hall.
Khedr, S. A., & Breakah, T. M. (2011). Rutting Parameters for Asphalt Concrete for Different Aggre-
gate Structures. International Journal of Pavement Engineering, 12 (1), 13–23.
Kottek, M., Grieser, J., Christoph Beck, C., Rudolf, B., & Rubel, F. 2006. World Map of The Köppen-
Geiger Climate Classification. Meteorologische Zeitschrift, Vol. 15, No. 3, 259–263.
NCHRP. (2004). Guide for Mechanistic-Empirical Design of New and Rehabilitated Pavement Struc-
tures. NCHRP Project 1–37 A. Champaign, Illinois: Applied Research Associates, Inc.
OFCM. (2005). Federal Meteorological Handbook No. 1: Surface Weather Observations and Reports.
Office of the Federal Coordinator for Meteorological Services and Supporting. Washington, D.C.:
U.S. Department of Commerce/ National Oceanic and Atmospheric Administration (NOAA).
Wunderground. (2013). Weather Underground: About Us. Retrieved October 10, 2013, from http://
www.wunderground.com/about/background.asp
942
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: During the Guss Asphalt (GA) pavement paving on the steel bridge, the uneven
temperature distribution has an adverse effect on the bridge structures. In this research, the
thermal effect of the steel bridge during the GA pavement paving was investigated. Firstly,
a thermal field model of steel box girder was developed based on the monitoring results of
the Tongling Yangtze River Bridge in China, and the thermal field characteristics of the
steel bridge deck were presented. Secondly, the thermal stress and deformation filed of the
steel box-girder at the critical time were investigated. Finally, the sub-model and buckling
mode were used to evaluate the stability of the stiffeners around the paving area under the
critical temperature stress. Results show that during the GA pavement paving, local thermal
deformation occurs in the steel bridge deck, and there is a hidden danger of instability in the
stiffener near the paving edge.
1 INTRODUCTION
The steel box girder is the common structure type for long-span bridges, and the asphalt
pavement is the thin-bedded structure paved on the steel bridge deck. The paving tempera-
ture of asphalt concrete is high (e.g. the paving temperature of Gussasphalt (GA) concrete is
220°C to 240°C), inducing the uneven temperature distributions in the bridge structures. The
uneven thermal field coupled with the high conductivity and anisotropic effect of the steel
bridge deck easily is liable to cause the local thermal deformation in the steel bridge deck, and
thereby affecting the performance and the stability of the bridge deck layered system.
Uneven thermal effects for bridges have always been the focus of many researchers. Mama-
douth and Amin (1983) calculated the 3D transient thermal field of some types of concrete
bridges by the analytic calculation approach. Similarly, Choi et al. (2011) used a theoretical
model of heat transfer and moisture diffusion to analyze the thermal stress of box girder
structure incited by the environmental loadings. However, the analytic calculation approach
is not suitable for long-span bridges. In addition, some researchers investigated the tempera-
ture distribution of bridges based the in-situ monitoring data (Dilger et al., 1983; Ding et al.,
2012). Although the monitoring data are objective to reveal the thermal field characteristics
of bridges, the data acquisition is a vast task, and the monitoring data could not describe
the effect of a single factor on the temperature distribution of bridge structures. With the
development of numerical technique, the Finite Element Model (FEM) began to be used to
analyze the effect of environmental parameters on the thermal field and thermal stress field
(Tong et al., 2001; Medani et al., 2008; Kim et al., 2009). The numerical technique makes it
easy to investigate the thermal behaviors of bridges, but the accuracy of numerical model and
parameters is hard to be determined. Therefore, numerical simulation combined with in-situ
monitoring has been a good choice in the thermal field studies (Xia et al., 2013; Rodriguez
et al., 2014).
It is summarized that the previous studies emphasized on the temperature differences
induced by the combined effects of ambient temperature variation, solar radiation intensity,
wind speed and humidity. However, the thermal behaviors incited by the high-temperature
943
asphalt pavement paving are usually neglected, especially when paving on the steel bridge
deck. In this paper, the thermal field characteristics of the long-span steel bridge during the
GA pavement paving were investigated by in-situ monitoring data and numerical simulation
results at first. Then the thermal stress field and thermal deformation field in the steel box-
girder during the paving were calculated. Finally, the stability of the stiffener was analyzed
considering the whole bridge stress field and the critical temperature load.
2 IN-SITU MONITORING
The in-situ monitoring was conducted in the approach span of the Tongling Yangtze River
Bridge (TL Bridge) in China (Figure 1). The approach span is a steel box-girder structure with
the bridge deck 33.5 m in width, and the bridge deck pavement is a double-layer asphalt pave-
ment structure, with 30-mm-thick GA concrete as the sublayer pavement and 30-mm-thick
Stone Mastic Asphalt (SMA) concrete as the upper pavement, and the methyl methacylate
(MMA) waterproof bonding system is used for the interlayer bonding. The temperature
changes of the bridge deck were monitored in real time during the GA concrete paving.
944
Figure 2. Cross-section positions and temperature monitoring area.
Figure 4. In-situ monitoring. (a) Temperature monitored by infrared thermal imager (b) Thermal
imagery of infrared thermal imager (c) Installing probes of thermocouple thermoscope (d) Temperature
monitored by thermocouple thermoscope.
Solar radiation intensity, Air temperature, Wind speed, Paving Temperature, Paving speed,
q (W/m2) Ta (ºC) v (m/s) Tp (ºC) V (m/min)
field of the chosen cross-sections at first, and then measuring the temperatures of measuring
points in the cross-section at the same time once every 5 minutes for 1 hour.
As can been seen in Figure 5, the temperature-time curves of the pavement reveals a down-
ward trend, and the temperature of Point which is at the pavement edge drops at the fast-
est rate. While the temperature of the steel bridge deck increases first and then decreases
over time, and reaches the peak at 121°C at around 20 min after paving. Comparing the
temperatures of the steel bridge deck in the paving area and non-paving area, it is observed
that Point ∼Point reach the highest temperature at about 20 min, while the temperatures
of Point and Point reach the highest at 25 min and 45 min respectively, and the Point
stabilizes constant in the first 15 min and then gradually rises, reaching the peak at about
55 min. Therefore, the transversal temperature change of the steel bridge deck experiences
distinct hysteresis.
In this section, a steel box-girder model developed by the FEM software ABAQUS 6.11 was
used to simulate the thermal field of steel bridge during GA pavement paving.
946
Figure 6. The simulation model.
Material parameters
Temperature boundary
conditions Steel GA concrete
947
Figure 7. The comparison of the simulation results and monitoring results. (a) Comparison of meas-
uring points in the pavement edge (b) Comparison of measuring point in the steel deck of the paving
area.
Figure 8. Temperature distribution surface plot of steel bridge deck at the critical time.
temperature curves are approximately antisymmetric. In addition, the surface plot experi-
ences fluctuations located around the stiffeners and the diaphragms, but the temperature of
the bridge deck around the stiffeners and the diaphragms is only 5°C∼6°C lower than the
other regions. Therefore, the influence of the stiffener and diaphragm on the thermal field
could be neglected.
In this section, the critical temperature load was applied to the steel box-girder model, and the
thermal stress field and thermal deformation field of the steel bridge deck were investigated.
Figure 9. The stress distribution of the steel box-girder. (a) The Von Mises stress (b) The transversal
stress (c) The longitudinal stress.
edge, and the steel bridge deck of the middle section has the maximum vertical displacement,
which is about 7.5 mm.
Above analysis shows that the stiffeners in the paving area bear the combined action of the
compression and tensile during the paving, which may has an adverse effect on the stability
of the stiffener, especially for the long-span cable-stayed bridge or the self-anchored suspen-
sion bridge. In this section, the sub-model (Zhao & Li, 2008) and buckling mode (Li, 2005;
Zhang, 2010) were used to analyze the stability of the stiffener.
Figure 12. The thermal stress distributions of the stiffeners. (a) The longitudinal stress distribution (b)
The vertical stress distribution.
Figure 13. The buckling modes figure of the stiffener. (a) The instability of the stiffener under critical
temperature stress (b) The instability of the stiffener under normal load.
951
Table 5. Ultimate bearing capacity of the stiffeners.
nomenon, which is due to the transversal bending moment caused by the temperature stress.
According to the results in Table 5, the critical temperature stress could weaken the ultimate
bearing capacity of the stiffeners, and the decreasing amplitude of the stiffener in the paving
edge (Stiffener 1#) reach the peak at 39.95%. Therefore, it is concluded that the temperature
stress during the paving has a dramatically adverse effect on the stability of the stiffener.
5.2.2 Stability analysis of the stiffeners under the coupling of the whole bridge stress field
and the critical temperature stress
During the GA concrete paving, apart from the temperature stress, the bending effect of the
whole bridge stress field also affects the stability of the stiffeners. Zhao (2010) simulated the
whole bridge stress field by the eccentric loading, and the eccentricity ratio of the stiffeners in
the bilateral paving area ranges from 0.2∼0.35. This research calculated the ultimate bearing
capacity of Stiffener 1# under the eccentric loading with the eccentricity ratios of 0.2, 0.25,
0.3 and 0.35, the calculation results are summarized in Table 6.
As can be seen in Table 6, the ultimate bearing capacity rises with the eccentricity ratio,
and the decreasing amplitude of the ultimate bearing capacity of Stiffener 1# under the
eccentric loading reaches 39.36%∼45.19%. Comparing the results in Table 5 and Table 6, the
ultimate bearing capacity under the eccentric loading is superior to that under the normal
load, and the ultimate bearing capacity under the coupling of the whole bridge stress field
and critical temperature stress is slightly better than that under the critical temperature stress.
That is due to the fact that compression stress in the stiffener would decrease under the
eccentric loading.
6 CONCLUSIONS
This study focuses on the thermal effect of steel bridge during the GA pavement paving, the
findings can be applied to the structural design of the steel box-girder. The conclusions are
obtained as follow:
1. The monitoring results show that the temperature of the steel bridge deck increases first
and decrease over time during the GA pavement paving, and the transversal temperature
change of the steel bridge deck experiences distinct hysteresis.
2. The simulation results show that the transversal temperature changing area of the steel
bridge deck converged within the scope which is 0.4 m on both sides of the pavement edge,
and the transversal temperature curves are approximately antisymmetric. The surface plot
of the steel bridge deck sees slight fluctuations due to the stiffeners and the diaphragms.
952
3. During the paving, the thermal stress mainly distributes in the paving area, and the maxi-
mum Von Mises stress of the steel bridge deck lies on the region around the pavement
edge. The thermal stress distributions of the stiffener in the paving area and the stiff-
ener in the paving edge are different, and there is a hidden danger of instability in the
stiffener.
4. Local thermal deformation occurs in the steel bridge deck during the paving. The bridge
deck in the paving area expands during the paving, and the maximum transversal displace-
ment is about 4 mm. The maximum vertical displacement of bridge deck lies near the
pavement edge in the middle section, which is about 7.5 mm.
5. Under the normal load, the decreasing amplitude of the ultimate bearing capacity of
Stiffener 1# caused by the critical temperature reaches 39.95%. However, under the eccen-
tric loading, the decreasing amplitude of the ultimate bearing capacity of Stiffener 1#
caused by the critical temperature reaches 39.36%∼45.19%.
ACKNOWLEDGMENTS
This study was supported by the National Natural Science Foundation of China (No.
51378122).
REFERENCES
Chen, W. & Ye, J.H. 2012. Experimental investigation on mechanic behavior of Q345 cold-formed steel
material at elevated temperatures. Journal of Building Structures 33(2): 41–49. (in Chinese).
Cho, S., Kikuchi, K. & Kawasaki, A. 2012. On the role of amorphous intergranular and interfacial lay-
ers in the thermal conductivity of a multi-walled carbon nanotube-copper matrix composite. Acta
Materialia 60(2): 726–736.
Choi, S., Cha, S.W., Oh, B.H. & Kim, I.H. 2011. Thermo-hygro-mechanical behavior of early-age
concrete deck in composite bridge under environmental loadings. Part 1: temperature and relative
humidity. Materials and Structures 44(7): 1325–1346.
Dilger, W., Ghali, A., Chan, M., Cheung, M. & Maes, M. 1983. Temperature stresses in composite box
girder bridges. Journal of Structure Engineering, ASCE 109(6): 1460–1478.
Ding, Y.L., Zhou, G.D., Li, A.Q. & Wang, G.X. 2012. Thermal field characteristic ananlysis of steel
box girder based on long-term measurement data. International Journal of Steel Structures 12(2):
219–232.
Gong, F., Bui, K., Papavassiliou, D.V. & Duong, H.M. 2014. Thermal transport phenomena and limita-
tions in heterogeneous polymer composites containing carbon nanotubes and inorganic nanoparti-
cles. CARBON 78: 305–316.
Kim, S.H., Cho, K.I., Won, J.H. & Kim, J.H. 2009. A study on thermal behavior of curved steel box
girder bridges considering solar radiation. Archives of Civil and Mechanical Engineering 4(3): 59–76.
Li, L.F. 2005. The analytical theory and model test research on local stability of orthotropic steel box
girder. Doctoral thesis. Hunan University. (in Chinese).
Liu, X., Medani, T.O., Scarpas, A., Huurman, M., & Molenaar, A.A.A. 2008. Experimental and numer-
ical characterization of a membrane material for orthotropic steel deck bridges: Part 2: Develop-
ment and implementation of a nonlinear constitutive model. Finite Elements in Analysis and Design
44(9–10): 580–594.
Medani, T.O., Liu, X., Huurman, M., Scarpas, A. & Molenaar, A.A.A. 2008. Experimental and numeri-
cal characterization of a membrane material for orthotropic steel deck bridges: Part 1: Experimental
work and data interpretation. Finite Elements in Analysis and Design 44(9–10): 552–563.
Qian, Z.D. & Liu, Y. 2012. Mechanical analysis of waterproof bonding layer on steel bridge deck under
bridge-temperature-load coupling effect. Journal of Southeast University: Natural Science Edition
42(4): 729–733.
Rodriguez, L., Barr, P. & Halling, M. 2014. Temperature effects on a box-girder integral-abutment
bridge. Journal of Performance of Constructed Facilities, ASCE 28(3): 583–591.
Sun, J., Li A.Q. & Ding, Y.L. 2009. Observation and research on temperature distribution in steel box
girders of Runyang Yangtze River Bridge. Journal of Highway and Transportation Research and
Development 26(8): 94–98. (in Chinese)
953
Tong, M., Tham, L.G., Au, F.T.K. & Lee, P.K.K. 2001. Numerical modelling for temperature distribu-
tion in steel bridges. Computers and Structures 79(6): 583–593.
Xia, Y., Chen, B., Zhou, X.Q. & Xu, Y.L. 2013. Field monitoring and numerical analysis of Tsing Ma
Suspension Bridge temperature behavior. Structural Control and Health Monitoring 20(4): 560–575.
Zhang, X.H. 2010. Local buckling properties of stiffened plates in steel box girders. Master thesis.
Hunan University. (in Chinese).
Zhao, L.H. & Li, T.C. 2008. An efficient finite element dynamic sub-modeling approach using the local
mesh refinement technique. Earth & Space 2008: 1–6.
Zhao, Q. 2010. Research on calculation methods of welding residual stress and compression stability of
U rib-stiffened plates. Doctoral thesis. Tongji University. (in Chinese).
954
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Hao Wang
Department of Civil and Environmental Engineering, Rutgers University, The State University
of New Jersey, USA
ABSTRACT: Based on numerical analysis, the temperature field and thawing depth of
embankments with different width were analyzed, and the influences on the thermal stability
of embankment by different protective measures were evaluated. The findings indicate that,
the embankment construction has serious disturbance on the thermal stability of permafrost
ground shown as obvious increasing of annual average ground temperature and maximum
thawing depth, and embankment with larger width leads to worse thermal stability. Rais-
ing embankment height, setting crush stone layer or Expanded Polystyrene (EPS) layer can
improve the thermal stability of embankment, however, the improvement is limited for wide
embankment with width exceeding 26 m. Combination of crush stone layer and EPS layer has
the best protective effects on the thermal stability of wide embankment and is recommended
as a protective measure for wide embankment of expressway in permafrost regions of Qinghai-
Tibet Plateau.
1 INTRODUCTION
It is proved that highway construction in permafrost regions severely changed the local con-
ditions such as micro relief, surface conditions and physical properties of shallow ground
(Peng et al. 2015). This causes inevitable disturbance to the energy exchange between the
atmosphere and the ground, correspondingly leading to thermal regime degradation of
permafrost underlying highway embankment (Wu & Niu 2013). Especially with asphalt
pavement used, the dark surface of the asphalt pavement absorbs a great deal of solar radia-
tion, increasing the temperature of pavement and negatively affecting the thermal regime
of underlying permafrost (Li & Sheng 2008). Since the physical and mechanical properties
of permafrost are closely related to its thermal regime, the thermal stability degradation
of underlying permafrost can obviously undermine the stability and durability of highway
embankment which can severely affect the normal operation and service life of highway
(Zheng et al. 2010). Studies showed that, more than 60% of the underlying permafrost along
the Qinghai-Tibet highway had experienced noticeable temperature increase, permafrost
table decline, and subsequent talik formation (Sheng et al. 2002, Wu et al. 2002, Wang et al.
2003). Filed observations showed that, structural failures of the highway occurred progres-
sively, which severely lowered the service ability of the highway. And nearly 85% of these
failures were accounted for roadbed thaw settlement induced by underlying permafrost
warming and thawing (Liu et al. 2002, Wu et al. 2000, Yu et al. 2002, Zhang et al. 2015, Niu
et al. 2005).
955
To maintain the service performance of highway in permafrost regions, a series of engineer-
ing methods have been studied and practiced to protect the underlying permafrost and keep
the thermal balance between the atmosphere and permafrost. Generally, the current protective
approaches for can be categorized into two types: the heat-insulating approaches and the heat-
cooling approaches. Typical heat-insulating approaches mainly include raising the height of
the subgrade, placing shading board over the subgrade slope, using heat-insulating materials
such as Expanded Polystyrene (EPS) in the subgrade, etc. (Wu et al. 2010, Chen et al. 2006,
Feng et al. 2003, Wang et al. 2008). Typical Heat-cooling approaches mainly include crushed
rock layer, ventilation ducts, and thermosyphons, etc. (Dong et al. 2010, Wu et al. 2008, Pan
& Wu 2002). Although many of the methods have been practiced in permafrost regions and
proved to be feasible to postpone the thermal degradation of permafrost, due to the complex
environmental conditions of permafrost regions, pavement distresses caused by thawing and
settlement of permafrost were still commonly observed in field. Therefore, understanding the
thermal regime and maintaining the thermal stability of permafrost under highways are still
the main focuses for highway construction in permafrost regions.
However, most of the previous researches related to the thermal regime and stability of
permafrost were based on narrow embankments usually with design width smaller than
20 m. Thus, to provide fundamental basis for expressway construction in permafrost regions
of Qinghai-Tibet Plateau, this study analyzed the thermal regime of permafrost under
wide embankment and evaluated the feasibility of different protective approaches for wide
embankment.
∂T ∂ ⎡ ∂T ⎤ ∂ ⎡ ∂T ⎤
C (T ) = βx T + β y (T )
∂t ∂x ⎢⎣ ∂x ⎥⎦ y ⎢⎣ ∂y ⎥⎦
(1)
∂w
C (T ) = cρ + ρw u
∂T (2)
∂w
βx ( ) ρw Dx u
∂T
x
(3)
∂w
βy ( ) ρw Dy u
∂T
y
(4)
where, kx and ky are the components of soil thermal conductivity at x and y direction respec-
tively (W⋅m−1⋅K−1); Dx and Dy are the components of water diffusivity at x and y direction
respectively (m2⋅s−1); ρ and ρw are the densities of soil and water respectively (kg⋅m−3); c is the
specific heat capacity of soil (J⋅kg−1⋅K−1); C(T) is the equivalent volumetric heat capacity of
soil (J⋅m−3⋅K−1); L is the volumetric latent heat of thawing (J⋅m−3); wu is the volumetric unfro-
zen soil moisture (m3⋅m−3); βx(T) and βy(T) are the components of the equivalent thermal
conductivity of soil at x and y direction respectively (J⋅m−1⋅s−1⋅K−1), T is temperature (ºC); t
is time (s); and x and y represent the horizontal direction and vertical direction in the cross
section of the embankment, respectively.
As frozen soil is a composite material, the mineral and organic components in the frozen
soil also contribute to its thermal properties (Farouki 1981). Depending on its physical state
(frozen or thaw) and thermal properties of components, the equivalent thermal properties of
frozen soil can be obtained in the following way:
956
⎪⎧(csu + w cw ) ρd T ≥ −θ f
C (T ) = ⎨ −( + ) (5)
⎩ρd cssff + d ci w0 + ρd (cw − ci ) w0θ f ( −T ) L ⋅ w θ ⋅ b ⋅ (− )
−b
b b
0 f T < −θ f
⎧λu θf
βx ( ) β y (T ) = ⎨ − (b + ) (6)
⎩λ f θ b
f b ( −T ) T < −θ f
where, csu and csf are the specific heat capacities of soil at thawing and frozen states, respec-
tively (J⋅kg−1⋅K−1); w0 is the initial volumetric unfrozen soil moisture (m3⋅m−3); ci and cw are
the specific heat capacities of ice and water, respectively (J⋅kg−1⋅K−1); ρd is the dry density of
soil (kg⋅m−3); θf is the absolute value of freezing temperature of soil; b is a constant related to
soil type; λu and λf are the thermal conductivities of the thawed and frozen soil, respectively
(W⋅m−1⋅K−1); and D is the water diffusivity (m2⋅s−1).
Part I II III IV
Material Gravel soil Pebbly clay Gravel loam Strong erosion mudstone
ρd /(kg/m3) 1800 1700 1300 1500
w0 /(%) 25 30 30 30
csu /(KJ/kg⋅ ºC) 0.79 0.84 0.84 0.84
csf /(KJ/ kg⋅ ºC) 0.71 0.73 0.75 0.75
cw /(KJ/ kg⋅ ºC) 4.182 4.182 4.182 4.182
ci /(KJ/ kg⋅ ºC) 2.09 2.09 2.09 2.09
λu /(W/m⋅ºC) 1.919 1.95 0.87 1.47
λf /(W/m⋅ºC) 1.980 2.69 1.22 1.82
D /( m2⋅s−1) 9.35 × 10–6 4.66 3.73 3.44
−θf /(ºC) −0.20 −0.10 −0.19 −0.05
L /(KJ/kg) 334.56 334.56 334.56 334.56
b 0.610 0.733 0.574 0.474
957
Table 2. Pavement structure and corresponding material parameters.
Depth /m 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Temperature /ºC 6.11 2.62 0.55 −0.40 −1.35 −1.57 −1.86 −2.03 −2.03 −2.09
Depth /m 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0
Temperature /ºC −2.12 −2.10 −2.15 −2.00 −2.25 −2.25 −2.25 −2.25 −2.25 −2.00
Month 1 2 3 4 5 6 7 8 9 10 11 12
Daily solar radiation 10.1 12.3 17.5 20.4 22.0 22.5 24 21.6 18.3 15.2 13 10.2
(MJ/m2d−1)
Effective ground 6.82 6.80 8.63 8.68 8.38 7.44 7.51 7.26 7.54 9.46 8.96 8.21
radiation (MJ/m2d−1)
thickness of 0.1 m was set at the depth of 1 m below pavement surface and the crush rock layer
with thickness of 1.5 m was set at the depth of 0.5 m below the natural ground surface. The
thermal parameters for EPS and crush rock are summarized in Table 3.
958
tively. Finally, user-defined DFLUX model and Film model were developed in ABAQUS to
define the thermal boundary conditions for the numerical model.
Figure 2. Temperature fields in August for embankments with different width: (a) 10 m, (b) 18 m, (c)
26 m, (d) 34 m, (e) 42 m, and (f) 50 m.
959
Figure 3. Temperatures at depth of 5 m along the centerline of embankments with different width.
ture keeps increasing. It also indicates that, at the same time point, wider embankment has
greater temperature. Based on temperature field analysis, the maximum thawing depth ten
years after embankment construction at the centerline and shoulder of embankments with
different width are summarized in Fig. 4. The thawing depth at the centerline of embank-
ments with width of 18, 26, 34, 42 and 50 m are 0.47, 2.29, 2.72, 3.12 and 3.77 m greater than
that of embankment with width of 10 m, respectively. The thawing depth at the shoulder of
embankments with width of 18, 26, 34, 42 and 50 m are 0.52, 0.96, 1.15, 1.47 and 1.51 m
greater than that of embankment with width of 10 m, respectively. It is clearly indicated
that the thawing depth increases with the embankment width increasing. Meanwhile, after
the embankment width exceeds 18 m, the embankment has more influences on the thawing
depth of the embankment centerline than on the thawing depth of the embankment shoul-
der. Therefore, larger embankment width results in bigger thawing depth and more seriously
uneven thawing problems of embankment.
For embankments with different width, the maximum thawing depth at the centerline and
shoulder of embankment versus embankment height are shown in Fig. 5. In the figures, by using
the natural ground surface as the zero line, the positive value of depth represents that maximum
thawing depth is in the natural ground while the negative value of depth means that the maxi-
mum thawing depth is within the embankment. It can be seen that, for all of the embankments,
both the thawing depth and thawing depth difference between centerline and shoulder decrease
with the embankment height increasing. When the embankment height reaches 6 m, the maxi-
960
Figure 5. Influences by embankment height on the thawing depth: (a) at centerline, and (b) at
shoulder.
mum thawing depth at the centerline and shoulder of embankments with width smaller than
26 m are within the embankment or close to the natural ground surface while the thawing depth
of embankments with width bigger than 18 m are still within the natural ground. Meanwhile,
the thawing depth differences between centerline and shoulder of embankments with width
bigger than 18 m are more obvious than that of embankments with width smaller than 26 m.
It is indicated that, increasing embankment height is helpful to improve the thermal regime of
the embankment section, however, the improvement is obviously weakened by the increasing of
embankment width, especially when the embankment width exceeds 18 m.
961
Figure 6. Temperature fields in August for different embankments: (a) normal embankment, (b) CS
embankment, (c) EPS embankment, and (d) Composite embankment.
It proves that all of the protective measures can improve the thermal regime of embankment
section, however, different protective measures have different improving effects.
The temperature variations during two decades at the depth of 5 m along the centerline
of different embankments are shown in Fig. 7. It can be seen that the temperature variation
with service time for all of the four embankments presents similar trends which shows that the
temperature changes periodically over time and the average annual temperature keeps increas-
ing. At the same time point, normal embankment has the highest temperature while compos-
ite embankment has the lowest temperature. The EPS embankment has similar temperatures
with the CS embankment before the 15th year after embankment construction and lower tem-
peratures than the CS embankment after the 15th year after embankment construction.
Based on the temperature field analysis, the thawing depth twenty years after embank-
ment construction at the centerline and shoulder of embankment versus embankment width
are shown in Fig. 8 for different embankments. It is clearly indicated that, with the same
embankment width and service time, the normal embankment has the biggest thawing depth
while the composite embankment has the smallest thawing depth. And the thawing depth
of EPS embankment is smaller than that of CS embankment. With the embankment width
increasing, the thawing depth of normal embankment keeps increasing rapidly while the
thawing depth of composite embankment slowly increases. For EPS embankment and CS
embankment, the thawing depth increase slowly with the embankment width increasing
before the embankment width reaches 26 m and then increase rapidly with the embankment
width increasing after the embankment width exceeds 26 m. Meanwhile, the thawing depth
962
Figure 8. The twentieth-year thawing depth vs embankment width for different embankments: (a) at
centerline, and (b) at shoulder.
differences between centerline and shoulder of EPS and CS embankments become more and
more obvious after the embankment width exceeds 26 m. Therefore, the protective effects of
EPS layer and crush stone layer can be obviously weakened by wide embankment with width
exceeds 26 m while the combination of EPS layer and crush stone layer has the best protec-
tive effects for the thermal regime of the wide embankment section.
4 CONCLUSIONS
In this study, the thermal regime of wide embankment and different protective measures for
wide embankment were investigated. The following conclusions can be drawn:
1. It is indicated that the embankment construction in permafrost regions has serious influ-
ences on the thermal regime of the underlying permafrost ground. The temperature filed
of the embankment section shows annually periodical variation and the annual average
temperature keeps increasing with the service time increasing. It is the important reason
for the degradation of the permafrost underlying embankment.
2. The results show that, at the same service time, embankment with larger width has higher
annual ground temperature and greater maximum thawing depth. Meanwhile, larger
embankment width results in more seriously uneven temperature filed within the embank-
ment section. Therefore, the thermal stability of embankment section gets worse with the
embankment width increasing. Expressway with wide embankment faces more serious
thawing and uneven settlement problems than the current highway.
3. It is found that increasing the embankment height is helpful to improve the thermal regime
of the embankment section. However, the improvement is limited. Both EPS layer and crush
stone layer can improve the thermal regime of embankment section. However, after the width
of embankment exceeds 26 m, the protective effects by EPS layer or crush stone layer are
obviously weakened by the increasing of embankment width. Since the fundamental of EPS
layer is to provide heat resistance for permafrost ground in hot reason and crush stone layer
aims to facilitate the heat dissipation of permafrost ground during cold season, the combina-
tion of EPS layer and crush stone layer has the best protective effect for wide embankment.
4. Further study will be conducted to investigate the thermal settlement of wide embankment
for expressway, and to evaluate other protective measures for expressway embankment.
ACKNOWLEDGEMENTS
The study is financially supported by National Science and Technology Support Program
(2014BAG05B04), National Natural Science Foundation of China (No. 51378006), Huoy-
963
ing dong Foundation of the Ministry of Education of China (No. 141076), Excellent Young
Teacher Program of Southeast University (2242015R30027), and State Key Laboratory of
High Performance Civil Engineering Materials (2014CEM008).
REFERENCES
Chen, J., et al. 2006. Discussion on protection measures of permafrost under the action of engineering
from the point of earth surface energy balance equation in Qinghai–Tibet plateau. Journal of Glaciol-
ogy and Geocryology 28: 223–228.
Dong, Y.H., et al. 2010. Using Perforated Ventilation Ducts to Enhance the Cooling Effect of Crushed-
rock Interlayer on Embankments in Permafrost Regions. Cold Regions Science and Technology 62:
76–82.
Farouki, O.T. 1981. The thermal properties of soils in cold regions. Cold Regions Science and Technology
5(1): 67–75.
Feng, W.J., et al. 2003. Application of Awning to Roadway Engineering in Permafrost Regions. Journal
of Geotechnical and Geoenvironmental Engineering 25: 567–570.
Harlan, R.L. 1973. Analysis of coupled heat - fluid transport in partially frozen soil. Water Resources
Research 9(5): 1314–1323.
Li, J.P. & Sheng, Y. 2008. Analysis of the thermal stability of an embankment under different pavement
types in high temperature permafrost regions. Cold Regions Science and Technology 54: 120–123.
Liu, J. & Tian, Y. 2002. Numerical studies for the thermal regime of a roadbed with insulation on per-
mafrost. Cold Regions Science and Technology 35(1): 1–13.
Liu, Y., et al. 2002. Deformation of highway roadbed in permafrost regions of the Tibetan Plateau.
Journal of Glaciology and Geocryology 24(1): 10–15.
Niu, F., et al. 2005. Engineering-related slope failure in permafrost regions of the Qinghai–Tibet Pla-
teau. Cold Regions Science and Technology 42(3): 215–225.
Pan, Y. & Wu, C.Z. 2002. Numerical Investigations and Engineering Applications on Freezing Expan-
sion of Soil Restrained two-phase Closed Thermosyphons. International Journal of Thermal Sciences
41: 341–347.
Peng, H., et al. 2015. Impact of permafrost degradation on embankment deformation of Qinghai−Tibet
Highway in permafrost regions. Journal of Central South University 22(3): 1079–1086.
Sheng, Y., et al. 2002. Thermal regime in the embankment of Qinghai−Tibetan highway in permafrost
regions. Cold Region Science and Technology 35(1): 35–44.
Taylor, G.S. & Luthin, J.N. 1978. A model for coupled heat and moisture transfer during soil freezing.
Canadian Geotechnical Journal 15(4): 548–555.
Wang, S.J. 2008. Highway construction technology in the permafrost region of China: 153–158. Beijing:
China Communications Press.
Wang, S., et al. 2003. The thermal stability of roadbed in permafrost region along the Qinghai−Tibet
highway. Cold Regions Science and Technology 37(1): 25–34.
Wu, Q. & Mi, H. 2000. Predictions and control proposes of frozen ground under asphalt pavement at
the high ground temperature section of the Qinghai−Xizang Highway. Hydrogeology and Engineering
Geology 27(2): 14–17.
Wu, Q., et al. 2002. A review of recent frozen soil engineering in permafrost regions along Qinghai−Tibet
highway, China. Permafrost and Periglacial Processes 13(3): 199–205.
Wu, Q. & Niu, F. 2013. Permafrost changes and engineering stability in Qinghai−Xizang plateau. Chi-
nese Science Bulletin 58(10): 1079–1094.
Wu, Q.B., et al. 2010. Long-term thermal effect of asphalt pavement on permafrost under an embank-
ment. Cold Regions Science and Technology 60: 221–229.
Wu, Q.B., et al. 2008. Analysis of Cooling Effect of Crushed Rock-based Embankment of the Qinghai–
Xizang Railway. Cold Regions Science and Technology 53: 271–282.
Yu, Q., et al. 2002. Analysis of the subgrade deformation of the Qinghai−Tibetan Highway. Journal of
Glaciology and Geocryology 24(5): 623–627.
Zheng, B., et al. 2010. Investigation for the deformation of embankment underlain by warm and ice-rich
permafrost. Cold Regions Science and Technology 60(2): 161–168.
Zhang, C., et al. 2015. Sensitivity analysis of longitudinal cracking on asphalt pavement using MEPDG
in permafrost region. Journal of Traffic and Transportation Engineering 2(1): 40–47.
964
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
1.1 Background
A typical flexible pavement cross section consists of an asphalt surfacing, a base, a sub-
base and subgrade. The base and subbase layers serve as the foundation of the pavement
as most of the loads are supposed to be borne by the top asphalt layer (Brown, 1996). The
granular materials are non-linear in nature and are found to be highly pressure sensitive
(Kim and Tutumluer 2006). The pressure dependent characteristic of granular materials can
be captured by elasto-plastic soil models. However, due to complexity this is seldom done
in the stress-strain analysis for pavement design. The complexity arises due to the fact that
computational time and effort required for an elasto-plastic layer topped with viscoelastic
layer (asphalt layer) would be very much involved. Hence, even till date the linearized elas-
tic theory is widely used for pavement stress-strain analysis within the context of pavement
design. The concept of resilient modulus (Mr) was introduced by Hveem et al. (1950). Lekarp
et al. (2000) discussed the relationship of resilient modulus with stress conditions, moisture
content, density and fines content of the sample and type and frequency of loading cycles.
The evaluation of resilient modulus is involved as it requires special equipments like cyclic
triaxial setup and analyzing the test data to fit into an appropriate resilient modulus model.
Hence, various empirical models were developed to relate resilient modulus to index proper-
ties of the soil (Zaman et al., 1994) and various alternate methods were suggested to evaluate
resilient modulus.
965
layers and hence depends more on the mechanical properties of the granular layers and the
existing subgrade.
The California Bearing Ratio (CBR) and Plate Load Test were the earliest and the most
common tests used for the characterization of pavement materials. These tests are typical
“strength” tests from which the load bearing capacity of the material is gleaned. The CBR
test involves a standard plunger penetrating a compacted granular material under specific
moisture condition at a constant displacement rate. It is essentially a shear test and the CBR
is supposed to be an index of the shear strength (Turnbull, 1950). It was gradually real-
ized that the CBR test was not an appropriate test to characterize granular materials as in
the CBR test the specimen is laterally confined allowing no lateral deformation whereas in
actual field situation lateral movements are possible. The CBR test lies in-between a fully
confined sample as in an oedometer test and laterally unrestrained sample free to undergo
deformations in lateral direction as in a triaxial test. The triaxial tests provides more control
on loading conditions and proper interpretation of the test data can help in predicting realis-
tic stress-strain characteristics of the materials.
Granular layers are subjected to repetitive loading in a pavement due to vehicular loading
and hence repeated and cyclic load triaxial tests were developed to characterize the material.
The repeated load triaxial test lead to the development of the concept of road bed Resilient
Modulus (Mr) (AASHTO, 1986) which is essentially determined using a cyclic triaxial test
set up. Resilient modulus incorporates both the effect of repeated loading (as the traffic
loads in pavements) and also the effect of varying confining pressures. At very low load lev-
els also granular materials undergo plastic deformations and these deformations accumulate
with cycles of loading. But after being subjected to considerable amount of repeated cycles
of loading and unloading, the accumulation of plastic deformations gradually reduces and
the material then exhibits mainly recoverable deformation. Groeger et al., (2003) discussed
briefly the history of the resilient modulus test. The current AASHTO T 307 is based on the
LTPP protocol P46. The LTPP protocol P46 is based on the first protocol AASHTO T274
for resilient modulus test developed in 1982. The protocol was developed on the basis of
vehicle speed, deflections and variation of vertical stress along with depth. This served as
the basis of the various combination of stress states considered in the resilient modulus test
protocol.
Though the concept of resilient modulus is well established, the testing and post-process-
ing protocol has been changing over the years. The earlier test procedures such as, the CBR
test was straightforward and had set simple methodology, even though it lacks a strong theo-
retical background. The test procedure is well established and simple. This is not the case
for the resilient modulus test protocol as various improvements have been added to the test
protocol. The test is complex when compared to other conventional tests such as the CBR,
static triaxial, dynamic cone penetration test, unconfined compression test and oedometer
test. The data obtained from the test cannot be directly used in design procedures but needs
to be interpreted and fit into an appropriate model to be used in pavement design software.
The resilient modulus is also an elastic modulus and is used in an elasticity theory based
framework to determine stresses and strains and cannot predict the development of plastic
strains in the pavement system. In thin pavements where the granular layer serves as the main
load bearing layer, it is not clear whether one can still use the resilient modulus based models.
Significant amount of research work carried out recently have clearly outlined the limitations
of resilient modulus test methods and alternate test methods are being explored.
Rahim and George (2004) carried out experiments to correlate Mr and soil index properties
which included the Atterberg’s limits and density but such correlations are case sensitive and
vary as the material varies. Kim et al., (2001) suggested an alternative Mr testing technique for
subgrade soils using a static triaxial compression test. Okada et al., (2002) performed cyclic
penetration tests on railway subgrade in order to evaluate the resilient modulus of the mate-
rial and this data was used to develop models to predict the elastic modulus and permanent
deformation of the subgrade material. Dawson et al., (2009) performed both conventional
cyclic resilient modulus test and Non-Destructive Testing (NDT) with the falling weight
deflectometer, and developed correlations between the Mr values obtained.
966
In this paper an alternate test, repeated load CBR test using the standard CBR test setup
was studied. The response of the material from the repeated load CBR test, evaluated in the
form of modulus was compared with the resilient modulus of the material as obtained from
the cyclic resilient modulus test.
2 EXPERIMENTAL STUDY
967
Figure 2. Conventional CBR test results for the three placement condition.
The available cyclic triaxial test setup facilitates testing of maximum sample sizes of
100 mm diameter and 200 mm height. As per the standard protocol for resilient modulus
testing, the diameter of sample to be used for the test should be equal to or greater than five
times the size of the largest particle size (AASHTO: T307–99, 2012). Hence the material
coarser than 19 mm sieve was removed from the sample. The sample was cast in a split mould
by method of static compaction as per AASHTO T307 protocol (AASHTO: T307–99, 2012).
The granular material was mixed with water and known quantity of soil was laid in the
mould in five layers. Each layer was hand compacted by rammer and at the end, the entire
sample was subjected to static compaction to the required volume.
For the repeated CBR test, stainless steel CBR mould of 150 mm diameter and 175 mm
height was fabricated. The sample was mixed with the required quantity of water and laid in
five layers in the mould, which were lightly hand compacted by rammer, followed by static
compaction to create a sample of 130 mm height.
each increasing test sequence the applied confining pressure and cyclic axial stress on the
sample increases leading to higher bulk stresses on the material. Hence it can be interpreted
that the material strain hardens and this is reflected by an increase in the modulus value.
σ Hd
H σ d 2 (L − H ) σ d ⎡ d (L − H ) ⎤
u= + →E = (1)
ED ED 2 uD ⎢⎣ D ⎥
⎦
where, d = plunger diameter = 50 (mm), H = height of the conical part (mm), D = diameter of
the CBR mould = 150 (mm) and L = height of specimen (mm).
In order to obtain H and L, an estimate has to be made about the angle of load spread-
ing. As the angle of spread increases, H would reduce leading to higher value of modulus.
This is physically also comprehendible that a higher angle of spread for load for the mate-
rial indicates that the material has higher stiffness. This angle affects the modulus value to
substantial extent but is difficult to determine. A value of 45° is a reasonable assumption for
granular material (Araya et al., 2010). Results with assumption of 45° angle of spread gave
better results comparable to the resilient modulus values and hence the assumption is veri-
fied. Using this assumption the E-values were evaluated from the vertical deformation using
equation 1 for the last reloading curve where the deformations are primarily elastic in nature.
Being elastic modulus, the corresponding vertical strain was back calculated from the modu-
lus and plunger stress values
The stress-penetration plots for the deformation controlled and load controlled repeated
load CBR test results for the wet side of OMC are shown in Figures 6 and 7, respectively. It can
be observed that the deformations nearly stabilize after 10 cycles of loading and unloading.
970
Figure 5. Conical and cylindrical deformation parts in a CBR test (Opiyo, 1995).
Figure 6. Stress-strain plot of displacement rate controlled repeated CBR test (wet side−9.5%).
Figure 7. Stress-strain plot of load controlled repeated CBR test (wet side–9.5%).
This is indicated by the reduction in the hysteresis in the loop as the number of loading cycles
increases which means that the accumulations of plastic deformations ceases and the material
primarily behaves in elastic manner and also indicated by the reduction in the variation of the
modulus vale with the increase in number of cycles in Figure 8. The test data is used to evalu-
971
Figure 8. Variation of CBR modulus with number of cycles for sample on wet side.
Figure 9. Last reloading curve for displacement rate and load controlled test for three placement
conditions.
Table 1. E-values for displacement and load controlled repeated load CBR tests and Mr values.
ate single modulus value from the slope of the last reloading curve for each soil condition. The
last reloading curves in the form of vertical stress-strain for all the three placement conditions
for both displacement rate and load controlled are depicted in Figure 9. The influence of
moisture content can be well understood from Figure 9. As the water content increases the
magnitude of strain increases and the slope of the last reloading curve reduces. The reduction
in the steepness of the slope, marks the reduction in the magnitude of the modulus value as the
material moves from dry to wet side. The same trend was observed for the resilient modulus
values as in Figure 4. Table 1 shows the modulus values for three placement conditions, evalu-
ated from the tenth reloading cycle. The magnitude of the modulus values is in order of the
Mr values and there is no drastic variation in the magnitude values.
972
3.3 Discussion
The resilient modulus test yields 15 resilient modulus values for the various stress conditions.
The modulus values should ideally be then used to evaluate the resilient modulus model
parameter, which eventually is used in a mechanistic empirical theory based pavement design
software. A wide variety of such models exist (Hjelmsted and Taciroglu, 2000), the most
advanced being the NCHRP model (NCHRP, 2004). The models use the resilient modulus
values and the stress conditions for which the modulus was evaluated and fit model parame-
ters to the data. Despite subjecting the material to 15 sequences of loading and post-process-
ing, the data collected to fit model parameters for various resilient modulus models, such an
approach is largely not implemented in any stress-strain computation in pavement engineer-
ing. As far as low volume roads are concerned, it suffices to have one modulus parameter
truly depicting the nature of material response. Hence the repeated CBR test serves as a
perfect blend of simple test procedure coupled with the effect of repeated loading on granu-
lar material to evaluate a single modulus depicting the true response of the material under
repetitive loading.
The CBR modulus values evaluated from the slope of the last (tenth) reloading cycle for
the three placement conditions can be seen to be an average of the resilient modulus values
evaluated for the 15 sequences for the corresponding three placement conditions. Table 1 tab-
ulates the range and average of the resilient modulus values of the 15 sequences for the three
placement conditions as obtained from the regular cyclic resilient modulus test. The bulk
stresses expected under the pavement for the average resilient/CBR modulus values using the
Hicks and Monismith model (Hicks and Monismith, 1971) whose model parameters where
determined using the cyclic resilient modulus test results have also been indicated. It can be
seen that the repeated CBR modulus values lie within the range of average resilient modulus
values with a maximum error of 6%. It can be stated that modulus values evaluated from the
load controlled CBR test closely replicate average resilient modulus values with maximum
error of about 5%.
4 CONCLUSIONS
An attempt is made in the present study to obtain resilient modulus using the conventional
California Bearing Ratio apparatus by subjecting the sample to repeated loading. The slope
of the last loading cycle is taken as the CBR modulus E. It was found that the E values
obtained are comparable to the average Mr values obtained from the cyclic triaxial apparatus.
Therefore, the repeated load CBR test can be used as an alternate method for estimation of
resilient modulus using the simple CBR test setup by performing repeated load test without
resorting to sophisticated cyclic triaxial apparatus. The gradation of material should be simi-
lar to the current used test material and care should be taken that the material is not sensi-
tive to deformation rate which in this case was ensured by performing tests on the material
at varying deformation rates. The choice for the angle of spread should be done cautiously.
Similar tests in future on other types of granular materials can lead to generalization of the
method to serve as an alternate for the determination of resilient modulus values.
REFERENCES
AASHTO.1986. Guide for design of pavement structures. Technical report, American Association of
State Highway and Transportation Officials, Washington, D.C., USA.
AASHTO: T274–82. 1986. Standard method of test for resilient modulus of subgrade soils. American
Association of State Highway and Transportation Officials, Washington, D.C., USA.
AASHTO: T307–99. 2012. Determining the resilient modulus of soils and aggregate materials. Ameri-
can Association of State Highway and Transportation Officials, Washington,D.C., USA.
Araya, A. A., Molenaar, A. and Houben, L. 2010. Characterization of unbound granular materials
using repeated load CBR and triaxial testing. Proceedings of Geo Shanghai International Confer-
ence, Shanghai, 3–5.
973
Brown, S. F. 1996. Soil mechanics in pavement engineering. Geotechnique, 46(3), 383–425.
Dawson, T. A., Baladi, G. Y., Sessions, C. P. and Haider, S. W. 2009. Back calculated and laboratory-
measured resilient modulus values. Transportation Research Record, 2094, 71–78.
Hicks, R. G. and Monismith, C. L. (1971). Factors influencing the resilient properties of granular mate-
rials. Transportation Research Record, 345, 15–31.
Hveem, G. T. S. 1950. Unbound granular base for roads.Faculty of Civil Engineering and Geosciences.
Delft University of Technology, Delft.
IRC: 37. 2012. Guidelines for design of flexible pavements. The Indian Road Congress, New Delhi,
India.
IS: 2720. 2006. Method of test for soils. The Indian Standards, New Delhi, India.
Kim, A. U.,Kweon, D. S. and Lee, G. C. 2001.Alternative method of determining resilient modulus of
subgrade soils using a static triaxial test. Canadian Geotechnical Journal, 38(1), 107–116.
Kim, M. and Tutumluer, E. 2006.Effect of stress-dependent modulus and Poisson’s ratio on structural
responses in thin asphalt pavements. Pavement Mechanics and Performance, 154, 29–36.
Lekarp, F., Isacsson, U. and Dawson, A. 2000. State of the art I: Resilient response of unbound aggre-
gates. Journal of Transportation Engineering, ASCE, 126(1), 66–75.
Long-Term Pavement Performance Protocol (LTPP). 1996. P46: Resilient modulus of unbound granu-
lar base/subbase materials and subgrade soils. United States Department of Transportation, Federal
Highway Administration, Washington, D.C., USA (http://www.tfhrc.gov/pavement/ltpp/pdf/p46.
pdf).
M-EPDG.2008. NCHRP-guide for mechanistic - empirical design of new and rehabilitated pavement
structures.National Cooperative Highway Research Program, Transportation Research Board, National
Research Council, Washington, D.C., USA.
MoRTH.2012. Specifications for road and bridge works. Technical report, Indian Roads Congress, New
Delhi, India.
National Cooperative Highway Research Program (NCHRP). 2004. Guide for mechanistic-empirical
design of new and rehabilitated pavement structures, Final Report (MEPDG). Project 1–37 A. (www.
trb.org/mepdg/guide.htm).
Okada, K. and Ghataora, G. S. 2002.Use of cyclic penetration test to estimate the stiffness of railway
subgrade. Independent Non-destructive Testing and Evaluation International journal, 35, 356–377.
Opiyo, T.O., 1995. A Mechanistic Approach to Laterite-based Pavements, in Transport and Road
Engineering (TRE).International Institute for Infrastructure, Hydraulic and Environment Engineering
(IHE): Delft.
Rahim, A. M. and George, K. P. 2004.Subgrade soil index properties to estimate resilient modulus.
Transportation Research Board, 83rd annual meeting CDROM, Washington D.C., USA.
Turnbull, W. J. 1950. Appraisal of the CBR method, Development of CBR flexible pavement design
method for airfields. A Symposium, Transaction American Society Civil Engineers, 547–554.
974
Pavements & Environment
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Climate zones are regions with homogeneous climatic conditions. In mecha-
nistic-empirical pavement design, climate is an important factor affecting pavement perform-
ance, so distress predictions are expected to be consistent across an identified geographic
area. This study aims to develop climate zones specifically for flexible pavement design for
the State of Michigan. In detail, annual mean air temperature, mean air temperature in hot
months, and mean air temperature in cold months were used to identify climate zones. The
critical threshold to determine a climate zone in this study was to keep the temperature within
the zone less than 2.8°F (1.6°C), a change which results in a relative distress difference of
about 10%. Nine climate zones were developed for Michigan taking into account both the air
temperatures and pavement distresses. The developed climate zones are expected to be help-
ful for future flexible pavement designs in Michigan.
1 INTRODUCTION
Climate zones are regions with homogeneous climatic conditions. The use of climatic regions,
where climates in given areas predict similar temperature and moisture profiles and result in
insignificant differences in performance prediction, have been used in previous studies. Cli-
matic zones are typically defined based on the landform, average temperature, precipitation,
and other climate variables. Other fields have used this approach for design.
Pavement design using the mechanistic-empirical approach has been well developed since
the release of the Mechanistic-empirical Pavement Design Guide (MEPDG) in 2002. There
are more than 800 weather stations in the USA, from which climatic information is collected
for the purpose of pavement design. Nevertheless, these weather stations are insufficient in
some states. There are 24 weather stations in Michigan, but these stations are not uniformly
distributed and a big area of gap region is found in the middle of the state. For those regions
without enough weather stations, climate zone can be a good alternative to address this con-
cern. It is expected that the pavement distresses in the same climate zone are similar if the
same pavement is designed. Li et al. developed climatic zones using the climate data from the
national climate data center in Oklahoma and found that the climatic zones can improve the
data accuracy, but it turned out that the data accuracy improvement was insignificant due to
the state’s flat geography.
According to the sensitivity analysis of pavement design to climatic variables in a bunch of
previous studies, air temperature is the most important factor affecting distress predictions.
Air temperature has been used to divide climate zones in previous studies, such as tempera-
ture in summer months, temperature in winter months, and extreme historical temperatures.
This study is motivated by the fact that climate zones are beneficial for pavement design,
especially for some regions without weather stations. The main objective of this study is to
977
create climate zones for flexible pavement design in Michigan. Air temperature and pavement
distress predictions will be utilized to achieve this goal. The climate zones from the three
parameters will be compared and combined to form a final climate zone.
3 RESEARCH METHODOLOGY
The temperatures and distress predictions from 39 weather stations were collected to create
climate zones for Michigan. Climate zones are created by air temperature and distress predic-
tions. The maximum air temperature difference within a climate zone will be determined by
the relationship between air temperature and the distress predictions.
978
Figure 1. The 39 weather stations in Michigan and the six representative stations (star shaped) selected
to determine temperature change threshold (map adopted from Google).
The average value from the six representative stations were calculated to analyze the cor-
relation between temperature change and distress prediction, as shown in Figure 2. It can be
observed that the distress predictions are overall linear with the change in temperature. With
a temperature change of 2°F, the relative distress prediction difference is about 0.8. Based
on the trend line of thermal cracking and AC rutting, the allowable temperature change to
cause a 10% difference is only about 0.9°F and 2.8°F, respectively. 0.9°F is unrealistic to be a
critical temperature change for the creation of climate zones. On the contrary, the 2.8°F for
AC rutting can be a good alternative because it is a moderate value and can meet the 10%
differences for all the distress predictions except for the thermal cracking. Considering that
thermal cracking prediction is highly fluctuated (from less than 100 ft/mile to more than
2000 ft/mile), it is acceptable to exclude thermal cracking in the determination of allowable
temperature change in a climate zone. Thus, 2.8°F is set as the allowable temperature differ-
ence within a climate zone.
979
Figure 2. The average distress prediction changes after a temperature change with -2°F, -1°F, 1°F, and 2°F,
respectively.
980
Figure 4. Climate zones based on mean air temperature in hot months.
files and are shown in Figure 3 through Figure 5. It should be noted that while temperature
increases overall with the decrease in latitude, the cold season temperature in the very north-
ern station is slightly higher than the two stations below it. This is mainly due to the big lake
effect because that station is surround by Lake Superior.
As mentioned above, the criterion of creating climate zones is that the maximum tempera-
ture difference within a climate zone is no higher than 2.8°F. According to this, six climate
zones were created based on mean annual air temperature, as shown in Figure 3; seven cli-
mate zones were created based on mean air temperature in hot months, as shown in Figure 4;
and seven climate zones were created based on mean air temperature in cold seasons, as
shown in Figure 5. The difference between Figure 3 and Figure 5 is that there is a micro-zone
in the bottom right corner in which the temperature in summer is visibly higher than the sur-
rounding regions. Additionally, it was observed from Figure 5 that the Upper Peninsula (the
northern part of the state as seen in the map) was divided into two zones because the cold
season temperature in the very left part is visibly lower than that in the right part. A final
climate zone map was created based on the combination of the maps in Figure 3 through
Figure 5, as shown in Figure 6.
This study explored the potential of creating climate zones for Michigan based on air tem-
perature and distress predictions. In detail, the mean annual temperature, mean temperature
in hot months, and mean temperature in cold months were utilized. Climatic information
from the 39 weather stations throughout the State were collected for climate zone divisions.
The main findings and conclusions can be drawn as bellow:
982
1. The allowable temperature variance within a climate zone was calculated as 2.8°F based
on the quantitative analysis on the effect of temperature on distress predictions.
2. Thermal cracking and AC rutting were found to be the most sensitive distresses to tem-
perature change. They were selected to create distress based climate zones because of this
and they correspond to the low and high temperature performances, respectively.
3. Nine climate zones were eventually created based on the combination of the zones from
air temperatures.
The climate zones from this study will be recommended to be used for pavement design
in Michigan. The same or similar pavement structures and materials can be selected for the
flexible pavement within a climate zone.
The research work was sponsored by the Michigan Department of Transportation (MDOT)
and Federal Highway Administration (FHWA). MDOT assumes no liability for its content
or use thereof. The contents of this report reflect the views of the contracting organization,
which is responsible for the accuracy of the information presented herein. The contents may
not necessarily reflect the views of MDOT and do not constitute standards, specifications,
or regulations. The authors appreciate the guidance and involvement of Michael Eacker of
MDOT as project manager, Andre Clover as project engineer, Pat Schafer (Pavements), David
Hoeh (Bay Region), Elise Kapphahn (ITS), Bob Miller (DTMB), Justin Schenkel (Pavement
Design), Wen Kuo (MDOT Statistical Advisor, RA), and Curtis Bleech (Focus Area Man-
ager) in the Research Advisory Panel (RAP). Their guidance and valuable comments on this
project are greatly appreciated.
REFERENCES
You, Z., et al., Improvement of Michigan Climatic Files in Pavement ME Design. 2015: Lansing, MI.
Li, Q., K.C.P. Wang, and J. Zhang, Development of Climate Zones for DARWin-ME, in TRB 92nd
Annual Meeting Compendium of Papers. 2013, Transportation Research Board: Washington DC.
Yang, X., Z. You, and J.E. Hiller, Sensitivity of Climate Inputs on Rigid Pavement Performance using
Pavement ME Design, in The 9th International Committee on Pavement Technology (ICPT). 2015:
Dalian, Liaoning, China.
Yang, X., et al., Sensitivity of Flexible Pavement Design to Michigan’s Climatic Inputs using Pavement
ME Design. International Journal of Pavement Engineering, 2016.
Tighe, S.L., et al., Using the MEPDG to Assess Climate Change Impacts on Southern Canadian Roads,
in 7th International Conference on Managing Pavement Assets. 2008, Transportation Research
Board: Calgary, Alberta, Canada.
Byram, D., et al., Sensitivity Analysis of Climatic Influence on MEPDG Flexible Pavement Perform-
ance Predictions, in Transportation Research Board 91st Annual Meeting. 2012, Transportation
Research Board: Washington DC.
Qiao, Y., et al., Examining Effects of Climatic Factors on Flexible Pavement Performance and Serv-
ice Life. Transportation Research Record: Journal of the Transportation Research Board, 2013.
2349(-1): p. 100−107.
Bandara, N., S. Henson, and K. Klieber, Creating a Climate Zone Map for Mechanistic Empirical Pave-
ment Designs, in T&DI Congress 2014. 2014, American Society of Civil Engineers. p. 682–691.
Harvey, J.T., A. Chong, and J.R. Roesler, Climate Regions for Mechanistic-Empirical Pavement Design
in California and Expected Effects on Performance. 2000, University of California-Berkeley, Pave-
ment Research Center: Richmond, CA.
Li, R., C.W. Schwartz, and B. Forman, Sensitivity of Predicted Pavement Performance to Climate Char-
acteristics, in 2013 Airfield and Highway Pavement Conference. 2013, American Society of Civil
Engineers: Los Angeles, CA.
Schwartz, C.W., et al., Sensitivity Evaluation of MEPDG Performance Prediction. 2011: Washington
DC.
983
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Air pollution has been a serious problem in China. Automobile exhaust gas
is an important source. Titanium dioxide (TiO2) has been long used as photocatalyst against
exhaust gas. One key problem lies in how to integrate it on pavement. In this paper, we stud-
ied solutions of carrying nanoparticles of TiO2 on the surface of pavement for an active deg-
radation of exhaust gases. Epoxy resin was selected as the matrix material. An epoxy resin fog
seal and an epoxy resin gravel thin layer topping were designed. The mechanical performance
and exhaust gas degrading effect of these materials were studied. A test section of was built
on a cement concrete pavement within a toll station. The concentration of exhaust gas in
the toll zone was tested before and after the treatment. The solution turned out to effectively
reduce the exhaust gas level and was strongly recommended for similar projects.
1 INTRODUCTION
Currently, air pollution is serious in China. The air quality of 17.2% urban area cannot meet
the national standard of Level II. The air pollution has become increasingly prominent in
the regions of Pearl River Delta, Yangtze River Delta, Beijing, Tianjin and Hebei, and the
number of haze days has accounted for 30% 50% in Guangzhou, Shenzhen, Shanghai and
other big cities. The air pollution caused by automobile exhaust has reached 50%, and the
concentration of carbon monoxide and nitrogen oxide in most large and medium cities is
more than the threshold, causing a high air quality index (PM2.5). Research has proven that
PM2.5 caused by automobile exhaust and industrial pollution is one of the culprits leading
to respiratory cancer. Seen from the composition of PM2.5 particles (Fig. 1), automobile
exhaust accounts for the largest proportion as 30%, which has caused serious pollution to
the environment. Vehicle emissions are important sources of urban PM2.5, including carbon
monoxide, nitrogen dioxide, hydrocarbons and photochemical smog pollutants, which seri-
ously affect the air quality, and directly endanger the public health.
Currently, exhaust emissions are increasing with the traffic volume, and the pollution
problem is becoming more and more serious. Especially in the highway toll station, exhaust
emissions is on a high level due to the large traffic volume, serious overload phenomenon,
frequent idle, acceleration and deceleration of vehicles. Therefore, it is urgent to put forward
an effective exhaust degradation technology to improve ambient air quality in toll stations
and to reduce the damage of harmful gas to the drivers, charging members and management
staff.
Nanometer photocatalytic material (TiO2), which is used as a new type of air purification
material, has great advantages, such as photocatalytic activity, chemical stability and recy-
clable property. At present, the photocatalytic material has realized industrialization in the
scope of environmental sanitation and environmental protection industries.TiO2 is gradually
applied in the transportation industry for the degradation of automobile exhaust gas, such
as coated on the highway noise barrier, street lamps and other areas, but seldom used on
pavement of toll stations.
985
Figure 1. Composition of PM2.5 particles. Figure 2. Automobile exhaust pollution.
In this paper, titanium dioxide is mixed in the matrix of epoxy resin. It is then coated on
the pavement of toll station in the way of the fog seal and thin layer topping for exhaust gas
degradation.
2 PHOTOCATALYSIS PRINCIPLE
The photocatalytic properties of nano TiO2 are closely related to the particle size in the following
ways: (1) the smaller the particle size, the larger the specific surface area for the acceptation of
ultraviolet ray and catalytic reaction; (2) the band gap width is bigger if the particle size is smaller
than 10 nm, meanwhile a stronger oxidation-reduction tends to be produced at the band edge
without free energy of charge transfer increased between the particle surface and the solvent;
(3) the smaller the particle size, the more possible the blue shift of the absorption wavelength;
(4) The smaller the particle size, the higher the surface energy. A large number of experimental
studies show that the smaller the particle size, the higher the photocatalytic efficiency. But if the
particle size is small to a certain extent, a second aggregation will appear which will lead to the
decline of catalytic efficiency. Therefore, it is appropriate to choose the anatase titanium dioxide
within 20–50 nm as the photocatalyst. The catalytic principle is shown in Fig. 3.
Cold-mix and weather resistant epoxy resin is selected as the base material for exhaust degra-
dation. Due to its good adhesion to the cement concrete pavement, it is not easy to peel off.
And the photocatalyst can be dissolved and dispersed easily in epoxy resin, so it will not be
washed away by rain water. The skid resistance and wear resistance are both perfect. Based
986
on the advantages of cold mixed epoxy resin, two schemes have been raised: epoxy resin fog
seal and epoxy resin thin layer topping.
987
Figure 6. Structure of epoxy resin thin layer topping.
Technical
Item Unit Result requirements
durability, high tensile strength, high breaking extension rate, good permeability for crack
healing, restoration of pavement structure strength and fine seal waterproof. Spreading grav-
els can improve the pavement anti-slide and anti-wear performance to ensure the driving
safety and comfort, as shown in Table 3.
988
Table 3. Test of anti-wear performance of thin layer topping.
Technical
Item Unit Result requirements
Figure 9. The exhaust degradation efficiency with variation of the amount of titanium dioxide.
989
Figure 10. The exhaust degradation efficiency with time variation.
Scheme
990
Figure 11. Field test value of exhaust concentration before the implementation of program.
Figure 12. Field test value of exhaust concentration after the implementation of program.
4. The thickness of fog seal solution spraying is so small as to call for high performance
requirements on the original pavement. However, the thickness of thin layer topping is
relatively bigger to be more effective to cure diseases of the original pavement and recover
the pavement structure strength. Therefore, this scheme is more suitable for the pavement,
whose diseases are a little more serious, with a relatively high cost.
According to the present situation of highway toll station, with the economic, technical
and other aspects to be considered, the two schemes are compared in Table 4.
4 FIELD TEST
A cement concrete pavement in the scope of a toll station with large traffic volume has been
selected as the test section. The concentration of exhaust gas in the toll station was detected
before and after the treatment in section I and II of the fog seal scheme and in section III of
the thin layer topping scheme, as shown in Figure 11 and Figure 12. From the results of field
tracking, CO, CH and NOx can be effectively degraded by both of the two schemes.
5 CONCLUSIONS
Based on the laboratory and field test on the two different schemes of applying nano-particle
of titanium dioxide in epoxy resin for exhaust gas degradation, the following conclusions can
be drawn:
1. The test scheme of the exhaust gas degradation with a portable exhaust gas analyzer can
meet various working conditions, both for laboratory use and for the detection in scene.
2. TiO2 photocatalyst has significant degradation effect on NOx, CH and CO from automo-
bile exhaust gas with a fast degradation rate and high efficiency within only 10 min.
3. The content of TiO2 has a great influence on the degradation efficiency. The degradation
efficiency reaches the highest when the TiO2 content is about 8%∼10%. For high degrada-
tion efficiency, the content of TiO2 should be strictly controlled to avoid an insufficient
991
contact with exhaust gas at too low content, and to prevent clumping of the nanoparticles
at too high content.
4. The degradation effect of the two schemes is roughly equal. And epoxy resin has good
adhesive property to the original concrete cement pavement. The construction of fog seal
is more convenient with lower cost, but this scheme calls for higher requirements on the
original pavement. However, it is more effective to cure distresses of the original pavement
and to recover the pavement structure strength by the scheme of thin layer topping. With
slightly higher cost and larger demand for labor, the thin layer topping scheme can greatly
improve the durability, skid resistance and wear resistance of the pavement.
REFERENCES
Dong Xiang. Pavement Performance of Composite Material with Rubber Powder Loading Photocat-
alyst for Pavement Automobile Exhaust Degradation [J]. Journal of Building Materials, 2011.12,
14(6).
Fu Jishima A. Photocatalytic Material and Applications [J]. Function and Materials, 2000, 20(1): 9–15.
Hua Tong, ShuxinOuyang. Nano-Photocatalytic Materials: Possibilities and Challenges [J]. Advanced
Materials, 2011,1–23.
Lei Yuzi. Research on the Degradation of Automobile Exhaust and Mitigation of Urban Heat Island
Effect of Bituminous Mixture [D]. Xi’an,China: Chang’an University, 2014.
Li Li, QianChunxiang. Study on the Removal of Nitrogen Oxides Emissions from Automobile Exhaust-
byPhotocatalytic Functional Concrete Road on Nanjing Yangtze River Bridge [J]. Journal of Henan
University of Science and Technology: Natural Science, 2009.2, 30(1).
Lu Juncai, Chen Huolin. Preparation of Modified TiO2 Particles and Photocatalytic Properties [J]. Joue-
nal of Wuyi University (Natural Science Edition). 2009.8, 23(3).
Mao Xiaoou, Dong Ying. Study on the Material for Exhaust Degradation Applied in Road Surface [J].
Shanxi Architecture, 2013.2, 39(5).
Muggli D S, Falconer J L. Role of Lattice Oxygen in Photocatalytic Oxidation on TiO2 [J]. Journal on
Catalysis, 2000, 191(2): 318–325.
Qi Feng, Li Fulin. Experiment Research on Curing Indoor Pollution by Photocatalyst[J]. Chemical Engi-
neer, 2010,179(8).
Tan Yiqiu, Li Luoke.Application Performance Evaluation on Material of Automobile Exhaust Degra-
dation in Asphalt Pavement [J].China Journal of Highway and Transport. 2010, 11.
Wang Xiaojing, Liu Chao. Preparation, Charaterization and Photocatalytic Performance of Porous
TiO2 Micro-beads [J]. Chinese Journal of Catalysis, 2008.4, 29(4).
Wei Peng. The Study on the Asphalt Pavement Degrading Tail Gas of Mobile [D]. Harbin,China: Harbin
Institute of Technology, 2008.
ZhaiZengxiu, ZouKehua. Experimental Research on Photocatoytic Degradation of Nano-TiO2 for Cok-
ing Wastewater [J]. Environmental Sanitation Engineering, 2011.10, 19(5).
ZhengGuanghong, Feng Xining. Study on Application of Nano Titanium Dioxide Photocatalyst [J]. Dye-
stuffs and Coloration, 2004.10, 41(5).
Zhang Wengang. Experimental Studies on Automobile Exhaust PhotocatalyticDegradationly Asphalt
Pavement Material [D]. Xi’an, China: Chang’an University, 2014.
Zhao Ming. Asphalt Pavement TiO2 Degradation of Vehicle Exhaust (NOx) Monitoring and Effective-
ness Evaluation Studies [D]. Harbin,China: Northeast Forestry University, 2013.
Zhang Xiaoxing. The Preparation of Photocatalytic Cement Paste and Research on Degradation of
Automobile Exhaust [D]. Changsha, China: Central SouthUniversity, 2014.
Zhu Huayu, Zhang Qiuyan. Research Progress on Photocatalytic Cement Materials [J]. Bulletin of
Chinese Ceramic Society, 2013.3, 32(3).
992
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Hongduo Zhao
Key Laboratory of Road and Traffic Engineering of Ministry of Education, China
Zhongpu Lin
Tongji University, Shanghai, China
Luyao Qin
Key Laboratory of Road and Traffic Engineering of Ministry of Education, China
1 INTRODUCTION
The asphalt pavement will sustain million times of vehicle loads during its life time. The
load of moving vehicle can cause stress, strain, deformation and vibration in pavement. The
energy from vehicle moving over pavement, if not harvested, can cause damage to the pave-
ment and lost as heat.
The piezoelectric technology is a potential way to harvest mechanical energy from asphalt
pavement. Piezoelectric material has the ability to convert stress into electricity. The work by
Haocheng Xiong(1997, 2012) suggested that piezoelectric material, among varies energy har-
vesting methods, is most suitable for asphalt pavement. The piezoelectric material, combined
with a specific structure, can be made into different kinds of piezoelectric transducer. A force
amplifying structure is a structure that can amplify the load applied to piezoelectric material.
By carrying heavier load, the piezoelectric material can produce higher voltage, which leads
to higher energy conversion efficiency.
The existing piezoelectric structures include Multilayer, Cymbal, Moonie, Bridge, THUN-
DER, MFC, Bimorph and so on. Cymbal, developed from Moonie, is studied extensively by
worldwide researchers. J.F Tressler(1995) built the vibrational model of Cymbal structure and
dynamic analysis is conducted base on that model. The road performance of Cymbal is tested
by Zhao Hongduo(2010), using ABAQUS to study the energy conversion efficiency of differ-
ent construction form of Cymbal. Tao Yujie(2013) design a Piles structure for piezoelectric
transducer. An Israeli corporation called IPEG claimed that they have developed a whole
system of harvesting energy from pavement in 2008 and an experimental road has been built.
This paper compared piezoelectric transducer with different force amplifying structures,
using index such as force amplifying coefficient k, energy density of unit ceramic volume dv
and energy density of unit site area ds. Finite element analysis is applied to analyzing different
structure and the performance of different transducer is presented.
993
2 RELATIONSHIP BETWEEN HARVESTED ENERGY AND APPLIED FORCE
Without an external electric field, the energy that piezoelectric material produce under repeti-
tive loading in the polarization direction is shown in the following equation:
1 2 1 2 ε rT ε 0 A 1 F32t p
E U3 C U3 = d3i g3i (1)
2 2 tp 2 A
There are different kinds of structure that can increase force. Thus a set of index is needed to
evaluate the performance of different transducers. Considering both the structure’s ability to
amplify force and the amount of material and space the structure use, the force amplifying
coefficient k, energy density of unit ceramic volume dv and energy density of unit site area ds
are used to evaluate the efficiency of the transducers(Lu, Y.Q. 2015).
The force amplifying coefficient k is calculated as follow:
From equation (2), it is noted that k is used to evaluate the structure’s ability to amplify
force. The higher k is, the heavier load will be applied to the piezoelectric material, which also
leads to higher voltage and more energy.
While in some case, a large k may require a large amount of space, which leads to low
energy conversion efficiency. Here we introduce energy density into the evaluation index.
Energy density can be divided into two parts, energy density of unit ceramic volume dv and
energy density of unit site area ds.
The energy density of unit ceramic volume dv is determined as follow:
dv mainly characterize the energy conversion efficiency of unite ceramic volume of a certain
structure. The energy density of unite site area ds is determined as follow:
994
the energy a transducer produces
ds = (4)
the amount of area a transducer
a covers
ds reflects the energy conversion efficiency of unite site area of a certain structure. Combining
the force amplifying coefficient and energy density, different structure’s ability to converge
energy can be fully evaluated.
There are plenty of structure that has the ability to amplify force. However, the depth of
asphalt pavement structure is around 20 cm, and the equivalent radius of traffic load is only
10.65 cm, which provides little space for the transducer. So complicate structures, such as
lever, gear and pulley, are not allowed in this case. This paper mainly discuss the structures of
Piles, Torus, Trapezoid and Bridge.
995
Figure 2. The 21 columns circle piles piezoelectric transducer.
cialty about Torus is the stiffness of the whole structure can be adjusted by changing the thick-
ness of two rings. In that way the Torus transducer’s stiffness can fit the pavement properly.
qLL2o
Tx = (5)
4t p (ti + tc )
Where L0, tp, ti and tc refer to the size of different part of bridge structure.
Through mechanical analysis of bridge structure (Li, D. H, 2011), stress concentration is
found on the edge of the bonding layer between cap and piezoelectric ceramic block. The
996
Figure 4. The trapezoid structure (mm).
stress concentration not only consume part of the energy from vehicle passing over the pave-
ment, but also result in low strength. The maximum tensile stress and shear stress can lead to
various kinds of failure. The stress concentration can be relieved by changing the geometry
of cap. So in order to improve the energy conversion efficiency of Bridge transducer, the arch
Bridge structure is designed. The detail is shown in figure 6.
Compared with Bridge structure, the arch Bridge structure move the bonding layer edge
near the main structure from the inner side to outer side of bonding layer. The stress con-
centration may well be moved to the edge of the ceramic and the force amplifying effect can
be improved due to increase of cap’s area and the decrease of angel between bevel edge and
lateral edge of cap.
In this part, FE analysis software ABAQUS is used to build FE model for different force
amplifying structure and analyze all the structure’s performance. The result includes three
index, the force amplifying coefficient k, energy density of unit ceramic volume dv and energy
density of unit site area ds.
The force amplifying coefficient k of different structures are shown as table 1:
As can be seen, Piles and two kinds of Bridge structure have a relatively high k. But the FE
result of Torus and Trapezoid differs greatly from the theoretically simplified one.
The FE result of Torus structure is around 70% of the theoretically simplified result. One
of the reason Torus structure fail to amplify force is that the two ring’s deformation takes up
part of energy, leaving less to be passed to piezoelectric ceramic block. The other reason is
that the vertical load dose not all transfer into lateral compressive stress. The stress distribu-
997
Table 1. FE analysis result of k.
Figure 7. The stress distribution of force transmission block in the lateral direction.
tion through the surface between force transmission block and ring in the lateral direction is
fetched from the FE model and shown as follows:
Figure 7 clearly shows that there are compressive stress and tensile stress in the force trans-
mission block at the same time. The area that is subjected to tensile stress takes up about half
of the whole surface. Maximum tensile stress reaches 11.16 MPa. This part of tensile stress
offset the compressive stress that is transmitted from the ring, and leaves even little to reach
the piezoelectric ceramic block.
As for Trapezoid structure, compared with theoretically simplified result, FE result is
reduced by two third. The lateral and vertical stress distribution through piezoelectric ceramic
block is shown as follow:
The area around point No.28 is subjected to a higher level of stress, which means the force
does not distribute evenly through the piezoelectric block and that affects force amplifying
998
Table 2. Energy density from FE analysis result.
effect. Another reason Trapezoid structure fail to amplify force is that much of load from
moving vehicle is offset by the static friction between force transmission block and piezo-
electric ceramic block. The stress statistics from the FE model shows that the angel between
vertical force and lateral force is 26°, proves that the vertical load did not all transfer to lateral
force, but largely offset by static friction.
The energy density of different structures is shown as follow:
The Piles structure collected the most energy, 6.95 mJ. However, ds result showed part
of the reason why Piles harvested the most energy is that it covers a large amount of area.
Torus and Trapezoid structure did not show great energy density mainly because they fail
to amplify force and the energy they converse is not enough. Two Bridge structures have
similar results. Arch Bridge has the best performance on energy density (dv and ds ) and the
second best performance on energy. Also it performs better than the normal Bridge structure
in all three index. In summary, the Arch Bridge structure is able to converse large amount
of energy given limited material and space, which makes it the optimum structure for energy
harvesting in asphalt pavement.
6 CONCLUSIONS
This paper analyzed the energy conversion efficiency of different piezoelectric transducer
with different force amplifying structure. Theoretically simplified calculation and FE simu-
lation technology is used to evaluate different structure. Base on the evaluation result, the
following conclusion are made:
1. A single force amplifying coefficient is not enough to fully evaluate a structure. The energy den-
sity is introduced to determine the overall energy conversion efficiency of different structure.
2. While the theoretically simplified result looks good, the FE analysis shows that Torus and
Trapezoid structure fail to amplify force, which makes them hard to harvest energy in the
pavement.
3. Piles structure has the highest force amplifying coefficient and a decent energy density. It har-
vest the most energy while takes up the more area than Bridge structure at the same time.
4. The performance of Bridge structure is similar. With the highest energy density of unit
ceramic volume dv and energy density of unit site area ds, arch Bridge structure performs
better in all three index than normal Bridge structure.
5. Overall, the arch Bridge structure shows the best energy conversion efficiency of all struc-
ture. Thus it is the most suitable structure to be used in the pavement.
REFERENCES
Li, D.H & Yao X. 2011.Research on Cymbal Piezoelectric Transducer Arrays. Piezoelectrics & Acous-
tooptics. 23(3):236–239.
Lu, Y.Q. 2015. Synergistic Mechanism and Optimization between Piezoelectric Transducer and Asphalt
Pavement. Shanghai.
999
Tressler, J.F & Dogan, A & Fernandez, J.F & Fielding, J.T & Uchino, K. Jr & Newnham, R.E. 1995.
Proceedings of the 1995 IEEE Ultrasonics Symposium, Nov. 710, Seattle, U.S.A.
Tao, Y.J. 2013. The structure and performance of piezoelectric transducer for harvesting energy in
asphalt pavement. Shanghai.
Xiong, H.C & Wang, L.B & Wang, D & Cristian, D. 1997. Piezoelectric Energy Harvesting from Defor-
mation of Pavements. Int. J. Pavement Res. Technol.5(5): 333–337.
Xiong, H.C & Wang, L.B & Wang, D & Cristian, D. 2012. Piezoelectric Energy Harvesting from Traf-
fic Induced Deformation of Pavement. International Journal of Pavement Research and Technology.
Vol.5: 333–337.
Yan, Z.F. 2012. Power Generation Performance of Cymbal Piezoelectric Vibration Energy Harvester.
Guangdong: Guangzhou.
Zhao, H.D & Yu, J & Ling, J.M. 2010.Finite element analysis of Cymbal piezoelectric transducers
for harvesting energy from asphalt pavement. Journal of the Ceramic Society of Japan, 118(10):
909–915.
Zhou, Y.F & Zhang, H.R. 2005. Research on Relation Between of surf ace Characteristics and fuel
Consumption. Highway, 01(1):30–36.
1000
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: As is well known, pavement features affect traffic noise. In more detail, rolling
noise depends on generation factors and acoustic absorption. This latter can be assessed
through in-lab (ISO 10534-2) and/or on-site (ISO 13472-1; ISO 13472-2) tests but several
issues arise when comparing measures carried out according to different measurement meth-
ods. Consequently, the objectives and scope of the study described into this paper were to
compare different methods in order to derive relationships to use in practical applications.
The following devices were used: in-lab Kundt tube; on-site Kundt tube; Adrienne device.
Based on measurements, absorption spectra were derived. Analyses and results focused also
on composite indicators and first-pick frequencies and values. Outcomes can benefit both
practitioners and researchers. Future research will address a number of issues arouse during
experiments and analyses and will address a better understanding of phenomena and reasons
which originated the relationships which emerged from the study.
1 INTRODUCTION
As is well known, noise pollution is one of the most important environmental problems in
Europe (see, 2002/49/CE; SMILE workshop 2003; Freitas et al, 2012; Praticò and Anfosso,
2012; Kihlman et al, 2014). Power unit and tyre-road contact are the main sources, and in the
mid-to-high speed range (above 40 km/h–80 km/h) the main contributor to traffic noise is
rolling noise (Sandberg and Ejsmont, 2002). These phenomena generate acoustical pressures
(Weyl-Van Der Poel’s equation) which generate loudness (Attenborough, 1983; Praticò, 2001;
Frías et al, 2011; Jimenez-Espadafor et al., 2011). Tyre/road noise may vary more than 15 dB,
based on tyre and pavement type, and noise reduction at the source can be more cost-effective
than treatments on the buildings or on the propagation path (e.g., noise barriers). Genera-
tion factors (pavement texture, tyre type, etc.) and acoustic absorption govern rolling noise
(Luong et al, 2014). In more detail, pavement composition, volumetric and surface proprie-
ties affect generation, absorption and propagation (Praticò, 2001; Praticò et al, 2010; Praticò
and Moro, 2011; Praticò and Vaiana, 2013; Praticò and Vaiana, 2012; Vazquez et al, 2016).
Based on the above, road agencies require satisfactory level of acoustic absorption, which
is usually quite appreciable in porous asphalt concretes.
Acoustic absorption can be assessed through in-lab (ISO 10534-2) and/or on-site (ISO
13472-1; ISO 13472-2) tests but several issues arise when comparing measures carried out
according to different measurement methods.
2 OBJECTIVES
Based on the above, the objectives and scope of the study described into this paper were to
compare different methods in order to derive relationships to use in practical applications.
1001
Figure 1. Main tasks.
3 DESIGN OF EXPERIMENTS
As abovementioned, the absorption coefficient was measured through three different devices:
AD: Adrienne device (ISO 13472-1, see also Londhe et al, 2009); KS: on site Kundt device
(ISO 13472-2, method for reflective surfaces); KL: in-laboratory Kundt device (ISO 10534-2,
see also Wolkesson, 2013). Note that the hypothesis which underlies this analysis is the homo-
geneity of the surface.
Figures 2–7 and Tables 1–7 illustrate main results and analyses.
By referring to the probability density function (PDF, y-axis) of the absorption coefficient
(x-axis) for the different devices, for each frequency, Figure 2 illustrates the main results.
In particular it shows the probability density function of the acoustic absorption coef-
ficient, for a given frequency (e.g., 250 Hz). Frequencies in the range 250–1600 Hz are
considered.
At 250 Hz, note that the in-lab device (KL) provides the lowest value (about 0.1) while the
on-site Kundt tube provides the highest value (about 0.3). Furthermore note that the high-
est PDF corresponds to the KS, which entails, in this case, that the standard deviation is the
1002
Figure 2. Probability density function of the absorption coefficient for the three devices, for each
frequency. Symbols. PDF: probability density function. AD: Adrienne; KS: on-site Kundt tube; KL:
in-lab Kundt tube.
lowest (highest precision). In contrast, the lowest PDF corresponds to the AD, which implies
the lowest precision. Based on these data, no evidence can be associated towards the best
accuracy (degree of closeness of measurements to true value).
When different frequencies are considered, the following observations can be pointed out:
i) the precision decreases in the range 250–800 Hz. Afterwards, it slightly increases in 800–
1600 Hz; ii) in-lab measures usually yield lower values of absorption; iii) extended surface
method (AD) usually yields higher values of acoustic absorption coefficient.
1003
Figure 3. Probability density function of the absorption coefficient, for each frequency domain. Sym-
bols. 400–630: frequency range (Hz); 800–1250: frequency range (Hz); PDF: probability density. func-
tion. AD: Adrienne; KS: on-site Kundt tube; KL: in-lab Kundt tube.
Figure 4. Curve envelopes of the absorption coefficient (maximum and minimum) for the three
devices. Symbols. α: absorption coefficient (dimensionless, defined in 0, 1). Max: envelope curve of the
highest values for the given frequency (i.e., 630 Hz), for the given device (i.e., AD). AD: Adrienne; KS:
on-site Kundt tube; KL: in-lab Kundt tube.
Figure 5. Average absorption coefficient and lower specification limits. Symbols. AD: Adrienne; KS:
on-site Kundt tube; KL: in lab Kundt tube; α: absorption coefficient.
As is well known, contract specifications often require to merge different frequencies into
an average value (e.g., ANAS, 2010: 400–630; 800–1250; 1600–2500 Hz). To this end, by refer-
ring to the probability density function of the absorption coefficients for the different devices,
for each frequency range, Figure 3 illustrates the main results (the range 1600–2500 Hz is not
considered because KS and KL don’t provide data for these frequencies).
Even when frequency ranges are considered, note that KL usually provides the lowest
values of absorption.
1004
Figure 6. Absorption coefficient in 400–630 Hz (averages). Symbols. AD: Adrienne; KS: on-site
Kundt tube; KL: in lab Kundt; dotted line: equality line.
Figure 7. Absorption coefficient in 800–1250 Hz (averages). Symbols. AD: Adrienne; KS: on-site
Kundt tube; KL: in lab Kundt; dotted line: equality line.
Table 1. Main results for the linear trendlines applied to Figures 6 and 7.
y = mx + q
Devices Range [Hz] m q R2 y = m 1 × m1
By referring to the highest and lowest values of the absorption spectra (peaks), for each
frequency, for each device, Figure 4 illustrates the main results.
In particular it shows the envelope curves for the three devices. By referring to the maxi-
mum values (Max), note that all the devices yield a maximum envelope curve which has a
relative minimum (or/and an appreciable variation of the first derivative towards zero) at
about 630 Hz. In contrast the three devices yield a relative maximum at about 500 Hz and an
absolute maximum at about 800–1250 Hz. It may be observed that the lower the frequency is
the higher the discrepancy among the three outputs becomes.
For the envelopes which refer to the minimum values (Min), note that local and global
extrema appear quite uncorrelated. For example the ADMin curve presents a local minimum
at about 1250 Hz which appears absolutely unrealistic and which doesn’t have any corre-
spondence with the remaining measurements.
1005
Table 2. Main statistics for Adrienne measurements.
Adrienne 250 Hz 315 Hz 400 Hz 500 Hz 630 Hz 800 Hz 1000 Hz 1250 Hz 1600 Hz
Average 0.24 0.31 0.36 0.43 0.58 0.68 0.60 0.45 0.46
St dev 0.09 0.05 0.05 0.14 0.15 0.17 0.19 0.14 0.11
St dev/average 0.39 0.17 0.14 0.32 0.25 0.25 0.32 0.32 0.25
Numb of meas 19 19 19 19 19 19 19 19 19
Min −0.09 0.13 0.26 0.27 0.35 0.31 0.23 0.06 0.28
Max 0.39 0.37 0.50 0.86 0.88 0.90 0.87 0.76 0.69
P (0.5) 0.24 0.31 0.36 0.40 0.57 0.70 0.56 0.47 0.45
Skewness −2.07 −2.03 0.86 2.03 0.36 −0.54 −0.10 −0.51 0.67
Observation n* n* p* p* q* n* q* n* p*
Kurtosis 7.29 5.27 3.24 4.50 −0.44 −0.41 −1.05 2.16 −0.15
Observation Leptokurtic Leptokurtic Leptokurtic Leptokurtic Mesokurtic Mesokurtic Platykurtic Leptokurtic Mesokurtic
Symbols. n*: negative skewed (that is to say, left-skewed, left-tailed, skewed to the left: The left tail is longer); p*: positive skew (i.e., right-skewed, right-tailed, skewed
to the right: The right tail is longer); q*: quasi-symmetric distribution; P(0.5): 50% percentile
1006
Table 3. Main statistics for on-site Kundt measurements.
On-site Kundt 250 Hz 315 Hz 400 Hz 500 Hz 630 Hz 800 Hz 1000 Hz 1250 Hz 1600 Hz
Average 0.30 0.26 0.25 0.27 0.33 0.47 0.58 0.54 0.41
St dev 0.03 0.03 0.04 0.09 0.13 0.16 0.17 0.20 0.12
St dev/average 0.09 0.11 0.17 0.32 0.39 0.34 0.29 0.36 0.30
Numb of meas 31 31 31 31 31 31 31 31 31
Min 0.24 0.20 0.17 0.16 0.17 0.23 0.27 0.25 0.20
Max 0.34 0.31 0.39 0.62 0.64 0.75 0.88 0.96 0.61
P (0.5) 0.30 0.27 0.24 0.24 0.29 0.44 0.56 0.59 0.44
Skewness −0.86 −0.39 1.11 2.34 0.84 0.40 −0.14 0.09 −0.20
Observation n* q* p* p* p* q* q* q* q*
Kurtosis 0.44 −0.56 2.58 7.27 −0.32 −1.00 −0.80 −1.12 −1.17
Observation Leptokurtic Leptokurtic Leptokurtic Leptokurtic Mesokurtic Mesokurtic Platykurtic Leptokurtic Mesokurtic
Symbols. n*: negative skewed (that is to say, left-skewed, left-tailed, skewed to the left: The left tail is longer); p*: positive skew (i.e., right-skewed, right-tailed, skewed
to the right: The right tail is longer); q*: quasi-symmetric distribution; P(0.5): 50% percentile
Table 4. Main statistics for in-lab Kundt measurements.
On-site Kundt 250 Hz 315 Hz 400 Hz 500 Hz 630 Hz 800 Hz 1000 Hz 1250 Hz 1600 Hz
Average 0.08 0.11 0.18 0.28 0.39 0.52 0.46 0.33 0.31
St dev 0.02 0.04 0.09 0.14 0.13 0.19 0.16 0.10 0.10
St dev/average 0.28 0.36 0.50 0.52 0.34 0.37 0.35 0.30 0.34
Numb of meas 19 19 19 19 19 19 19 19 19
Min 0.05 0.06 0.09 0.12 0.16 0.21 0.15 0.20 0.18
Max 0.12 0.22 0.44 0.64 0.63 0.79 0.77 0.61 0.58
P (0.5) 0.07 0.10 0.14 0.22 0.40 0.61 0.48 0.33 0.31
Skewness 0.39 1.08 1.55 1.34 0.04 −0.19 0.04 1.16 0.91
Observation q* p* p* p* q* q* q* p* p*
Kurtosis −1.14 0.45 2.07 1.30 −0.52 −1.37 −0.31 1.99 0.61
Observation Platykurtic Mesokurtic Leptokurtic Leptokurtic Platykurtic Platykurtic Mesokurtic Leptokurtic Leptokurtic
Symbols. n*: negative skewed (that is to say, left-skewed, left-tailed, skewed to the left: The left tail is longer); p*: positive skew (i.e., right-skewed, right-tailed, skewed
to the right: The right tail is longer); q*: quasi-symmetric distribution; P(0.5): 50% percentile
1007
Adrienne On-site Kundt In-lab Kundt
Range 400–630 Hz 800–1250 Hz 1600–2500 Hz 400–630 Hz 800–1250 Hz 400–630 Hz 800–1250 Hz
Average 0.46 0.58 0.54 0.28 0.55 0.29 0.42
St dev 0.09 0.10 0.07 0.08 0.12 0.13 0.14
St dev/average 0.19 0.18 0.14 0.31 0.22 0.45 0.33
Numb of meas 19 19 19 31 31 19 19
Min 0.34 0.34 0.36 0.17 0.32 0.10 0.17
Max 0.64 0.74 0.65 0.59 0.74 0.54 0.67
P(0.5) 0.45 0.61 0.54 0.26 0.57 0.25 0.38
Skewness 0.46 −1.00 −0.75 1.84 −0.37 0.56 0.02
Observation q* n* n* p* q* p* q*
Kurtosis −0.61 0.72 0.42 4.62 −0.82 −0.81 −0.79
Observation Platykurtic Leptokurtic Mesokurtic Leptokurtic Platykurtic Platykurtic Platykurtic
Threshold 0.25 0.50 0.25 0.15 (+) 0.43 (+) 0.15 0.30
Ratio 1.84 1.16 2.16 1.87 1.28 1.93 1.4
Symbols. n*: negative skewed (that is to say, left-skewed, left-tailed, skewed to the left: The left tail is longer); p*: positive skew (i.e., right-skewed, right-tailed, skewed
to the right: The right tail is longer); q*: quasi-symmetric distribution; P(0.5): 50% percentile. (+): working hypothesis.
Table 6. Conclusions.
AD KS KL
Figure 5 shows how measurements carried out by the three abovementioned devices match
with lower specification limits (reference: ANAS, 2010).
The same pavement presents slight issues when analysed through the methods AD and
KL, while it would be considered totally compliant in the case KS.
Figures 6 and 7 refer to the comparison among the three devices in the range 400–630,
800–1250 and report linear regressions.
In more detail, Figure 6 refers to the estimate of the acoustic absorption coefficient in the
range 400–630 Hz and highlights how linear regressions (with zero intercept or not) compare
with equality line (dotted line) and data (averages).
Note that even if non-zero intercept linear regressions yield the best results (highest
R-square values), when Kundt-type measurements are considered (i.e., KS-in situ, and KL-in
the laboratory), zero-intercept curve has a slope quite close to one and curve seems to explain
an appreciable percentage of data variations.
This occurrence is very relevant because of the high speed of KS with respect to KL meas-
urements and call for further research and investigation. Importantly, it refers to a domain
in which Adrienne-type measurements can present accuracy and precision issues and draw-
backs (see Figure 3).
Figure 7 refers to the range 800–1250, where human hearing is generally considered to be
far more sensitive to sounds. Note that KL measurements are matched quite well by both
AD and KS. In more detail, KS and AD curves seem to overestimate lower values even if
they explain 50–60% of data variance. Importantly, zero-intercept curves have a slope quite
close to one.
Tables 2–7 summarise the main statistics of data gathered.
Table 1 shows how y-intercepts (q) are quite high if compared to the range of α (0.1). In
contrast, m1 values (slopes when the intercept is assumed zero), are quite encouraging except
than for AD. Note that in the case of zero-intercept, the coefficient of determination resulted
always very low (<0.2).
Tables 5 refers to ranges of frequency.
As for dispersion indicators (see tables 2–5), note that standard deviation of the three indi-
cators give information about how much spread are the data and yield that the three devices
are comparable in terms of precision. Unfortunately, standard deviation is affected by the
magnitude of the data (average), and higher values yield higher standard deviations. In order
to limit this bias effect the coefficient of variation has been derived (see again tables 2–5).
Based on the results obtained it is possible to observe that the in-lab measurements (KL)
result less precise than the remaining two methods, because of the coefficient of variation
in the range 0.3–0.5 (see tables 4 and 5). In contrast, in terms of symmetry (skewness), the
on-site method provides outputs which are, on average, quite symmetric (see table 3) except
that for 500 Hz. In contrast, both the Adrienne and the in-lab method are right-skewed and
usually present a longer right tail., i.e., a few, very high values (see tables 2 and 4). In terms
of kurtosis, the three methods perform similarly, with a certain tendency towards leptokurtic
distributions. Importantly, in table 5 a working hypothesis is formulated about the threshold
value for the on-site measurements carried out through the KS device. This hypothesis origi-
nates from the assumption of having a ratio (in 400–630 and in 800–1250 Hz) which complies
with the ones obtained through the remaining two devices. Under the hypothesis of statistical
significance, it turns out that having the above thresholds (+) would imply, on average, a com-
mon ranking of pavement acoustic absorption, based on one out of the three devices.
1008
5 CONCLUSIONS
The work presented in this paper was carried out by considering a very high amount of data
and further analyses and checks are necessary. The following preliminary conclusions may
be drawn. The rationale which underlies the on-site Adrienne method is that the quantity of
energy reflected by the pavement surface depends on its acoustic absorption. This device per-
forms measurements in 0–4000 Hz, with a coefficient of variation of 0.1–0.4. This method is
time-consuming (about 30 minutes are necessary to perform one test). Its performance seems
to better match KL for high frequencies more than for low frequencies.
The rationale behind on-site Kundt tube is that pressures into the tube are affected by
the sample absorption. It is supposed to work properly only for reflective surfaces (as per
standard specifications). This notwithstanding its correlation with the in-lab device (which
is usually supposed to represent the best method) is relevant and appreciable. Importantly,
zero-intercept linear regression (KS versus KL) works quite properly and provides a straight-
forward tool for practitioners.
In-lab impedance tube is based on the same principle as on-site Kundt tube. Such as the
Adrienne method, it is time consuming (coring, transport, drying, core fixing in the tube).
Issues in fixing the core inside the tube can occur and can generate false response. Based
on previous studies, it seems to provide an outstanding estimate of the real spectrum of
absorption coefficient, as per theoretical models estimates (Stinson, Bereinger, etc.), even if
other validations of this assumption are required. Importantly, it has the worst coefficient
of variation.
Authors are aware that coring process might have affected results (Howard and Doyle,
2013; Praticò et al, 2014). Even if further studies are needed, from this study (see table below)
an overall scenario emerges in which the best accuracy is associated with the worst precision
(KL), the worst accuracy is associated with a questionable precision (AD), and the least
appreciated device (KS) provides reasonable (even if to correct), high-speed measurements.
REFERENCES
Attenborough, K. 1983. Acoustical characteristics of rigid fibrous absorbents and granular materials,
JASA, Vol.73.
Capitolato speciale d’appalto ANAS S.p.A. 2010 – Norme Tecniche.
European Commission. Directive on environmental noise, 2002/49/CE, 25th June 2002, DOCE L 89,
7/18/02.
Freitas, E., Mendonça, C., Santos, J.A., Murteira, C., Ferreira, J.P. 2012. Traffic noise abatement:
How different pavements, vehicle speeds and traffic densities affect annoyance levels, Transportation
Research Part D 17 321–326.
Frías, M., Jiménez-Mateos, J.M., Pfretzschner, J., Olmeda, J., Rodríguez, R.M., Sánchez de Rojas, M.I.
2011. “Development of blended cement mortars with acoustic properties using petroleum coke”,
Construction and Building Materials 25, pp 1086–1092.
Howard, I.L., Doyle, J.D. 2013. Investigating the consistency of asphalt density measurement methods
over a wide range of air voids. Journal of Testing and Evaluation, 42 (3).
Jimenez-Espadafor et al., 2011. Optimal design of acoustic material from tire fluff, Materials and Design
32, pp 3608–3616.
Kihlman, T., Kropp, W., Lang, W. 2014 Quieter cities of the future. Lessening the severe health effects
of traffic noise in cities by emission reductions. International Council of Academies of Engineering
and Technological Sciences, CAETS.
Luong, J., Bueno, M., Vázquez, V. F., Paje S. E. 2014. Ultrathin porous pavement made with high viscos-
ity asphalt rubber binder: A better acoustic absorption? Applied Acoustics 79 117–123.
Niranjan Londhe, Mohan D. Rao, Jason R. Blough. 2009. Application of the ISO 13472-1 in situ tech-
nique for measuring the acoustic absorption coefficient of grass and artificial turf surfaces, Applied
Acoustics, Volume 70, Issue 1. Pages 129–141.
Praticò, F.G. 2001. Roads and loudness: a more comprehensive approach, Journal of Road Materials
and Pavement Design, Volume2, issue 4/2001.
1009
Praticò, F.G., Anfosso-Lédée F. 1994. Trends and Issues in Mitigating Traffic Noise Through Quiet
Pavements, SIIV – 5th International Congress – Sustainability of Road Infrastructures, Procedia –
Social and Behavioral Sciences, 2012. E201994 – issn: 1887–0428.
Praticò, F.G., Vaiana, R., A study on volumetric versus surface properties of wearing courses, Construc-
tion and Building Materials, Volume 38, January 2013, Pages 766–775.
Praticò, F.G., Moro, A., Ammendola, R. 2010. Potential of fire extinguisher powder as a filler in bi-
tuminous mixes, Journal of Hazardous Materials, 173 (1–3), pp. 605–613.
Praticò, F.G., Moro, A. 2011, In-lab and on-site measurements of hot mix asphalt density: Convergence
and divergence hypotheses, Construction and Building Materials, 25 (2) 2011, pp. 1065–1071.
Praticò, F.G., Vaiana, R. 2013, A study on volumetric versus surface properties of wearing courses,
2013, Construction and Building Materials, 38, pp. 766–775.
Praticò, F.G., Vaiana, R. 2012, Improving infrastructure sustainability in suburban and urban areas: Is
porous asphalt the right answer? And how?, 2012, WIT Transactions on the Built Environment.
Pratico, F.G., Vaiana, R., Moro, A. 2014. Dependence of volumetric parameters of hot-mix asphalts on
testing methods. Journal of Materials in Civil Engineering. Volume 26, Issue 1, Pages 45–53.
Sandberg, U., and Ejsmont, J. A. 2002. Tyre / Road noise reference book, Informex, Kisa, Sweden.
SMILE workshop (2003). Guidelines for road traffic noise abatement. Berlin, Germany. <http://www.
smile-europe.org/>.
Vázquez,V. F., Luong, J., Bueno, M., Terán, F., Paje, S. E. 2016, Assessment of an action against envi-
ronmental noise: Acoustic durability of pavement surface with crumb rubber. Science of the Total
Environment 542 (2016) 223–230.
Wolkesson M. 2013. Evaluation of impedance tube methods. A two microphone in-situ method for
road surface and the three microphone transfer function method for porous materials. Department
of Civil and Environmental Engineering, division of Applied Acoustics, Vibroacoustics Group. Chalmers
University of Technology. Goteborg, Sweden 2013.
1010
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Zhi Suo
School of Transportation Engineering, Tongji University, Shanghai, China
Beijing Collaborative Innovation Center for Metropolitan Transportation, Beijing University of Civil
Engineering and Architecture, Beijing, China
Lijun Sun
School of Transportation Engineering, Tongji University, Shanghai, China
1 INTRODUCTION
The traditional method for the recycling of aged asphalt is to add mineral oil to the asphalt
in order to supplement the missing light oil component but there are still many shortcom-
ings. In view of the shortcomings of the traditional asphalt recycling technology and the
characteristics of vegetable oil, in the last 1–2 years the attempt to develop a recycling agent
with vegetable oil instead of mineral oil has emerged in foreign countries. They begun to use
vegetable oil instead of mineral oil to develop recycling agents. Even so, the research on the
technology of vegetable oil asphalt recycling remains in the macro level.
In this paper the aging of asphalt is simulated in the lab and vegetable oil recycling agent
is used for the aged asphalt recycling. Two new technical evaluation indexes, aging index and
recycling index, are proposed based on the method of Fourier infrared spectrum analysis. The
relationships between the two indexes and the basic physical properties of the aged asphalt and
vegetable oil recycled asphalt are investigated. Finally the two proposed indexes are used to
evaluate the aging degree of asphalt and the recycling effect of vegetable oil recycling agent.
1011
2.2 Preparation of aged asphalt
In this paper, we use the method of laboratory aging to prepare the aged asphalt. Specific
types and preparation methods are as follows:
Aged asphalt A: The 70# base asphalt is aged 75 min in the Rolling Thin Film Oven
(RTFOT);
Aged asphalt B: At first the 70# base asphalt is aged 75 min in the RTFOT, and then it is
placed in a Pressure Aging Vessel (PAV) during a total cycle of 20 hours.
Aged asphalt C: At first the 70# base asphalt is aged 75 min in the RTFOT, and then it is
placed in the PAV during a total of two cycles of 40 hours.
Specification
Test items Test results requirements
Absorption
peak position Absorption peak
1012
2.5 Asphalt physical performance test
In this paper, the physical properties of aged asphalt and vegetable oil recycled asphalt are
tested too. The measured indexes include penetration, ductility, softening point and Brook-
field viscosity. The ductility test was carried out at the temperature of 10°C. The test tempera-
ture of the rotary viscosity is 135°C.
3 TEST RESULTS
CII s =0
V=
CII s OBase
where: A = the integral area of the absorption peak; CI = relative integral area of absorption
peak; V = aging index.
The required absorption peak integral area data are obtained through the OMNIC analy-
sis software. The analysis results are shown in Table 4.
1013
Table 4. Aging index of asphalt.
ent amounts of plant oil recycling agent (2%, 4%, 6%, 8% and 10%, respectively) were tested.
The results are shown in Figure 2.
Figure 2 shows that after adding vegetable oil recycling agent to the aged asphalt, a new
absorption peak was created in the infrared spectrum at position 1745, i.e. a carbony(C = O)
stretch vibration absorption peak of aldehydes, ketones and acids. This absorption peak is not
in the matrix asphalt (Figure 1) and aged asphalt. It is explained by the fact that compared
with aged asphalt, the functional groups of the asphalt have some changes due to the addi-
tion of the vegetable oil. Obviously, with the increase of the amount of recycling agent, the
sulfoxide group (S = O) functional groups does not decrease but the carbony (C = O) absorp-
tion peak intensity of aldehydes, ketones and acids gradually increases. The explanation is
that the recovery of the chemical composition of asphalt is not achieved by adding vegetable
oil, which may be due to the different functional groups in vegetable oil and asphalt. Given
the above situation, based on the Carbony (C = O) stretch vibration absorption spectra of
aldehydes, ketones and acids, the Recycling index U is proposed as the technical evaluation
index of the recycling effect of plant oil. The Recycling index is calculated as follows:
Ac = 0
CII c = 0 =
Ac H
1014
Table 5. Recycling index of recycled asphalt.
Softening 135°C
Penetration 10°C ductility point viscosity
135°C
Penetration 10°C ductility Softening point viscosity
With the increase of the amount of recycling agent, the performance of asphalt is gradu-
ally improved. It is believed that the physical properties of the asphalt can be recovered when
the amount of recycling agent is 4%-5%. In addition, the ductility of asphalt has greatly
improved.
1016
Figure 7. Relationship between recycling index U and penetration.
1017
1 < V ≤ 1.6: slight aged, the penetration of the asphalt was reduced about 0–30%;
1.6 < V V ≤ 2.5: medium aged, the penetration of the asphalt was reduced about
30%-60%;
2.5 < V: serious aging, the penetration of the asphalt was reduced more than 60%.
5 CONCLUSIONS
There is a good correlation between the aging index V and the recycling index U with the
physical properties of asphalt, which means that it is effective to characterize the degree of
aging and recycling of asphalt by V and U in the molecular structure.
The aging degree of asphalt was defined according to the change of the aging index V: 1
<V ≤ 1.6: slight aging,1.6 < V ≤ 2.5: medium aging, 2.5 < V: serious aging.
It is believed that the complete recycling of the asphalt can be realized when the recycling
index U of recycled asphalt is about 0.025.
REFERENCES
Zhang Zheng-qi, LIANG Xiao-li, LI Ping, Evaluation method of asphalt aging properties [J]. Journal
of Traffic and Transportation Engineering, 2005, 5(1):1–5.
Kallas B F. Flexible pavement mixture design using reclaimed asphalt concrete [R]. Report No. FHWA/
RD-84/088, Final Report, 1984.
Shen J, Amirkhanian S, Jennifer A M. Effects of rejuvenating agents on superpave mixtures containing
reclaimed asphalt pavement [J]. Journal of materials in civil engineering. 2007(5):376–384.
Mohanty A K, Misra M, Drzal L T. Sustainable Bio-composites from renewable resources: opportuni-
ties and challenges in the green materials world [J]. Journal of Polymers and the Environment, 2002,
10:19–26.
CAO Rong-ji, CHEN Rong-sheng, XU Zheng-lin, Evaluation on Performance of Asphalt Rejuvenator
Agent [J], Journal of Highway and Transportation Research and Development, 2007, 24(3): 43–45.
LvWei-min, Recycling Principle of asphalt and technical requirement of recycling agent, Petroleum
asphalt [J], 2007, 21(6): 1–6.
WANG Shou-zhi,ZHU Guo-jun,XIAO Min1,JIA Jun-hong, Photo oxidation Aging Mechanism of
SBS Modified Asphalt Based on Infrared Spectra Technique[J],Science and Technology of Overseas
Building Materials, 2008, 29(6): 86–90.
LI Jing;LIU Yu;ZHANG Xiao-ning, Microscopic Analysis on the Aging Mechanism of Asphalt, Bul-
letin of Chinese ceramic society [J], 2014, 33(6): 1275–1281.
CHEN Hua-xin, CHEN Shuan-fa, WANG Bing-gang, The aging behavior and mechanism of base
asphalts, Journal of Shandong University(Engineering) [J], 2009, 39(2).
RUI Li-jun, Analysis of the Mechanism of Asphalt Aging Based on Infrared Spectra Technique by
Simulating Nature Aging, highway engineering [J], 2014, 39(5): 307–310.
N. Dupuy, L. Duponchel. Classification of Edible Fats and Oils by PCA of FTIR [J]. Food Chemistry,
1996, 57(2): 245–251.
Nigen C, et al. Rejuvenating agent and process for recycling of asphalt [P]. US2010/0041798 A1.
2010.2.
1018
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Song Liu
Southeast University, Nanjing, P.R. China
Nanxiang Zheng
Chang’an University, Xi’an, P.R. China
Jun Yang
Southeast University, Nanjing, P.R. China
1 INTRODUCTION
As an effective way to lower freezing point of water, deicers are widely used to remove snow
and ice on pavement surface in cold regions. Traditional chloride-based deicers are most
commonly used for pavement snow and ice control, a huge amount of chloride salt were scat-
tered over pavement every year. In Beijing, for example, in the recent years, more than ten
thousand tons of deicers are used annually, of which the majority are chloride-based deicers.
However, long-term investigations in the last decades have confirmed that chloride-based
deicers have negative effects on water, soil, and vegetation, and corrosion induced by chloride
ion was blamed for the environmental pollution (Fay and Shi, 2012; Li et al., 2015). Thus
acetate-based deicers were introduced to be environmentally friendly deicers, and it was used
as a substitute for chloride salts in urban roads.
Acetate-based deicers were firstly used for snow removal of runway, so as to protect the
airplane from corrosion caused by chloride ion. While, another problem appeared with large
use of acetate-based deicer. In Nordic country, it was found that durability of runway asphalt
pavement declined dramatically after deicers being changed to acetate-based salts (Edwards
et al., 1999). Field investigation showed that acetate-based deicers tend to soften asphalt bind-
ers and accelerate stripping and ravelling of asphalt mixture (Edwards and Aurstad, 2000).
Subsequently, a lot of research have been done to explore mechanism of pavement disease
caused by using of new deicer types. Tiessen reported the emulsification of asphalt binder
caused by sodium acetate through laboratory (Tiessen, 2008). A study completed by Notting-
ham Transportation Engineering Centre concluded that inclusion of deicers increased water
susceptibility, leading to an increased risk of early life failure of the pavement surface course
(Wright, 2013). It was also reported that deicers promoted stripping of asphalt mixture by
1019
reducing asphalt-aggregate adhesion (Riviera et al., 2013). Hassan found deicer solution
with concentration in the range of 1–2% caused the maximum damage (Hassan et al., 2014).
Laboratory freeze-thaw cycling of asphalt mixture cores in deicer solutions was conducted
to simulate the damage of deicers on asphalt mixture during winter months, however, the
results of indirect tensile test did not successfully explain the deicer-induce damage (Hassan
et al., 2002).
The main objective of this study is evaluate the effect of sodium acetate on asphalt mix-
ture comparing to chloride-based deicers. Three types of deicers were evaluated in this study,
namely sodium chloride, calcium chloride and sodium acetate. Firstly, modified water immer-
sion test was conducted by immersing asphalt-coated aggregate samples in saturated deicer
solutions, stripping percent was adopted as the indicator to roughly characterize its effects on
asphalt-aggregate adhesion, and a possible emulsification mechanism was proposed on the
basis of image analysis. Then asphalt mixture samples were subjected to freeze-thaw cycling
while immersed in 2% deicer solutions. recently several research recommended the Cantabro
test has potential to evaluate the durability of dense-graded asphalt (Doyle and Howard,
2014), hence Cantabro test was adopted to characterize loss of cohesion during the freeze-
thaw cycling process. Finally, the results of water immersion test and Cantabro test were inte-
grated to comprehensively evaluate performance of sodium acetate against sodium chloride
and calcium chloride.
2 TEST MATERIALS
2.1 Deicers
The deicers selected for this investigation were sodium chloride (SC), calcium chloride (CC)
and sodium acetate (SA). Among them, sodium chloride and calcium chloride are the most
commonly used chloride-based deicers, sodium acetate is usually used in some urban area as
environmental deicer. All these deicers are only available for use as deicer solutions, which are
prepared using distilled water and pure solid chemicals.
2.2 Aggregate
Two types of aggregate were used: alkaline limestone and acidic granite. Crushed lime-
stone and granite are widely used as aggregate for pavement asphalt mixture due to its good
mechanical properties and extensive sources. Aggregate particles in good grain shape and
sized 9.5–13.2 mm were particularly selected to prepare asphalt-coated aggregate samples for
adhesive test. Selected aggregate particles were washed through 9.5 mm sieve size, and then
dried in 105 °C oven overnight to a constant weight. Those particles that were damaged or
cracked were removed.
1020
Table 1. Properties of SBS modified asphalt binder.
Sieve size (mm) 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075
3 TEST PROGRAM
1021
container until submerging entire sample into deicer solutions or distilled water. Before the
starting of freeze-thaw cycling, samples were all immersed in deicer solutions as well as dis-
tilled water at room temperature for 24 hours. The objective of this initial immersion was to
ensure that that a relatively high degree of saturation could be reached before the first freeze-
thaw cycle. To ensure that the asphalt mixture sample reach an internal temperature lower
than −15 °C,it took 48 hours for one freeze-thaw cycle, including 36 hours freezing in −18 °C
refrigerator and 12 hours thawing at room temperature. During the test process, 30 freeze-
thaw cycles were conducted in total, deicer solutions were refreshed every 5 freeze-thaw cycles.
For each group, a quarter of sample were taken out at the end of 5, 10 and 20 cycle, and the
last a quarter of samples were removed out when all 30 cycles were completed.
1022
The stripping percent provides a rough quantified index for evaluation of stripping sever-
ity, while the details of stripping are neglected. As a measure to make up for the above
shortcomings, graphical analysis was adopted to detect the difference in stripping patterns.
Figures 2–3 shows the close-up of stripping caused by different deicers. It can be found that
the stripping of aggregate caused by SA was different from SC and CC, the stripping of
asphalt film caused by SC and CC was in the pattern of flake-like stripping, while stripping
caused by SA was in the pattern of spots. The stripping spots macroscopically indicated the
effect of SA on emulsification of asphalt: due to emulsification, asphalt film would be pen-
etrated firstly in the area where the film was relatively thin; then deicer solutions entered into
the interface of asphalt and aggregate, and emulsifying area expanded outwards from the
penetrated point; finally, every penetrated point would expanded to a stripping spot.
1023
cially the cohesion loss of surface layer of samples. Therefore, Cantabro loss was used as the
indicator to characterize cohesion of asphalt mixture. After accomplishment of specified
freeze-thaw cycles, samples were then subjected to Cantabro test. The measured Cantabro
loss after different freeze-thaw cycles is shown in Figure 3.
Freeze-thaw damage is a kind of physical damage caused by freezing pressure which is
produced by volume change of porous water during the process of freeze-thaw cycling. As
shown in Figure 3, comparing to unconditioned samples, freeze-thaw cycling while immersed
in distilled water (DW) resulted in obvious increasing of Cantabro loss as well as sodium
chloride (SC), calcium chloride (CC) and sodium acetate (SA) solutions. All three types of
deicers caused larger Cantabro loss than distilled water, suggesting that addition of deicers
aggravated freeze-thaw damage of asphalt mixture. Samples immersed in sodium chloride
(SC) and calcium chloride (CC) solutions had very close Cantabro loss in Cabtabro test.
While the Cantabro loss of samples that had been immersed in sodium acetate solutions was
greater than that immersed in sodium chloride and calcium chloride solutions, which proved
that sodium acetate caused more cohesion loss than sodium chloride and calcium chloride
during freeze-thaw cycling process.
Moreover, it can be observed that Cantabro losses increase with increasing freeze-thaw
cycles, and its growth rate decreases with the extending of freeze-thaw cycles. The Canta-
bro loss increases rapidly when the first five freeze-thaw cycles were completed, but it only
increases a little when freeze-thaw cycles increases from 20 to 30. The difference between
deicer types also varies with freeze-thaw cycles, the more freeze-thaw cycles samples were
exposed to, the larger difference is observed.
1024
5 CONCLUSION
This study evaluate effect of sodium acetate on asphalt mixture comparing to chloride-based
deicers. Modified water immersion test was used to investigate loss of asphalt-aggregate
adhesion due to invasion of deicers, Cantabro loss was adopted to characterize loss of cohe-
sion during freeze-thaw cycling in deicer solutions. The test results shows that invasion of
sodium acetate and chloride-based deicers both resulted in loss of asphalt-aggregate adhe-
sion, the stripping caused by sodium acetate was more serious than chloride-based, and the
type of aggregate also had some influence on the effect. Moreover, Cantabro test was found
to have good potential for evaluation of cohesion loss caused by freeze-thaw damage. The
results of Cantabro test suggested that sodium acetate can cause more cohesion loss than
chloride-based deicers at the same conditions, and cohesion of asphalt mixture was found to
gradually deteriorate as freeze-thaw cycles increasing. However, due to limitation of asphalt
binder and mixture types, further study is needed to validate the effectiveness of Cantabro
test to evaluate cohesion of dense-graded asphalt mixture.
REFERENCES
Doyle, J.D. & Howard, I.L. Evaluation of the Cantabro Durability Test for Dense Graded Asphalt.
Geo-Frontiers 2011@sAdvances in Geotechnical Engineering, 2014. 4563–4572.
Edwards, Y. & Aurstad, J. 2000. The influence of runway de-icing agents on the durability of asphalt
concrete pavements for airfields. Road Materials & Pavement Design, 1(4).
Edwards, Y., Rollen, J., Lange, G., Aurstad, J. & Nilsen, T.N. 1999. Durability problems on Nordic
airfields—the influence of de-icing agents on asphalt concrete pavements. Statens väg—och trans-
portforskningsinstitut., VTI notat 24A–1999.
Fay, L. & Shi, X. 2012. Environmental Impacts of Chemicals for Snow and Ice Control: State of the
Knowledge. Water Air & Soil Pollution, 223, 2751–2770.
Hassan, Y., Abd, E.H., A O, El-Desouky, A., Mostafa, A. & Farha, M. Effects of new and conventional
deicing chemicals on airfield asphalt concrete pavement. proceedings of the 6th international confer-
ence on the bearing capacity of roads and airfields, Lisbon, Portugal, 24–26 JUNE 2002., 2002.
Hassan, Y., Halim, A.O.A.E., Razaqpur, A.G., Bekheet, W. & Farha, M.H. 2014. Effects of Runway
Deicers on Pavement Materials and Mixes: Comparison with Road Salt. Journal of Transportation
Engineering, 128, 385–391.
Li, F., Zhang, Y., Fan, Z. & Oh, K. 2015. Accumulation of De-icing Salts and Its Short-Term Effect on
Metal Mobility in Urban Roadside Soils. Bulletin of Environmental Contamination & Toxicology, 94,
525–531.
Riviera, P.P., Santagata, E. & Baglieri, O. Effect of anti-icing chemicals on stripping of asphalt concrete
mixtures for airport runway wearing courses. Airfield and Highway Pavement 2013@sSustainable
and Efficient Pavements, 2013. 1229–1239.
Tiessen, P. 2008. Laboratory Investigation of Acetate-Based Deicing/Anti-Icing Agents Deteriorating
Airfield Asphalt (With Discussion). Journal of the Association of Asphalt Paving Technologists, 77,
773–793.
Wright, M. 2013. A study of the effects of adding ice retardant additives to pavement surface course
materials. University of Nottingham.
1025
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
Emulsified asphalt cold recycling technology is widely used to protect the environment and
save resources (Berthelot C et al, 2007). Emulsified asphalt for cold recycling should not
only meet the emulsified asphalt and mixture performance index, but should also meet engi-
neering needs (Hilbrich S L et al, 2008; Du Xue-jie, 2012; Mallick R B et al, 2002). Given
the distance between a construction site and factory, an emulsified asphalt cold recycled
mixture should have good workability (compactibility), even after an extended transporta-
tion time(Lian Pengbao et al, 2013). When the emulsified asphalt cold recycling technique is
applied on urban roads, the emulsified asphalt cold recycled mixture should have high early
strength for traffic can not be interrupted. When curing during bad weather, the strength
of the emulsified asphalt cold recycled mixture should grow faster, and the mixture should
have good integrity. Early strength, workability and integrity were called early performances,
which limited the use of emulsified asphalt cold recycling technology (Xia Ping et al, 2012;
Lu Wei-min, 2009). The main influencing factor on early performance is the emulsion break-
ing speed and the strength growth speed, which relate to the type of emulsifier (Xiao Jing-
jing et al, 2015; Zhong Meng-wu et al, 2008). However, there is no optimization method for
the performance of emulsifier production and construction.
2 OBJECTIVE
The objective of this research is to develop a procedure for evaluating emulsifier optimization
based on the early performances of the mixture for emulsified asphalt cold recycling, when
execution conditions are special (such as a short curing time, long-time transport, or poor
curing conditions).
1027
3 EXPERIMENTAL CONSIDERATIONS
3.1 Material
Asphalt from Qin huang-dao, which is commonly used in Beijing, was selected. The test
results on the asphalt areas follows. The asphalt meets the Chinese specification for AH-70.
Seven kinds of cationic emulsifiers made in China and abroad were selected and named
emulsifier A∼G. Emulsified asphalt was made by batch-type colloid mill. To investigate the
influence of different kinds of emulsifiers on the early performance of emulsified asphalt
cold recycled mixtures, the content of all the emulsifiers was 4% to the emulsified asphalt.
The liquid soap was adjusted to the appropriate pH valve with HCl or NaOH. HCl was
used in this paper to adjust the pH valve of the liquid soap.
The emulsified bitumen colloidal particle surface charge increases with lower pH. When
mixing, the charge adsorption increases, which increases the speed of emulsified asphalt
demulsification. The pH valve was 2.0∼2.5.
During the production of emulsified asphalt, the asphalt should be heated to a liquid state.
The temperature of the matrix asphalt should be 135°C∼145°C, and the temperature of the
liquid soap should be about 45°C.
The reclaimed asphalt pavement (RAP) material was obtained from the Beijing area. The
new aggregates from Hebei were limestone. The emulsified asphalt content, water content,
and cement content of the emulsified asphalt cold recycled mixture were 3.5%, 1.6%, and
1.5%, respectively. The gradation is shown in Table 1.
1028
Table 2. Test plan of early performances.
a mixture depend on the emulsifier characteristics and are very important to engineering
application.
The test plan is shown in Table 2.
Workability: When the transport time was long, the workability of the emulsified asphalt
cold recycled mixture should be good. The mixture should have good flow ability and good
construction workability between production and the end of construction. Otherwise, it
would be hard to unload and compact. The speed of strength forming is related to the emul-
sified asphalt’s breaking speed. The gyratory compaction test was designed to evaluate the
workability of a mixture. The test steps areas follows:
1. About 50 kg emulsified asphalt cold recycled mixture was prepared.
2. According to the Standard Test Methods of Bitumen and Bituminous Mixtures for High-
way Engineering T0736-2011, the gyratory compacted specimens were made with 1 kg
mixture. The diameters of the specimens were 100 mm. The number of compaction was
20. The rest of the mixture was sealed by a plastic bag.
3. The gyratory compacted specimens were made with 1 kg mixture every 0.5 h. Parallel tests
were taken three times each time until the specimen height increased extensively or the
time of compaction was more than 7 hours.
4. The specimens and the uncompacted mixture were cured in a ventilation oven at 60°C
for 48 h.
5. The bulk density (ρn) and theoretical maximum density of the mixture (ρmt) were tested.
The void ratio of the specimen was calculated by the following formula, and curves of the
void ratio to the compaction time were drawn:
Early strength: When emulsified asphalt cold recycled mixtures are used in urban roads,
such as in Beijing, traffic has to be unclosed after construction. There will be vehicle load
during curing, so the early strength of the mixture should be good. The breaking speed and
strength growth of the emulsified asphalt that relate to the type of emulsifier should be rapid.
A cohesion test was performed to evaluate the early strength of the emulsified asphalt cold
recycled mixture (Yu Hao, 2012). The instrument of the cohesion test is shown in Figure 1.
1029
The test steps of early strength were as follows:
1. According to the Standard Test Methods of Bitumen and Bituminous Mixtures for High-
way Engineering T0736-2011, the gyratory compacted specimens were made with a 3 kg
mixture. The size of the specimenswere150 * 75 ± 2 mm. The number of compaction was
40.
2. The specimens were cured in a constant temperature and humidity curing box at 20°C and
90% relative humidity for 4 h.
3. According to ASTM (D1559), a cohesion test was taken when the curing was over.
The cohesion of the specimen was calculated by the following formula:
C M 0.031H + 0.00269H2 )/ D
M/( (2)
1030
Figure 2. Particle size analysis results.
shown in Figure 3. The fine aggregate adhered evenly over the surface of the coarse aggre-
gate. The colour of the aggregate surface was light brown.
Emulsifier type C D E F G
Figure 6. The influence of aggregate type on Figure 7. The influence of cement content
early strength. on early strength.
the influence of cement content on early strength. The results show that the aggregate type
and cement content has a great influence on the early strength as well. When the emulsi-
fied asphalt cold recycled mixture has higher requirements on open traffic that cannot be
achieved through formulation adjustment, the cement content should be raised.
1032
Figure 8. Curing of the coring test specimen.
Emulsifier type A B C D E F G
Integrity for core samples Loose Loose Broken Intact Broken Intact Intact
Curing of the coring test specimen is shown in Figure 8. Samples of the coring test are
shown in Figure 9. The influence of emulsifier type on integrity is shown in Table 4. The
samples of the emulsified asphalt cold recycled mixtures with emulsifiers E, F, and G are
intact; the others were loose or broken. This means the mixtures with emulsifiers E, F, and
G have good integrity.
Early performances
Test
Items Unit Standard EA1 EA2 EA3 methods
Items 1 2 3 Standard
The pavement performances of mixtures with compound emulsifiers are shown in Table 8.
Test results show that the pavement performances of mixtures with compound emulsifiers
were qualified. The formula of the emulsifiers meets engineering requirements.
1034
5 CONCLUSIONS
This article provided an emulsifier optimization method to produce three kinds of emulsified
asphalt that were suitable for long transport times, opening traffic rapidly, and poor curing
conditions.
The gyratory compaction test was designed to evaluate the workability of the mixtures.
The test results showed that emulsifiers C and D could be used when the transport time was
long. The cohesion test was used to evaluate the early strength of emulsified asphalt cold
recycled mixtures. The sequence of the cohesion of the mixtures with different emulsifiers
was F > G > E > D > C > A > B. The coring test was designed to evaluate the integrity and
the strength growth speed of the emulsified asphalt cold recycled mixtures. The mixtures with
emulsifiers E, F, and G had good integrity.
Laboratory tests and engineering applications showed that emulsified asphalt 1 was suit-
able for long transport times. Emulsified asphalt 2 was suitable for urban roads that need to
be opened to traffic rapidly. Emulsified asphalt 3 was suitable for poor curing conditions.
The emulsifier content and the pH value of soap liquid also had certain influence on the
early performances of emulsified asphalt cold recycled mixtures, which was not considered
in this article.
REFERENCES
Berthelot, C., Marjerison, B., Houston, G., et al. Mechanistic comparison of cement- and bituminous-
stabilized granular base systems[J]. Transportation Research Record, 2007(2026): 70–80.
Du Xue-jie. Improve the effect of emulsified asphalt emulsion[J]. Shanxi Architecture. 2014,
40(24):124–127.
Hilbrich, S.L. & Scullion, T. Evaluation of the laboratory mix design and field performance of an
asphalt emulsion and cement stabilized full-depth reclamation project in Texas [C]. In: Proceedings
of 2008 annual transportation research board meeting (C/CD). Washington (DC): National Research
Council, National Academy of Science, 2008.
Lian Pengbao, Fu Yuesong. Development and Effect of Emulsified Asphalt Stabilizer [J]. Petroleum
Asphalt. 2013, 27(5): 29–32.
Lu Wei-min. Key Technology and Application of Emulsified Asphalt Cold-recycling [J]. Petroleum
Asphalt. 2009, 23(3):29–34.
Mallick, R.B., Teto, M.R., Kandhal, P.S., et al. Laboratory study of full depth reclamation (FDR)
Mixes [J]. Transportation Research Record, 2002(1813): 103–110.
Xia Ping, Zhang Guo-hua, Li Sheng-qiang. The Effect of Emulsified Asphalt Formulation on Recycling
Asphalt Mixture Performance [J]. Science Technology and Engineering. 2012, 12(21): 5392–5395.
Xiao Jing-jing, Li Zhi-han, Sha Ai-min. Laboratory Test Study of Cement Emulsified-Asphalt Mix-
ture Performance Based on Emulsifier Complex Technology [J]. Journal of Chongqing Jiaotong
University(Natural Science). 2015, 34(3):52–56.
Yu Hao. A study on Design Method of High-performance Plant Cold Recycling Emulsified Asphalt
Mixture [D]: [ Master Degree Thesis]. Zheng Zhou: Zheng Zhou University. 2012.5.
Zhong Meng-wu, Wu Chao-fan, Yu Yong-sheng. A Study on Design Method of Cold Reclaimed Mix-
ture Using Emulsion Asphalt and Cement [J]. Highway. 2008(1): 195–199.
1035
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
In recent years, environmental issues regarding reducing energy consumption, reducing CO2
emissions and managing wastes have been increasingly articulated and have been gaining
more attention worldwide. One of the most important trends in road materials and pavement
engineering is the use of Cold Asphalt Mixes (CAMs) in roads construction.
The performance of CAM is intimately related to the properties and proportions of mate-
rials that are used in the mixture and to the curing condition. One of the most common types
of CAM is cold bitumen emulsion mixture (CBEM) treated with cement (Needham, 1996,
Thanaya, 2003, Oruc et al., 2007, Niazi and Jalili, 2009, Bocci et al., 2011). In general, the
incorporation of cement into CBEMs can increase: stiffness modulus, resistance to perma-
nent deformation, resistance to fatigue cracking and resistance to moisture damage (Need-
ham, 1996). The CBEM requires a certain time which is necessary to cure and build up the
ultimate mechanical properties such as strength and stiffness. This process is termed “curing”
and is a process whereby the CBEM gradually gains both strength and stiffness over time.
This process accompanied by emulsion breaking, moisture loss and/or hydration of cemen-
titious compounds in case of CBEMs treated with cement. It is a well-established fact that
the curing process has a significant effect on the mechanical properties and performance of
CBEMs (Jenkins, 2000).
Despite the fact that a wide range of studies have been undertaken to investigate the effect
of incorporation of cement into CBEMs, considerable issues still need to be addressed. In
particular, there is a lack of clarity regarding the influence of the curing process on the per-
formance of CBEMs treated with cement. This is because of the complex combination of
three phenomena acting together during the curing process: emulsion breaking, moisture loss
and hydration of cementitious compounds (García et al., 2013, Cardone et al., 2014, Serfass
et al., 2004). It is important that the bitumen emulsion breaking process is achieved as soon
as possible after emulsion application. However, it must not occur until after the completion
of the mixing and compaction phases. Accordingly, the presence of water after emulsion
breaking can negatively affect early strength gain. Increasing the curing temperature leads to
1037
an increase in the rate of water evaporation, resulting in an increase in the strength gain proc-
ess. Furthermore, the presence of cement accelerate the emulsion breaking process, increase
the rate of bitumen coalescence and reduce the amount of evaporable water (Needham,
1996). More strictly, hydration of cementitious compounds is linked to moisture loss; the
chemical reactions that take place during the hydration of cementitious fillers require the
presence of water and do not require any moisture loss. However, the increasing of curing
temperature results in an increase in the amount of water evaporation.
The main objective of the present study therefore, is to investigate the level of impact
of climatic parameters such as curing temperature and humidity in addition to the impact
of curing time and the presence of active filler (cement) on the curing process in CBEMs.
The effect of such factors on the curing process has been evaluated in the laboratory in
terms of indirect tensile stiffness modulus, water loss evaluation and binder characterization
before and after curing. A series of mixtures having different amounts of cement (0, 1, 3, 5%)
and cured at constant temperature (5, 20, 40ºC) has been evaluated. Moreover, two relative
humidity levels (less 50% and higher 85%) have been adopted in this study.
2 MATERIALS CHARACTERIZATION
2.1 Aggregate
The aggregate used in this study was crushed limestone. The physical properties of the aggre-
gate were: apparent density 2.70 Mg/m3; absorption 0.4%; Los Angeles Coefficient 28. The
gradation of the aggregate (Figure 1) was within the limits of 0/14 mm size dense graded
surface course, according to European Committee for Standarization (2005a).
2.3 Fillers
Two types of filler material were used in the CBEMs; natural limestone filler (LF) and Ordi-
nary Portland cement (OPC). The OPC used in this study was CEM I 52.5R. Scanning elec-
tron microscopy (SEM) was used for determining the morphology of these two fillers, as
shown in Figure 2. SEM analysis was implemented under a resolution of 3–4 nm and an
accelerated voltage of 15 kV.
1038
Figure 2. Morphology of (a) natural limestone filler and (b) ordinary Portland cement.
1039
between 19 and 23ºC. The specimens were stored over a steel mesh to avoid any contact with
the saturated NaCl solution and the chamber sealed to keep the moisture inside. A schematic
diagram and photograph of the chamber are shown in Figure 3.
3 EXPERIMENTAL PROGRAMME
1040
Table 2. ITSM test configuration based on BS EN 12697-26.
Item Range
Curing Curing
Mixture code Test time (days) temperature (ºC) Relative humidity
with increasing curing temperature and time. This gain was rapid for specimens cured at
higher temperatures (40ºC). This could be attributed to rapid water loss at higher tempera-
ture. Also, in cement treated mixes (1C-CBEM, 3C-CBEM, 5C-CBEM), the stiffness modu-
lus increased with increasing curing time. However, this gain was significantly influenced
by interaction with both moisture loss and cement hydration. This leads to the interesting
observation that improvement in the stiffness modulus did not continue after 28 days for 5ºC
and 40ºC cured mixes while the 20ºC cured specimens showed some further development.
1042
Figure 6. Evolution of ITSM at 20°C with curing time for (a) 0C-CBEM (b) 1C-CBEM (c) 3C-CBEM
and (d) 5C-CBEM.
in the microstructure (Rovnaník, 2010). Thus, further research is needed to study the effect
of curing temperature on the microstructure of cement treated CBEM. As a general trend,
with increase in moisture loss the stiffness increased. Therefore further analysis of the rela-
tionship between the moisture loss by evaporation and stiffness modulus was carried out, see
Figure 7.
It may be observed in Figure 7 that when a certain moisture loss is considered the cur-
ing temperatures have different impacts on the stiffness modulus for the same mixtures. On
the one hand, mix 0C-CBEM-40ºC showed far better stiffness than 0C-CBEM-20ºC or
0C-CBEM-5ºC with the same moisture loss at the end of the curing period. This can be
explained by the fact that specimens cured at 40ºC are subject to ageing over the curing
period which is consistent with the results from binder characterisation before and after the
curing, see section 3.3.
On the other hand, the stiffness modulus of mixes 1C-CBEM, 3C-CBEM and 5C-CBEM
showed a marked influence of moisture loss under different curing temperatures. At the same
cement content, the stiffness modulus of specimens cured at 20ºC was higher than those
cured at 5ºC or 40ºC.
In order to understand more deeply the effect of moisture loss on the moisture content of
CBEM, the moisture content is divided into two types. The first type is the evaporated mois-
ture (MEvaporation) which is calculated based on the difference between the initial weight
and the weight after the end of the curing period. The second type is the trapped moisture
(MTrapped) inside the mix due to (1) the water physically adsorbed on the surface of aggregate/
filler, (2) the water trapped in the closed pores within the bitumen and (3) the water used dur-
ing cement hydration. The MTrapped was calculated by subtracting the evaporated water from
the initial water content present in the CBEM specimens. The amount of moisture required
for the hydration of cement (MHydration) was calculated based on equation (1) recommended by
Mehta and Monteiro (1995)
1043
Figure 7. The relationship between moisture loss and stiffness modulus under different curing tem-
peratures (a) 0C-CBEM (b) 1C-CBEM (c) 3C-CBEM (d) 5C-CBEM.
MEvaporation MTrapped
M Hydration ( . . ) α k ×C (1)
where α is the degree of hydration of the cement paste, C is the cement content within the
mixture and k is the fraction of Portland cement clinker in the cement. In table 4 it is assumed
that k = 0.87 (average of the clinker content range) and α = 0.95 at 100 days of curing.
The moisture types are summarized in Table 4. It was found that the amount of MTrapped
inside the CBEM cured at 5ºC and 20ºC is more than MHydration whereas the MTrapped inside the
CBEM cured at 40ºC is less than MHydration. It is therefore evident that the moisture loss by
evaporation affects the hydration of cement, particularly at high curing temperature (40ºC).
This will hinder the hydration process. This result is in agreement with the finding of Car-
done et al. (2014). Moreover, the calculation in Table 4 indicates that most CBEMs still main-
tained some residual water (MTrapped - MHydration). This moisture may possibly be trapped inside
the surface pores of the aggregate or inside the bituminous mortar (García et al., 2013).
A three-way analysis of variance (ANOVA) was conducted to determine the effect of
cement content (0, 1, 3, 5%), curing temperatures (5, 20, 40ºC) and curing times (3, 7, 10, 14,
28, 54, 84 days) on the stiffness modulus of CBEMs. The result showed that these factors are
1044
statistically significant at the 99% confidence level. The p-values obtained for these factors
are less than 0.001. Also, the result of ANOVA analysis shows that cement content presented
the largest F-values of 314.078, followed by curing temperatures of 72.234 and curing times
of 9.357. This means that the cement content and curing temperatures had a higher signifi-
cant effect on stiffness modulus of CBEM, compared with curing times.
1045
4.3 Bitumen characterization before and after curing
To study the effect of curing temperature on bitumen properties, bitumen was extracted from
0C-CBEM samples cured for 28 days at 5, 20 and 40ºC. Softening point and Frequency
sweep tests were carried out and Figure 9 illustrates the complex modulus (G*) master curves
at a reference temperature of 20°C.
It can be noticed from Figure 9 that the differences in complex modulus of the recovered
bitumen are significant, particularly at low frequency, which indicates that the bitumen age-
ing effect differs considerably between different curing temperatures. Moreover, the effect
of curing temperatures on the softening point of the binder was also evaluated. Before mix-
ing and applying the curing protocols, the softening point of the original base binder was
52.8ºC. After applying the curing protocols the softening points of extracted binder were
55.6, 56.2, and 59.2ºC for 0C-CBEM-5ºC, 0C-CBEM-20ºC and 0C-CBEM-40ºC, respec-
tively. Over all, based on softening point and frequency sweep test results it may be con-
cluded that the binder properties changed appreciably during the curing process. Hence, it
is evident that water loss and the cement hydration are not the only mechanisms involved
during curing of CBEMs. High curing temperature is also responsible for additional stiff-
ness gain due to ageing.
5 CONCLUSION
The effects of several contributory factors on the curing process of CBEM were investigated.
These factors were the curing time, curing temperature, cement content and relative humid-
ity. On the basis of the laboratory test results, the following conclusions were derived:
1. An Increase in curing temperature facilitated the evaporation of moisture in the CBEMs
leading to improved mechanical and performance properties
2. The results also showed the role of cement and its usefulness in positively improving the
stiffness modulus especially during the early life of CBEMs.
3. High curing temperature was responsible for additional stiffness gain by increasing the
binder stiffness due to ageing and by increasing the moisture loss by evaporation during
the curing process. However, at high curing temperature the moisture loss by evaporation
may hinder the hydration of cement.
4. Further investigation is needed to study the effect of curing temperature on the micro-
structure of cement treated CBEM.
ACKNOWLEDGMENTS
The authors would like to acknowledge the University of Nottingham which has funded this
research through the Dean of Engineering Research Scholarship for International Excel-
lence. Also the authors wish to thank Mr. Rami Al Faqawi and Mr. Mustafa Abou Foul for
their help and assistance.
REFERENCES
BOCCI, M., GRILLI, A., CARDONE, F. & GRAZIANI, A. 2011. A study on the mechanical behav-
iour of cement–bitumen treated materials. Construction and Building Materials, 25, 773–778.
CARDONE, F., GRILLI, A., BOCCI, M. & GRAZIANI, A. 2014. Curing and temperature sensitivity
of cement–bitumen treated materials. International Journal of Pavement Engineering, 1–13.
EUROPEAN COMMITTEE FOR STANDARIZATION 2005a. BS EN 4987: Coated macadam for
roads and other paved areas. Part 1: Specification for constituent materials and for mixtures. London,
UK.: British Standards Institution.
1046
EUROPEAN COMMITTEE FOR STANDARIZATION 2005b. BS EN 12697-4 Bituminous mixtures
—Test methods for hot mix asphalt. Part 4: Bitumen recovery: Fractionating column. London, UK.:
British Standards Institution.
EUROPEAN COMMITTEE FOR STANDARIZATION 2012. BS EN 12697: Bituminous mixtures:
Test methods for hot mix asphalt. Part 26: Stiffness. London, UK.: British Standards Institution.
FU, P., JONES, D., HARVEY, J.T. & HALLES, F.A. 2009. Investigation of the curing mechanism of
foamed asphalt mixes based on micromechanics principles. Journal of Materials in Civil Engineering,
22, 29–38.
GARCÍA, A., LURA, P., PARTL, M.N. & JERJEN, I. 2013. Influence of cement content and environ-
mental humidity on asphalt emulsion and cement composites performance. Materials and Structures,
46, 1275–1289.
JENKINS, K.J. 2000. Mix design considerations for cold and half-warm bituminous mixes with empha-
sis of foamed bitumen. PhD Thesis, Stellenbosch University.
LIN, J., WEI, T., HONG, J., ZHAO, Y. & LIU, J. 2015. Research on development mechanism of early-
stage strength for cold recycled asphalt mixture using emulsion asphalt. Construction and Building
Materials, 99, 137–142.
MEHTA, P.K. & MONTEIRO, P.J. 1995. Concrete: Structure, Properties, and Materials, New York,
USA, 2nd Edition. McGraw-Hill.
NASSAR, A.I., THOM, N. & PARRY, T. 2016. Optimizing the mix design of cold bitumen emulsion
mixtures using response surface methodology. Construction and Building Materials, 104, 216–229.
NEEDHAM, D. 1996. Developments in bitumen emulsion mixtures for roads. PhD thesis, University of
Nottingham.
NIAZI, Y. & JALILI, M. 2009. Effect of Portland cement and lime additives on properties of cold in-
place recycled mixtures with asphalt emulsion. Construction and Building Materials, 23, 1338–1343.
ORUC, S., CELIK, F. & AKPINAR, M.V. 2007. Effect of cement on emulsified asphalt mixtures. Jour-
nal of Materials Engineering and Performance, 16, 578–583.
PUERTAS, F., MARTíNEZ-RAMíREZ, S., ALONSO, S. & VÁZQUEZ, T. 2000. Alkali-activated fly
ash/slag cements: Strength behaviour and hydration products. Cement and Concrete Research, 30,
1625–1632.
ROVNANÍK, P. 2010. Effect of curing temperature on the development of hard structure of metakao-
lin-based geopolymer. Construction and Building Materials, 24, 1176–1183.
SERFASS, J.-P., POIRIER, J.-E., HENRAT, J.-P. & CARBONNEAU, X. 2004. Influence of curing on
cold mix mechanical performance. Materials and Structures, 37, 365–368.
THANAYA, I.N.A. 2003. Improving the performance of cold bituminous emulsion mixtures (CBEMs):
incorporating waste. PhD thesis, University of Leeds.
1047
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Nowadays, the haze is an urgent pollution problem in big cities, especially
in Beijing China. The major resources of haze are motor vehicle exhaust and resuspended
dust. In this research, an intelligent pavement material with ecological asphalt modifiers was
developed for resuspended dust control, the components of the asphalt modifier absorb the
water in the air automatically and intelligently under designed temperature and humidity
ranges. The water film is formed on the surface of road, which can absorb the dust in order to
decrease the hazes. A standard equipment and method is developed to evaluate the efficiency
of dust control and materials performance. To balance the relationship between the pave-
ment performance and efficiency of dust control, the aggregates particle distribution and
optimum modifier contents are determined. Finally, the intelligent pavement material was
applied in the design of a pavement reconstruction and the economic benefit of the project
was analyzed.
1 INTRODUCTION
1.1 Background
Nowadays, the haze is an urgent pollution problem in big cities especially in Beijing China, a
city with about 30 million peoples. ‘2014 Beijing Environment Bulletin’ issued in April 2015
(listed in Table 1) reveals that the annual average contents of PM2.5 and PM10 exceed the
national standard seriously. It has strong impact on the health of citizens.
Currently, there are two main methods to control the haze all over the world: one is the
back-end purify the polluted air, the other is the front-end control the source of pollution.
The data from Beijing Environmental Protection Bureau shows that the dust and motor vehi-
cle exhaust are the major sources of haze in Beijing is 45.4%. Compared with the technology
of the back-end purify the polluted air, the front-end control the source of pollution is more
beneficial and efficient. “To solve the haze, controlling the dust firstly” should be the princi-
ple of city environment management.
1049
Table 1. The average of main pollutant concentration in Beijing.
Sweeping, water flushing and use of chemical suppressants are common ways to dimin-
ish resuspended dust. In Stockholm, Norman didn’t find a great loss of PM 10 after intense
sweeping in spring [2]. These results are consistent with the study which was conducted by
Kuhns [3]. Düring used a vacuum equipped with a particle filler cleaning street from Mon-
day to Thursday finding no significant reduction of PM 10 levels on dry days [4]. Fitz and
Bumiller, Kantamaneni illustrated a decrease of PM level, but none of these studies con-
clusively demonstrated the effectiveness of sweeping on reducing PM 10 [5]. In Hamilton,
a relatively frequent combination of mechanical and vacuum sweeping with water flushing
resulted in air quality improvements. Chang tested a combination of sweeper and washer
finding this process was effective at mitigating total suspended particles, but it was short
lasted no more than 4 hours [7].
Furthermore, America, Russia, Canada, Japan, Germany, include of China, conducted
lots of researches on dust control. The different types of dust-depressor were brought out
by different researchers, which are listed in Table 2 (Zhang, 2000). However, due to the
1050
weak adhesion between pavement surface and dust-depressor, the dust-depressor is easy
to lose under the traffic loads and that it also pollutes soil and destroys the growing of
plants.
In view of challenges, present road dust control technology has a lot of shortages, such as
poor timeliness, pollution, waste of water resources. The results of this research are going
to provide an intelligent pavement with initiative ability of dust control. It can promote the
effect of dust control, prolong the controlling time, and save the water source. The most
important is that this technology is green pollution-free.
2.1 Materials
Asphalt binder
SK 70# SBS modified asphalt binder was obtained in this study. The physical properties
are presented in Table 3.
2.1.1 Aggregate
Aggregates produced by BeijingYuGou CO.LTD were chosen as experimental material.
The stone was cleaning and drying before test. The test results were given in the Table 4.
1051
Table 5. The gradation of aggregate.
Size/mm 0.075 0.15 0.3 0.6 1.18 2.36 4.75 9.5 13.2 16
Grading 9 14 18 23 30 50 45 80 100 100
upper limit
Grading 5 6 9 13 15 20 30 60 90 100
lower limit
Grading the 7 10 13.5 18 22.5 35 37.5 70 95 100
median
Composite 7.7 9.1 11.3 17.2 24.7 34.6 39.2 70.9 97.6 100
gradation
Figure 2. Thickness of water under optimal additive amount and summer condition.
one group, the dust of 20 g/m2 each time was sprinkled on the surface of specimen to simulate
one time sand storm. In another group, the dust of 15 g/m2 each time was sprinkled on the
surface of specimen to simulate ten days dust in normal time. The wind blew for ten minute
to simulate the wind. The weight of specimen was measured each time. The test stopped until
the weight of specimen reached a stable level. The weight of dust absorbed by ecological
pavement was computed by subtract the initial weight of specimen from the measured weight
of specimen. The results of asphalt mixture with three types ecological modifiers are drew
in Figure 5.
It is clear to see that the pavement made with ecological asphalt mixture has ability to
hold dust to avoid forming haze. During the three types of ecological modifiers, the effect
of type C is the best, and the saturation dust weight is 100–105 g/m2. The effect of type A
is worst, and the saturation dust weight is 50–52 g/m2. The effect of type B is in the middle,
and the saturation dust weight is 88–89 g/m2. The pavement with ecological modifiers of type
1053
Figure 3. Thickness of water on specimen before and after scouring.
C can defense 6 times sand storm. In normal time, it can hold dust accumulated during the
90 days.
After saturation test, the wheel tracking load was conducted on the specimen with satura-
tion dust. The aim of this test is checking the ability for holding the dust of ecological pave-
ment under the traffic load. The dust weights absorbed on specimen before and after loading
are showed in the Table 6.
It can be seen from the Table 6 that although the weight of dust deduced after conduct-
ing the wheel tracking load, there are still large amount dust absorb on the specimen. The
residual rate is 87%-90% which means that the ecological pavement can hold most dust under
the traffic loading to avoid formation of floating dust. Regular clean can avoid the void cov-
ered with dust.
1054
Figure 5. Weight of dust adsorbed and haze decreased on pavement surface.
1055
Table 8. Test results of pavement performance with Type B modifier.
It can be seen from tables above, asphalt pavement performance with different kinds of
ecological modifiers can meet standard requirement totally. From experimental data we can
find that though mixture absorbs moisture in the air and keeps moisture membrane in the
surface, the moisture damage resistance still fulfill criterion.
Jingliang Road is in Daxing district, Beijing. The original road is 15 meters wide, two-way
and four lanes. The pavement is destroyed seriously, so reconstruction is needed. The photos
and cross section of Jingliang Road are shown in Figure 6 and Figure 7 respectively. The
cross section after reconstruction design is shown in Figure 8.
As shown in Figure 9, the ecological modifier is used on surface layer. 120 tons of eco-
logical modifiers are needed per kilometers, and the cost is 2400 thousand Yuan. The total
project cost increases by 10%. After the project is finished, in one year (180 days was consid-
ered to sprinkle water), 5.67 tons of dust will be decreased per kilometer, and 78,400 tons of
1056
Figure 7. The cross section of original road.
water resource, 95.18 tons of gasoline, 1,155,064 Yuan will be saved per kilometer, in which
the cost of labor is 109,897 Yuan.
4 CONCLUSIONS
REFERENCES
[1] Shou-Bin Fan, Gang Tian, Jianping Qin, Gang Li. ‘Emission characteristics of road dust fall in
Beijing’, Journal of Environmental Engineering (2010), Vol. 4, No. 3, pp. 629–632.
[2] Norman M, Johansson C. Studies of some measures to reduce road dust emissions frompaved
roads in Scandinavia. Atmos Environ 2006; 40: 6154–64.
[3] Kuhns H, Etyemezian V, Green M, Hendrickson K, McGrown M, Barton K, et al. Vehiclebased
road dust emissions measurements—Part II: effect of precipitation,wintertime road sanding, and
1057
street sweepers on inferred PM10 emission potentialsfrom paved and unpaved roads. Atmos Environ
2003; 37: 4573–82.
[4] Düring I, Zippack L, Bächlin W, Lohmeyer A. Auswertung der Messungen des BLUMEwährend
der Abspülmaßnahme am Abschnitt Frankfurter Allee 86, 2004.
[5] Fitz DR, Bumiller K. Determination of PM10 emission from street sweepers. J Air WasteManage
2000; 50: 181–7.
[6] DobroffF. Region of Hamilton–Wentworth Air Quality Program. Street cleaning initiative, 1999.
[7] Chang YM, Chou CM, Su KT, Tseng CH. Effectiveness of street sweeping and washing forcontrol-
ling ambient TSP. Atmos Environ.
1058
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Micro-surfacing mix has been wildly used in pavement maintenance for its
water proofing, anti-skidding and wear resistance. However, micro-surfacing has higher
noise than common asphalt pavement and its noise can’t be measured effectively and accu-
rately. Therefore, this study designs a new indoor noise test method based on wet track
abrasion instrument. Firstly, the noise is measured to find out the noise of micro-surfacing
is higher than contrastive pavement. Texture Depth (TD) and the exposure rate index of
micro-surfacing are proposed as the evaluation indexes to describe the surface properties of
micro-surfacing. Secondly, a new indoor noise test method is developed based on wet track
abrasion instrument. Thirdly, the low noise gradation is determined based on orthogonal
test design method. Finally, the indoor noise of recommended gradation was validated. The
result shows that the indoor noise of recommended micro-surfacing gradation can be low-
ered about 2.5dBA than common micro-surfacing gradation.
1 INTRODUCTION
1059
noise is the largest A level of sound pressure level. But the cost of this method is huge and the
measurement strip must be strictly required (Nina, 2005).The CPX method is mainly to meas-
ure the sound pressure near the surface of the road surface, and get the change of the density
of the air in the measurement area, so as to obtain the sound power spectrum and calculate the
equivalent and continuous A sound level of the test section (Robert, 2004).
There is a method to measure indoor noise. Drum method is a fire sound measurement
method that use a fixed drum which diameter is 3.7 m, measuring the test road paved around
the drum, in and the loading device for use, the tire contacts the drum with a certain pressure
and the tire can run with the pressure from the drum. Then we can gain the relative noise
between the tire and pavement by microphone fixed at the drum (Narayanan, 2005). But this
method cost very highly so that it is not easy to be extended.
However there is still lacking of efficient and economic indoor noise test method. If we
want to evaluate the noise level of one kind micro-surfacing before pave it, so that an indoor
noise test method is needed.
2 METHODOLOGY
The study designs a new indoor noise test method based on wet track abrasion instrument to
evaluate the relative noise. By conducting orthogonal experimental design with three factors
and three levels, the study determines a recommended gradation micro-surfacing mix with
lower TD. In the end the new indoor noise test method is conducted to prove the noise reduc-
ing performance of recommended gradation micro-surfacing mix.
The noise of actual micro-surfacing sections was measured before the indoor noise test to
determine the noise level of micro-surfacing. The noise of micro-surfacing relates to the serv-
ice time, measurement environment and vehicle speed (Lee, 1997). So that the study selected
2 test sections of different service time to collect noise data (G102 state road and Harbin-
Yichun freeway). The test section of G102 state road was a new micro-surfacing and the
section of Harbin-Yichun freeway was constructed three years. The test sections all had good
road condition and representativeness. The study measured both vehicle interior noise and
vehicle exterior noise.
Figure 2. Vehicle interior noise of G102 state road (left) and Harbin-Yichun freeway (right).
The noise measurement result of G102 state road and Harbin-Yichun freeway are shown
in Figure 2.
Figure 2 shows that vehicle interior noise from micro-surfacing is higher than common
pavement and vehicle interior noise from micro-surfacing increases with vehicle speed
increases.
Figure 3 shows that vehicle interior noise adding values of micro-surfacing reduces with
increasing of service time. And we can easily see that the trend of reduction of noise adding
values slow down over service time. The phenomenon reflects that the surface of micro-sur-
facing is going smooth so that the surface difference between micro-surfacing and contrastive
pavement reduce.
1062
Figure 5. Indoor noise testing rig.
We can easily see that the second time and the third time have good correlation between
TD and noise, but the fourth time and the fifth time are disorganized. Between NO.1 and
NO.2 the second time owns the best correlation between the noise values and TD, and the
surface of the specimen is not abraded too much. So that NO.1 time is the best one.
The study tends to gain a type of low noise micro-surfacing through changing the gradation
of micro-surfacing mix with orthogonal experimental design (ISSA, 2010).
1063
5.1 Evaluation indexes
5.1.1 Texture Depth
The method utilized to measure TD of pavement for this study is handmade spreading the
sand. TD of pavement is a key indicator to reflect pavement’s macroscopic structure and it is
also the main criteria for the formulation of pavement maintenance (Roe, 1988).
Table 1 shows that TD reduce with micro-surfacing service time increases. We all know that
both TD and noise added value reduce over time, and noise added value has good relativity with
TD. So that one suppose was made that we can gain low noise micro-surfacing by reduce TD.
Is
ERI = × 100% (1)
It
where Is = the aggregate number of pixels not covered by sand; It = the total number of pixels
of the image.
month mm mm
1064
Basalt produced by Acheng Fanrong Quarry from Heilongjiang Province is used as aggre-
gate, with 80% sand equivalent and it is larger than 65% that standard requires.
Levels A B C
7 mm 4.75 mm 2.36 mm
1 85.0 75.0 46.0
2 90.0 81.9 53.8
3 95.0 85.0 65.0
TD
1 2 3 mm
1065
Table 4. Recommended micro-surfacing gradation.
Size
Passing rate 9.5 7 4.75 2.36 1.18 0.6 0.3 0.15 0.075
% 100.0 95.0 81.9 46.0 38.0 26.1 15.0 10.0 5.4
6 PERFORMANCE TEST
After determining the recommended micro-surfacing gradation we use the indoor noise test
method to evaluate the noise reduction performance of recommended micro-surfacing mix
and then evaluate its road performance.
1066
Table 5. Gradation comparison.
Passing rate
%
Gradation
type 9.5 7 4.75 2.36 1.18 0.6 0.3 0.015 0.075
Recommended 100.0 95.0 81.9 46.0 38.0 26.1 15.0 10.0 5.4
gradation
Contrastive 100.0 85.0 75.0 46.0 38.0 26.1 15.0 10.0 5.4
gradation
Run
time Leq Leq mean
Test
Gradation Number time s dBA dBA
Asphalt-aggregate ratio
Wet abrasion test 1h Wet abrasion value 750 175 341 ≤540
g/m2
6h Wet abrasion value 1117 352 540 ≤800
g/m2
Load wheel test Sand adhered amount 431 610 559 ≤450
g/m2
Transverse rut deformation 1.2 2.0 1.8 ≤5
%
Longitudinal rut deformation 0.73 0.97 0.91 --
mm
6 CONCLUSION
The study designs an indoor noise test method based on wet track abrasion. The noise meas-
urement was conducted to find out the correlation for the noise values and TD. The recom-
mended gradation micro-surfacing mix is determined by orthogonal experimental design,
and its noise reduction and road performance meet requirements.
1. The noise of micro-surfacing has good correlation with TD, and it reduces with the TD
reduces. Besides the ERI has good correlation with texture depth and it reduces with
increasing of TD within the proper range.
2. The study increase the 7 mm size to control large particle size aggregate, the recommended
micro-surfacing gradation is determined with orthogonal experiment.
1067
3. the indoor noise test method of micro-surfacing mix based on wet track abrasion instru-
ment is explored. The micro-surfacing mix specimens are used to simulate micro-surfac-
ing. Wet track abrasion instrument simulate the vehicle loads, and real-time spectrum
analyzer is used to measure the noise between tire and pavement.
4. The recommended micro-surfacing mix can reduces 2.5dBA compared with common micro-
surfacing mix and its road performance meets the requirements of the specification.
ACKNOWLEDGEMENTS
This study was supported by National Natural Science Foundation of China (No. 51478152).
The authors would like to acknowledge their financial support.
REFERENCES
American Association of State Highway and Transportation Officials (AASHTO). 2012. Provisional
standard test method for measurement of tire/pavement noise using the on-board sound intensity
(OBSI) method. Standard TP 76–12, Washington, D.C.
Ezoddin, G. & Jonsson, J.A. et al. 2014. Equilibrium sampling through membrane based on a hollow
fiber for determination of naproxen and diclofenac in sludge slurry using Taguchi orthogonal array
experimental design. Desalination and Water Treatment 52(13): 2472–2480.
Thomas, B. & Doug, H. et al. 2005. Influence of pavement surface type on tire/pavement generated
noise. Journal of Testing and Evaluation 33(2): 94–100.
International Slurry Surfacing Association. (ISSA). 2010. Recommended Performance Guideline for
Micro Surfacing. Guideline A143, Annapolis, MD.
Hencken, J. & Haas, E. et al. 2014. Pavement noise analysis of pavement preservation techniques in
New Jersey. Transportation and development Institute: 702–710.
Lee, J.J. & Ni, A.E. 1997. Structure-borne tire noise statistical energy analysis model. Tire Science and
Technology25 (3): 177–186.
Lindenmann, H.P. 2006. New findings regarding the significance of pavement skid resistance for road
safety on Swiss freeways. Journal of Safety Reasearch37: 395–400.
Narayanan, N. 2005. Tire -pavement interaction noise: recentresearch on concrete pavement surface
type and texture. 8th International C onference on Concrete Pavements.
Nina, R. 2005. Measuring of noise and wearing of quiet surfaces. Helsinki University of Technology.
Robati, M. &Cater, A. et al. 2013. Repeatability and reproducibility of micro-surfacing mixture design
tests and effect of aggregates surface areas on test results. Australian Journal of Civil Engineering
11(1): 41–55.
Robert, B. 2004. An Introduction to tire/pavement noise of asphalt pavement. Purdue University.
Roe, P.G. & Tubey, L.W. et al. 1988. Surface texture depth measurements on some British roads. Trans-
port and Road Research Laboratory 143.
Torija, A.J. 2011. Relationship between road and railway noise annoyance and overall indoor sound
exposure. Transportation Reasearch(16): 15–22.
1068
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Lihong He
Department of Materials Science and Engineering, Chongqing Jiaotong University, P.R. China
Xuejuan Cao
Department of Materials Science and Engineering, Chongqing Jiaotong University, P.R. China
1 INTRODUCTION
Phase Change Materials (PCM) can store and release large amounts of heat energy during
phase change while their own temperature keep constant [1]. PCM have heat energy storage
and temperature control characteristics and can solve the inconsistency between energy sup-
ply and demand in time and space so that they have great prospect in the domain of energy
efficiency [2]. Asphalt is a typical temperature-sensitive material, when the environment tem-
perature changes suddenly, the asphalt pavement easily caused a series of temperature dis-
eases, such as high-temperature rutting and low-temperature cracking and these diseases have
a strong impact on the service life, road ability, security of pavement [3]. When PCM applied
in the asphalt pavement, they could absorb and release heat energy as the temperature of the
asphalt pavement reached phase change temperature, reduce the temperature change range
and cope with the temperature changes accordingly so that the related diseases could be alle-
viated [4]. In recent years, a lot of research work about this subject had been carried out by
the scholars both at home and abroad. Polyethylene Glycol, which was suitable for the heat
storage and temperature control of asphalt pavement, was selected by Zhang Yibo [5] et al.
in accordance with the PCM characteristics and the requirements for the asphalt pavement
materials. Polyethylene Glycol was directly applied in asphalt and hot-mix asphalt mixture
1069
by Hu Shuguang [5] and Cao Changbin [6]. The temperature sensibility of the asphalt was
reduced and the phase change heat storage asphalt mixture could control the temperature
in a certain degree. However, Polyethylene Glycol was a solid-liquid phase change material,
and was directly blended in the asphalt or asphalt mixture. Flowing liquid was created dur-
ing the phase change, which had negative effect on the temperature stability of the asphalt
and asphalt mixture. The composite shape-stabilized phase change material made by Ma
Biao [7–8] with physical absorption-membrane cladding method was applied in the asphalt
mixture, and the good temperature control effect was obtained, but the phase change mate-
rial leaked after it was changed into liquid, causing the decrease of its high-temperature sta-
bility. The impact of paraffin/expanded graphite shape-stabilized phase change material on
the temperature sensibility of the asphalt was studied by Chen Meizhu et al. [9–11], and the
result showed that it had some cooling effect on the cold-mix asphalt concrete, but was not
suitable for the hot-mix asphalt concrete. Preliminary research was conducted by X. Cocu
et al in Belgium using the heat released by low-melting paraffin phase change material when it
was frozen to slow down the condensation of the pavement, but the temperature change was
rather small. The current research mainly focused on the impact of the phase change materi-
als or shape-stabilized phase change materials on the technical performance of asphalt and
asphalt mixture, while there was still little research about the functional mechanism, phase
change behavior and phase change heat storage principle.
This paper employed Polyethylene Glycol (PEG) as the phase change material, and porous
silicon dioxide (SiO2) as the shape-stabilized carrier, and then PEG/SiO2 Shape-Stabilized
Phase Change Material (SSPCM) was made by the sol-gel method. In order to investigate the
application feasibility of SSPCM in the asphalt mixture, it was mixed with melting asphalt,
and asphalt-SSPCM mixture was made. Then the phase change heat storage property of
SSPCM was observed and its principle was analyzed, which provided a theoretical basis for
the application of SSPCM in the asphalt mixture.
2 EXPERIMENTAL
2.1 Materials
Polyethylene Glycol (PEG), provided by Jiangsu Haian Petrochemical Plant, the average
molecular weight was 2000, the initial temperature of phase change was 42.4°C, the peak
temperature was 63.5°C, and the enthalpy was 188.1 J/g.
Silica solution, produced by Zhejiang Yuda Chemical Co., Ltd, the solid content of SiO2
was 29.3%, pH was 9.9. SK70# matrix asphalt.
1070
Figure 1. Appearance shapes of PEG/SiO2 SSPCM (a), asphalt-SSPCM before (b) and after (c)
heating.
2.3 Characterization
The pore size, BET and pore volume of silicon dioxide based on its absorption to nitrogen
were determined by V-Sorb 2800 P BET and pore size analyzer (Beijing Jinaipu Science and
Technology Co., Ltd., China).
The surface features of SiO2 and PEG/SiO2 were observed by KYKY-2800B SEM (Beijing
Kyky Technology Co., Ltd., China).
The thermal properties and thermal stabilities of Asphalt-SSPCM were measured by simul-
taneous thermal analyzer STA 449C (DSC/TG, Netzsch, German) The protective gas and
purge gas were high purity nitrogen, and their flow velocities were 20 mL/min and 40 mL/min
respectively. The temperature increase rate was 10°C/min. The test temperature for DSC and
TG were 25~200°C and 25~600°C at a heating rate 10°C/min respectively.
The crystallization properties of Asphalt-SSPCM were characterized by X’PERT-PRO X-ray
powder diffractometer (XRD, PANalytical, Netherlands) The XRD patterns were collected at a
scanning speed of 4°/min in the 2θ range 10∼70° using Cu Kα radiation (λ = 1.54 Å, wavelength).
1072
Table 2. Thermal physical parameters of asphalt-SSPCM.
1# 2# 3# 4#
can be concluded that the residues are mainly inorganic absorption carrier SiO2, and a small
amount of the incomplete-decomposed residue of asphalt and PEG. As the stable chemical
properties of inorganic absorption carrier SiO2 in the temperature range of the TG analysis,
there would be no thermal decomposition reaction and the residues are mainly contributed
by SiO2. The lightweight components of asphalt are evaporated at 100°C or so, as the tem-
perature increases, the low molecular components will be decomposed and loss the masses,
and small amount of residues will be remained. PEG curve shows that the residue mass is
4.28%, which may be caused by the impure components.
In conclusion, Asphalt-SSPCM can meet the technical requirements for asphalt pavement
materials, which possesses suitable phase change temperature, higher enthalpy and good
thermal stability [5].
4 CONCLUSIONS
PEG/SiO2 SSPCM was prepared by using SiO2 as the carrier and PEG as the phase change
material by sol gel method. Then it was mixed with melting-asphalt to obtain Asphalt-
SSPCM with different PEG content. This solved the leakage problem of the application of
PEG solid-liquid phase change material fundamentally.
The mass fraction of PEG decided the heat storage capacity of Asphalt-SSPCM. With
the increase of the mass fraction of PEG, the enthalpy of phase change of Asphalt-SSPCM
increased, but lower than the theoretical enthalpy. The thermal stability of Asphalt-SSPCM
with different PEG content was fairly good.
There were crystallized PEG existing in the PEG/SiO2 SSPCM structure of the asphalt.
With the decrease of the PEG content, the crystallized PEG features weakened.
Because of the high enthalpy of phase change of PEG/SiO2 SSPCM in asphalt and its good
thermal stability, the phase change temperature matched with the temperature of the high-
temperature diseases of asphalt pavement. What’s more, the inorganic porous substrate SiO2
in PEG/SiO2 SSPCM played as a framework, which offered fixed shapes and certain mechan-
ical properties. It can be applied in the asphalt mixture as alternative fine aggregate and
formed the controllable phase change heat storage asphalt mixture so as to actively respond
to the environment temperature change and achieve heat storage and cooling effect.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the supports from the National Natural Science Fund
project of China (Grant No. 51508063).
1074
REFERENCES
Harald Mehling, Luisa F. Cabeza. Heat and Cold Storage with PCM [M]. Berlin, Germany: Springer-
Verlag Berlin Heidelberg Press, 2008.
Zhang Ren-yuan. Phase Change Material and Phase Change Energy Storage Technology [M]. Beijing:
Science Press, 2009.
Hu Shu-guang, Li Qian, Huang Shao-long et al. Feasibility Study of PEG Applied in Asphalt Mixture
[J]. Pavement, 2009(7): 291–295.
Ma Biao, Wang Xiao-man, Li Chao et al. Analysis of Application Prospect of Phase Change Materials
in Asphalt Concrete Pavement [J]. Pavement, 2009(12): 115–118.
Zhang Yi-bo, Zhu Hong-zhou, LI Jing-ruo et al. Selections of Phase Change Materials in Heat Stor-
age & Cooling Asphalt Pavement [J]. Journal of Zhengzhou University (Engineering Version), 2012,
33(3): 10–14,18.
Cao Chang-bin, Luo Yang-ming, Li Wen-hu et al. Heat Storage Impact of PEG on Asphalt and Asphalt
Mixture [J]. New Chemical Material, 2013, 41(4): 137–139.
Li Chao, Ma Biao, Wang Xiao-man, et al. Experimental study on self-adjustable temperature asphalt
mixture [C]. Proceedings of International Workshop on Energy and Environment in the Develop-
ment of Sustainable Asphalt Pavements. Xi’an: Xi’an Jiaotong University Press, 2010: 65–71.
Ma Biao, Wang Sha-sha, Li Jin. Study on Application of PCM in Asphalt Mixture [J]. Advanced Mate-
rials Research, 2011, Vols. 160–170: 2625–2630.
Chen Mei-zhu, Hong Jing, WU Shao-peng, et al. Optimization of phase change materials used in asphalt
pavement to Prevent rutting [C]. 2011 International Conference on Information Science, Automation
and Material System, Advanced Materials Research, 2011, V219–220: 1375–1378.
Chen, M.Z.,Wan, L., Wu, S.P., et al. Effect of phase change material on temperature susceptibility of
asphalt [C]. Proceedings of the International Conference on Heterogeneous Material Mechanics, 3rd,
Shanghai, China, 2011: 576–579.
Chen Mei-zhu, Xu Guang-ji, Wu Shao-peng, et al. Preliminary study on asphalt mortar containing
shaped- stabled phase change material [C]. 1st International Congress on Advanced Materials 2011.
Advanced Materials Research, 2011, V306–307: 1702–1706.
Cocu, X., D. Nicaise, S. Rachidi. The Use of Phase Change Materials to Delay Pavement Freezing [C].
Transport Research Arena, Brussels. 2010, 1–13.
Li Jing-ruo, He Li-hong, Liu Tang-zhi Liu, CAO Xue-juan, ZHU Hong-zhou*. Preparation and Char-
acterization of PEG/SiO2 composites as shape -stabilized phase change materials for thermal energy
storage [J]. Solar Energy Materials and Solar Cells, 2013, 118: 48–53.
He Li-hong*, Li Jing-ruo, Zhou Chao, Zhu Hong-zhou, Cao Xue-juan, Tang Bo-ming*. Phase change
characteristics of shape-stabilized PEG/SiO2 composites using calcium chloride-assisted and tem-
perature-assisted sol gel methods [J]. Solar Energy, 2014, 103: 448–445.
Wang Wei-long, Yang Xiao-xi, Fang Yu-tang, et al. Preparation and performance of form-stable poly-
ethylene glycol/silicon dioxide composites as solid–liquid phase change materials [J]. Applied Energy,
2009, 86: 170–174.
Wang Wei-long, Kang Hui-ying, Yang Xiao-xi et al. Performance Study of PEG/SiO2 Composite Phase
Change Heat Storage Materials [J]. Functional Materials, 2007, 38(10): 1652–1654.
Yang Huazhe, Feng Li-li, Wang Chong-yun, et al. Confinement effect of SiO2 framework on phase
change of PEG in shape-stabilized PEG/SiO2 composites [J]. European Polymer Journal, 2012, 48:
803–810.
1075
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Xuejuan Cao
School of Material Science and Engineering, Chongqing Jiaotong University, Chongqing, China
Chen Chen
Department of Civil and Environmental Engineering, Washington State University, Pullman,
Washington, USA
Xiaoyu Yang
School of Material Science and Engineering, Chongqing Jiaotong University, Chongqing, China
Boming Tang
School of Civil Engineering, Chongqing Jiaotong University, Chongqing, China
ABSTRACT: Nanometer TiO2 powders was modified by doping with urea as a doping source.
The photocatalytic activity of Nitrogen-doped titanium oxide (N-doped TiO2) was evaluated under
different nitrogen content by the photodegradation test of methylene blue (MB). The results show
that when the dosage of N elements is 2%, the degradation rate of N-doped TiO2 reaches a maxi-
mum. From the X-ray Diffraction analysis, it is found that the position of the crystal plane diffrac-
tion peak of the doped crystal shifts to a small angle, which indicates that the N enters the crystal
lattice and causes distortion, increases the photocatalytic activity center. The modified nano TiO2
is dispersed in the silica sol under the condition of ultrasonic vibration, and the results of Scanning
Electron Microscope (SEM) show that the dispersed TiO2 particles are smaller and dispersed evenly
in the dispersion. The cement-based specimens loaded nano TiO2 dispersion were respectively pre-
pared by direct mixing method, sprinkle paving method and spraying method. The photocatalytic
activity of the specimens was evaluated. It can be concluded that the NO degradation rate of three
kinds of specimens increases with the increase of nano TiO2 amount; the specimens prepared by
direct mixing method and sprinkle paving method have lower NO degradation rate because TiO2
is wrapped in cement and stone and distributed unevenly; NO degradation rate of the specimens
prepared by spraying method can reach 74.7% when the amount of TiO2 is 15 g/m2.
1 INTRODUCTION
With the rapid development of transportation industry and the use of fossil fuels, dust, sulfur
oxides, nitrogen oxides, carbon oxides and other substances into the atmosphere, the quality
of the atmosphere is seriously deteriorated. Air pollution in the city is the most serious prob-
lem. It mainly comes from the exhaust emissions of vehicles. The main components of auto-
mobile exhaust are Carbon Monoxide (CO), Nitrogen Oxides (NOx), hydrocarbons (CH),
solid suspended particles, etc. Due to air pollutants gathered on the road for a long time, so
some road researchers began to develop a kind of photocatalytic performance pavement with
the ability to degrade automobile tail gas.
In 1997, Japan made photocatalyst load on the paving bricks to form the first photocatalytic
functional way functional Road, which can degrade the automobile exhaust along the street(1).
Ballari et al. mixed the dry TiO2 powder with cement to prepare a photocatalytic cement specimen
according to the ordinary concrete preparation method. The results show that NOx was effec-
1077
tively degraded by the specimen under UV radiation, and the degradation efficiency is proportion
to the light intensity and inversely proportion to the gas relative humidity (2–4). In 2006, the Ital-
ian laid a 500 m long photocatalytic concrete pavement, and the laying area reached 12000 m2 (5).
The monitoring analysis showed that for the degradation rate of NOx, this environment-friendly
pavement can make it up to 45%. In 2011, Hassan evaluated the environmental benefits of the
application of TiO2 in warm mix asphalt (6). In 2004, Zhao et al. sprayed the nano TiO2 aque-
ous solution on the surface of cement concrete, and evaluated the effect of the water saturation
and concrete ages and strength on photocatalytic performance (7). In 2005, the Qian’s team laid
5000square meters of photocatalytic road in the toll station square of Nanjing Yangtze River
Bridge (8). In 2010, Tan applied the nano TiO2, which in the form of coating, sprinkle paving,
direct mixing, to the asphalt mixture in the preparation of photocatalytic specimens. The results
show that the specimens prepared by sprinkle paving method and coating method have better
photocatalytic degradation effect of NOx pollution (9). In 2013, Chen carried out the spraying
test of nanometer TiO2 in the porous concrete, and he confirmed that its photocatalytic effect is
better than the application effect of the ordinary road. Now, most of Photocatalytic pavements
exist problems such as low efficiency (10). On the one hand, the band gap of pure TiO2 is big and
the carriers are easy to composite, which lead to the fact that light energy utilization and photo-
catalytic degradation efficiency is low; on the other hand, nanoparticles are easily reunited, which
results in the decrease of its activity. Therefore, in this paper, the nanometer TiO2 was modified
by nitrogen doping in this paper, and the modified nano TiO2 was dispersed in the silica sol under
the condition of ultrasonic vibration to form the nano dispersion. Finally, it was applied to the
cement concrete specimens, and the photocatalytic effect was evaluated.
2 EXPERIMENTAL
2.1 Materials
Urea, Sichuan Lutianhua group; Nano TiO2, Anatase, Anhui province Xuancheng Jingrui
New Material Co. Ltd.; Cement, PO.42.5 ordinary portland cement, Chongqing Lafarge
cement Co. Ltd.;
Machine-made sand as fine aggregate. The main properties of cement and fine aggregate
are shown in Table 1 and Table 2, respectively.
Compressive Flexural
Soundness Setting time/h strength/MPa strength/MPa
Le chatelier
Test items Fineness/% soundness test Initial set Final set 3d 28d 3d 28d
Standard < Must be qualified > > > > > >
10 3 6 10 32.5 2.5 5.5
requirement
Results 2.1 Qualified 3.25 6.00 27.2 55.5 6.5 10.0
1078
quality of nitrogen dioxide in the proportion of nano TiO2 to add urea. The mixture at 85° for
drying and grinding, and muffle furnace calcines for 3 hours at 500° to prepare N-TiO2.
C0 − C A0 − A
η= = × 100% (2)
C0 A0
In the formula: C0, A0 = the initial concentration and the absorbance of the solution,
respectively; C, A = the Concentration and the absorbance of the solution after reaction,
respectively.
1079
2.6 Scanning Electron Microscope analysis (SEM)
The distribution state and particle size of nano TiO2 particles before and after being scattered
were tested by KYKY-2800B scanning electron microscope. SEM uses a very narrow electron
beam to scan tested samples and produce secondary electrons. Furthermore, the surface mor-
phology and size of samples can be measured by using the secondary electrons to enlarge the
characteristics of surface morphology.
Figure 1. Cement-based specimens preparation. (a) Schematic diagram of spray (b) Comparison dia-
gram of specimens before and after spraying.
1080
could be spread evenly over the surface of the specimens which had been molded. Among
them, the nano TiO2 dosage was 10 g/m2, 25 g/m2, 50 g/m2.
3. Spraying method
The cement mortar specimens were molded according to highway engineering cement and
cement concrete test code (JTG E30–2005), and a certain amount of dispersion was sprayed
on the surface of the specimens by a spraying machine when the specimens had been cured
under standard curing conditions for three days. The amount of dispersant was 5 g/m2, 10 g/
m2, 15 g/m2, respectively.
1081
around the substituted position, which leads to the lattice distortion. The lattice distortion
makes the crystal surface generate defects, and inhibits the recombination of the photo pro-
duced electron hole, at the same time increases the photocatalytic activity centers.
Figure 5. SEM of the dispersion and undispersed TiO2 slurry. (a) TiO2 dispersion (b) Undispersed
TiO2 slurry.
1082
From Fig. 5(a), it is known that the TiO2 dispersed particles distribute uniformly in the
dispersion and the particle size is small, however, in Fig. 5(b), the phenomenon of agglom-
eration is serious in the TiO2 which has not been dispersed, and its particle size is huge. The
characteristics of nano materials and the results of electron microscopy show that the disper-
sion of TiO2 particles can be greatly improved by dispersing treatment and mixing, which can
effectively prevent the particles from agglomeration and decrease the particle size and then
enhance the photocatalytic degradation efficiency.
Figure 6. Effect of TiO2 content on the degradation rate of NO (direct mixing specimens).
1083
Figure 7. Effect of TiO2 content on the degradation rate of NO (sprinkle paving specimens).
is greatly improved. On the one hand, nano TiO2 of spraying specimens is distributed in the
specimen surface, which can be more fully exposed to the light source and the NO. On the
other hand, the TiO2 particle size of dispersion that is used in spraying specimens is small
and the particles distribute uniformly. Therefore, the photocatalytic reaction position on the
surface of spraying specimens is greatly increased, and the physical and chemical properties
of nano materials and the electron diffusion velocity are enhanced, which can make the NO
degradation rate of spraying specimens reach the highest under the condition of very little
spraying amount.
4 CONCLUSIONS
1. Nanometer TiO2 powders was modified by doping with urea as a doping source. The deg-
radation rate of MB solution increases and then decreases with the increase of N content;
when the dosage of N elements is 2%, the degradation rate reaches a maximum.
2. From the X-ray Diffraction analysis, it is found that the position of the crystal plane
diffraction peak of the doped TiO2 crystal is shifted, which indicates that the N element
1084
enters the crystal lattice and causes distortion, increases the photocatalytic activity center,
and the photocatalytic performance is improved.
3. It can be found by SEM that the modified nano TiO2 can be uniformly dispersed in the
silica sol under the condition of ultrasonic vibration, and the particles are small.
4. The cement-based specimens loaded nano TiO2 dispersion were respectively prepared by
direct mixing method, sprinkle paving method and spraying method. And it was com-
pared that the degradation rate of NO under different preparation methods. The results
show that the NO degradation rate of three kinds of specimens increases with the increase
of nano TiO2 amount, and then tends to be gentle; the NO degradation rate of the speci-
mens prepared by spraying method is highest and it can be up to 74.7%.
REFERENCES
Ballari, M. & Brouwers, H. Full Scale Demonstration of Air-purifying Pavement[J]. Journal of Hazard-
ous Materials, 2013, 254(255): 406–414.
Ballari, M., Hunger, M., Hüsken, G., et al. Modelling and Experimental Study of the NOX Photocata-
lytic Degradation Employing Concrete Pavement with Titanium Dioxide[J]. Catalysis Today, 2010,
151(1): 71–76.
Ballari, M., Yu, Q. & Brouwers, H. Experimental Study of the NO and NO2 Degradation By Photocata-
lytically Active Concrete[J]. Catalysis Today, 2011, 161(1): 175–180.
Bolte, G., Bittnar, Z., Bartos, P., et al. Nanotechnology in Construction[M]. [S.l.]: Springer Berlin Heidel-
berg, 2009: 55–61.
Chen Meng, Jin Li-sheng & Chu Jiang-wei. Performance study on the decontamination of NOx from
vehicle base on photocatalytic active concrete road. Journal of Wuhan University of Technology, 2013,
35(2): 61–65.
Fujishima, A., Hashimoto, K. & Watanabe, T. TiO2 Photocatalysis Fundamentals and Applications.
Tokyo: BKC Inc, 1999.
Hassan, M., Mohammad, L.N., Dylla, H., et al. A Breakthrough Concept in the Preparation of Highly-
sustainable Photocatalytic Warm Asphalt Mixtures[C]//Nsf Engineering Research and Innovation
Conference, Atlanta, Georgia, [S.l.]: [s.n.], 2011.
Li li & Qian Chun-xiang. Removal of Nitrogen Oxides in Vehicular Emissions on Photo-catalyst treated
Portland Cement Concrete Road of Nanjing No. 3 Yangtze River Bridge. Journal of Henan Univer-
sity of Science and Technology Natural Science, 2009, 30(1):49–52.
Tan Yi-qiu, Li Luo-ke & Wei Peng. Application Performance Evaluation on Material of Automobile
Exhaust Degradation in Asphalt Pavement. China Journal of Highway and Transport, 2010, 23(6):
21–27.
Zhao Lian-fang, Qian Chun-xiang & Fu Da-fang. Preparation and study on the photocatalytic con-
crete. Journal of Functional Materials, 2004, 35 (Supplement):2546–2549.
1085
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The objective of this study is to quantify the noise generated due to tyre-
pavement surface interaction for different modes and to develop a noise prediction model for
each mode with the use of old and new tyres by considering the factors affecting tyre-pave-
ment surface interaction noise. Noise levels were measured as a function of vehicle type, vehi-
cle speed, pavement temperature, direction of wind, condition of vehicle tyre with respect to
age, condition of the pavement and type of pavement. The tyre-pavement interaction noise
was collected on eight asphalt pavements and four cement concrete pavements, for four dif-
ferent modes of test vehicles using a sound level meter by control pass by method. The noise
data was processed to develop noise prediction models for each mode in each survey location
and a combined tyre-pavement interaction noise model is developed for each mode sepa-
rately for both asphalt and cement concrete pavements.
1 INTRODUCTION
Sound which exceeds the acceptable level and creates annoyance is known as noise. Exposure
to the high level noise will results in stress on the auditory system which can lead to psycho-
logical disorders. In urban areas, transportation sector is one of the major contributors to
noise pollution. Urbanization increased the demand for transport, thus increasing the road
based vehicles which in turn increased the noise pollution. Specifically, highway traffic noise
has become an important environmental concern as traffic related noise pollution accounts
for nearly two-third of the total noise pollution in cities. Even though, practices like installa-
tion of noise barriers can reduce the noise levels, cost incurred is more. Thus, more appropri-
ate solution is to reduce noise at the source level itself.
In general, noise from the motor vehicle is classified into power train noise, tyre-pavement
interaction noise and aerodynamic noise. As the engine speed of the vehicle increases, noise
from the exhaust and drive train increases. Similarly, as vehicle speed increases noise from
tyre road interaction increases due to increased passage of air circulating within the tread
grooves of the tyre in contact with the pavement. Aerodynamic noise is influenced by the
body geometry of the vehicle and is high at the higher vehicular speeds. Tyre pavement noise
dominates the engine and exhaust noise at higher speeds which essentially depends on tyre
characteristics, pavement characteristics and vehicular characteristics. Thus, the evaluation
of tyre pavement noise is done by considering all the characteristics influencing the genera-
tion of tyre/road noise. Environmental parameter like temperature is also taken into account
in quantifying the tyre road noise, since it changes the pavement temperature over time which
in turn affects tyre stiffness leading to tyre road noise.
2 REVIEW OF LITERATURE
Parameters affecting the generation of tyre pavement interaction noise have to be identified in
order to mitigate the noise at the source level. The percentage of contribution in generating the
1087
tyre pavement noise is not constant for the parameters which are affecting tyre pavement noise.
Therefore, thorough study has been made in order to identify these parameters before quantify-
ing the noise levels. Several works have been done in past on tyre pavement noise generation and
grouped the factors affecting the generation of noise levels at the tyre pavement interface under
three headings:environmental, pavement and traffic characteristics. Pavement characteristics indi-
cate both the surface characteristics and mix properties. Kandhal (2004)had presented study on
mitigating measures of tire/road noise by using an open-graded asphalt rubber mix.The author
identified that conventional Asphalt Concrete (AC) pavements having isotropic surface texture
generating lesser tyre pavement noise compared to Cement Concrete (CC) pavements having ani-
sotropic surface texture. Bennert et al. (2005) observed the effect of nominal aggregate size of the
AC materials in mitigating the tyre road noise in comparison with CC pavements. Neithalath et al.
(2006) evaluated the effect of aggregate size of pavement on noise generation by using the stress
distribution model and reported that variation in aggregate sizes for producing enhanced porosity
concrete is proven to be effective in reducing the tire pavement noise. Harvey et al. (2006) reported
that due to higher airvoid contents and permeability, open-graded mixes reducing the tire/pave-
ment noise up to 4.5 dB (A) in comparison with other asphaltic mixes. Paul (2006) reported that
quiet, safe, long-lasting pavements can be constructed with hard sand and with use of conventional
textures. Aditya et al. (2012) reported that cement concrete pavement showing an increase of 4 to
5 dB (A) noise levels compared to bituminous pavement. It is due to the fact that cement concrete
pavement will have high surface stiffness and is not capable of absorbing sound energy. Randen-
berg & Miljkovic (2012) reported that porous pavements reduces the splash and spray due to voids
in pore structure which leads to less stick-snap and results in less tyre pavement noise generation.
Syamkumar et al. (2013) developed an noise prediction model for different modes of vehicles on
asphalt and rigid pavements and observed an increase in noise level with the increase in vehicle
speed and vehicle size irrespective of pavement condition. Mak & Loh (2015) had developed a sim-
ple tyre/road noise model linking acceleration and vehicle speed and observed that proposed model
is potentially sound enough for estimating the tyre/road noise on asphalt pavements effectively.
Walsh & Krylov (2006) reported a study on investigation on the air pumping effect at the
tyre/road interface due to the tread block mechanism. The air squeezing out of the com-
pressed grooves and rushing into the expanded grooves gives rise to air fluctuations which in
turn raises the tyre pavement noise. The tyre pavement noise interaction is also influenced by
atmospheric conditions in field. It is due to the fact that noise propagation depends on the
density and speed of air which again depends on the temperature. Pavement temperature has
been taken into study since it affects the stiffness of tyre and viscoelastic nature of asphalt
which influences the tyre road noise generation.
In India, Central Pollution Control Board (CPCB) provides guidelines for restriction of
noise levels in public places as shown in Table 1. Noise generated from vehicular movement is
major contributor to over all noise in public places. CPCB also provides guidelines on noise
limits for various vehicles applicable at manufacturing stage in which the maximum noise
generated from any vehicle shall not exceed 80 dB (A). Since the major contribution of noise
from traffic is due to tyre pavement interaction and the power train noise, mitigating the tyre
pavement noise at the source will make the better environment. In this study, noise levels are
1088
quantified by considering various modes of transport with old and new tyres separately. It is
because of the fact that worn out and new tyre mechanism will influence the tyre pavement
noise generation due to variation in the air pumping within the tread block mechanism,
roughness of the tyre in contact with the pavement and stiffness vibrations within the tyre.
Based on this, appropriate noise prediction models are developed for each study location
including asphalt and cement concrete pavements for each mode and a combined noise pre-
diction model is developed for each mode with old and new tyres separately.
3 METHODOLOGY
The study methodology contains selection of parameters that affect tyre-pavement noise,
identification of survey locations to conduct tyre-pavement noise surveys, field studies which
include measurement of noise levels with respect to traffic surveys. The different surveys like
type of pavement surface, temperature, roughness, texture depth have been included in this
study. Quantifying the noise based on these parameters, tyre-pavement noise prediction models
are developed. Methods available for noise measurements are classified into far-field measure-
ments and near field measurements. In the far field method, microphones are placed along
the side of a roadway. These measurements are further classified into statistical pass-by and
controlled pass-by methods. The statistical pass by method consists of placing the microphone
at a distance of 1.2 to 1.5 m depending on the standards followed in the respective countries.
In this study, noise levels were measured using controlled pass-by method using pocket Sound
Level Meter (SLM) with real time analyzer having a noise measuring range of 30 to 130 dB (A).
In the controlled pass by method, noise from the selected single test vehicle is measured. In the
test procedure, tyre-pavement interaction noise is measured by eliminating the noise generated
from engine propulsion and vehicle exhaust systems by allowing the test vehicle to move at the
specified speed and switching off the engine soon after the desired speed was attained.
The peak noise levels produced by the selected vehicles were measured on identified test
sections with the instrument set to A-weighted filter mode which represents that instrument
is set to the human sensitivity of ear while recording the noise in dB (A). The test vehicles
considered in this study are: bus, car, Auto-Rickshaw (AR), and two wheeler (2 W). All the
test vehicles selected are in good condition. In order to measure the total noise from the test
vehicle, vehicle is driven at the target speed and by using sound level meter; noise is meas-
ured in dB (A). For recording the tyre pavement noise alone by eliminating the engine noise,
vehicle is driven at the same target speed and the engine is switched of soon after vehicle
attains desired speed. From methodology adopted, noise level captured is the combination of
aerodynamic and tyre pavement noise. But, the aerodynamic noise is neglected because it is
only experienced by the person in the vehicle and its effect is very less compared to remaining
noises. The noise levels were measured on eight asphalt pavements and four cement concrete
pavement mid-block sections in the tri-cities of Kazipet, Hanamkonda, and Warangal in
Telangana State, India. The measurement sites were selected such that the background noise
is eliminated. The test variables considered in this study are given in Table 2 and the list of
survey locations are given in Table 3.
1089
Table 3. List of survey locations selected in this study.
Different test vehicles with old and new tyres were selected and noise levels were measured at
different speeds using controlled pass by method on asphalt concrete and cement concrete pave-
ments with varying vehicle and pavement characteristics. Variation of Tyre Pavement Interac-
tion Noise (TPIN) levels measured on AC pavement (road no.1) for 2 W and car in dry and wet
conditions with new and old tyres are shown in Figure 1 and Figure 2, respectively. Similarly,
variation of TPIN levels measured on CC pavement (road no. 9) for 2 W and car in dry and wet
conditions with old and new tyres are shown in Figure 3 and Figure 4, respectively. Noise levels
were quantified by taking into account the following variables: vehicle type, vehicle speed, pave-
ment temperature, direction of wind, condition of vehicle tyre with respect to age, condition
of the pavement and type of pavement. Variation of tyre pavement noise levels are carefully
analyzed for all these variables on AC and CC test sections.Individual models are developed for
all the modes separately for old and new tyres because tyre condition influence the generation
of noise with the governing parameters including both pavement and vehicle parameters. Air
passage through the tine mechanism present on the tyre and the pavement surface in contact
with the tyre will influence the noise generation due to the amount of air that can pass through
the tread block mechanism while the vehicle is in motion. As this tread block mechanism will
be different for old and new tyres, separate models are considered.
The trend of each independent variable as a function of the TPIN is analyzed for all
the twelve roads. The observations of above trends are summarized here. The TPIN lev-
els increased with an increase in speed of the vehicle. As the vehicle speed increases, the
number of tyre revolutions per unit time increases implying that the speed with which the
tyre comes in contact with the pavement surface increases resulting in higher impact with
the pavement surface and also an increased pumping in and out of air from the wheel treads.
The wind direction was measured on a scale of 0 to 180° wherein the wind might be blow-
ing either towards or away from the point where the SLM was placed. As the wind blows
away from the receiver, the noise levels decreased and the noise levels increased as the wind
blows towards the receiver. TPIN levels decreased with increase in pavement temperature.
The increase in noise levels in the case of AC pavements with a reduction in temperature
is essentially due to increased stiffness of asphalt concrete. Asphalt exhibits brittle elas-
tic behavior at lower temperatures whereas at higher temperatures, the material exhibits
a significant viscous behavior resulting in increased noise levels at lower temperatures. In
CC pavements, the noise levels increased with decrease in pavement temperature. Since the
1090
Figure 1. TPIN for 2 W on road 1. Figure 2. TPIN for car on road 1.
cement concrete stiffness will not depend much on temperature, the increased noise lev-
els during night time may be attributed to increased hardness of the vehicle tyre. In other
words, the tyres become less noisy at higher temperature due to weakening of the rubber,
making noise radiation by tyre vibration less efficient. Further, the temperature of the trans-
mission path also affects the noise levels because of the dependence of the density and
pressure of the transmission medium on the temperature. Wet pavements produce more
noise compared to dry pavements because of the water splash effect. Vehicles with new tyres
produced more noise than vehicles with old tyresboth in wet and dry condition. It may be
due to the fact that air sprinkling with the tyre which is major source for tyre road noise in
wet pavements is dominant with the new tyres due to the tyre tread block and high tyre cav-
ity presence between new treads and sharp tine dimensions. Sensitivity analysis was done for
two-wheelers and bus separately on road−1. Because for bus, wet and dry condition are not
considered for analysis because similar results were obtained for both new and old tyre on
both dry and wet pavements. But, 2 W, AR and car have shown considerable variation with
new and old tyres on wet and dry pavements respectively. Regression model was developed
by taking all independent variables that affect tyre-pavement interaction noise. To check the
interdependency between the independent variables (surface condition, speed of the vehicle,
direction of wind, pavement temperature, and air temperature) different regression models
were developed by neglecting one independent variable at each time. From the results of
sensitive analysis, it was observed that the air temperature and pavement temperature are
interdependent and, so air temperature can be neglected.
1 2W 91.16 (4.92) −2.08 (0.46) 0.29 (3.23) −0.04 (1.08) −2.08 (1.73) 0.97
AR 204.21 (2.78) −5.06 (1.09) 0.33 (7.17) −0.03 (0.92) −3.58 (1.11) 0.97
Car 233.90 (3.73) −5.33 (3.21) 0.26 (6.17) −0.11 (1.14) −4.73 (1.16) 0.99
Bus 86.65 (1.49) — 0.06 (2.69) −0.02 (0.72) — 0.66
2 2W 224.32 (2.01) −6.47 (2.45) 0.37 (10.02) −0.19 (0.06) −4.78 (1.02) 0.98
AR 162.78 (2.97) −4.48 (2.01) 0.42 (8.02) −0.08 (1.39) −2.62 (0.24) 0.96
Car 13.61 (1.51) 4.88 (0.98) 0.44 (7.58) −0.04 (0.76) 2.08 (0.87) 0.96
Bus 27.00 (1.94) — 0.80 (2.98) 0.18 (0.23) — 0.90
3 2W 180.32 (1.90) −2.94 (0.45) 0.38 (14.64) 0.09 (0.76) −3.85 (1.62) 0.98
AR 135.36 (4.27) −0.68 (4.13) 0.47 (6.16) 0.03 (0.28) −2.10 (1.48) 0.95
Car 248.44 (1.99) −6.47 (2.34) 0.35 (5.67) 0.61 (1.16) −9.11 (1.09) 0.94
Bus 77.28 (2.70) — 0.63 (1.93) −0.07 (0.26) — 0.87
4 2W −14.83 (0.53) 4.76 (1.97) 0.50 (9.03) 0.06 (2.44) 1.85 (1.22) 0.97
AR 140.68 (1.55) −3.95 (1.02) 0.32 (3.72) −0.10 (0.27) −1.70 (0.42) 0.91
Car 124.19 (2.40) 0.53 (2.90) 0.37 (4.64) −0.11 (0.25) −1.05 (1.02) 0.97
Bus 144.58 (1.4) — 0.93 (1.93) — −0.63 (0.70) 0.96
5 2W 2.82 (1.83) 13.05 (2.12) 0.56 (9.72) −0.01 (0.43) 1.56 (0.83) 0.98
AR 163.98 (0.94) −1.11 (6.36) 0.56 (5.53) −0.06 (2.26) −2.71 (1.91) 0.88
Car 71.31 (1.24) −0.20 (1.23) 0.40 (4.49) 0.05 (1.32) −0.30 (2.08) 0.92
Bus 102.23 (0.80) — 0.70 (0.95) −0.26 (0.65) — 0.75
6 2W 200.76 (5.86) −6.20 (2.78) 0.39 (15.08) −0.05 (0.93) −5.05 (1.36) 0.98
AR 316.88 (3.23) 1.15 (0.98) 0.52 (8.27) −0.38 (2.61) −4.10 (2.35) 0.96
Car 150.68 (2.56) 5.64 (4.15) 0.40 (5.73) −0.24 (2.04) −0.96 (1.93) 0.94
Bus 25.43 (0.83) — 0.70 (2.77) 0.48 (1.37) — 0.78
7 2W −6.68 (2.76) 13.16 (1.23) 0.53 (11.06) −0.01 (0.62) 2.33 (1.98) 0.98
AR 95.85 (2.71) −6.81 (0.94) 0.59 (5.49) −0.08 (1.32) −0.34 (2.17) 0.94
Car 292.39 (2.42) 8.80 (2.10) 0.09 (3.68) 0.19 (0.61) 8.39 (0.72) 0.95
Bus 15.93 (2.86) — 0.27 (1.98) 0.27 (0.97) — 0.78
8 2W −62.82 (1.83) 11.80 (5.78) 0.53 (10.59) 0.09 (2.07) 3.22 (0.20) 0.95
AR −326.83 (3.19) 16.05 (0.23) 0.75 (6.16) −0.04 (2.33) 11.90 (2.51) 0.92
Car −89.22 (1.58) 10.33 (4.89) 0.48 (3.31) −0.01 (0.59) 4.88 (0.69) 0.91
Bus 1.00 (1.65) — 0.53 (2.64) 0.47 (1.66) — 0.80
9 2W 84.74 (1.94) 6.01 (3.45) 0.51 (14.66) −0.19 (0.82) 0.07 (2.11) 0.95
AR 488.50 (2.34) −19.73 (1.34) 0.26 (3.38) −0.17 (2.40) −10.83 (1.02) 0.96
Car −301.53 (2.18) 10.41 (6.78) 0.55 (5.44) 0.17 (0.31) 0.17 (0.15) 0.93
10 2W 92.66 (3.50) −2.21 (4.56) 0.37 (7.77) 0.02 (2.36) −0.94 (1.75) 0.96
AR 476.06 (0.97) −3.93 (1.34) 0.45 (5.37) −0.35 (1.21) −10.27 (0.83) 0.92
Car −294.84 (2.16) 21.55 (0.67) 0.57 (3.20) 0.12 (0.69) 9.53 (1.03) 0.93
Bus 48.25 (1.99) — 0.47 (3.29) 0.17 (1.56) — 0.77
11 2W 46.58 (0.87) 6.50 (2.10) 0.63 (2.20) −0.26 (0.34) 1.37 (0.46) 0.94
AR 281.31 (2.08) −10.20 (0.89) 0.55 (4.67) 0.73 (0.10) −10.73 (1.69) 0.91
Car 30.22 (2.10) 4.85 (3.12) 0.34 (7.61) 0.09 (1.37) 0.94 (1.06) 0.88
12 2W 14.01 (3.34) 4.87 (1.23) 0.57 (6.33) −0.12 (0.96) 1.90 (1.81) 0.95
AR 37.63 (1.67) 2.61 (2.10) 0.77 (5.23) 0.41 (2.26) −1.66 (0.56) 0.94
Car −294.35 (1.29) 16.63 (0.95) 0.40 (4.31) −0.02 (0.59) 10.73 (1.01) 0.86
Bus 196.70 (1.23) — 0.11 (4.83) −0.80 (2.77) — 0.98
1092
TPIN (car) = a0 + a1* Pavement surface condition + a2* Speed of the vehicle
+a3*Direction of wind + a4 * Pavement temperature (3)
TPIN (Bus) = a0 + a1* Speed of the vehicle + a2*Direction of wind (4)
where, a0, a1, a2, a3, a4 are the regression coefficients.
A TPIN model is developed for each mode with new and old tyres separately and model
developed for each mode with new tyres in each stretch by considering most significant vari-
ables and the regression coefficients for all the 12 roads are shown in Table 4.
From the Table 4, it is observed that there is a significant contribution of the independent
variables in tyre road noise generation is known as observed t-values are less than the table
t-values. The comparison between measured and predicted TPIN levels on asphalt pavement
for various modes with new and old tyres are shown in Figures 5–12. Similarly, the compari-
son between the measured and predicted TPIN levels for various modes with new and old
tyres on cement concrete pavement are shown in Figures 13–20. It can be seen from these
figures that the developed models closely predicts the TPIN levels.
Figure 5. Measured and predicted noise levels Figure 6. Measured and predicted noise levels
for 2 W with new tyres on AC pavement. for 2 W with old tyres on AC pavement.
Figure 7. Measured and predicted noise levels Figure 8. Measured and predicted noise levels
for AR with new tyres on AC pavement. for AR with old tyres on AC pavement.
Figure 9. Measured and predicted noise levels Figure 10. Measured and predicted noise lev-
for car with new tyres on AC pavement. els for car with old tyres on AC pavement.
1093
Figure 11. Measured and predicted noise lev- Figure 12. Measured and predicted noise levels
els for bus with new tyres on AC pavement. for bus with old tyres on AC pavement.
Figure 13. Measured and predicted noise levels Figure 14. Measured and predicted noise lev-
for 2 W with new tyres on CC pavement. els for 2 W with old tyres on CC pavement.
Figure 15. Measured and predicted noise levels Figure 16. Measured and predicted noise levels
for AR with new tyres on CC pavement. for AR with old tyres on CC pavement.
Figure 17. Measured and predicted noise levels Figure 18. Measured and predicted noise levels
for car with new tyres on CC pavement. for car with old tyres on CC pavement.
1094
Figure 19. Measured and predicted noise levels Figure 20. Measured and predicted noise lev-
for bus with new tyres on CC pavement. els for bus with old tyres on CC pavement.
5 CONCLUSIONS
In this study, TPIN models are developed for each mode with old and new tyres for both asphalt
and cement concrete pavements. The conclusions drawn from this study are listed below:
• It is observed that, tyre-pavement noise increased with the speed irrespective of the mode.
At the initial speeds, total noise for bus is more and gradually decreased with increase in
speed which is due to the fact that at initial condition, vehicle engine noise for heavy vehicle
is more which is reverse for two-wheeler, auto-rickshaw and car.
• With increase in speed of the bus from 20 to 50 kmph, the noise levels increased by about
12 dB (A) for bus with old tyre in dry condition. Same vehicle produced about 15dB (A)
more noise in case of new tyre in dry condition. It is due to the fact that tyre surface con-
tact area for bus will be more due to which new tyre with high friction will generate more
noise at high speeds.
• The noise levels on cement concrete pavement are 4 to 5 dB (A) more than that on asphalt
concrete pavement for auto-rickshaw, 3 to 7 dB (A) for car for both old and new tyres. It
is due to the fact that more acoustical energy will be dissipated in AC pavements due its
viscoelastic nature.
• With increase in pavement temperature, there is a reduction in tyre-pavement interaction
noise possibly due to softening of vehicle tyre rubber.
• For all the modes, wet pavements produced more noise as compared to dry pavements due
to the water splashing effect at the tyre pavement interface irrespective of the vehicle and
tyre condition.
• The speed of vehicle and wet condition of pavement are affecting the noise generation with
old and new tyred vehicles, which can be mitigated by speed restrictions, better drainage
properties of pavement.
REFERENCES
Aditya, G., Naseer, P., Venkat, S., Jagadish, C.S. and Chowdary, V. 2012. Studies on noise generated
due to tyre-pavement surface interaction, Proceedings of International Conference on Environmentally
Sustainable Urban Ecosystems, CD-ROM, Guwahati, India.
Bennert, T., Hanson, D., Maher, A., and Vitillo, N. 2005. Influence of pavement surface type on tire/
pavement generated noise, Journal of Testing and Evaluation, Vol. 33, No. 2, pp. 1–20.
Harvey, J., Qing, L., Erwin, K. and Ongel, A. 2007.Comparison of surface characteristics and pave-
ment/tire noise of various thin asphalt overlays, Transportation Research Board 86th Annual Meeting
(CD-ROM), Washington, D.C.
Kandhal, P.S. 2004. Asphalt pavements mitigate tyre/pavement noise, Arizona Milepost, Auburn, pp.
22–31.
Mak, K.L. and Loh, W.K. 2015. Statistical tyre/road noise modeling in Hong Kong on polymer modi-
fied friction course. Science, Technology and Development, 34 (2): 101–104.
1095
Neithalath, N., Garcia, R., Weiss, J. and Olek, J. 2006. Tire-pavement interaction noise: recent research
on concrete pavement surface type and texture, Fifth International Conference on Concrete Pavement
and Rehabilitation, Purdue University, Vol.2, pp. 107–124.
Noise limits for vehicles at manufacturing stage, Central Pollution Control Board, New Delhi, 2005,
http://cpcb.nic.in/divisionsofheadoffice/pci2/Noise-vehicle.pdf, last accessed May 13, 2015.
Paul, W. 2006. Concrete solutions for quieter pavements on existing roadways, Proceedings of National
concrete pavement technology centre, Iowa State University, U.S.
Randenberg, M., Miljkovic, M. 2012. The noise reducing asphalt pavements for the urban areas in ger-
many.” International Journal of Pavement Engineering, vol.13, Issue 6, pp. 569–576.
Sivaraj, K.P., Subramani, T. and Kavitha, M. 2012.Modeling of traffic noise pollution,International
Journal of Engineering Research and Applications, Vol. 2, No. 3: pp. 3175–3182.
Syamkumar, A., Aditya, K. and Chowdary, V. 2013. Development of mode-wise noise prediction mod-
els for the noise generated due to tyre-pavement surface interaction, Advanced Materials Research,
Vol.723, pp. 50–57.
The noise pollution (regulation and control) rules, Central Pollution Control Board, New Delhi, 2000,
http://cpcb.nic.in/divisionsofheadoffice/pci2/noise_rules_ 2000.pdf, last accessed May 13, 2015.
Walsh, S. and Krylov, V. 2006. Experimental investigations into the air pumping effect at the tyre/road
interface, Proceedings of the Institute of Acoustics, Vol. 28.
1096
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The use of Reclaimed Asphalt (RA) or Reclaimed Asphalt Pavement (RAP)
in flexible pavements results in energy savings, material cost savings and conservation of
natural resources. In order to maximize the benefits/savings from RA use, the tendency now-
adays is to incorporate as much RA as possible in recycled asphalt mixtures. In this study
a reference and two recycled asphalt concrete mixtures produced and tested. The recycled
mixtures produced with 50% RAP and two different penetration grade bitumen (50/70 and
70/100). The aggregate gradation and binder content was kept the same in all mixtures tested.
The performance of all mixtures was determined in terms of stiffness (IT-CY) and fatigue
resistance (ITT-CY), at 20 °C. Stiffness was determined at four testing temperatures (0°C,
10°C, 20°C and 30°C) and at five levels of risetime (40 ms, 64 ms, 124 ms, 155 ms and 200 ms)
per testing temperature. Results have shown that stiffness increases with the incorporation of
50% RAP. The highest stiffness was obtained by the mixture with 50% RAP and 50/70 bitu-
men. The recycled asphalt concrete mixture with 50% RAP and 70/100 bitumen showed the
best fatigue performance of all mixtures tested.
1 INTRODUCTION
With the constantly reduced supply of virgin materials, increasing production costs and
shortage of landfill space for disposing reclaimed materials, the use of recycled asphalt mix-
tures has become of major concern in the pavement industry. Recycled asphalts are produced
from reclaimed asphalt pavement (RAP) with the addition of virgin materials (aggregates
and bitumen). In order for the recycled asphalt to be ‘cost-effective, perform well, and be
environmentally sound’ the Federal Highway Administration (FHWA) recommends the ‘use
of recycled mixtures in the construction of highways to the maximum economical and prac-
tical extent possible with equal or improved performance’ (Copeland 2011). It is considered
that the most economical use of RAP is in the intermediate and surface layers of flexible
pavements because the less expensive binder of the reclaimed material (RAP) can replace a
portion of the more expensive virgin binder (Copeland 2011).
In hot recycling the percentage of RAP usually used varies between 10% and 30%. According
to several studies, recycled mixtures with the above RAP contents perform similarly to con-
ventional mixtures (Kandhal et al. 1995, McDaniel et al. 2000, Widyatmoko 2008, McDaniel
et al. 2002, Shah et al. 2007, Li et al. 2008). However, due to the shortage of raw materials and
increased environmental awareness, in many countries, higher RAP contents have started to be
used. As of 2011, more than 40 state agencies in US allow the use of more than 30% RAP (Cope-
land 2011). Apart from practical problems encountered during mixing, the use of high RAP
content is also limited by the necessity of achieving comparative field performance of pavements
constructed with conventional asphalts. An excellent review of very high-content RAP (>40%)
use has been made by Zaumanis & Mallick (2015). According to their review there are issues
that yet need to be addressed such as the development of a methodology to evaluate the blending
of RAP and virgin binder in the laboratory and also the development of a fundamental test for
evaluation of cracking resistance of asphalt mixes with high RAP contents.
1097
Fatigue performance of recycled asphalts, especially when high RAP contents are incor-
porated, is of major concern. The general concept is that the aged stiff RAP binder typically
increases stiffness of the recycled mixture (Al Qadi et al. 2012, West et al. 2013), can cause
fatigue damage (Shah et al. 2007, Daniel et al. 2010), and low temperature brittleness (Ter-
rel et al. 1992). McDaniel et al. (2000) concludes that at higher RAP contents, the indirect
tensile test results and beam fatigue results showed increase in stiffness, which would lead in
cracking if no adjustments in mix design were made. Bennert & Dongre (2010) have reported
that in high RAP contents the mix stiffness and consequently the magnitude of cracking
will largely depend on the degree of blending between the virgin and RAP binders. West et
al. (2013) has examined the use of recycled mixes with 55% RAP. The research showed that
the recycled mixtures stiffness measured by dynamic modulus at different temperatures and
frequencies increased by 25%-60% compared with virgin mixtures. Additionally the recycled
mixtures fracture energy was examined. Virgin mixtures showed better results than the recy-
cled mixtures. A study on long-term pavement performance (LTTP) for overlays of ∼20 years
and 30% RAP content by West et al. (2011) showed that fatigue, longitudinal and transverse
cracking are the distresses that occur more often in RAP mixtures. The study however con-
cluded that the pavements with recycled mixtures performed better than or equal to virgin
pavements for the majority of cases.
Contrary to all the above, the studies by Al-Qadi et al. (2012) and McDaniel et al. (2012)
showed increased fatigue life for mixtures containing 40% RAP or more compared with con-
ventional mixtures. In these studies beam fatigue and Simplified Viscoelastic Continuum
Damage (S-VECD) procedures were respectively used for evaluating fatigue performance.
Similar beam fatigue results were obtained by Shu et al. (2008). West et al. (2012) developed
relationships between laboratory results and test truck findings which suggest that mixtures
with 50% RAP are expected to have better fatigue performance than the virgin control mix.
Considering all the above the objective of this study was to evaluate and compare the stiff-
ness and fatigue performance of recycled asphalt concrete mixtures with maximum aggregate
size 19 mm (AC19) in the laboratory. The reclaimed asphalt (RAP) content of the recycled
mixtures was 50%, so to see the effect of high RAP content on the performance of the recy-
cled mix. Additionally two types of virgin binders were used in the recycled mixtures; a 50/70
penetration grade bitumen and a 70/100 penetration grade bitumen, so to see also the effect
of the virgin binder type on the performance of the recycled mixture.
2 EXPERIMENTAL WORK
The experimental work of this study included firstly the design of an AC-19 mm mixture (ref-
erence mixture –AC19-ref), secondly the design of the recycled mixture with 50% RAP so to
have the same gradation and binder content as the reference mixture and thirdly the testing
of all mixtures in terms of stiffness and fatigue performance.
In particular the AC19-ref mixture was composed by limestone aggregates and 50/70 pen-
etration grade bitumen and was designed using the Superpave mix design (Asphalt Institute
2001). The recycled mixture with 50% RAP was designed so to have the same gradation and
binder content as the AC19-ref mixture and was composed by RAP material, virgin lime-
stone aggregates and virgin binder. Two types of virgin binders were used for the recycled
mixture; 50/70 penetration grade bitumen and 70/100 penetration grade bitumen. Table 1
shows the three mixtures used in this study.
1098
3 MATERIALS
25 100.00
19 96.95
9.5 77.68
4.75 56.45
2.36 34.58
1.18 20.48
0.6 13.51
0.3 9.44
0.075 6.45
RAP bitumen properties
RAP bitumen content 4.39% by weight of RAP
Penetration (pen) 24
Softening point (°C) 66.8
Dynamic viscosity at 150°C(Pa.s) 0.955
A B C
Aggregate
sieve size (mm) Passing %
1099
Table 4. Virgin aggregate characteristic properties.
Property
4 MIX DESIGN
1100
Figure 1. Mix design gradation.
Optimum Bulk
Bitumen specific
content* gravity Voids VMA VFA Stability Flow
(%) (gr/cm3) (%) (%) (%) (kN) (mm)
Recycled mix
Reference mix with 50% RAP
Sieve size
(mm) Passing %
25 100.00 100.00
19 96.95 96.65
9.5 77.68 77.92
4.75 56.45 54.18
2.36 34.58 32.77
1.18 20.48 21.39
0.6 13.51 15.38
0.3 9.44 11.38
0.075 6.45 7.11
For the reference and the recycled mixture of this study, thirty six (36) in total gyra-
tory compacted cylindrical specimens (100 mm in diameter) were produced for stiffness
and fatigue performance. Prior stiffness and fatigue testing the volumetric properties,
1101
aggregate gradations and binder characteristic properties were determined, so to verify
that the produced mixtures were complying with the designed ones. After verification of
the above, the stiffness and fatigue performance of the produced specimens was deter-
mined in accordance with CEN EN 12697-26 (2012) and CEN EN 12697-24 (2012)
respectively.
5.3 Properties of the recovered bitumen from reference and recycled mixtures
The binder properties of the recovered bitumen from the reference and recycled mixes are as
shown in Table 10.
Mixtures produced
1102
Table 10. Measured properties of recovered bitumen.
Property
Table 11. Determination of the penetration and softening points of the recycled
mixtures.
Bitumen properties
As it can be seen from Table 9 the incorporation of 50% RAP has hardened the bitumen
of the reference mixture by one category as specified by CEN EN 12591 (2009), regardless of
the type of virgin bitumen added.
Penetration and softening point of recycled mixtures can also be determined by two equa-
tions proposed by CEN EN 13108-1(2008) when the penetration and softening point of the
virgin and RAP binder are known. These two equations proposed are as follows:
P 2 = (a b ) log Penmixi
a log Pen1 b log Pen (1)
TR & B MIX
I aT
TR & B1 bT
TR & B 2 (2)
where Pen1 is the penetration of the binder recovered from the reclaimed asphalt mix, Pen2
is the penetration of the added binder, Penmix is the calculated penetration of the recycled
mixture, TR&B 1 is the softening point of the binder recovered from the reclaimed asphalt mix,
TR&B 2 is the softening point of the added binder and TR&B MIX is the calculated softening point
of the recycled mixture.
By implementing equation 1 and 2 for the recycled mixtures used in this study the resulting
penetration and softening point values are shown in Table 11.
As it can be seen from Table 11, the theoretical determination of the penetration and sof-
tening point values are quite close to those measured.
6 STIFFNESS RESULTS
The determination of stiffness was carried out according to CEN EN 12697-26, Annex C,
IT-CY at four temperatures: 0°C, 10°C, 20°C and 30°C. Additionally, stiffness was also meas-
ured at five different risetimes; 40 ms, 64 ms, 124 ms, 155 ms and 200 ms. Figures 2, 3, 4 and
5 show the stiffness results per risetime for each testing temperature for the reference and
recycled mixes.
As it can be seen from Figures 2 to 5, the stiffness, in all cases, decreases as risetime
increases, regardless to the testing temperature used.
In all cases recycled mixtures showed higher stiffness values than the reference mix. Addi-
tionally, mixtures’ ranking with respect to their stiffness remains unchanged regardless of the
testing temperature, at any level of risetime. The AC19-50%RAP-50/70 mixture exhibited the
higher stiffness values regardless of the testing temperature, at any level of risetime.
1103
Figure 2. Stiffness values of mixtures at 0°C.
1104
Table 12. Stiffness values of mixes at 20 °C.
Stiffness (MPa)
Risetimes
Figure 6. Fatigue lines of reference mixture and mixtures with 50% RAP.
Table 12 shows the stiffness values at 20 °C (most commonly testing temperature used) for
all levels of risetime and AC mixtures tested. As it can be seen with the addition of 50% RAP
the stiffness increased by almost 100% or more, in comparison to the stiffness of the reference
mix, regardless of risetime level chosen. Needless to say that recycled mixtures with 50/70
virgin bitumen showed higher stiffness than recycled mixtures with 70/100 virgin bitumen.
7 FATIGUE RESULTS
Although requirements for asphalt concrete fatigue testing recommend the use of two-point
or four-point bending tests (CEN EN 13108-20 2006), the indirect tensile test was used in
this study. The AC-20 cylindrical specimens were tested at three levels of stress using four
specimens at each level. A repeated haversine load was applied with 0.1 sec loading time and
0.4 sec rest time. The test was conducted at 20°C. The fatigue data were analyzed by running
a regression analysis to determine the fatigue relationship parameters in the following form:
n
⎛ 1⎞
Nf k ×⎜ ⎟ (3)
⎝ e0 ⎠
where Nf is the number of load applications, k and n are material constants and e0 is the ten-
sile strain in με at the center of the specimen.
Figure 6 shows the results of the fatigue test.
1105
As it can be seen from Figure 6, the AC-50%RAP-70/100 mixture exhibited the best
fatigue performance of the three mixtures tested. As far as the AC-50%RAP-50/70 recy-
cled mixture it seems that it exhibits slightly worse fatigue performance than the reference
mixture, when the initial tensile strain is greater than 100 microstrain. Figure 6 also shows
that the use of softer virgin binder has improved the fatigue performance of the recycled
mixtures.
8 CONCLUSIONS
Based on the results obtained in this study the following conclusions can be drawn.
1. The incorporation of 50% RAP has resulted in hardening the bitumen of the recycled
mixtures by one category compared to the bitumen category of the reference mix.
2. The values of penetration and softening point of the recovered binders from recycled
mixtures as determined by the equations proposed by CEN EN 13108-1 were found to be
very close to those measured in this study.
3. Stiffness of all mixtures decreased as risetime increased regardless of the testing tempera-
ture used.
4. Ranking of mixtures with respect to their stiffness remains unchanged regardless of the
testing temperature at any level of risetime.
5. Recycled mixtures showed higher stiffness values than the reference mix, at any level of
risetime.
6. At 20°C testing temperature the stiffness of the recycled mixtures increased by almost
100% or more in comparison to the stiffness of the reference mix.
7. The recycled mixture with 50/70 binder showed the highest stiffness value regardless of the
testing temperature, at any level of risetime.
8. The use of softer virgin binder has improved the fatigue performance of the recycled
mixtures.
9. The recycled mixture with 70/100 virgin binder showed better fatigue performance than
the reference mixture.
As a general conclusion the use of RAP increases the stiffness of the resulted (recycled)
mix which will cause reduction of required thickness or increase of design life.
However, the use of RAP may reduce fatigue performance of the resulted mix with respect
to mixture without RAP if attention is not given to the selection of proper grade of virgin
bitumen.
REFERENCES
Al Qadi I.L., Aurangzeb Q., Carpenter S.H., Pine W.J., Trepanier J. 2012. Impact of high RAP contents
on structural and performance properties of asphalt mixtures. Illinois Center for Transportation,
Report Number FHWA-ICT-12-002.
Asphalt Institute SP-2. 2001. Superpave mix design. Superpave series No. 2 (SP-2). Lexington, USA:
Asphalt Institute.
Bennert, T. and Dongre, R., 2010. Backcalculation method to determine effective asphalt binder proper-
ties of recycled asphalt pavement mixtures. Transportation Research Record: Journal of the Transpor-
tation Research Board, 2179, 75–84.
CEN EN 12591. 2009. Bitumen and bituminous binders - Specifications for paving grade bitumens. EN
Standard. Brussels: CEN.
CEN EN 12697-1. 2005. Bituminous mixtures -Test methods for hot mix asphalt -Part 1: Soluble binder
content. EN Standard. Brussels: CEN.
CEN EN 12697-3. 2005. Bituminous mixtures -Test methods for hot mix asphalt -Part 3:Bitumen
recovery:Rorary evaporator. EN Standard. Brussels: CEN.
CEN EN 12697-24 +A1. 2012. Bituminous mixtures-Test methods for hot mix asphalt-Part 24: Resist-
ance to fatigue. EN Standard. Brussels: CEN.
1106
CEN EN 12697-26. 2012. Bituminous mixtures-Test methods for hot mix asphalt-Part 26: Stiffness. EN
Standard. Brussels: CEN.
CEN EN 1426. 2007. Bitumen and bituminous binders - Determination of needle penetration. EN
Standard. Brussels: CEN.
CEN EN 1427. Bitumen and bituminous binders -Determination of the softening point -Ring and Ball
method. EN Standard. Brussels: CEN.
CEN EN 13108-1:2006/AC. 2008. Bituminous mixtures-Material specifications-Part 1: Asphalt Con-
crete. EN Standard. Brussels: CEN
CEN EN 13108-20. 2006. Bituminous mixtures –Material specifications -Part 20: Type Testing. EN
Standard. Brussels: CEN.
CEN EN 13108-21. 2006. Bituminous mixtures - Material specifications - Part 21: Factory Production
Control. EN Standard. Brussels: CEN
CEN EN 13302. 2010. Bitumen and bituminous binders- Determination of dynamic viscosity of bitu-
minous binder using a rotating spindle apparatus. EN Standard. Brussels: CEN
Copeland, A. 2011. Reclaimed asphalt pavement in asphalt mixtures: state of practice. Research, Devel-
opment and Technology Turner-Fairbank Highway Research Center, Report number FHWA-HRT-
11-021.
Daniel, J., Pochily, J., and Boisvert, D. 2010. Can more reclaimed asphalt pavement be added? Study of
extracted binder properties from plant produced mixtures with up to 25% reclaimed asphalt pave-
ment. Journal of the Association of Asphalt Paving Technologists, 2180, 19–29.
Kandhal, P.S., Rao, S.S., Watson, D.E., Young, B. 1995. Performance of recycled hot mix asphalt mix-
tures in state of Georgia. National Center for Asphalt Technology, NCAT Report 95-01.
Li, X., Marasteanu, M., Wiiliams, R., Clyne, T. 2008. Effect of RAP (proportion and type) and binder
grade on the properties of asphalt mixtures. Transport Res Rec: J Trasport Res Board 2051: 90–97.
McDaniel, R., Soleymani, H., Anderson, R., Turner, P., Peterson R. 2000. Recommended use of
reclaimed asphalt pavement in the superpave mix design method. NCHRP Web Document 30
(Project D9-12): Contractor’s final report.
McDaniel, R., Soleymani, H., Shah, A. 2002. Recommended use of reclaimed asphalt pavement in the
superpave mix design method: technician’s manual. National cooperative Highway Research Pro-
gram (NCHRP) Report 452. Transportation Research Board of the National Academies, Washing-
ton D.C.
McDaniel R., Shah A., Huber G. 2012. Investigation of low- and high- temperature properties of
plant-produced RAP mixtures. Research, Development and Technology Turner-Fairbank Highway
Research Center, Report number FHWA-HRT-11-058.
Shah, A., McDaniel R., Huber G.A., Gallivan V. 2007. Investigation of properties of plant produced
rap mixtures. Transport Res Rec: J Trasport Res Board 1998: 103–111.
Shu X., Huang B., Vukosavljevic D. 2008. Laboratory evaluation of fatigue characteristics of recycled
asphalt mixture. Constr Build Mater 22 (7):1323–1330.
Terrel, R., Joseph, P., and Fritchen, D. 1992. Five year experience of low temperature performance of
recycled hot mix. Transportation Research Record: Journal of the Transportation Research Board,
1362, 56–65.
West, R., et al. 2011. A comparison of virgin and recycled asphalt pavements using long-term pavement
performance SPS-5 data. In: Transportation research board 90th annual meeting. Washington, DC:
Transportation Research Board,Paper No. 22-3865.
West R., Timm D., Willis R., Powell B., Tran N., Watson D., Sakhaeifar M., Brown R., Robbins M.,
Vargas-Nordcbeck A., Leiva Villacorta F., Guo X., Nelson J. 2012. Phase IV NCAT pavement test
track findings. National Center for Asphalt Technology, NCAT Report 12-10.
West, R., Willis, J.R., and Marasteanu, M., 2013. Improved mix design, evaluation, and materials man-
agement practices for hot mix asphalt with high reclaimed asphalt pavement content. Auburn, AL:
National Center for Asphalt Technology, Report Number NCHRP 752, Project Number 09-46.
Widyatmoko, I. 2008. Mechanistic-empirical mixture design for hot mix asphalt pavement recycling.
Constr. Build Mater 22 (2):77–87.
Zaumanis, M., Mallick, R.B. 2015. Review of very high-content reclaimed asphalt use in plant-pro-
duced pavements: state of the art. International Journal of Pavement Engineering. Vol. 16, Issue 1,
pp. 39–55.
1107
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: In order to study the performance of warm mix recycled asphalt mixture,
and mold the specimens in accordance with the best dosage of each file of recycled asphalt
pavement, the water stability, low temperature performance and high temperature stability
under different dosages of recycled asphalt pavement were studied and analyzed through
the freeze-thaw splitting strength, strain energy density and rutting test. The results showed
that: with the increase of recycled asphalt pavement the water stability of warm mix recycled
asphalt mixture first increased and then decreased. When the content of the recycled asphalt
pavement is 40%, the freeze-thaw splitting strength ratio reached a maximum and the low
temperature performance of Warm mix recycled asphalt mixture and common warm mix
asphalt mixture were generally at the same level. With the increase of recycled asphalt pave-
ment, the high temperature stability of warm mixture asphalt was improved.
1 INTRODUCTION
At present, asphalt pavement recycling technology includes hot recycled and cold recycled.
The hot recycled asphalt mixture generally refers to the mixing and paving of asphalt and
mineral materials (RAP and some of the new aggregate) in a high temperature state (RAP
refers to recycled asphalt pavement), whereas the cold recycled asphalt mixture usually refers
to the emulsified asphalt or diluted asphalt mix with mineral materials (RAP and some of
the new material) in a normal temperature state, then paving on the road[1]. At present, the
researchers only examine the RAP with screening and filing, but how to determine the pro-
portion of each grade of the RAP that have filed is not examined. The quality control of
RAP is the key to the performance of recycled asphalt mixture[3]. According to the material
composition of the AC-13 and AC-20, of reasonable blending schemes and methods to form
specimens were determined. Water stability, high temperature stability and low temperature
properties of warm mix recycled asphalt mixture with different dosages of RAP are studied.
1109
Table 1. Physical properties of old aggregate.
Table 2. The blending method of each file of asphalt mixture with different dosages of RAP.
Table 3. The testing results of asphalt mixture with different dosages of RAP.
1110
As it can be seen from Table 3, when the content of RAP is 30%, comparison of the VMA
and VFA, the 1# option, 2# option and 3# option have little difference. For the index of
porosity, porosity of the 1# option is larger, and the difference between 3# and 2# is not
significant. For the index of flow value, the difference of the flow value of specimens about
1# and 3# options is big, and it does not meet the requirements of the standard. When the
content of RAP is 30%, the 2# option is more suitable. When the content of RAP is 40%,
the 2# option is the best with respect to the porosity and other indexes. When the content
of RAP is 50%, recycled asphalt mixture is only use 1# and 2# options, combined with the
experimental results of Marshall, the 2# option is the most appropriate.
According to the AC-13/AC-20 materials and the results of Marshall test, we chose the #
2 option to mold specimens. We studied water stability, high temperature stability and low
temperature properties of warm recycled asphalt mixture with different dosages of RAP, and
then we compared with the performance of hot mix recycled asphalt mixture.
Table 4. The freeze-thaw splitting test results of warm mix recycled asphalt mixture.
1111
asphalt mixture’s structural strength. TSR of warm recycled mixture reached the maximum
when the content of RAP is 40%. The water stability performance is decreased with the RAP
content of 50%, but it can still meet the requirements; it is recommended to add recycling
agent to recover performance of the old asphalt when the content of RAP is greater than
40%. It is shown that the water stability performance of warm mix recycled asphalt mixture
can be improved within a certain range.
ε
dw c
dv ∫0
= σ (ε )d ε
σ, ε are corresponding to the stress and strain, εc is the strain value corresponding to the
maximum stress value in the formula. The bigger the critical strain energy density is, the big-
ger the required energy is to destroy the material and the low temperature cracking resistance
performance is better. The critical value of dw/dv is the area under the actual uniaxial stress
and strain related curve when it is broken, which can be determined experimentally.
(1) The design of the experiment
It is in accordance with the oil-stone ratio of the AC-20 warm mix recycled asphalt mixture
and the target gradation to form rutting specimens. Then the rutting specimens were cut into
trabecular specimens with the size of 250 mm × 30 mm × 35 mm. On the basis of the test
procedures, the low temperature bending test of AC-20 warm mix recycled asphalt mixture
was carried out and we studied the change of low temperature bending performance for two
kinds of warm mix recycled mixtures through comparing the tests.
The failure strain of common asphalt mixture should be bigger than 2000 με accord-
ing to the standard from Table 5 it can be seen that the low temperature performance of
AC-20 warm mix recycled asphalt mixture cannot meet the standard requirements; the
strain gradually decreased with the increase of RAP, while the flexural tensile strength
is increased first and then decreased with the increase of RAP. It is illustrated that the
low temperature performance of warm mix recycled asphalt mixture and warm mix com-
mon asphalt mixture have little difference when the content of RAP was less than 50%.
Figure 1 shows the change trend of strain energy density of warm mix recycled asphalt
mixture.
From the test results it can be seen that the strain energy density of the mixture increased
first and then decreased with the increase of RAP. The bending tensile strain was declining
and ultimate flexural strength was increasing at the same time with the increase propor-
1112
Table 5. The low temperature bending test results of warm mix recycled asphalt mixture.
tion of RAP, something that led to this result. The reasons for this phenomenon are that
the warm-mix agent does not change the performance of the old and new bitumen, the
asphalt in RAP is aged seriously, and the performance improvement of recycled asphalt
only depends on adding new asphalt into the mixture—which is not enough. With the
increasing content of RAP, the influence is gradually more prominent; the new addition
of asphalt to the improved performance of the mixture is still less than the bad effect of
RAP, therefore the low temperature performance of the mixture decreased. Combining
the strength and deformation, the strain energy density of warm mix recycled asphalt mix-
ture can be described as increasing at first and then decreasing with the increase of RAP.
of RAP is meeting the requirement of more than 1000 times/mm. The dynamic stabil-
ity of recycled asphalt mixture is lower than that of hot mix recycled asphalt mixture.
According to the existing literature[7,8], the main reason for this phenomenon is that the
heating temperature of hot mix asphalt mixture is higher than that of warm mix asphalt in
the production process and the aging of asphalt is more serious compared to warm mixing
process. This leads to a slightly higher dynamic stability of hot mix asphalt mixture.
2. With the increase content of RAP the high temperature stability of warm mix recycled
asphalt mixture is improved, which is mainly due to the merge of new and aged asphalt.
The asphalt in the recycled mixture hardens and its stiffness increases, which illustrates the
high temperature performance of warm mix recycled asphalt mixture is better.
4 CONCLUSION
1. The blending methods of RAP in recycled asphalt mixture is put forward. Recycled
asphalt mixture is crushed, sieved, it is in the grading range, the blending methods of
RAP is recommended in accordance with the 2# option.
2. Combination of strength and deformation, the strain energy density of warm mix recy-
cled asphalt mixture can be described as increasing first and then decreasing with the
increase of RAP.
3. With increasing the content of RAP, the high temperature stability of warm mix recycled
asphalt mixture is improved, which is mainly due to the merge of new and aged asphalt. Then
the asphalt in the recycled mixture would harden and its stiffness increase, which illustrates
that the high temperature performance of warm mix recycled asphalt mixture is better.
REFERENCES
Qing Yongchun, Huang Songchang, Xu Jian, et al.2010, Performance of SMA mixture based on
Evotherm-DAT warm mix asphalt technology [J]. Journal of building materials. 13(01): 32–35(in
Chinese).
Huang Xiaoming, Zhao Liyong & Jiang Cheng. 2001, Test and analysis of recycled mixture for old
asphalt pavement [J]. Journal of geotechnical engineering. 23(4): 468–471 (in Chinese).
Michaelm, Martip E. & Andrew Mielke. 2002. Synthesis of Asphalt Recycling in Minnesota Final
Report [R]. Minnesota Local Road Research Board.
Geng Jiuguang, Dai Jinliang & Chen Zhongda. 2008. Whole process evaluation of low tempera-
ture anti-cracking property of hot-Recycled asphalt mixture[J]. Journal of Wuhan University of
Technology(Traffic science and Engineering)32(6):1029–1033 (in Chinese).
Cai Siwei & Cai Ming.1999. On the application of continuum damage theory to fiber reinforced con-
crete [M]. Beijing:People’ Communications Press. 38–39 (in Chinese).
Qing Yongchun & Huang Songchang. 2009. Aging of asphalt from warm mix asphalt mixture dur-
ing construction [J]. Journal of Tongji University(Edition of Natural Science). 37(9):1200–1202
(in Chinese).
Zhang Zhen, Liu Liping, Tang Wen, et al. 2009. Research on performance of Evotherm warm-mix
asphalt [J]. Journal of building materials.12(4):438–501 (in Chinese).
1114
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Ruxin Jing
Delft University of Technology, Delft, The Netherlands
Jingwei Liu
Harbin Institute of Technology, Harbin, China
ABSTRACT: In view of the demand for ecological road projects, this research focuses on
the filtration layer added in porous pavement. The material for a filtration layer is required
to meet road sewage purification and bear the load of road conditions. The addition of a
filtration layer in porous pavement could reduce the pollutants in roadway runoff, and pro-
tect ground water to be polluted. The water sample for treatment was prepared, and then the
performances of different materials of the filtration layer and multi-material combination fil-
tration layer were tested, respectively. The results showed that the concentration of pollutants
in roadway runoff was apparently reduced when a filtration layer was added to the porous
pavement, which suggests its promising effects in reducing groundwater pollution.
1 INTRODUCTION
With the widely urbanizing development in China, the area of pavement impervious to water
is also increasing rapidly. The problems related with pavement impervious to water appear
frequently, so more and more countries begin to conduct many research works to resolve
these problems, by which porous pavement is thought as a promising option. However, the
application of this technology encounters the problem that how to control pollution of road-
way runoff.
Roadway runoff coming from rainfall, especially initial rainfall, contains many organic
and inorganic pollutants, leading to the COD, BOD5, SS and concentration of petroleum far
exceeding standards for discharge of pollutants[1] [2]. The urban waterways and groundwater
become polluted if the roadway runoff is discharged directly into water. The addition of a
filtration layer into the porous pavement could efficiently reduce the concentration of pol-
lutants by entrapment and adsorption processes. This research highlights that the addition
of a filtration layer which is plays playing a positive role in controlling pollution for roadway
runoff and has a promising potential in ecological road projects [3].
2 EXPERIMENT MATERIAL
1115
Table 1. The pollutant indexes of the water sample prepared.
tap water. The amount of additives were determined according to observed values and lit-
erature results [5]. The pollutant indexes of the raw water prepared manually were stable and
easy to control. The pollutant indexes of the raw water prepared are listed as Table 1.
After the raw water prepared, the raw water was filtered by the surface layer and basic layer
of standard specimens. The pollutant indexes after filtrating treatment are shown in Table 2.
The raw water was preserved at a low temperature for experiment use.
The raw water prepared was preserved at a low temperature for experiment use.
3 EXPERIMENT METHODS
In this study, the pollution indexes of the roadway runoff are concentrations of COD, SS, Pb
and Zn. The determination of COD was adopted the dichromate method set by GB11914–89,
while the determination of SS was used the gravimetric method. The concentration of Pb
and Zn were measured by ICP-AES (Inductive Coupled Plasma Emission Spectrometer).
The experimental setup was developed to simulate the purification process of porous pave-
ment with a filtration layer. The experimental setup contains several parts: the water box, to
provide stable water flow; the support, to hold the pipes and the water box; the sprinkler, to
attain water samples which are filtered by different filtration layers; the permeable layer is
made to simulate the pavement; and there also have flow meter to measure the velocity of the
water flow, and the valve to control it. And the experimental setup is shown in Figure 1.
Average room
Index Flow (ml/s) Filtration velocity (m/h) temperature (°C) Initial PH
(The material was washed for several times and then dried before being used in this
study).
through the filtration layer. Water samples were taken after flowing out of the filter by three
minutes. The water sample was then added into the sample bottle, subsequently COD, SS and
concentration of Pb and Zn were examined.
2. With regards to the removal of COD, fine sand and volcanic rock has low dependence
on the thickness of the filtration layer, indicating the removal rate of COD not increasing
with the increase of the thickness of the filtration layer. The above data shows a good
effect can be achieved by using a thinner filtration layer. On the contrary, the removal rate
of zeolite, activated carbon, and cinder is highly dependent on the thickness of the filtra-
tion layer, while a thinner filtration layer of these three materials couldn’t realize a good
effect, which could only be achieved when the filtration layer is thick enough.
As shown in Figure 3, the following conclusions are obtained:
1. Fine sand has a high SS removal rate, and the removal rate exceeded 90% when the thick-
ness of the filtration layer is only 10 cm. Activated carbon, cinder and volcanic rock had
moderate removal rate of SS. Zeolite has the worst performance; the removal rate of cer-
amsite is worse than fine sand, but better than the other four materials.
2. For fine sand and ceramsite, the removal of SS has high dependence on the thickness of
the filtration layer, and a good effect can be expected using thinner filtration layers of
these two materials.
1118
Figure 4. The removal rate of Pb in single-material experiments.
1119
As shown in Figure 5, there are two conclusions putting forwards as follows.
1. The removal rate of Pb is in agreement with that of Zn for the selected materials except for
the activated carbon.
2. Activated carbon presents little capacity of removing Zn. According to existing research,
the absorption of Zn by the activated carbon was significantly affected by PH. At PH < 6,
Zn is hardly absorbed by activated carbon, while at PH > 7, absorption of Zn by activated
carbon is increased quickly with the rise of PH [7].
As shown in Figure 6, we can draw following statements:
1. Zeolite performs worst in several kinds of materials, so it is not suitable for a porous pave-
ment filter layer.
2. Some heavy metals such as Zn, are hardly absorbed by activated carbon at low PH. Owing
to the aggravation of pollution in recent years, the roadway runoff is generally weakly
acidic or even strongly acidic, so activated carbon can’t satisfy the general requirements.
3. Volcanic had bad decontamination effect on some heavy metals like Pb and Zn. Thus, it is
not suitable for the filter layer material.
1120
Table 4. Orthogonal design table.
1 10 0 10 Cinder above
2 20 0 20 Cinder above
3 0 10 10 Cinder above
4 20 0 20 Ceramsite above
5 0 20 20 Cinder above
6 10 20 0 Ceramsite above
7 20 10 0 Fine sand above
8 20 10 0 Ceramsite above
9 0 20 20 Fine sand above
10 10 10 20 Cinder above, ceramsite beneath
11 20 20 10 Cinder above, ceramsite beneath
Figure 7. The removal rate of COD in multiple- Figure 8. The removal rate of SS in multiple-
material experiments. material experiments.
We can draw the following conclusions from the above four figures:
1. If the thickness of the filtration layer was uniform, the combination of different materi-
als had a better effect in removing pollutants than one material. That maybe because the
combination of materials can fully utilize the advantages of different materials.
2. The combination of different materials can improve the structure of the filtration layer to
a large extent. Different sub-layers have different porosity and absorption properties, so
the whole layer would not be blocked in a short time.
In conclusion, a multiple-material filtration layer is more persistent and effective, and is
recommended for the adoption of porous pavement.
5 CONCLUSION
1. Nowadays, more and more problems are caused by impervious pavement. So it is neces-
sary to do some job in providing a practicable way to solve the pollution, and from the
research we can draw a conclusion that we can use some material to make up filtration
layers.
2. And from the results, we comparing the selected six materials, find that three of them
proved fit to use in the filtration layer of porous pavement, besides, maybe more kinds
of material can be used and have better property. And we will do further research about
this.
3. It is clear that the combination of materials can improve the capacity of removing pollut-
ants and prolong the period of validity, but we only do little work on this, for the mixture
1121
Figure 9. The removal rate of Pb in multiple- Figure 10. The removal rate of Zn in multiple-
material experiments. material experiments.
of material can be complex, and the time is tight, but we think that it will be the trend of
filtration layer structures of porous pavement and more research is needed.
4. The combination of different materials can improve the structure of the filtration layer to
a large extent. Different sub-layers have different porosity and absorption properties, so
the whole layer will not be blocked in a short time. A multiple-material filtration layer is
more persistent and effective.
5. For further application, we concluded that in different regions, at different temperature
and moisture, the function will quite different from other situation, so to different regions,
we should do the same research, but it seems impossible. So we think orthogonal design
experiment is a good way, through this method, we can consider different influence factors
together and even draw a graph about all these factors.
REFERENCES
[1] Jianqiang ZHAO, Min YAN and Shan LIU. Investigation of the pollution in urban runoff.[J]
China’s water supply and drainage, 2001, 17(1), 33–35.
[2] GB8978-96, “Synthetical Draining Standard of Sewage”[S]. The People’s Republic of China.
[3] Lizhu HOU, Shaoyuan FENG and Yueyuan DING. Environmental effect and prospect of porous
pavement in urban areas.[C] Chinese Society of Agricultural Engineering, Beijing, China, 2005,
234–240.
[4] Ling JIANG and Yaomin QIN. The studies of using soil layer to purify the urban rainwater for
artificially supplying groundwater.[J] Journal of Soil and Water Conservation, 2005, 19(1), 94–97.
[5] Jianqiang ZHAO, Shan LIU, Yingling LIU and Youping LI. Analyzing the quality characteristics
of urban road surface runoff rainwater.[D] Xi’an Highway University, 1999, 19, 31–33.
[6] Hongshan WANG, Xiaoguang XIE, Decheng FENG. Decontamination effect evaluation of
permeable pavement filter layer materials.[J] Engineering Mechanics, 2012, S2:190–194.
[7] Taihong SHI and Bing TANG. Active carbon absorption water Zn2+ and the characteristics of
complexes.[J] Industrial Water Treatment, 2000, 10, 12–15.
1122
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Rui Li, Jiexue Cui, Bin Xue, Dongliang Kuang & Jianzhong Pei
Chang’an University, Xi’an, Shaanxi, China
ABSTRACT: In order to study the TiO2 photocatalysis material to purify the automobile
exhaust in tunnel, soluble gel method was used to prepare nano-TiO2 photocatalyst, with
different dosage of Fe3+ to modify nano-TiO2, and the sample was made after being calcined
under 500ºC condition. Doping Fe3+ can capture the electrons and holes so that it can help to
reduce the recombination probability of electron hole pairs and improve the photocatalytic
performance of nano-TiO2. Compared with the purification efficiency of nano-TiO2 without
Fe3+, the purification efficiency of nano-TiO2 with 0.1% Fe3+ to CO, CO2, HC and NOx can
be improved by 0.6%, 0.5%, 2.3% and 8.2% respectively under the weak UV light condition.
1 INTRODUCTION
In highway tunnel, there are a large number of automobile exhausts which contains a lot of
carbon monoxide (CO), nitrogen oxides (NOx), hydrocarbons (HC), as well as some solid
phase. These harmful exhaust gases can not be easily discharged in the tunnel, which cause
the growing contamination and affects traffic safety and human health[1]. Semicoductor pho-
tocatalyst which achieved good catalyzation of organic pollutants was found in the 1970s[2].
At present, photocatalytic degradation of pollutants using the nano-TiO2 has become one of
the active research directions. The photocatalytic activity of nano-TiO2 is relatively high, but
the visible light utilization rate is still low compared with other semiconductor photocatalyst.
The utilization rate of light should be further improved to promote the practical performance
of nano-TiO2 light catalyst. Nano-TiO2 doped with a small amount of metal ions can make it
become a photoproduction electronic-hole for shallow potential capture trap, which prolongs
recombination of electrons and holes and improves the photocatalytic performance of nano-
TiO2[3].Some doping can also reduce the forbidden band width of nano-TiO2, enlarge its
range of light absorption and improve the utilization of sunlight[4]. There are research groups
reported that the photocatalytic performance of nano-TiO2 doped with Fe3+ can improve
the difficulty in degradable of organic pollutants in water and nitrogen oxide (NOx) in the
atmosphere, however, the research on automobile exhaust purification is seldomly found[5,6],
especially in the automobile exhaust purification in the tunnel.
Nano-TiO2 doped with Fe3+ is prepared using the sol-gel method, four major automobile
exhausts that compose of HC, CO, CO2 and NOx are used for testing the performance of
purification. This article discussed the photocatalytic activity of nano-TiO2 doped with Fe3+
and analyzed the mechanism of Fe3+ in improvement of the photocatalytic properties based
on data of infrared spectroscopy (IR) and X-ray diffraction (XRD) analysis that used as a
measurement tools in this work.
1123
2 EXPERIMENTAL
M + e → M ( n− ) + (1)
M + h → M ( n+ ) + (2)
(n−1)+ n+
Where n is oxidation number, and only the band energy level of M /M is lower than
that of nano-TiO2’s conduction band energy level, the energy level of M(n+1)+/Mn + is higher
than that of nano-TiO2’s, the reaction may occur.
The surface treatment of nano-TiO2 evenly applied into an organic glass and then put it in
photocatalytic reaction device of homemade exhaust gas. Ultraviolet light (287 nm) is used
as the light environment, and the actual vehicle is used to generate exhaust. The nano-TiO2
photocatalytic activity can be observed by the measured concentration of four different gases
before and after 1 h. XRD can analyze catalyst crystal and estimate the grain size and IR can
analyze surface properties of nano-TiO2.
3 MATERIAL ANALYSIS
1124
Fe3 ) and 13.6 nm (0.1%wt Fe3 ) respectively. We can not find the characteristic peaks of
Fe2O3, Fe3O4, FeO and other iron oxide crystals in XRD. It can be speculated that the crystals
are highly dispersed on the surface of nano-TiO2 or in the crystal lattice. Due to the presence
of iron valency diversity, nano-TiO2 doped with Fe3+ can become both the electronic capture
and the capture of the hole, it can capture the electrons and holes at the same time, which
reduces the electron hole on the composite risk and improves the photocatalysis activity of
nano-TiO2.
1125
4 RESULTS AND DISCUSSION
4.1 Nano-TiO2 mixed with Fe3+ raise the efficiency of the purification
The degradation function is selected as a suitable index to evaluate the maximum gas deg-
radation rate. Computation formula is as follows:
=( 0 − t ) / C0 (3)
Table 1. The gas purification efficiency of nano-TiO2 with different content of Fe3+.
CO CO2 HC NOX
1126
Figure 4. The concentration variation of: a) CO, b) CO2, c) HC, d) NOX.
5 CONCLUSIONS
The nano-TiO2 with Fe3+ and without Fe3+ were successfully prepared, we can see from the
results that doping Fe3+ can improve the photocatalytic performance of nano-TiO2.Purifica-
tion efficiency of nano-TiO2 with 0.1% Fe3+ to CO, CO2, HC and NOx can be improved by
analyzing the exhaust degradation effect of nano-TiO2 under the weak UV condition. It
attributed this to Fe3+ can capture electrons and holes simultaneously so that it can reduce the
recombination probability of electron hole pairs, and helping to improve the photocatalytic
activity of nano-TiO2. It can be drawn from the IR spectrum that the function of doping Fe3+
is not very effective to change easy-hydroxylated properties of nano-TiO2.
ACKNOWLEDGEMENTS
This work were supported by the National Natural Science Foundation of China (grant no.
51378073 and 51408048), the Key Program for International Science and Technology Coop-
eration Projects of Shaanxi province (grant nos. 2012KW-09 and 2014KW10-03), the China
1127
Postdoctoral Science Foundation(grant no. 2015M570805), the Fundamental Research
Funds for the Central Universities of Chang’an university(grant nos. 310821153502 and
310821152005) and the Research Fund for Doctoral Program of Higher Education of
China (grant no. 20120205120010).
REFERENCES
1. X. Yan, W. Guo, K. Song, M. Huo and J. Wang: ‘Preparation of Nano-TiO2 Film on TiO2 And Pho-
tocatalytic Activity’, Adv. Energy Mater., 2011, 22,196–199.
2. Y. Sakata, T. Yamamoto and T. Okazaki: ‘Visible light response of titania photocatalyst containing
copper ion’, J.Chem.Lett., 1998, 324, l253–1254.
3. X. Yu and H. Wei: ‘Layer-by-layer assembly of TiO2 colloids onto diatomite to build hierarchical
porous materials’, Adv.Coll.Inter., 2008, 323, 326–331.
4. H.B. Yuan, J.H. Zhang, R.J. Yu and J. Su: ‘Synthesis of rare earth sulfides and their UV vis absorp-
tion spectra’, J. Rare Earths, 2009, 27, 308–311.
5. S. Montieone, R. Thfeu, A.V. Kanaev, E. Scolan and C. Sanchez: ‘Quantum size effect in TiO2 nano-
particles: does it exist’, J.Appl.Sur.sci., 2000, 162, 565–570.
6. A. Fujishima and K. Honda: ‘Electrochemical photocatalysis of water at a semiconductor electrode’,
Nature, 1972, 238:37–41.
7. J. Museat, V. Swamy, N.M. Harrison: ‘First-Principles calculations of the Phase stability of O2’, Phys.
Rev.B, 2002, 65, 224112–224127.
8. Jianzhong Pei, Weisi Ma, Rui Li, Yanwei Li and Hongzhao Du: ‘Preparation and Photocatalytic
Properties of TiO2-Al2O3 Composite Loaded Catalysts’, Journal of Chemistry, 2015, 806568.
1128
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: This study aimed to develop tire-pavement contact models and predict noise
with different tire configurations and surface characteristics using a coupled Finite Element
Modeling (FEM) and Boundary Element Modeling (BEM) approach. Radial truck tire mod-
els were developed and validated for the traditional dual-tire assembly and the new wide-base
tire. The sound pressure spectrum and the overall noise level were compared considering the
effects of pavement surface texture, noise measurement methods, and loading conditions. The
results show that the wide-base tire is quieter than the dual tire assembly at 4–6 dB(A). On
the other hand, the higher surface texture generates the greater noise level under the same
testing condition. The overall noise level decreases as tire inflation pressure increases or tire
load decreases at typical operation conditions. The findings indicate that tire vibration noise is
highly related to tire-pavement contact behavior including contact stresses and contact areas.
1 INTRODUCTION
In recent years, the trucking industry has developed innovations of tire technology for pur-
poses of improving the efficiency of vehicle operations and fuel consumption while minimiz-
ing tire wear. The interest of using single wide-base tires as an alternative to conventional
dual-tire assemblies has gradually increased recently. Historically, dual-tire assemblies have
been used to provide the largest footprint to adequately distribute the axle load onto the
pavement surface. However, compared to conventional dual-tire assembly, wide-base tires
have the potential to offer several benefits, including improved fuel efficiency, increased haul-
ing capacity, reduced wheel cost, and superior ride and comfort (Al-Qadi and Elseifi 2007).
The effects of wide-base tires on road infrastructure is receiving considerable attention
and eliciting widespread interest in pavement researchers, particularly since the introduction
of new generations of wide-base tires (wide-base 445 and 455) (Al-Qadi et al. 2002, Greene
et al. 2010; Wang and Al-Qadi 2011). These studies concluded that the new wide-base 445 or
455 tire could cause greater or lesser pavement damage potential than the dual-tire assembly,
depending on pavement structure and failure mechanism. However, researches on the impact
of wide-base tires on environment, especially tire-pavement noise is quite limited. Noise,
defined as unwanted or excessive sound, has become one of the greatest sources of nuisance
affecting environmental and social sustainability. While noise emanates from different sources,
traffic noise is perhaps the most pervasive and difficult source to avoid (Mark 2006).
With the development of automobile technology, the exhaust and engine noise are greatly
controlled. Instead, the tire-pavement interaction noise becomes dominant. As a major compo-
nent of the vibrating source of the tire-pavement noise, tires play a significant role in active noise
control. Although the effect of tire configurations on tire-road noise can be studied using field
testing or large-scale laboratory tests, the testing requires considerable manpower, materials, and
financial resources (Hanson et al. 2004). The testing results may be susceptible to the interference
from a variety of factors, such as environmental factors, traffic conditions, etc. Hence, it is desired
to develop robust numerical simulation models for evaluating the impact of tire configurations on
1129
tire-pavement noise. This will benefit the development of environment-friendly tires as well as the
design of low-noise pavement surface.
This study aimed to develop tire-pavement contact models and predict tire-pavement noise
with different tire configurations and surface characteristics using a coupled Finite Element
Modeling (FEM) and boundary element modeling (BEM) approach. Tire models were devel-
oped for the traditional dual-tire assembly (275/80R22.5) and the new generation of wide-
base tire (445/50R22). The tire surface accelerations under excitation of pavement surface
texture were obtained from modal analysis of FEM. The solutions for the radiation sound
fields caused by tire vibration were solved using the BEM analysis. The tire-pavement model
was validated with the measured tire deflections, contact stresses, and overall noise level. The
developed tire-pavement interaction model was used to predict the sound pressure spectrum
and the overall noise caused by different tire configurations considering the effects of pave-
ment surface texture, noise measurement methods, and tire loading conditions.
1130
Figure 1. Tire finite element meshes (a) 2-D 275 tire mesh; (b) 3-D 275 tire mesh; (c) 2-D 445 tire mesh;
and (d) 3-D 445 tire mesh.
(u [M ] + u [C ] + [K ]){ } = 0
2
(1)
( −ω [ ] + [ ]){ } = 0
2
(2)
Where, [M ] is the mass matrix; [C ] is the damping matrix; [ K ] is the stiffness matrix; u is
the eigenvalue; and {φ } is the eigenvector, the mode of vibration.
Generally, tire-pavement noise can be described as two mechanisms: the mechanical vibra-
tions of the tire at the lower frequencies, and the aerodynamic phenomenon at the higher
frequencies (Sandberg and Easement 2002). The tire vibration noise was focused in this
study because it is mostly related to pavement engineering application and human hearing
sensitivity.
Steady state linear dynamic analysis predicts the linear response of a structure subjected
to continuous harmonic excitation. The set of eigenmodes extracted in the previous eigenfre-
quency step were used to calculate the steady state solution as a function of frequency of the
applied excitation. The structural dynamics equation for steady-state response under a har-
monic excitation is shown in Equation 3. The acceleration response obtained in the FE analy-
sis was used as a boundary condition for a BEM analysis of the transmitted noise field.
2
([ K ] [[C
C] [ M ]{U } {F } (3)
Where, {U} is the nodal displacement vector of structure; and [K] is the load vector of
external excitation.
1131
trum represents the texture level in A-weighted decibel relative to the reference texture amplitude
at different wavelengths (spatial frequencies). The texture level was calculated using Equation 4.
Where, Lx is texture level in dB(A); ax is root mean square magnitude of surface profile
variation in the frequency band; arefe is reference texture magnitude (10−6 m); and x is center
frequency of one-third-octave frequency band.
The most widely used approach for random vibration analysis is using Power Spectral
Density (PSD) functions to represent loading and response (Lu and Greenberg 2000; Lu
2003). In this study, to take into consideration of pavement texture excitation, PSD func-
tions of texture profiles after Fourier transformation were used in tire-pavement interaction
model. After the step of modal frequency analysis, the response of the system is used for
random response analysis under excitation of surface texture spectrum in the contact patch.
Figure 2 shows the representative texture spectra of EA surface and HFS. It clearly shows
that the EA surface exhibits the greater texture level compared to that of HFS.
(∇ + )p=0 (5)
∂p ∂n = − i ρw
wvn (6)
Where, n is the normal vector pointing outside from the acoustic volume; i = −1 is the
imaginary unit; p is the acoustic fluid density and vn is the normal velocity.
From the theory of Green’s functions, Equation 7 can be recast into the Helmholtz integral
for an exterior boudnary value problem (Wu 2000).
⎛ ∂pp ∂G ⎞
p(P) = ∫ ⎜G −p ⎟d Γ (7)
Γ
⎝ ∂ n ∂n ⎠
1132
Figure 3. Boundary element model of tire-pavement system for (a) dual-tire assembly and (b) wide-
base tire with field points in the CPX method.
Where, G is the singular fundamental solution; and r is the distance between the field
point.
The acceleration responses obtained in the FE analysis were used as boundary conditions
for analysis using the BEM. For a given acceleration field on the tire surface, the BEM cal-
culates the values of pressure and normal acceleration at all the boundary nodes and field
points. The outer surface elements of the tire were selected to constitute the boundary ele-
ment. In order to avoid sound leakage, additional elements were added to the rim. For the
dual-tire assembly, a symmetric reflection panel was built to simulate the noise effect from
the second tire. The distance between the symmetric panel and tire was 310 mm based on the
spacing between dual tires.
In this study, the locations of field points to calculate sound pressure were determined fol-
lowing the measurement locations used in the Close-Proximity (CPX) method (ISO11819-2
2000) and the On-Board Sound Intensity (OBSI) method (AASHTO TP76 2010). As a near-
field technique, the OBSI method measures tire-pavement noise using a phase-matched pair
of microphones. However, the CPX method measures noise levels using sound pressure from
microphones at five locations. The boundary element model of tire-pavement system with
field points in the CPX method is illustrated in Figure 3. The overall sound pressure and
noise level is reported as the average of sound pressure and noise from all the field points,
respectively, for the CPX and OBSI method.
4 MODEL VALIDATION
1133
Figure 4. Comparisons between measured and predicted deflections for (a) 275 tire and (b) wide-base
445 tire.
Figure 5. Comparisons between measured and predicted vertical contact stresses for (a) 275 tire and
(b) wide-base 445 tire.
the wide-base tire. These trends were consistent with the data reported in the previous studies
(Wang and Al-Qadi 2011; Hernandez et al. 2014). The good agreements in the measured and
calculated tire deflections and tire-pavement contact stresses indicate that the proposed tire
model is adequate for tire-pavement interaction analysis.
1134
ing effect, other than the excitation of tire rubber deformation due to surface texture. The
computed sound pressure levels at the one-third octave frequency bands are calculated using
Equation 8, as shown in Figure 6. The data show that the sound pressure level of the wide-
base tire is lower than that of the dual-tire assembly across the frequency range.
The overall noise level can then be calculated from the sound pressure levels at different
frequency bands using Equation 9. The overall noise levels for two tire configurations were
compared in Table 1. Two common noise testing methods, CPX and OBSI, were considered
in the analysis for comparison.
Where, Lpi is sound pressure level in dB(A); pi is acoustic pressure; pref is reference sound
pressure (2 × 10−5 pa); Lp is overall noise level; and i is center frequency of one-third-octave
frequency band.
The results show that the wide-base tire is quieter than the dual tires. Differences of
4–6 dB(A) were observed between two tire configurations on the high-texture surface. The
effect of tire configuration on the overall noise level is affected by two main factors, the
number of tires and tire-pavement contact behavior. The tire-pavement contact behavior
includes the distributions of contact stresses at the tire-pavement interface and the area of
contact patch. One possible reason that wide-base tire generates lower noise levels is that
wide-base tire has more uniform distribution of contact stresses with the smaller peak stress
value (as shown in Figure 2), but has the similar contact area with the dual-tire assembly. The
greater tire-pavement contact stresses may cause the greater tire vibration and sound pressure
level, which determines the noise level.
The sound pressure levels at the field measurement points used in the CPX and OBSI
methods were compared in Figure 6. As expected, the OBSI method results in higher noise
levels than the CPX method because the microphone is closer to the tire in the OBSI method.
As shown in Table 1, the overall noise levels measured using the OBSI method were about
3 dB(A) and 1.4 dB(A) higher than the noise level measured using the CPX method, respec-
tively, for dual-tire assembly and wide-base tire. This trend is consistent with findings from
the field testing using different noise testing methods (Oddershede 2013). It is noted that the
noise of dual-tire assembly might be more sensitive to the testing method (such as the loca-
tions of microphones around the tire) as compared to the noise of wide-base tire.
Figure 6. Sound pressure levels for dual-tire assembly and wide-base tire (MPD = 0.6 mm).
1135
Table 1 Comparison of noise level for different tire configurations
and testing methods (MPD = 0.6 mm).
1136
Table 2. Comparison of noise level for different tire configurations
and texture levels.
Figure 9 shows the influence of tire loads on overall noise levels. It was found that when the
inflation pressure was constant, the increase of load causes noise increase for both tire con-
figurations. Similarly, the reason might be that the greater load increases both contact area
and tire-pavement contact stresses as observed from the tire-pavement interaction model,
which could cause the increase of tire noise. In general, the results indicate that the variation
of tire noise at different levels of tire load and inflation pressure depends on the combined
effect of tire contact stress and contact area.
6 CONCLUSIONS
This study developed tire-pavement contact models to predict tire-pavement noise caused
by different tire configurations and pavement surface characteristics using a coupled FEM-
BEM approach. The tire—pavement interaction model was validated with measured tire
deflections, tire-pavement contact stresses, and overall noise level. The following conclusions
can be concluded from the analysis:
1. The results show that the wide-base tire is quieter than the dual-tire assembly due to dif-
ferent contact stress distribution patterns in the similar contact area. Differences in the
overall noise level caused by two tire configurations were found to be 4 to 6 dB(A).
1137
2. Considering the effects of testing methods on noise, the overall noise levels measured using
the OBSI method are about 3 dB(A) and 1.41 dB(A) greater than the ones measured using
the CPX method, respectively, for dual-tire assembly and wide-base tire.
3. The pavement surface type with the greater surface texture generates the greater noise level
under the same testing condition. For the two types of pavement surfaces with MPDs of
3.3 mm and 0.6 mm, the difference in tire noise can reach 4 dB(A).
4. It was found that tire inflation pressure and load influence tire-pavement interaction noise
in different ways depending on tire-pavement contact behavior. The overall noise level
decreases as tire inflation pressure increases or tire load decreases at typical tire operation
conditions.
The findings indicate that tire vibration noise is highly related to the tire-pavement contact
behavior including contact stresses and contact area. Although tire vibration noise at the
lower frequency band is mostly related to pavement engineering application and human hear-
ing sensitivity, further studies should be conducted to consider other noise mechanisms, such
as air pumping and horn effect, in the tire-pavement interaction modeling.
REFERENCES
Al-Qadi, I.L., A. Loulizi, I. Janajreh and T.E. Freeman. 2002. Pavement Response to Dual Tires and
New Wide-Base Tires at Same Tire Pressure,” Transportation Research Record, No.1806, pp. 38–47.
Al-Qadi, I.L., and M.A. Elseifi. 2007. New Generation of Wide-Base Tires: Impact on Trucking Opera-
tions, Environment, and Pavements, Transportation Research Record, No. 2008, pp. 100–109.
AASHTO TP 76-10.2010. Measurement of Tire/Pavement Noise Using the On-Board Sound Intensity
(OBSI) Method, American Association of State and Highway Transportation Officials.
Brebbia, C.A., J. Telles, J., and L. Wroble. 1984. Boundary Element Techniques: Theory and Applications
in Engineering, Springer-Verlag, Springer Berlin Heidelberg.
Ghoreishy, M.H.R., Malekzadeh, M., and H. Rahimi, 2007. A Parametric Study on the Steady State
Rolling Behavior of a Steel-Belted Radial Tire, Iranian Polymer Journal, No.16, pp. 539–548.
Greene, J., U. Toros, S. Kim, T. Byron., and B. Choubane. 2010. Impact of Wide-Base Tires on Pave-
ment Damage, Transportation Research Record, No. 2115, pp. 82–90.
Hernandez, J.A., Gamez, A. and I.L. Al-Qadi. 2014. Introducing an Analytical Approach for Predicting
3D Tire-Pavement Contact Load, Transportation Research Record, No. 2201, pp. 75–84.
ISO/CD 11819-2. 2000. Acoustics –Measurement of the influence of road surfaces on traffic noise—
Part 2: The close-proximity Method, SO, Geneva, Switzerland.
Knothe, K. 2001. Advanced Contact Mechanics-Road and Rail, Vehicle System Dynamics, Vol. 35, pp.
361–407.
Kelvin C.P. Wang. 2011. Elements of Automated Survey of Pavements and a 3D Methodology, Journal
of Modern Transportation, Vol.19, pp. 51–57.
Lu, S, Greenberg, B.S. 2000. Dynamic response of linear system to moving stochastic sources, Journal
of Sound and Vibration, Vol.229, pp. 957–972.
Lu, S. 2003. Simulation of pavement roughness and IRI based on power spectral density, Mathematics
and Computers in Simulation, Vol. 61, pp. 77–88.
Mark, F. 2006. Highway Traffic Noise in the United States-Problem and Response, Publication FHWA-
HEP-06-020. FHWA, U.S. Department of Transportation.
Oddershede, J. 2013. CPX – OBSI Relation in Tire/Road Noise Measurement Results, Report 463, Dan-
ish Road Directorate.
Padovan, J. 1997. On Standing Waves in” Tires”, Tire Science and Technology, Vol. 5, pp. 83–101.
Sandberg U., J.A. Easement. 2002. Tire/Road Noise Reference Book. Informex, Kisa, 2002.
Schwanen, W., H.M. Leeuwen, A.A.A. Peeters, G.J. Blokland, H.F. Reinink, and W. Kropp. 2007. Acous-
tic Optimization Tool RE3: Measurement Data Kloosterzande Test Track, Delft, The Netherlands.
Wong, J.Y. 1993. Theory of Ground Vehicles, John Wiley & Sons, Inc., New York.
Wu, T.W. 2000. Boundary element acoustics: Fundamentals and computer codes. WIT Press, United Kingdom.
Wang, H. and I.L. Al-Qadi. 2011. Impact Quantification of Wide-base Tire Loading on Secondary Road
Flexible Pavements, Journal of Transportation Engineering, Vol. 137. No.9, ASCE, pp. 630–639.
Wang, H., I.L. Al-Qadi, and I. Stanciulescu. 2012. Simulation of Tire-Pavement Interaction for Predict-
ing Contact Stresses at Static and Various Rolling Conditions, International Journal of Pavement
Engineering, Vol. 12, pp. 310–321.
1138
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: In Guangzhou, a part of an expressway has passed through some water sen-
sitive areas. In order to protect the environment of these areas, a runoff treatment system
was designed. Based on this system, an experiment was designed to test the effect the system
has in treating the runoff pollution. By doing a model test in laboratory to know the effect
when using different materials and different combinations the system has on runoff pollut-
ants removal, so to optimize the treatment system. The results show that a combination of
gravel and activated carbon has a good capability to remove the runoff pollutants. Especially
to the suspended solids, more than 68% could be removed. The removal of the heavy metals
(Pb, Zn, Cu) ranges from 6.67% to 51.46%, and the COD is between 33% and 75% which
both have a large range of variation. As the concentration of BOD5 is just between 2 mg/L
and 6 mg/L, a little vary could lead to a great change of the removal. But the removal of oil
is between 45% and 67%, which is relatively stable.
1 INTRODUCTION
A road which is under construction or operating could create lots of pollutant like solids, nutri-
ents, heavy metals, oil, lipid, bacteria and so on (Stotz 1987). Some of them could be dissolved
or taken away by road runoff, and bring the pollution problem when the water flows into rivers,
lakes, or groundwater. Now there are some measures on engineering that used in treating the
road runoff pollution. Such as vegetation control (Deletic et al. 2006), retention ponds (Peng
et al. 2014), oxidation ponds (Tang et al. 2014), artificial wetlands (Shutes et al. 1999; ZHAO
et al. 2013), infiltration system (Bai et al. 2011) and united control measures (He et al. 2010).
In Guangzhou, an integrated road runoff treatment system is used in a section of express-
way that runs through water source protection areas. The system is mainly consisting of a
treating pool and a biochemical pool. The whole treating process is: physical filtration pool→
physical adsorption pool→ chemical treating pool→ biochemical pool or natural water sys-
tems, as shown in Figure 1. In order to study the treating pool’s effect on runoff when using
1139
different combinations of materials. This paper tells a simulation test on the treating pool,
which chooses several different kinds of materials and it makes combinations to study the
removal performance on pollutants in road runoff.
2 MATERIALS SELECTION
3 TEST METHODS
1140
activated carbon. Then we take a certain amount of sediments that was collected from left
edge of the road surface, dispersed and dissolved into water. In this case, the concentration of
each pollutant is different in each experiment when using different kind of activated carbon.
We divide the mixture into six evenly during the stirring (so it could avoid solids settling in
the water). We take one of them as the initial sample (detect the initial concentration), the rest
five samples for experiment. After stirring and pouring a sample into the physical filtration
pool slowly we collect the water after treated, as shown in Figure 3.
4.1 Treatment effect of each size gravel combined with coal columnar activated carbon
After the water quality test on collected samples, the test results of each indicator are shown
in Table 1.
According to the test results, the initial pH value of the sewage is 7.3 which is slightly alka-
line water. But the value of the water treated has a small decrease that reaches 7.2∼7.0, which
tends to be more neutral, so the material combinations play a regulatory role on water’s pH.
The test results of SS, BOD5, COD, heavy metals (Zn, Pb, Cu) and petroleum are represented
as shown in Figure 4 in the form of changes in the concentration curve and the removal rate.
From the figures several points can be concluded: a combination of gravel and coal columnar
activated carbon can brings about 70% removal on suspended solids in road runoff, and the
smaller the size of gravel chosen, the higher the removal; the removal rate on BOD5 reaches
25% to 50%; the removal rate on COD is less than 50%; the removal on heavy metals in sew-
age is not high, ranging from 6.67% to 44%, where the removal on Pb is slightly better than
Table 1. Sewage treatment by different size of gravel combined with coal columnar activated carbon.
1141
Figure 4. Changes in the concentration of pollutant index.
Cu and Zn; the removal on petroleum reaches 30% to 60% which gradually increases with the
decrease of the gravels’ size choose.
4.2 Treatment effect of same size gravel combined with each kind of active carbon
We used size of 0.3 mm gravel, which has the best removal when combined with coal colum-
nar activated carbon, combined with coconut shell granular activated carbon and analytically
pure granular activated carbon separately to do the test. Table 2 shows the results.
The removal of each indictor under the three combinations is presented by histogram, as
shown in Figure 5.
It can be seen that the removal on pollutants in the runoff is good when using the 0.3 mm
size gravel combined with the three types of activated carbon, especially the removal on sus-
pended solids. In general, the 0.3 mm size gravel combined with analytically pure granular
1142
Table 2. Treatment effect on runoff by 0.3 mm particle size of gravel combined with different types of
activated carbon.
activated carbon has the best removal efficiency, and the 0.3 mm size gravel combined with
coconut shell granular activated carbon is better than combined with coal columnar activated
carbon.
5 CONCLUSION
REFERENCES
Bai, Y. & Zuo, J. et al. 2011. Urban Non-point Source Pollution Control by Runoff Retention and Fil-
tration Pilot System. Chinese Journal of Environmental Science, 09:2562–2568.
Deletic, A. & Fletcher, T.D. 2006. Performance of Grass Filters used for Stormwater Treatment a Field
and Modelling Study. Journal of Hydrology, 317 (3–4): 261–275.
He, J. & Qian, G. et al. 2010. Research Progress of Treatment Technology on Pavement Runoff Pollu-
tion. Environmental Science & Technology, S2:430–434+439.
Peng, S. & Huang, J. et al. 2014. Reach of Ffiltration Type Bioretention Pool on Runoff Control. Water
& Wastewater Engineering, (6):38–42.
Shutes, R.B.E & Revitta, D.M. et al. 1999. The Design of Vegetative Constructed Wetlands for the
Treatment of High way Runoff. The Science of the Total Environment, 235 (1–3): 189–197.
Stotz, G. 1987. Investigations of the Properties of the Surface Water Run-off from Federal Highways in
the FRG. The Science of the Total Environment, 59:329–337.
Tang, A. & Wan J. et al. 2014. Integrated Mode of Source Controlling-wastewater intercepting-re-
cycling for Rural Non-point Source Pollution. Chinese Journal of Environmental Engineering,
05:1761–1768.
Zhao, W. & L(u), X. 2013. Review of Using Constructed Wetland and Ecological Pond System for
Road Runoff Treatment. Journal of Highway and Transportation Research and Development,
02:133–139.
1144
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Aboelkasim Diab
Department of Civil Engineering, Aswan University, Aswan, Egypt
Cesare Sangiorgi
Department of Civil, Chemical, Environmental and Materials Engineering, University of Bologna,
Bologna, Italy
Amr M. Wahaballa
Department of Civil Engineering, Aswan University, Aswan, Egypt
1 OVERVIEW
Hot Mix Asphalt (HMA) has been used as the most common material for asphalt pave-
ments construction, since 1900. For providing sufficient workability of HMA mixes, and
proper coating of aggregates, both the asphalt binder and the aggregates are heated to high
temperatures ranging between 140 and 180°C which results in high energy (fuel) costs and
greenhouse gases emissions. As a result of the 1997 Kyoto treaty on climate change, Euro-
pean countries were confronted with greenhouse gas reduction requirements (Prowell et al.
2011). WMA technologies were developed to support the objective set by the Kyoto Protocol
to reduce the emissions of carbon dioxide during the production and compaction of asphalt
mixes. These technologies can either lowering the viscosity of the asphalt binder or improve
the workability of the mix; thus, the production and compaction temperatures can be lower,
as compared to those needed for conventional HMA. Generally, WMA mixes are produced
at temperatures ranging from 20 to 60°C lower than conventional HMA (D’Angelo et al.
2008). However, the benefits of WMA technologies are not limited to the heart of potentially
mitigating emissions and reducing fuel consumption; more benefits will be discussed later in
detail.
1145
Table 1. List of commonly used WMA technologies.
Asphalt production
Temperature
(or reduction ranges),
Technology or additive Manufacturer Description °C
1146
Aspha-Min Eurovia GmbH (Germany) 0.3% by mass of the mixture (20–30)
Advera WMA Zeolite PQ Corporation (USA) 0.25% by mass of the mixture 120
Foaming (Water-based Processes)
WAM Foam Shell (UK) and Kolo-Veidekke 2–5% water by mass of hard binder 100–120
(Norway)
LEA—Low Energy Asphalt LEA-CO (France) 3–4% water introduced with fine sand 100
Double—Barrel Green Astec Industries (USA) ∼2% water by mass of binder 116–135
Terex WMA system Terex (USA) ∼2% water by mass of binder 130
Gencor Ultrafoam GX Gencor Industries Inc. (USA) 1.25–2% water by mass of binder 110–120
Accu-Shear Stansteel (USA) combination of water and/or additives 122–158
(dependent on the additive/manufacturer)
Aquablack WMA Maxam Equipment Inc. (USA) 1.5% –3.0% water by mass of binder 125–140
LT Asphalt (Nynas Low Nynas (Netherlands) Foam binder with hydrophilic additive the 90
temperature asphalt) amount of which 0.5–1.0% by mass of binder
LEAB Royal BAM Group (Netherlands) Foam binder with a special additive (0.1% 90
by mass of binder)
2 WMA TECHNOLOGIES
Examples of the available WMA products and processes are listed in Table 1. More infor-
mation and summary of the recent related studies of the corresponding technologies are
presented in the following sections.
1147
therm 3G is newly developed and falls in this category as a water-free version of Evotherm
that can be introduced to the asphalt binder or at the asphalt mix production plant. The
Evotherm ET is known as a binder-rich water-based emulsion with approximately 70% sol-
ids. It is capable of reducing the mixing temperatures by approximately 38°C. The water in
emulsion turns into steam during mix production and facilitates the mixing and compaction
processes. The emulsifiers in the Evotherm are adsorbed onto the aggregate surface with a
long hydrocarbon tail extending beyond the aggregate surface which in-turn promotes the
interfacial adhesion between the binder and aggregate surfaces (Chowdhury and Button
2008). The Rediset is a surfactant-based chemical additive produced by Akzo Nobel with
the same objective of reducing the interfacial friction between thin films of the asphalt
binder and coated aggregates, improving the workability and allowing for mixing and com-
paction at reduced temperatures. The additive includes built-in anti-strip agents to promote
the interfacial adhesion between aggregate and asphalt binder. The third example in this
category is Cecabase RT produced by CECA (France). The additive has the same hypoth-
esized mechanism to produce WMA as that of surfactants such as Rediset. Iterlow T is a
liquid chemical additive, when added to asphalt binder, allows for the production of WMAs
at temperature around 120°C.
The WMA technologies have been developed to lower the production and placement tem-
peratures of asphalt mixes, which offer several benefits over the conventional HMA. The
most important benefit of WMA is the possibility to reduce the greenhouse emissions. This is
achieved by the reduced production temperature of WMA. Figure 1 shows the reported emis-
sion reduction results from WMA European practice conducted by WMA technical working
group for selected EU nations. In addition, the reduction of production temperatures provides
energy savings as compared to HMA. However, this mostly depends on the production tem-
perature and the kind of fuel used. Hassan (2009) reported that WMA provides a reduction
of 24% in air pollution and a reduction of 18% on fossil fuel consumption as compared to
HMA. In a study performed by Barthel et al. (2004), a reduction in energy consumption by
30% was reported, only 5.6 ltrs was sufficient as compared to 8 ltrs of oil per ton of mix, by
using a synthetic zeolite. A mixing plant operator reported the difference between the fuel
consumption of HMA and WMA in a low volume pavement construction project in Alaska
(Saboundjian et al. 2011). Interestingly, a consumption of 0.5 gallons of fuel per ton of WMA
produced as opposed to 1.5 gallon of fuel per ton needed for HMA production. The reduction
of emissions is also beneficial to the working crew health and to the people in the surrounding
areas of production and paving sites. This adds other benefits to mitigate the emission prob-
lems resulting from a plant site in urban areas. On the other hand, the production of WMA
mixes at lower temperatures reduces the aging of the asphalt binder and tends to improve the
pavement structural flexibility, which reduces susceptibility to fatigue and temperature crack-
ing at early stages. This helps further improvement of the pavement longevity and reducing
the potential costs for restoring the asphalt overlay (Perkins 2009). Use of higher amounts of
RAP is reported to be feasible in WMA mixes (Guo et al. 2014). This is because of a higher
workability of WMA mixes during production and compaction. Asphalt mixes containing up
to 90% RAP was reported to be produced using WMA technology with a reasonable work-
ability and less efforts needed for compaction, which translates to additional energy saving
(Drüschner 2009). This helps utilizing the RAP materials and saving landfill space, reduction
of virgin aggregate consumption and the energy used for mining (Dini-Almeida et al. 2016).
Less compaction effort is needed for WMA because of the higher workability of the resulted
mix. The impact of this benefit becomes more dominant when working in cold weather
Figure 1. Reduction in plant emissions with the use of WMA for selected EU nations (Andersen 2007,
Brosseaud 2007, Moen 2007).
1150
because of the smaller difference between the mix and ambient temperature as compared to
HMA. Therefore, a wider field construction window can be achieved by using WMA. This
permits an extended paving season and/or paving during nights, ensuring a significant eco-
nomic benefit for paving contractors. D’Angelo et al. (2008) reported better road density of
WMA produced using Sasobit® at ambient temperatures ranging from +1 to + 3°C for field
trial sections, compared with those of HMA mixes in Germany. Longer hauling distances
are promised for WMA mixes because of the possibility to compact the mix at lower tem-
peratures without compromising the workability. The WMA technology provides another
benefit for urban or high maintenance roads that need early traffic opening because less time
is necessary for cooling the mixture since the initial temperature is lower as compared to
HMA. The WMA technology provides indirect economic effects such as the reduction of the
mobilization costs and longer paving season. Additional indirect benefit is less wear of the
asphalt plant due to the reduced mixing temperature (Hughes et al. 2009, Ball 2010). Finally,
laboratory testing methods can be directly applied to WMAs to assess the potential benefits
of these mixtures with both traditional and advanced procedures (Dondi et al. 2013).
It is necessary to ensure that the WMA has the same or better mechanical characteristics and
long-term performance as HMA in order to reach a widespread implementation. The tech-
nology has been applied to produce many types of asphalt mixes, including dense graded,
stone mastic, porous, and mastic asphalt. Also, it has been used with modified binders as well
as different types of aggregates and RAP amounts for a variety of layer thicknesses and traf-
fic levels. Notwithstanding the benefits of WMA, there are concerns about the application
of this technology. This is mainly because the technology is relatively new and insufficient in-
situ performance data are available. Although several research studies have been conducted
on different WMA mixes both in the laboratory as well as in the field, comparing WMA
performance with reference HMA mixes, concerns still exist, particularly with respect to rut-
ting and moisture susceptibility of WMA.
The rutting concern of WMA comes from the fact that decreased mixing temperature of
WMA may lead to incomplete drying of aggregates, insufficient coating with asphalt binder
and decreased oxidative hardening of asphalt binder (Goh and You 2008). This might be
mitigated by adding adhesion promoting agents, or by initially choosing a harder binder
grade. Potential rutting problems require careful evaluation of asphalt in laboratory, since the
rheological properties of asphalt binder is the main contributing factor to rutting perform-
ance (Roque et al. 1987). However, WMA that is produced by adding waxes generally showed
better resistance to rutting than the reference HMA. This can be attributed to formation of
the lattice structure in asphalt binder below the crystallization point of wax, which stiffens
the asphalt at in-service temperatures (Zaumanis and Haritonovs 2010). Ruckel et al. (1983)
studied the rutting behavior of foamed WMA mixes and concluded that the rutting failure
occurs usually weeks after the construction not years. This is possible due to insufficient cur-
ing (water dissipation) of the foamed WMA mix. Copeland et al. (2010) reported the rutting
performance of control HMA and WMA mixes based on the Flow Number. The WMA
was produced by foaming the asphalt binder by adding 2% water based on the weight of the
binder and all mixes contained 45% RAP. It was reported that even with the addition of 45%
RAP, the WMA was still more susceptible to rutting than HMA with RAP.
Moisture damage results in failures of adhesive bonding at the binder-aggregate interface
and/or cohesive bonding within the asphalt binder (Hicks 1991). The loss of adhesion (i.e.,
stripping) is caused by the breakage of the interfacial adhesive bond between aggregate sur-
face and asphalt binder, primarily due to the action of water and water vapor (Jo et al. 1997).
The loss of cohesion is primarily caused by the action of moisture within the asphalt binder,
which causes softening, and hence loss of stability of the mix. The moisture susceptibility
of WMA pavements becomes a greater problem than HMA because the WMA binders may
be softer and some technologies depend on utilizing water (i.e., water-bearing additives or
1151
water-based foaming processes) as a workability aid (Abbas and Ali 2011). The presence of
water after foaming and even the incomplete drying of aggregates due to reduced produc-
tion temperatures can impair the binder-to-aggregate bonding and may lead to an increased
susceptibility to moisture damage. This problem may be successfully mitigated with addition
of active adhesion agents. Diab et al. (2014a, 2014b) concluded a decrease of the moisture
damage of the asphalt-aggregate in terms of the cohesive and adhesive bonding, with the use
of the nano-sized hydrated lime as an additive to foamed WMA. However, it is extremely
important to test the compatibility of the chemical additives used in producing WMA asphalt
mix with aggregates and asphalt binders (Lamperti et al. 2015).
Increasing awareness over the human impacts on the environment such as global warming,
carbon footprint, conservation of the natural resources and focus on energy consumption, are
likely to stimulate interest in a wider use of WMA technologies, in the near future. Despite the
decade-long history of the use of WMA in the early sites in Germany and Norway, this tech-
nology has not been applied to its full potential in the U.S. and other countries, so far. This
is mainly due to the lack of information over the advantages and disadvantages of different
types of WMA technologies and concerns over the long-term performance of the pavements
constructed using WMA. Although wealth of knowledge available in the pertinent literature
on the properties and laboratory performance of WMA, still many of the laboratory stud-
ies draw contradictory conclusions. Also, lack of long-term field performance data for this
type of pavement material as it can be found for HMA, may lead to hesitations using this
technology. Therefore, there is a need for a comprehensive literature search and summarizing
the outcomes of different studies on WMA in a systematical way. In view of this objective, a
selection of the available literature on WMA based on the recent publications and findings
in this area was reviewed in order to identify the areas in which each of these technologies
may be used with success. The current review was undertaken with a focus on the various
categories of the WMA technology, benefits and drawbacks of their use. Based on the review
of the previous studies and the discussions presented in this paper, the following aspects may
be highlighted as follows:
1. Generally, the production and placement temperatures of an asphalt mix can be reduced
from 20 to 60°C, depending on the WMA technology used. According to an increasing
number of laboratory and field experiences, this temperature reduction leads to savings in
energy cost, cutting emissions, providing a healthier work environment for the construc-
tion crew, early traffic opening, and a wider field construction window and radius.
2. Use of higher amounts of RAP is reported to be feasible in WMA mixes. Along with new
plant mixing solutions a wide part of the ongoing research deals with new additives and
rejuvenators to improve binders blending and effectiveness.
3. The concerns over the rutting of WMA mixes may be addressed by using an adhesion
promoting agent and/or by choosing a harder binder type (bumping the PG grade up). In
this case rheological properties of the asphalt binder should be studied.
4. Among WMA mixes, those produced by adding waxes generally showed a better rutting
performance, even compared to their reference HMA mixes.
5. The moisture-induced damage potential is the most important concern associated with
the WMA mixes. More specifically, WMA technologies involving water-bearing additives
and those produced using water injection techniques are more prone to moisture dam-
age. However, this issue can be mitigated by using an effective anti-stripping agents (e.g.,
hydrated lime).
6. Construction of test section(s) using different WMA technologies and performance moni-
toring of these test sections side by side with laboratory tests are key elements to under-
standing the long-term performance of the WMA mixes.
1152
7. The economic benefits of WMA should be evaluated together with environmental ben-
efits, especially if stricter emission standards are implemented. Therefore, the complete
estimation of the potential economic benefits of using WMA technologies may not be
feasible if it is not jointly evaluated with the environmental regulations and additional
costs and offsets of WMA production.
REFERENCES
Abbas, A.R. and A. Ali (2011). “Mechanical Properties of Warm Mix Asphalt Prepared Using Foamed
Asphalt Binders.” Final Report FHWA/OH-2011/6, Federal Highway Administration, Springfield,
Virginia.
Andersen, E. (2007). “WAM Foam® An Environmental Friendly Alternative to Hot Mix Asphalt.”
Norwegian Public Roads Administration, Norway, Presentation to WMA Scan Team.
Anderson, R.M., G. Baumgardner, R. May and G. Reinke (2008). “NCHRP 9-47: engineering proper-
ties, emissions, and field performance of warm mix asphalt technologies.” Interim Report, National
Cooperation Highway Research Program, Washington, DC, USA.
Ball, G. (2010). “Environmental and Financial Costs and Benefits of Warm Asphalts.” NZ Transport
Agency Research Report 404.
Barthel, W., J.P. Marchand and M.V. Devivere (2004). “Warm Asphalt Mixes by Adding a Synthetic
Zeolite.” 3rd Eurasphalt & Eurobitume Congress, pp.1241–1249.
Brennen, M.T., A.G. Altschaeffl and L.E. Wood (1983). “Laboratory Investigation of the use of Foamed
Asphalt for Recycled Bituminous Pavements.” Transportation Researh Record No 911, pp. 80–87.
Brosseaud, Y. (2007). “Warm Asphalt – Overview in France.” LCPC, France, Presentation to WMA
Scan Team.
Chowdhury, A. and J. Button (2008). “A Review of Warm Mix Asphalt.” Texas Transportation Insti-
tute. Springfield, Virginia: National Technical Information Service, Technical report.
Copeland, A., D. John, R. Dongre, S. Belagutti and G. Sholar (2010). “Field Evaluation of High
Reclaimed Asphalt Pavement/Warm Mix Asphalt Project in Florida: A Case Study.” Paper No.
10-3792, Transportation Research Board Annual Meeting CD-ROM, Transportation Research
Board of the National Academies, Washington, D.C.
Csanyi, L.H. (1957). “Foamed asphalt in bituminous paving mixtures.” Highway Research Record 160:
pp.108–122.
D’Angelo, J., E. Harm, J. Bartozsek, G. Baumgardner, M. Corrigan, J. Cowsert, T. Harman, M. Jam-
shidi, W. Jones, D. Newcomb, B. Prowell, R. Sines and B. Yeaton (2008). “Warm Mix Asphalt: Euro-
pean Practice.” FHWA report no: FHWA-PL-08-007. 68p.
Diab, A., Z. You, R. Ghabchi and M. Zaman (2014b). “Effects of Regular-Sized and Nanosized
Hydrated Lime on Binder Rheology and Surface Free Energy of Adhesion of Foamed Warm Mix
Asphalt.” Journal of American Society of Civil Engineers.
Diab, A., Z. You, Z. Hossain and M. Zaman (2014a). “Moisture Susceptibility Evaluation of Nano-
sized Hydrated Lime-Modified Asphalt-Aggregate Systems Based on Surface Free Energy Concept “.
Transportation Research Board and Journal of Transportation Research Record, Washington, D.C.
Dinis-Almeida, M., Castro-Gomes, J., Sangiorgi, C., Zoorob, S.E., Afonso, M.L. (2016). “Performance
of Warm Mix Recycled Asphalt containing up to 100% RAP”. Construction and Building Materials,
112: pp. 1–6.
Dondi, G., Pettinari, M., Sangiorgi, C., Zoorob, S.E. (2013). “Traditional and Dissipated Energy
approaches to compare the 2PB and 4PB flexural methodologies on a Warm Mix Asphalt”. Con-
struction and Building Materials, 47: pp. 833–839.
Drüschner, L. (2009). “Experience with Warm Mix Asphalt in Germany.” Sønderborg: NVF-rapporter,
pp. 11–58.
Goh, S.W. and Z. You (2008). “Resilient Modulus and Dynamic Modulus of Warm Mix Asphalt.”
GeoCongress 2008, Geosustainability and Geohazard Mitigation, Geotechnical Special Publication
(GSP 178), American Society of Civil Engineers, 1000–1007.
Goh, S.W. and Z. You (2011). “Moisture Damage and Fatigue Cracking of Foamed Warm Mix Asphalt
Using a Simple Laboratory Setup.” ASCE T&DI Congress 2011: Integrated Transportation and
Development for a Better Tomorrow, Proceedings of the First T&DI Congress 2011.
Guo, N., Z. You, Y. Zhao, Y. Tan and A. Diab (2014). “Laboratory performance of warm mix asphalt
containing recycled asphalt mixtures.” Construction and Building Materials, 64: pp. 141–149.
1153
Hassan, M.M. (2009). “Life-Cycle Assessment of Warm-Mix Asphalt: Environmental and Economic
Perspectives.” Proceedings of 88th transportation research board.
Hicks, R.G. (1991). “NCHRP Synthesis of Highway Practice 175: Moisture Damage in Asphalt Con-
crete.” TRB, National Research Council, Washington, D.C.
Hughes, T., J.K. Davidson and A. Cormier (2009). “Performance of Warm Mix Technology in the
Province of New Brunswick.” Proceedings, 54th Annual Conference, Canadian Technical Asphalt
Association, Vol. LIV, Moncton, New Brunswick, Canada, pp. 149–172.
James, R. (1965). “Metering of steam-water two-phase flow by sharp-edged orifices.” Inst. Mech. Engi-
neers Proc, v. 180, pt. 1, no. 23, pp. 549–572.
Jenkins, K., M.V.D. Ven and J.D. Groot (1999). “Characterisation of Foamed Bitumen.” 7th Confer-
ence on Asphalt Pavement for Southern Africa: South Africa.
Jo, M.C., A.R. Tarrer, Y.W. Jeon, S.J. Park and H.H. Yoon (1997). “Investigation of the effect of aggre-
gate pretreatment with anti-strip agents on the asphalt-aggregate bond.” Petroleum Science and
Technology, 15(3&4), pp. 245–271.
Kristjanssdottir, O., S.T. Muench, L. Michael and G. Burke (2007). “Assessing potential for warm-mix
asphalt technology adoption.” Transportation Research Record, 2040, pp. 91–99.
Lamperti, R., Grenfell, J., Sangiorgi, C., Lantieri, C., Airey, G.D. (2015). “Influence of waxes on adhe-
sion properties of bituminous binders”. Construction and Building Materials, 76:404–412.
Li, X., H. Wang, C. Zhang, A. Diab and Z. You (2015). “Characteristics of a Surfactant Produced
Warm Mix Asphalt Binder and Workability of the Mixture.” journal of Testing and Evaluation,Vol.
44, No. 6.
Masson, J.F., L. Pelletier and P. Collins (2001). “Rapid FTIR method for quantification of styrene–
butadiene type copolymers in bitumen.” J Appl Polym Sci;79: pp. 1034–41.
Middleton, B. and R.W. Forfylow (2009). “An Evaluation of Warm Mix Asphalt Produced with the
Double Barrel Green Process.” Transportation Research Board 88th Annual Meeting. Washington,
D.C.
Moen, Ø. (2007). “Warm Mix Asphalt (WMA) International Scanning Tour.” Norwegian Public Roads
Administration, Norway, Presentation to WMA Scan Team.
Perkins, S.W. (2009). “Synthesis of Warm-mix Asphalt Paving Strategies for use in Montana Highway
Construction.” Final Project Report. Western Transportation Institute, Montana State University,
Bozeman, MT, USA.
Prowell, B.D., G.C. Hurley and B. Frank (2011). “Warm-Mix Asphalt: Best Practices.” National Asphalt
Pavement Association, Lanham, Md, USA, 2nd edition.
Roque, R., M. Tia and B.E. Ruth (1987). “Asphalt rheology; relationship to mixture, Asphalt rheology
to define the properties of asphalt concrete mixtures and the performance of pavement.” ASTM STP
941. O.E. Briscoe, cd., ASTM, Philadelphia, pp. 3–27.
Rowe, G.M., G.L. Baumgardner, G. Reinke, J. D’Angelo and D.A. Anderson (2009). “Evaluation of
the BBR Test with Mixtures Containing Waxes.” Binder Expert Task Group Meeting. San Antonio,
Texas.
Ruckel, P.J., S.M. Acott and R.H. Bowering (1983). “Foamed-asphalt paving mixtures: Preparation of
design mixes and treatment of test specimens.” Transportation Research Record 911: pp. 88–95.
Saboundjian, S., J. Liu, P. Li and B. Brunette (2011). “Late-Season Paving of a Low-Volume Road
with Warm Mix Asphalt, An Alaskan Experience.” Transportation Research Record: Journal of the
Transportation Research Board, no. 2205, pp. 40–47.
Saleh, M.F. (2004). “New Zealand Experience with Foam Bitumen Stabilization.” Transportation
Research Record 1868, TRB, National Research Council, Washington, D.C.
Sasol (2008). “Sasobit” in Warm Mix Open House.” Morro Bay, CA.
Zaumanis, M. and V. Haritonovs (2010). “Research on Properties of Warm Mix Asphalt.” Scientific
Journal of Riga Technical University, Construction Science. pp. 77–84.
1154
Pavements & Innovation
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Jianjun Liang
Harbin Municipal Engineering Design Institute, Harbin, China
1 INTRODUCTION
With the big background of building a resource-conserving and environment-friendly society, how
to improve the use function of the road, how to provide safer, more comfortable, more environ-
mental road surface characteristics, and how to realize the sustainable development of engineering
project have become the hot problems and new goals of road engineering under the new period.
Throughout the technological frontier of road engineering, with the characteristics of large
voids, the porous asphalt pavement (PA) have the advantages of reduction of splash and
spray as well as risk of hydroplaning and wet skidding, and improvement of pavement mark-
ings visibility in wet weather (Liu K-W, 2010). In addition, porous asphalt pavement provide
improved riding quality and noise reduction effectiveness as compared to dense-graded HMA,
consequently, porous asphalt pavement can yet be regarded as a kind of high performance,
high quality of functional pavement structure type (Button JW, 2004, Allex E, 2011).
The porous asphalt mixture is a special type of hot mix asphalt (HMA) characterized by
the use of a high quality open-graded aggregate to obtain: (i) a coarse granular skeleton that
develops stone-on-stone contact and (ii) a high connected air voids (AV) content. However,
it is paramount that these pavements perform as intended, and which requires proper design
of the mixture, among other variables (Bradley J. Putman, 2012).
The mix design and construction of a PA mixture should lead to a compacted mixture
with an equilibrium density and optimum internal structure (proper size, distribution, and
connectivity of AV and adequate aggregate contact and interlock). However, Huber (2000)
reported that there was not a unified PA mix design procedure in the United States in 2000.
A similar situation was reported later for the design practice of European PA mixtures. Thus,
different agencies and countries use diverse criteria for selecting the optimum asphalt binder
content. Particular approaches applied for mix design of permeable friction course (PFC)
and PA mixtures are summarized by Alvarez et al (2006) and in NCHRP Report 640 (Cooley
1157
Jr LA, 2009). It is can be found that these mix design procedures define the optimum asphalt
binder content depending on the volumetric properties.
The volume indicators play an important role in the process of asphalt mixture design.
Both traditional Marshall (Marshall Design method and Superpave mixture design method,
have strict regulation on requirements of the volume indicators for asphalt mixtures. In the
1990s, the Superpave mixture design method, the important achievements of the strategic
highway research program (SHRP), takes volume design as an important process of design-
ing asphalt mixture, of which purpose is to control the volume parameters within the rea-
sonable scope to guarantee a good performance of asphalt mixture. Furthermore, studies
(Liu Honghai, 2007) indicate that the volume parameters, such as compaction void, void in
mineral aggregate and asphalt interstitial rate determine the high temperature performance,
the water damage resistance and durability performance. The volume parameters are inter-
dependence and mutual restriction.
At present, the volume design method is the dominant method of asphalt mixture design
at home and abroad. Of course, there is no exception to porous asphalt mixture, the volume
parameter selection reasonably or not in the process of design will significantly influence
the performance of the mixture. The admirable performance of porous asphalt pavement
depends on the formation of coarse aggregate skeleton of stone—stone contact and the
residual voids gaining from a moderate amount of bitumen filling the aggregate skeleton
gap. Therefore, the volume parameters are of vital importance.
In this paper, taking the porous asphalt mixture as the research object on the basis of
existing research, firstly, it studies the relationship among the volume parameters of porous
asphalt by means of mathematical deduction and theoretical analysis and obtains the quan-
titative relation among the volume parameters. The change law and mechanism of volume
parameters are analyzed based on the indoor tests. Secondly, on the basis of the relationship
and the change law among volume parameters, it has built the technology of determining
asphalt content without making the Marshall specimens and puts forward the new design
method for porous asphalt pavement.
1158
Where: VV is the volume of air voids in bituminous mixture, VMA is the volume of voids
in mineral aggregate in bituminous mixture, Vbe is the volume of effective asphalt, Vba is the
volume of asphalt absorbed by aggregate, Vsa,, Vsb, Vse are the apparent volume, bulk volume
and effective volume, respectively, Vt is the volume of corresponding to theoretical maximum
density of mixture.
100 + Pb
γt =
100 Pb
+ (1)
γ se γ b
⎛ γ ⎞
VV = ⎜1 − f ⎟ × 100 (2)
⎝ γt⎠
100 × Pb × γ f
VA = (3)
100 + Pb × γ b
γf 100 × Pb × γ f
M = 1−
VMA × 100 + (4)
γt 100 + Pb × γ b
γf
VCAmix = 1 − × Pca × 100 (5)
γ ca
100
Making γ f × γ ,
100 + P b
⎛ γ ⎞
Then, VMA = ⎜1 − dγ
× 100, and VMA ) × γ ssb (6)
⎝ γ ssb ⎟⎠
Vbe VMA
V −VV (7)
Pfa × 100 1
Vfa = γ f × × (8)
100 + Pb γ fa
Plugging the equation (6) into the equation (8) gets the following equation (9).
1159
Pfa P
Vfa = γ × = 1 − 0.01 × VMA ) × γ ssb × fa (9)
γ ffa γ ffa
Pfi × 100 1
Vfi = γ f × × (10)
100 + Pb γ fi
Plugging the equation (6) into the equation (10) gets the following equation (11)
Pfi P
Vfi = γ × = 1 − 0.01 × VMA ) × γ ssb × fi (11)
γ ffi γ ffi
Where: Vfa is fine mineral aggregate volume in bituminous mixtures, Pfa is percent of the
fine mineral aggregate in mineral aggregate, γ fa is synthetic bulk specific gravity of fine min-
eral aggregate, Vfi is filler aggregate volume in bituminous mixtures, Pfi is percent of filler
aggregate in mineral aggregate, γ fi is synthetic bulk specific gravity of filler.
So according to the volume composition of compaction porous asphalt mixture:
Plugging the equations (10, 12) into the equation (12) gets the following equation (13)
⎛P P ⎞ ⎧ ⎛P P ⎞⎫
VCA mix = × ⎜ fa + fi ⎟ + ⎨1 − 0.01 × γ ssb × ⎜ fa + fi ⎟ ⎬ × VMA (13)
⎝ γ fa γ ffi ⎠ ⎩ ⎝ γ fa γ ffi ⎠ ⎭
ssb
It can be seen from the equation (13), the value of VCA mix consists of two parts. One
part is related to the mineral aggregate gradation, which can be calculated by the proportion
and density of fine aggregate and filler in mixture. The other part is a constant times of the
VMA. Visibly, the VCAmix is closely related to the VMA under the certain raw materials and
aggregate gradation.
Bulk specific gravity 2.743 2.870 2.864 2.803 Specific gravity 1.020
Sieve, mm 2.36–1.18 1.18–0.6 0.6–0.3 0.3–0.15 0.15–0.075 filler
Apparent specific gravity 2.717 2.724 2.691 2.686 2.696 2.746
1160
Figure 2. The aggregate gradation curves.
1161
Figure 5. The change law of VCAmix to bitumen-aggregate ratio.
Sum of square
Source for III type df Mean square F Sig.
a
Correction model 1125.913 19 59.259 197.188 .000
Intercept 57888.548 1 57888.548 192629.072 .000
Gradation 1087.740 4 271.935 904.887 .000
Pb 2.426 3 .809 2.691 .054
Error 19.233 64 .301
Total 61284.500 84
Total of correction 1145.146 83
Sum of square
Source for III type df Mean square F Sig.
the gradation has more significant influence on VCAmix than the asphalt content, which is
consistent with the theoretical analysis above. The conclusion can be drawn that in a cer-
tain range of asphalt content, the VMA relies on aggregate gradation. In order to study the
change mechanism of the VMA and VCAmix, SPSS18.0 was employed to conduct ANOVA.
Table 2 and Table 3 list the statistical analysis results.
In table 2, under the significance level of 5.0%, the gradation change has significant effect
on the VMA, the P value of the bitumen-aggregate ratio is 0.054 larger than 0.05, namely,
there is no significant influence on the VMA under certain asphalt content, which fundamen-
tally reveals the change mechanism of VMA. Table 3 shows under significance level of 5.0%,
the gradation and asphalt content both have notably influence on VCAmix, whereas, grada-
tion affects significantly. Thus, in the design process of porous asphalt mixture, it is reasonable
1162
to get required VMA and skeleton interlocking structure by adjusting the aggregate gradation.
In addition, under the premise that has obtained rational VMA, then can be allowed to change
the asphalt content to get the good performance for porous asphalt mixture design. What’s
more, VMA can be rationally estimated without preparation of Marshall specimens.
3 APPLICATION RESEARCH
⎛ 1 1⎞
Pba = ⎜ − ⎟ × γ b × 100 (14)
⎝ γ sb γ sse ⎠
Pbe
The volume of effective asphalt binder is calculated by the following equation (15),
Pbe × 100 1
Vbe = γ f × × (15)
100 + Pb γ b
Plugging the equation (6) into the equation (15) gets the following equation (16),
Pbe
Vbe = γ × =( − × VMA ) × γ ssb (16)
γb
Pbe =
Vbe × γ b
=
( VMA VV) × γ b (17)
( − × VMA ) × γ ssb ( − × VMA ) × γ ssb
Finally,
Pb Pba + Pbe
⎛ 1
Pb = ⎜ −
1⎞
× γ b × 100 +
( VMA VV) × γ b + 0.1%
(18)
⎝ γ sb γ sse ⎟⎠ ( − × VMA) × γ ssb
3.2 Application
3.2.1 Aggregate gradation design
The reference (Liu Honghai, 2007) indicated that the key factor to the value of VV is the
passing rate of 2.36 mm sieve. In view of the above research conclusion, this paper builds
the linear relationship between VV and the passing rate of 2.36 mm sieve (Figure 6.) through
taking example by the realization method of VV in Japan. Then it designs three different
designing VV gradation curves by adjusting the passing rate of 2.36 mm sieve which allows
1163
Figure 6. The relationship between VV and passing rate of 2.36 mm sieve for porous asphalt mixture.
Sieve, mm 16 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075 Designing
VV %
Designing
Gradation VV, % VMA, % Pba, % Pbe, % Pb, %
the fluctuation change rate is 3%. Three specific design gradation programs are in the fol-
lowing table 4.
1164
It can be seen the values of VV, MS and draindown meet the specification (JTG F40–2004)
requirements for different porous asphalt mixture.
4 CONCLUSION
Base on the calculation formula of the volume parameters in the specification, the internal
natural relationship among the volume parameters was analyzed by analyzing the volumetric
structure characteristic of porous asphalt mixture. The mathematical relationship among the
volume parameters was obtained, which indicates that VCAmix is closely related to the VMA
under the certain raw materials and aggregate gradation.
The indoor test results show the VMA value increases along with the gradation being
coarser under the same asphalt content, the change amplitude of the VMA value is extremely
small and the gradation has more significant influence on VCAmix than the asphalt content.
Besides, the ANOVA results of VMA and VCAmix demonstrate it is reasonable to get required
VMA and skeleton interlocking structure by adjusting the aggregate gradation. In addition,
VMA would be estimated without preparation of these specimens, design of hot mix asphalt
to obtain both an optimum asphalt content and necessary void content would be rational.
On the basis of the relationship and the change law among volume parameters, it has built
the formula of determining asphalt content without making the Marshall specimens. The
formula was verified properly by designing porous asphalt mixtures. We would like to pro-
pose a volumetric design method of PA by using this formula instead of amount of tests.
REFERENCES
Allex E. Alvarez, Amy Epps Martin, Cindy Estakhri. 2011. A review of mix design and evaluation
research for permeable friction course mixtures. Journal of Construction and Building Materials,
25:1159–1166.
Alvarez AE, Epps Martin A, Estakhri C, Button JW, Glover C, Jung SH. 2006. Synthesis of current
practice on the design, construction, and maintenance of porous friction courses. Report No. FHWA/
TX-06/0-5262-1. College Station, TX: Texas Transportation Institute-Texas A&M University.
Button JW, Fernando EG, Middleton DR. 2004. Synthesis of pavement issues related to high-speed
corridors. Report No. 0-4756-1. College Station, TX: Texas Transportation Institute-Texas A&M
University.
Bradley J. Putman, A.M.ASCE; and Laura C. Kline. 2012. Comparison of Mix Design Methods for
Porous Asphalt Mixtures. Journal of Materials in Civil Engineering, 24:1359–1367.
Cooley Jr LA, Brumfield JW, Mallick RB, Mogawer WS, Partl M, Poulikakos LD, et al. 2009. Construc-
tion and maintenance practices for permeable friction courses. NCHRP Report 640, Washington,
DC.
Dr. Takemi Inoue, Dr. Yasuo Gunji, Prof. Hirokazu Akagi. 2004. Rational Design Method of Hot Mix
Asphalt Based on Calculation VMA [J]. 3rd Eurasphalt & Eurobitume Congress Vienna—Paper 080.
2011. “Standard Test Methods of Bitumen and Bituminous Mixtures for Highway Engineering” JTG
E20-2011, Beijin: China communication press.
Huber G. 2000. Performance survey on open-graded friction course mixes. Synthesis of highway prac-
tice 284. Transportation Research Board, National Research Council, Washington, DC.
Liu K-W, Alvarez AE, Epps Martin A, Dossey T, Smit A, Estakhri C, et al. 2010. Performance, design,
construction, and maintenance. FHWA/TX-10/0–5836−1. College Station, TX: Texas Transportation
Institute-Texas A&M University.
Liu Hong-hai, Fan Peng-yun, Yue Peng-cheng, Wang Jing, Zhou Li-ping. 2007. The Factors of Influ-
encing Mixture Volume Parameter of Asphalt and Construction Quality Control. Journal of Wuhan
University of Technology, 29(9).
Suresha, S N George, Varghese Shankar, A U Ravi. 2009. Evaluation of Properties of Porous Friction
Course Mixes for Different Gyration Levels. Journal of Materials in Civil Engineering, 21:789–796.
1165
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
T. Al-Mansoori
School of Civil Engineering, Nottingham Transportation Engineering Centre,
University of Nottingham, Nottingham, UK
R. Micaelo
School of Civil Engineering, Nottingham Transportation Engineering Centre,
University of Nottingham, Nottingham, UK
DEC, FCT, Universidade NOVA de Lisboa, Caparica, Portugal
CERIS, Universidade de Lisboa, Lisboa, Portugal
A. Garcia
School of Civil Engineering, Nottingham Transportation Engineering Centre,
University of Nottingham, Nottingham, UK
ABSTRACT: Maintenance and preservation of asphalt pavements is one of the main issues
affecting the budgets of Highway Administrations. Asphalt roads life expectancy could be
extended by including encapsulated rejuvenators that automatically break and spread healing
agents into the microcracks of aged bitumen. With this purpose, a new type of capsules have
been developed and added into asphalt mixture. The objective of this article is to investigate
the size, morphology, mechanical stability, thermal stability of the new type of capsules and
their softening effect on bitumen. The results show that capsules can survive the high temper-
atures and stresses of the mixing process. Moreover, it was found that the oil can diffuse into
bitumen and change its viscosity if the capsules are broken. These capsules containing heal-
ing agents can be a promising improvement towards the development of clever pavements.
1 INTRODUCTION
According to García et al. (2010), highway industry uses 110 million tons of bitumen annually
around the world. This indicates the importance of this material for construction and develop-
ment of highway pavements. The most significant problem related to the use of bitumen in road
pavements is ageing, which starts directly after mixing the bitumen with asphalt materials and
continues during transport and laying on the roads, as well as during service life (Lu and Isacsson,
1998). The ageing of bitumen usually happens due to the mechanism of oxidation and losing vol-
atiles from its composition, which causes a stiffness and viscosity increment (Holleran et al., 2006,
Lu and Isacsson, 1998). Holleran et al. (2006) state that the ageing problem of bitumen results
in pavement failure, such as surface raveling and cracking, particularly reflective cracking. These
problems increase the cost of maintenance and preservation of highway pavements. Currently,
there are two methods to improve the lifetime of roads. These are (1) recycling asphalt pavements
(RAP), currently limited because of bitumen ageing, which allows between 30% and 40% of RAP
to be used in the pavement (Shen et al., 2007), and (2) spreading rejuvenators over the road sur-
face, which focuses on rejuvenating the chemical composition of asphalt mixture on-site.
Using rejuvenators is probably the only method that restores the original properties of
aged bitumen (Karlsson and Isacsson, 2003). Karlsson and Isacsson (2003) state the differ-
ence between the softening agents and rejuvenating agents. They found that softening agents
such as asphalt flux oil, lube stock and slurry oil are used to reduce the viscosity of the aged
1167
bitumen. Rejuvenators, on the other hand, which consist of lubricating and extender oils
containing a high proportion of maltene fractions, are used for the purpose of reconstituting
the bitumen chemical composition (Shen et al., 2007). Therefore, the essential aim of using
rejuvenators is to restore the chemical equilibrium of bitumen that was lost during ageing.
Generally, rejuvenators must be highly aromatic so that both hardening and temperature
susceptibility can be improved (Peterson et al., 1994). Rejuvenators are used to increase the
peptizing ability of the maltene phase, therefore, they should be composed in a way that
allows them to perform this requirement (Venable et al., 1983).
The novelty of this article is to propose a new type of capsule containing rejuvenators (oil)
to be used in asphalt self-healing process. These capsules contain more oil and are easier to
make than capsules shown in previous researches. The objective of this paper is to explain
how to make these capsules, and characterize their different sizes and oil content, the effect
of polymer amount and diffusion of oil into bitumen.
2.1 Materials
The capsules had an inner core made of calcium-alginate polymer that contained the rejuve-
nator material and a polymeric material, and an external shell made of epoxy-cement.
The rejuvenator material used in the capsules was sunflower cooking oil (CO), which is a
commercial sunflower oil supplied by East End (UK). The smoke point of the oil was 227°C,
while the flash point was 315°C.
The polymeric material in the inner core of the capsules was composed of (1) sodium algi-
nate, which is an anionic polysaccharide widely distributed in the cell walls of brown algae,
it was supplied by Sigma-Aldrich; and (2) a calcium source, provided by Sigma-Aldrich as
anhydrous, granular pellets of calcium chloride, of 7 mm diameter, and 93% purity.
The outer shell was composed of liquid epoxy resin (Gurit SP106 Multi-purpose Epoxy
System Resin and Gurit SP106 Multi-purpose Epoxy System Fast Hardener) and micro-
cement with average particle size ranged from 5 to 10 μm. The epoxy resin worked as a binder
for the cement particles. Any other crushed filler could be used as an outer shell for the cap-
sule; cement was selected in this research simply because of its availability in the laboratory.
1168
Figure 1. Capsules. To the left, inner core of the capsule. To the right, capsule with epoxy-cement
shell.
by hand, in a mass ratio of 14% (14 g of epoxy for each 100 g of calcium-alginate capsules)
until the capsules become totally covered by epoxy.
Later, 7 steel balls of 2 cm diameter were placed in a plastic container, together with
approximately 1 kg of micro-cement. To this, approximately 250 g of capsules and epoxy
were added. Then, the container was closed, and vigorously shaken by hand in a circular
movement for 15 s.
After this, the capsules, cement, and steel balls were separated in a 2 mm sieve, and let to
rest for 12 hours. The external coating process was repeated three times for each batch of
capsules to improve their strength. At the end of the process, the capsules were similar to that
shown in Figure 1, Right.
1169
failure at a loading rate of 0.2 mm/min. The tests were performed at room temperature using
an Instron Model 5969.Ten different types of capsules were tested.
Figure 2. Variation of the capsule diameter according to different amount of oil in capsules.
1170
3.2 Capsules morphology
The capsules’ morphology was investigated using an Environmental Scanning Electron Micro-
scope (ESEM). Figure 3a shows a section of the capsule where the core and the external shell
are easily identified. Figures 3b and 3c show a zoom-in of the core, core-shell interface and
shell. These areas are numbered 1 to 3 in the Figures. The core has a polymer structure with
micro-pores where the oil is entrapped. Figure 3c shows a section of the epoxy-cement shell.
The structure of the shell is very dense, which is highly positive for preventing the oil from
coming out the capsule. While the core is almost perfectly circular, the final capsule including
the epoxy-cement shell is not: The images show that the core and shell are not bonded at the
interface. There is thin air layer at the interface with small and discrete points of contact.
Figure 3. Capsule morphology; (a) general overview of the capsule; (b) zoom-in of the core and the
core-shell interface; (c) zoom-in of the shell.
1171
Figure 4. TGA curves for the capsule (300 ml oil) and its components.
⎧ Mo M p + M c M e = 1
⎪
⎪ ⎛M M ⎞
⎪ ρccap .⎜ o + p ⎟ = Vo + p
⎪ ⎝ ρo ρp ⎠
⎨ (1)
⎪ ⎛M M ⎞
⎪ ρccap .⎜ c + e ⎟ = Vc + e
⎝ ρc ρe ⎠
⎪
⎪⎩0.86.Mc + 0 07 Mp + 0.05.Me = Mr
Where, M o is the mass proportion of oil; M p is the mass proportion of polymer; M c is the
mass proportion of cement; M e is the mass proportion of epoxy resin; M r is the remaining
mass proportion of the capsules after heating to 1000°C; ρcap a is the density of the capsules;
ρo is the density of the oil (g/cm3); ρ p is the density of the polymer (g/cm3); ρc is the density
of cement (g/cm3); ρe is the density of the epoxy (g/cm3); Vo + p is the volume proportion of the
core in capsules; Vc + e is the volume proportion of the shell in capsules.
Table 1 presents the obtained composition in terms of weight and volume proportion for
the capsules made with 300 ml of oil. These capsules have 40%M, or 64%V, of oil which is sig-
nificantly higher than presented by other different capsules proposed in literature (Garcia
et al. 2010). This factor is highly positive for the asphalt rejuvenation ability of these capsules.
1172
made with 100 g and 200 g of oil had the highest average values of peak load, about 95N,
while for the other capsules the load ranged from 48 to 66N. It is concluded that although
the amount of oil leads to an increase of size the capsules are not necessarily stronger. The
mechanical strength results from the polymer structure and size, which depends on the poly-
mer and oil proportion. Note that during mixing and compaction capsules may also develop
shear stresses. However, these stresses are difficult to tests and methods to evaluate them, e.g.
discrete element methods (DEM) modelling should be applied.
1173
oil immediately after the capsules broke, and the peak of change being between 6 h and 12 h.
This is a clear signal that oil diffused from the capsules into the bitumen. The reduction in
aromatic index after 12 h is likely to be related with the oil diffusion equilibrium in the test
sample because only one capsule was used in each specimen.
4 CONCLUSIONS
In this paper, a new encapsulated self-healing system for asphalt pavements has been explained
and designed. These capsules contain sunflower oil as a rejuvenator. The capsules comprise
a core with a sponge-like inner core embedding the oil in its pores and an external epoxy-
cement shell. The strength of the inner core is variable and can be controlled by the ratio of
oil/polymer during capsule production. The external shell of epoxy-cement gives a very high
strength to the capsules, which is required to resist the high temperatures and stresses during
asphalt mixing and compaction procedures.
The size of the capsules varies with the oil content, from 2 to 7 mm of average diameter.
From the compression tests it was concluded that there is an optimum strength defined by
the amount oil in the capsules. Hence, the size and strength of these capsules can be designed
for each asphalt mixture by changing the oil and polymer content.
Moreover, the TGA test results showed that these capsules can survive the temperatures
used for asphalt mixing, with just 2% mass loss at 200ºC.
Finally, the capsules release the oil after the breakage of the external shell. The oil diffusion
in bitumen was evaluated from the change in FTIR profile. The bitumen was changed after
6 hours of the capsule being broke.
This is an on-going investigation in asphalt self-healing with encapsulated agents and the
next step is to test these capsules in asphalt mixture.
REFERENCES
Garcia, Á., Schlangen, E., Van De Ven, M. & Sierra-Beltran, G. 2010. Preparation of capsules contain-
ing rejuvenators for their use in asphalt concrete. Journal of hazardous materials, 184, 603–611.
Holleran, G., Wieringa, T. & Tailby, T. 2006 Rejuvenation treatments for aged pavements. Transit New
Zealand and New Zealand Institute of Highway Technology (NZIHT) Annual Conference, 8th, 2006,
Auckland, New Zealann.
Karlsson, R. & Isacsson, U. 2003. Investigations on bitumen rejuvenator diffusion and structural stabil-
ity (with discussion). Journal of the Association of Asphalt Paving Technologists, 72.
Lamontagne, I., Durrieu, F., Planche, I., Mouillet, V. & Kister, J. 2001. Direct and continuous methodo-
logical approach to study the ageing of fossil organic material by infrared microspectrometry imag-
ing: application to polymer modified bitumen. Analytica chimica acta, 444, 241–250.
Lu, X. & Isacsson, U. 1998. Chemical and rheological evaluation of ageing properties of SBS polymer
modified bitumens. Fuel, 77, 961–972.
Mouillet, V., Lamontagne, I., Durrieu, F., Planche, J. & Lapalu, L. 2008. Infrared microscopy investiga-
tion of oxidation and phase evolution in bitumen modified with polymers. Fuel, 87, 1270–1280.
Peterson, G., Davison, R., Glover, C. & Bullin, J. 1994. Effect of composition on asphalt recycling agent
performance. Transportation Research Record, 38–46.
Shen, J., Amirkhanian, S. & Aune Miller, J. 2007. Effects of rejuvenating agents on superpave mixtures
containing reclaimed asphalt pavement. Journal of Materials in Civil Engineering, 19, 376–384.
Venable, R.L., Petersen, J., Robertson, R. & Plancher, H. 1983. Investigation of factors affecting asphalt
pavement recycling and asphalt compatibility. Department of Energy, Laramie, WY (USA). Laramie
Energy Technology Center.
1174
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Martin Jasso
Department of Civil Engineering, University of Calgary, Calgary, Canada
Dusan Bakos
Department of Plastics and Rubber, Slovak University of Technology, Bratislava, Slovakia
1 INTRODUCTION
For almost a century, flexible (asphalt) pavements have been the unchallenged material for
the construction of the road surfaces (Zanzotto et al.1987). In order to improve the engineer-
ing properties of asphalt binder, it is modified by some types of polymers, or other additives
e.g. polyphosphoric acid, and some other materials such as fibres, lime, asbestos, etc. The
first mentioned group, however, offers the best option for enhancement of thermo-mechan-
ical properties of asphalt. Apart from these, the introduction of a polymer can bring other
advantages e.g. enhanced resistance to oxidation, stripping and prolonged in-life service of
pavement (Lessueur 2009). According to their chemical structure and properties, polymers
used in the asphalt industry can be divided into the three main groups: elastomers, plastom-
ers and reactive polymers. In the asphalt industry, the amount of polymer in polymer modi-
fied asphalt (PMA), for paving applications, is usually in the range of 2 to 5 wt% (Polacco
et al 2004a). From the large diversity of available polymers only few of them provide satisfac-
tory results and are commercially used as asphalt modifiers.
One of the most important factors is the miscibility of polymer with asphalt which is given by
the comparable solvating parameter, thus only materials with similar level of polarity will have a
propensity to blend, e.g. when styrene-butadiene-styrene (SBS) is added into asphalt it is swollen
by the absorption of the light oily fraction of maltenes (Blanco et al 1995, Chen & Huang 2007,
Adejeji et al. 2006, Rossi et al. 2013). The volume of the polymer phase can be increased up to 10
times with respect to the volume of the original material. From this point of view, total miscibil-
ity is undesirable and only partial miscibility, in which the polymer and asphaltene-rich phases
coexist on a microscale equilibrium, is preferred (Polacco et al. 2005, Jasso et al. 2013).
Commercially, the most successful polymer modifiers are thermoplastic elastomers, espe-
cially SBS. Addition of SBS into the high shear mixer leads to dispergation of polymer pellets
1175
into the small particles and thus the polymer surface in the continuum asphalt phase is larger.
Therefore the contact between these two components is increased and the development of a
three-dimensional polymeric network in an asphalt blend occurs (Vonk & Hartemink 2004).
In the literature a large numbers of publications can be found dealing with the mechanical,
rheological and chemical properties of PMAs e.g. (Airey 2004, Vlachovicova et al. 2007, Pol-
acco et al. 2007, Wloczysiak et al. 1997, Jasso et al. 2012, Fawcett & McNally 2001, Becker
et al. 2003).
Polyolefins are also to some degree used as asphalt modifiers. The typical examples are
polyethylene (Yusof et al. 2013) and polypropylene (Pérez-Lepe et al. 2005). The non-polar,
saturated structure results in a poor compatibility of asphalt with polymer, the separation of
polymer during hot storage occurs and subsequently the effect of the modifier in such PMAs
is completely lost. This negative aspect of polyolefins together with unsatisfactory enhance-
ment of the engineering properties of the asphalt blend led to the modification of the main
polyolefin chain by functional groups, i.e. ethylene-vinyl acetate (Giavarini et al. 2004), or
terpolymers of ethylene, glycidylmethacrylate and an ester group (usually methyl, ethyl or
butylacrylate) (Polacco et al. 2006). The second mentioned group, called reactive polymers
(RET), has been studied as possible asphalt modifiers (Polacco et al. 2006). The presence of
functional groups not only enhances the compatibility with asphalt but it also possibly leads
to a chemical reaction with asphalt.
Unfortunately, due to the chemical complexity of asphalt, which is further complicated by
addition of polymer, the character of the chemical reactions with RET (curing) is very hard
to predict. Thus one has to consider all possible interactions of reactive centers of RET with
asphalt. Based on this fact the content of RET has to be limited due to the risk of gelation,
thus the content of RET in asphalt blend is typically in the range of 1.0 to 2.5 wt% (Guo et al.
2001). Although the application of RET in asphalt blends is rather restricted, the superior
enhancement of the asphalt performance at high service temperatures, as well as stability
during hot storage, have been reported (Polacco et al. 2004b).
This paper compares the rheological characteristics of blends of SBS and Elvaloy with
asphalt. In the case of SBS no crosslinking was used, neither was PPA used in the case of
Elvaloy modifier. Presently, industrial technologies usually use these third materials. Herein,
we thus purposely focus on the basic theoretical comparison of both PMAs, in order to
understand the basic interactions between asphalt and polymer modifiers.
D1 98 2 –
D2 97 3 –
D3 96 4 –
D4 95 5 –
E1 99 - 1
E2 98.5 - 1.5
E3 98 - 2
E4 97.5 - 2.5
ments from −30 to 0°C were performed by using rectangular torsion bars with dimensions of
25.5 ± 0.1 × 12 × 2.65 mm.
In dynamic experiments, the frequency interval was either from 0.1 to 100 rad/s or 0.01
to 100 rad/s, in order to construct master curves. Small amounts of each PMA were slowly
heated to the testing temperature (approximately 10 minutes) and then placed into the rheom-
eter. The samples were left at the testing temperature for 30 minutes before the appropriate
test was started, in order to reach thermal equilibrium. For dynamic experiments below 0°C
with the rectangular torsion bar geometry, the hot asphalt was poured into custom molds
preheated to 160°C for 5 minutes. Samples were left at ambient temperature for one hour,
then trimmed and stored in the freezer before testing. The testing was done within 24 hours
in order to avoid the steric hardening of asphalt and its blends.
The limits of linear viscoelasticity were determined for each sample at the appropriate
temperature by running strain sweeps. The optimal strain was suggested as 80% of the maxi-
mum strain within the linear viscoelastic region. Viscosity measurements were conducted at
50 °C; and, the rate of shear was varied from 10−4 to 100 s−1 using a 25 mm plate-plate geom-
etry. The testing was done at least three times at the test temperature and the reported results
are the averages of these measurements.
The multiple stress creep and recovery tests were performed in the stress controlled rhe-
ometer at two constant stress levels; 100 and 3200 Pa. Every test consisted of creep with a
duration of 1 second, followed by recovery (zero stress level) for 9 seconds. For each stress
level 100 cycles were completed at the testing temperature of 50°C, again using a 25 mm
plate-plate geometry.
Asphalt blends with SBS without crosslinking agent have a tendency to separate during
storage at high temperatures. Cooling the samples to ambient or low temperatures eliminated
the macroscopic separation of SBS. Before the rheological measurements, all samples were
stirred thoroughly in order to avoid possible separation. The testing temperatures for all
tests were far below the curing temperature; therefore, any kinetic effects should have been
avoided.
1177
Dynamic frequency sweep tests were performed in the range of frequencies from 0.01 or
0.1 to 100 rad/s in the temperature interval from −30 to 70°C. When such an experiment is
repeated at varying temperatures, the data can be shifted to the master curves and described
by discrete relaxation (Maxwell) spectra {λi,Gi, gi}. These spectra were calculated from the
experimental values of G’(ω) and G’’(ω) (Ferry 1961), by using the commercial software IRIS
for the selected reference temperature. The horizontal shifting factors are usually described
(as in our case) by the WLF equation (Ferry 1961).
The significance of time-temperature superposition is based on the theoretical possibil-
ity of obtaining the relaxation modulus on the time interval (0, ⊥) by conducting the small
amplitude oscillatory experiments at different temperatures. The master curves of storage
moduli G ′ ( ) at the reference temperature Tr = 50 °C for base asphalt, Kraton® D1101 and
Elvaloy® 4170 can be seen in Figure 1. The three basic regions of viscoelastic behavior of
base asphalt were reported elsewhere (Polacco et al. 2004b). The most substantial changes
were in both cases observed at frequencies lower than 104 rad/s (intermediate temperatures).
Here the onset of the shoulder of G ′ ( ), a phenomenon observed usually for highly entan-
gled polymer systems, was evident. The magnitudes of G ′ ( ) were increasing with increas-
ing concentration of polymers. The magnitudes of G ′ ( ) for Elvaloy® 4170 were generally
higher than those in asphalt blends modified by Kraton® D1101. The role of both polymers
was clearly visible at higher temperatures. The Kraton® D1101 asphalt blends reached the
terminal region at a higher frequency (lower temperature) than asphalt modified by Elvaloy®
4170.
The master curves of phase angle for the studied PMAs blends can be seen in Figure 2.
For the PMAs modified with Kraton® D1101 (Figure 2) a local minimum followed by a local
maximum was observed when the concentration of polymer was higher than 3 wt% while for
Elvaloy® 4170 (Figure 2) the same phenomenon was observed at a concentration higher than
1.5 wt%. As indicated by the shoulder of G’ master curves, the maximum and minimum can
Figure 1. Storage moduli for PMAs: upper—Kraton® D1101, lower—Elvaloy® 4170, Tr = 50°C.
1178
Figure 2. Loss tangent for PMAs: left—Kraton® D1101, right—Elvaloy® 4170, Tr = 50°C.
Figure 3. Loss moduli for PMAs: upper—Kraton® D1101, lower—Elvaloy® 4170, Tr = −20°C.
point to the plateau region, thus to the creation of a polymer network in these PMAs. This
effect is of course desirable and could lead to enhancement of asphalt resistance to deforma-
tion. On the other hand, in base asphalt as well as in PMA with low concentration of poly-
mer, the phase angle maxima and minima are difficult to observe, thus indicating a system
with weak internal structure. The Kraton® D1101 PMAs had generally lower values of phase
angle than blends containing Elvaloy® 4170, however, these samples showed the tendency to
flow at lower temperatures.
1179
Since the time and the temperature are interdependent, it is possible to transform the fre-
quency domain to the temperature domain for the chosen test frequency. Thus from the
WLF relation (Ferry 1961) one obtains:
1180
Figure 4. Viscosity functions at 50°C for PMAs: upper—Kraton® D1101, lower—Elvaloy® 4170,
T = 50°C.
of shear creep (duration of 1 s) and recovery (lasting 9 s). The results of the accumulated
compliance function for Kraton® D1101 at stress level of 100 Pa can be seen in Figure 5. As
expected, at the lowest concentration of Kraton® D1101 (2 wt%) the recovery was very small
and comparable to the recovery of the base asphalt. As the content of polymer was increased,
the recovery increased as well. The lower accumulated creep compliance the larger the recov-
ery was. The deviation of the shape of the creep compliance curve was observed in the sample
with 5 wt% of D1101. The accumulated compliance function at the stress level of 3200 Pa for
Kraton® D1101 is shown in Figure 5 as well. It can be seen that the higher stress level com-
pletely destroyed the internal structure of the studied asphalt blends and the recovery time of
9 s was not sufficiently long enough for the rebuilding of the internal structure.
A quite different behavior was observed in PMAs modified with Elvaloy® 4170 tested at
the same conditions (Figure 6). The values of accumulated creep compliance and the recov-
ery were almost identical for both stress levels. Here again, the increasing concentration of
polymer led to lower values of accumulated creep compliance function and higher recovery.
This observation is a confirmation of the earlier discussed properties of the asphalt blends
modified by Elvaloy® 4170.
When the testing is done in the linear viscoelastic domain it is possible to use the following
simple form of the shear compliance (Ferry 1961, Tschoegl 1993)
N
⎛ − ⎞
t
J (t ) = J g + ∑Ji ⎜1 − e Λi ⎟ + t / η0 (2)
i =0 ⎝ ⎠
In (2), Jg is the glassy compliance (usually a very small number), the pair of real numbers
{ i, Ji} represents the discrete retardation spectrum and η0 is the zero-shear viscosity of the
1181
Figure 5. Creep compliance J(t), T = 50°C, 100 cycles for PMAs modified with Kraton® D1101: left -
100 Pa, right - 3200 Pa.
Figure 6. Creep compliance J(t), T = 50°C, 100 cycles for PMAs modified with Elvaloy® 4170: left -
100 Pa, right - 3200 Pa.
tested material. Strictly speaking the number of elements of the discrete retardation spectrum
has to be larger than 3, in order to describe the compliance functions (J(t) or J*(ω )) on large
intervals (theoretically from zero to infinity). In our case the data of accumulated compli-
ance, Ja, (Wasage 2011) are given on a relatively small time interval thus the three-element
retardation spectrum was satisfactory for the description of the accumulated compliance
function, e.g. see Figure 7.
Interestingly enough, in this case the linear viscoelastic form of J(t) worked well even
though high stress was used. The already mentioned stress levels of 100 and 3200 Pa were
adopted according to the AASHTO specification (American Association of State Highway
and Transportation Officials 2009). The lower stress level should keep the values of accumu-
lated strain in the domain of linear viscoelasticity and the viscosity in (2) should be the limit-
ing viscosity for shear rates approaching zero. All experimental data obtained from MSCR
experiments were fitted to (2).
The second stress level (3200 Pa) cannot generally be considered to be small, however the
accumulated compliance function can be described by the same equation when the viscosity
function depends on the level of the applied stress (Wasage 2011). Generally, for straight-run
asphalts, at higher temperatures, the elastic properties are supressed, thus the recovery during the
9 s is very small. Such a material might be highly susceptible to accumulated deformation at these
conditions. The same phenomenon was observed in sample D4 at 3200 Pa. It is possible, that this
sample would after a sufficiently long period of time reach the equilibrium creep compliance,
a characteristic behaviour for a true viscoelastic solid material in which ηt0 0. The higher the
applied stress is, the larger the shear rate of the steady flow contribution to the creep seemed to
be. Subsequently, with increasing shear rate the viscosity function generally decreases.
1182
Figure 7. Accumulated creep compliance Ja(t) for E4, T = 50°C, 3200 Pa. Fit of 100 cycles to Eq.(7),
N = 3. ν experimental data, — fit.
4 CONCLUSION
The impact of two different polymer modifiers on the rheological properties of soft conven-
tional asphalt was investigated. The performance of the first modifier Kraton® D1101 (not
cross-linked with sulfur) was compared to the reactive elastomeric terpolymer Elvaloy® 4170
(without presence of PPA).
From the thermo-rheological point of view, the influence of polymers (present in low con-
centrations) was insignificant at low temperatures as observed by similar values of Tg. In this
regard, the impact of polymers probably could be better demonstrated by other material
characteristics of the studied PMAs e.g. tensile strength and others.
In small amplitude oscillatory shear, the onset of the shoulder of G ′ ( ) was observed for
both studied polymers in asphalt. The magnitudes of G ′ ( ) of Elvaloy® 4170 were generally
higher than those of Kraton® D1101 modified asphalt blends. The plateau region of phase
angle, the indicator of the creation of a polymer network in asphalt blends, was observed in
both studied systems. This leads to a greater resistance of PMAs to the shear deformations
at elevated temperatures.
The steady-state viscosity measurements showed peculiar behavior of PMAs with Kra-
ton® D1101 at higher concentrations. The observed shear thickening can be explained by the
creation of internal structure of swollen polymer particles. These measurements also showed
possible creation of asphaltene-Elvaloy® 4170 complex with variable mobility and strength.
For all studied PMAs the shift to the lower shear rates with increasing concentration of the
polymer, where η0 is noticeable, was observed.
The results obtained in multiple stress creep and recovery protocol showed extraordinary
resistance of Elvaloy® 4170 modified blends to accumulated shear deformation under both
applied stress levels. On the other hand, the internal structure of Kraton® D1101 blends were
destroyed under the higher level of stress, leading to higher accumulated deformation.
The shear rheology of these two different asphalt modifiers points to Elvaloy as a possible
competing asphalt modifier when compared to SBS, especially at higher temperatures. Due
to the limiting loading, hard prediction of curing and the possibility of the gelation, the use
of RET is, of course, restricted.
REFERENCES
Adedeji, A., Grünfelder, T., Bates, F.S., Macosko, C.W., Stroup-Gardiner, M., Newcomb, D.E. 1996. Asphalt
modified by SBS triblock copolymer: Structures and properties. Polym Eng Sci. 36: 1707–1723.
Airey, G.D. 2004. Styrene butadiene styrene polymer modification of road bitumens. J. Mat Sci. Vol.
39: 951–959.
1183
American Association of State Highway and Transportation Officials. 2009.
Becker, M.Y., Muller, A.J., Rodriguez, Y. 2003. Use of rheological compatibility criteria to study SBS
modified asphalt. J. App Polym Sci. Vol. 90, Issue 7: 1772–1782.
Blanco, R., Rodríguez, R., García-Garduño, M., Castaño, V.M. 1995. Morphology and tensile proper-
ties of styrene–butadiene copolymer reinforced asphalt. J Appl. Polym. Sci. 56: 57–64.
Chen, J.S., Huang, C.C. 2007. Fundamental characterization of SBS-modified asphalt mixed with sul-
fur. J. Appl. Polym. Sci. 103, 2817–2825
Fawcett, A.H., McNally, T. 2001. Blends of bitumen with polymers having a styrene component. Polym
Eng Sci. Vol. 41: 1251–1264.
Ferry, J.D. 1961. Viscoelastic Properties of Polymers. Wiley, New York.
Giavarini, C., De Filippis, P., Santarelli, M.L., Scarsella, M. 2004. Production of stable polypropylene-
modified bitumens. Fuel Vol.75, Issue 6: 681:686
Guo, Q., Figueiredo, P., Thomann, R., Gronski, W. 2001. Phase behavior, morphology and interfacial
structure in thermoset/thermoplastic elastomer blends of poly(propylene glycol)-type epoxy resin
and polystyrene-b-polybutadiene. Polymer 42: 10101–10110
Jasso, M., Bakos, D., MacLeod, D., Zanzotto, L. 2013. Preparation and properties of conventional
asphalt modified by physical mixtures of linear SBS and montmorillonite clay. Const and Build Mat.
Vol.38: 759–765
Jasso, M., Bakos, D., Stastna, J., Zanzotto, L. 2012. Conventional asphalt modified by physical mixtures
of linear SBS and montmorillonite. Appl Clay Sci. Vol 70: 37–44.
Lesueur, D. 2009. The Colloidal Structure of Bitumen: Consequences on the Rheology and on the
Mechanisms of Bitumen Modification. Adv.in Coll and Interf Sci. Vol. 145: 42–82
Lu, X., Issacsoon, U., Ekblad, J. 2003. Influence of polymer modification on low temperature behav-
iour of bituminous binder and mixtures. Materials and structures Vol.36: 652–656
Polacco, G., Berlincioni, S., Biondi, D., Stastna, J., Zanzotto, L. 2005. Asphalt modification with differ-
ent polyethylene-based polymers. European polymer journal. Vol.41, Issue 12: 2831–2844
Polacco, G., Stastna, J., Biondi, D., Antonelli, F., Vlachovicova, Z., Zanzotto, L. 2004b. Rheology of
asphalts modified with glycidylmethacrylate functionalized polymers. J. Colloid Interface Sci Vol.280
(2): 366–373.
Polacco, G., Stastna, J., Biondi, D., Zanzotto, L. 2006 Relation Between Polymer Architecture and Non-
linear Viscoelastic Behavior of Modified Asphalts. Curr Op in Coll & Inter Sci. Vol. 11(4): 230–245
Polacco, G., Stastna, J., Michalica, P., Biondi, D., Cantù, M., Zanzotto, L. 2007. Memory functions in
polymer modified asphalts. J. Appl Polym Sci. Vol. 104: 2330–2340.
Polacco, G., Stastna, J., Vlachovicova, Z., Biondi, D., Zaznzotto, L. 2004a. Temporary networks in
polymer-modified asphalts. Polym Eng Sci. Vol. 44, Issue 12: 2185–2193.
Pérez-Lepe, A., Martínez-Boza, F.J., Gallegos, C. 2005. Influence of Polymer Concentration on the
microstructure and rheological properties of highdensity polyethylene (HDPE)-modified bitumen.
Energy Fuels 193:1148–52.
Rossi, D., Filippi, S., Merusi, F., Giuliani, F., Polacco, G. 2013. Internal structure of bitumen/polymer/
wax ternary mixtures for warm mix asphalts. J. Appl. Polym. Sci. 129: 3341–3354
Tschoegl, N.W., Emri, I. 1993. Generating line spectra from experimental responses, part II. Storage and
loss functions. Rheol Acta 32: 322–327.
Vlachovicova, Z., Wekumbura, Ch., Stastna, J., Zanzotto, L. 2007. Creep characteristics of asphalt mod-
ified by radial styrene–butadiene–styrene copolymer. Constr Build Mat. Vol. 21, Issue 3: 567–577.
Vonk, W., Hartemink, R. 2004 SBS-modified binders, also cost effective in hot climates. 8th Conference
on Asphalt Pavements for Southern Africa. ISBN 1–920–01718–6
Wasage, T.L.J., Stastna, J., Zanzotto, L. 2011. Rheological analysis of multi-stress creep recovery
(MSCR) test. International Journal of Pavement Engineering. 12: 561–568
Wloczysiak, P., Vidal, A., Papirer, E., Gauvin, P. 1997. Relationships between rheological properties,
morphological characteristics, and composition of bitumen–styrene butadiene styrene copolymers
mixes. I. A three-phase system. J. Appl Polym Sci. Vol. 65: 1595–1607.
Yusof, N.I.M., Mounier, D., Marc-Stéphane, G., Hainin, M.R., Airey, G.D., Benedetto, H. 2013. Mod-
elling the rheological properties of bituminous binders using the 2S2P1D Model. Constr and Build
Mat. Volume 38: 395–406
Zanzotto, L., Foley, D., Rodier, C.E., Watson, R.D. 1987. Modified Asphalts – Are You Really Coming?
Canadian Technical Asphalt Association Proceedings. Vol. 32: 92–117.
1184
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Chiara Riccardi
University of Pisa, Italy
ABSTRACT: Fatigue failure is one of the major distresses for asphalt pavements at inter-
mediate temperatures. Accurate assessment of fatigue properties of asphalt mastic is of great
importance for ensuring satisfactory performance and the structural integrity of asphalt
pavements. In this paper, Linear Amplitude Sweep and Time Sweep tests in strain controlled
mode are performed on two different asphalt mastics using the Dynamic Shear Rheometer.
A comparison between the two methods is presented, and the experimental results are ana-
lysed based on the viscoelastic continuum damage theory. The fatigue life, Nf, obtained from
the two test procedures returns significantly different values. This is most likely associated to
the type of failure induced by the two diverse loading modes, which is visualized in the spe-
cific crack growth patterns. Therefore, the LAS test should be intended as relative tolerance
tests for damage rather than an effective fatigue test.
1 INTRODUCTION
Fatigue cracking is one of the major distresses of asphalt pavement caused by the accumulation
of damage under traffic load repetition. The fatigue properties of asphalt mixtures are strongly
related with those of binders and mortars. For this reason in recent years different test methods
were proposed to study the fatigue performance of binders and with the objective of linking
them to the fatigue resistance of asphalt mixtures (Soenen et al., 2004). During the NCHRP
9–10 project (NCHRP Report 459 2001), the application of Dynamic Shear Rheometer (DSR)
time sweep tests in strain and stress controlled mode was proposed to evaluate the fatigue resist-
ance of binder with promising results. However, time sweep tests present a significant dura-
tion; for this reason, Johnson et al., 2010 introduced an accelerated procedure, called Linear
Amplitude Sweep tests (LAS) (AASHTO TP 101–14 2014) that consists on a frequency sweep
followed by an amplitude sweep in which the strain amplitude is rapidly increased during the
tests in order to achieve an accelerated damage progression in the binder sample.
Since the behavior of asphalt mixtures are greatly influenced by the mortar that bond
the coarse aggregate, fatigue testing on asphalt mortar is an interesting option which can be
potentially used in place of costly and time consuming experimentation on mixtures speci-
mens. In this paper time sweep and LAS tests were performed on mortars composed by dif-
ferent binders at a same volume percentage of two different aggregate particles and the Visco
Elastic Continuum Damage (VECD) approach was used to analyze the test results.
According to study by Johnson, 2010 the damage intensity (D) can be calculated at data point
N by summing the incremental damage for each data point i. using equation 1:
1185
α
N 1+ α 1
D(t ) ≅ ∑ ⎡⎣ΠI d 0
(| G* | sin
i i − | G * | si i )⎦
⎤ (ti ti 1 )1++α (1)
i =1
where: Id = initial undamaged dynamic shear modulus divided by a modulus of 1 MPa, γ0 = applied
shear strain amplitude, |G*| = dynamic shear modulus, δ = the phase angle and α = 1+1/m, where
m is the steady state log-log slope of the master curve of the complex modulus.
For a cyclic fatigue test, |G*| sinδ is linked to the damage intensity (D) through a power
law (Kim et al., 2006):
si δ
| * | sin C0 − C1 ( )C2 (2)
Substituting equation 2 into equation 3 and making the derivative respect to D, equation
4 is obtained:
dW
= −π I dC1C2 ( D )C2 −1 (γ max )2 (4)
dD
Considering the following model for damage growth in viscoelastic materials proposed by
Park et al. (1999):
α
dD ⎛ ∂W ⎞
= − ⎟ (5)
dt ⎝ ∂D ⎠
Introducing equation 4 into equation 5 and rearranging the equation the following fatigue
law equation can be derived:
f D f )k
N= (γ max )−2α (6)
k ( I dC1C2 )α
where N is the number of cycles to failure, k is equal to 1+(1-C2)α, f is the loading frequency,
Df is the damage accumulation at failure. In the present work according to the study by John-
son (2010) the parameter Df was chosen equals to the damage intensity corresponding to a
35% reduction in |G*|sinδ. Equation 6 can be simplified as shown by Kim et al., 2006:
Nf A(γ max )B (7)
Four different asphalt binders and two different types of aggregate particles were used in the
present study for preparing samples of asphalt mortar. The properties of the asphalt binders
are summarized in Table 1.
1186
Table 1. Properties of the asphalt binders.
Pen Grade 75 52 62 46
Ring and Ball Temperature 44 50 48 54
Performance Grade 58–22 64–28 58–18 64–22
Mixing fresh binders Type 1 and Type 3 with aggregate particles smaller than 0.149 mm,
having density 2.002 g/cm3, and binder Type 2 and Type 4 with the fine aggregate particles
with density 2.708 g/cm3, four different asphalt mortars were produced. All mortars pre-
sented a volume fraction of fine aggregate particles equals to 48.7% by the total mortar.
The entire set of mortars was tested using the DSR in the classical configuration of paral-
lel plate with 8 mm in diameter and 2 mm gap. Frequency and temperature sweep tests were
conducted in order to generate Master curves of complex modulus and phase angle of the
undamaged material in the linear viscoelastic range. From the master curve of the complex
modulus (Figure 1) is possible to determine the parameter α in Equation 1.
Time sweep tests in strain controlled mode were then performed at 25°C together with
LAS tests using the DSR in the same configuration.
Time sweep tests consist in applying a strain controlled cyclic loading up to 360,000 cycles
at a constant frequency of 10 Hz. Three different strain levels were used (0.5%, 1%, 1.5%) and
two replicates for each strain level were tested.
LAS tests (AASHTO TP 101–14 2014) consist in a frequency sweep tests, used to first
determine the undamaged linear viscoelastic properties, at 0.1% strain level over a range of
frequencies from 0.2 to 30 Hz, followed by an amplitude sweep at 10 Hz in strain controlled
conditions in which the strain level is increased during the test to accelerate damage in the
sample. The strain level is increased linearly from 0 to 30% over the 3100 cycles of loading.
4 RESULTS
Figure 2. Complex modulus versus time during time sweep at different strain levels.
Figure 3. |G*|sinδ versus D at different strain levels for the mortar type 1.
1188
Table 2. Time sweep fit coefficients.
Mortar C0 C1 C2 α
Figure 4. |G*|sinδ versus D for LAS tests for the different mortars.
Mortar C0 C1 C2 α
1189
tests, at the beginning the crack growth rate increases with crack length, while when a critical
depth is reached the crack growth rate decrease with increasing crack length. Indeed, in the
LAS tests in each step the crack growth rate decreases and this is not consistent with fatigue
crack propagation. This implies that the LAS test does not induce typical fatigue damage.
Nevertheless, LAS test can be used as a damage tolerance test rather than a fatigue test as
Hintz et al., 2013 have highlighted.
In the present work a comparison between the fatigue laws obtained using the two differ-
ent methods is also presented. In Tables 4 and 5 the parameters of the fatigue laws obtained
from time sweep tests and LAS tests are summarized, respectively. As shown, the value for
the estimated fatigue life for each material differs consistently; however, the fatigue ranking
of the materials is the same.
Moreover, plotting the predicted number of cycles from LAS tests versus the predicted
number of cycles from time sweep tests a strong linear correlation between the two test meth-
ods can be observed as shown in Figure 5.
Mortar A B Nf at 1% Nf at 10%
Type 1 3.32.107
−4.54 1.41.107
957.5
Type 2 5.46.107 −4.20 5.46.107 3445
Type 3 1.15.106 −4.82 1.15.106 17.4
Type 4 6.24.106 −4.42 6.24.106 237.24
Mortar A B Nf at 1% Nf at 10%
Figure 5. Number of cycles at failure predicted at 1% applied strain from LAS and time sweep tests.
1190
The correlation shown in Figure 5 is very promising and for this reason further investiga-
tion of the relationship between LAS and time sweep tests for asphalt mortar is needed.
In this study four different asphalt mortars were tested with the DSR using two different
methods: LAS tests and time sweep tests at three different strain levels. The VECD theory
was applied to the results of both the test methods. The characteristic curves derived from
time sweep and LAS tests do not overlap; however, a good correlation between the number
of cycles predicted with the two experimental approaches was found. The differences between
the two procedures are due to the different damage propagation and, in particular, to the dif-
ferent crack growth. In fact, in time sweep test an increase of the crack growth is observed at
the beginning and after reached a maximum point, that correspond to a critical depth of the
crack, the rate of the crack growth starts to decrease.
On the LAS tests the same trend was not found due to the very complex response that does
not allow a stable crack growth. On the basis of these results, the LAS tests can be referred
as a “damage tolerance” tests rather than a fatigue tests. Different fatigue laws were found
for the two test methods, but a good correlation between the number of cycles at failure pre-
dicted with time sweep and LAS tests was observed. Moreover, the application of the VECD
theory to time sweep tests at different strain levels generate characteristic curves that overlap
and this is an indicator of the possibility to define a unified fatigue approach.
REFERENCES
AASHTO TP 101-14 (2014). Estimating Damage Tolerance of Asphalt binders using the Linear Ampli-
tude Sweep. American Association of State Highway and Transportation Officials.
Bahia, H.U., Hanson, D.I., Zeng, M., Zhai, H., Khatri, M.A., and Anderson, R.M. (2001) Charac-
terization of Modified Asphalt Binders in Superpave Mix Design. NCHRP Report 459. National
Academy Press Washington, D.C.
Hintz, C., and Bahia, H. 2013. Simplification of Linear Amplitude Sweep Test and Specification Param-
eter. Journal of the Transport Research Board. Vol. 2370. DOI: 10.3141/2370–02.
Johnson, C. 2010. Estimating asphalt binder fatigue resistance using an accelerated test method. PhD
thesis, University of Wisconsin-Madison, Madison, USA.
Kim, Y.R., Lee, H.J., Little, D.N., and Kim, Y.R. 2006. A simple testing method to evaluate fatigue
fracture and damage performance of asphalt mixtures. Journal Assn. Asphalt Paving Technologists,
Vol. 75, 755–788.
Park, S., W., Kim, R.Y., and Schapery, R.A. (1996). A viscoelastic continuum damage model and its
application to uniaxial behavior of asphalt concrete, Journal of Mechanics of materials, Vol. 24, Issue
2, 241–255.
Soenen, H., De la Roche, C., and Redelius, P. 2004. Predict Mix Fatigue tests from binder fatigue prop-
erties, measured with the DSR. 3rd Euroashalt &Eurobitume Congress, Vienna.
1191
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Ying-Haur Lee
Department of Civil Engineering, Tamkang University, New Taipei City, Taiwan
Jui-Wen Chen
JW Eco-Technology Pavement, Ping-Tai, Ltd., New Taipei City, Taiwan
ABSTRACT: Pervious pavements are generally weak in material strength and can be easily
clogged. Thus, they were only used in sidewalks, squares, and parking lots, but not for vehicular
loads. An innovative eco-technology recently developed in Taiwan uses air-circulated aqueduct
frames and impervious Portland cement concrete on top of an aggregate (crushed stone or gravel
base) layer to form a structured permeable concrete pavement (called JW eco-technology pave-
ment) can alleviate such problems. The main features include: (1) The aggregate layer may serve
as a detention reservoir to increase flood control capabilities; (2) The stored water may lower
pavement surface temperature in summer and help to reduce heat island effects; (3) This eco-
technology can prevent water accumulation on pavement surface to improve pedestrian and driv-
ing safety; and (4) Enriching the ecological system. Through proper designs, the structured JW
pavement can possess adequate load bearing capacities as conventional concrete pavements do.
1 INTRODUCTION
Since surface and subsurface “water” has been treated as a primary enemy in engineering designs
over the years, the concept of traditional pavement design is generally based on providing surface
and subsurface drainage and preventing water infiltration and pumping. The concept of sustain-
able development was advocated by U.N. World Commission on Environment and Development
(WCED) in the report of “Our Common Future” in 1987. Impervious coverage of ground sur-
face due to urban development often results in heat island effect, flood, drought, air pollution,
haze and other environmental problems, which are the obstacles of sustainable development.
To mitigate and adapt to global warming, heat island effects, and other extreme climate
changes, the construction of green buildings has become very popular recently (IPCC, 2012).
As one of the criteria of green buildings, water-permeable pavements are highly recommended
by LEED (Leadership in Energy & Environmental Design, U.S. Green Building Council),
EEWH (Green Building Certification System in Taiwan, Taiwan Green Building Council)
and other similar green building evaluation systems in the world (USGBC, 2010). By treating
“water” as a valuable resource, on the other hand, a permeable pavement allows rain or snow
water sinking into the pavement structure layers so as to retain (or store) the water.
However, pervious pavements are generally not structurally sound, weak in material
strength, and can be easily clogged. Thus, they were only used in sidewalks, squares, and
parking lots, but not for vehicular loads. The primary objective of this study is to intro-
duce an innovative eco-technology recently developed in Taiwan which uses air-circulated
aqueduct frames and impervious Portland cement concrete on top of an aggregate (crushed
stone or gravel base) layer to form a structured permeable concrete pavement (called JW eco-
technology pavement) to alleviate such problems (Chen, 2004; Lee, et al., 2015).
1193
1.1 Conventional rigid pavement design concept
Traditionally, Portland cement concrete (Rigid) and asphalt concrete (Flexible) pavements
predominantly use impervious surface layers with roadside gutters to alleviate the effect of
rainfall. As compared to a flexible pavement, a rigid pavement, consisting of an impermeable
concrete slab (surface course), subbase, and subgrade soil, has stronger structural capability
and can distribute load stress to larger area (Huang, 2004).
The concept of rigid pvement design is generally based on providing surface and subsur-
face drainage and preventing water infiltration and pumping. The concrete slab thickness is
not highly influenced by subgrade strength (Huang, 2004). The primary reason of providing
a subbase layer is to prevent pumping, rather than solely to increase the structural strength of
a rigid pavement (Huang, 2004). A well-designed rigid pavement normally has higher bear-
ing capacity, longer service life, and lower maintenance fees. However, after several years of
service, as depicted in Figure 1, a rigid pavement may still have distresses such as faulting and
pumping due to the combination effects of loadings, climatic conditions, and water infiltra-
tion (FHWA, 2003).
1194
loads (Wu, 2005; Wang, et al., 2010). Figure 2 depicts some example distresses of pervious
pavements and block pavements.
Regardless of the fact that “water” is our major enemy and at the same time “water” can
also be treated as a valuable resource, the primary function of a permeable pavement is to
introduce “water” into the pavement structure layers. A real permeable pavement should be
designed to be “real permeable” for life if at all possible as depicted in Figure 3. Nevertheless,
porous material pavements may be permeable only on the surface and its permeability may
be greatly reduced with time. By allowing water permeating through the pavement layers and
storing in the underneath layers, these types of pavement may experience premature failures
due to the softening of subgrade soil. Besides, the performance of a pervious pavement is
much worse than that of a normal impervious material due the voids in the porous material.
1.3 Existing successful cases of permeable surface layer for heavy load
A permeable pavement is desired to have the following features, which can: (1) retain water
and provide storm water storage; (2) breathe and enrich the ecosystem; (3) distribute the
load to larger area; (4) sustain heavier vehicular load; (5) be maintenance free or require little
maintenance, in order to cope with global warming, heat island effects, and other extreme
1195
climate changes. From an engineering point of view, is it too much to ask from a permeable
pavement? Let’s first look at the following two existing successful cases of permeable surface
layer applications for heavy load.
The crawlerway at Kennedy Space Center (Wikipedia, 2014) as shown in Figure 4 is
designed to sustain 17 million pounds (7.7 million kg) of load, which may possibly be the
heaviest loading on record and the crawler speed is at less than 1 km/hr. The crawlerway
consists of the following layers: the top surface layer is 10∼20 cm river gravel, the base layer
is 120 cm graded crushed stone, the subbase layer is 76 cm select fill plus 30 cm compact fill,
and the subgrade soil. Note that the primary structural layers of this pavement are permeable
and at the same time they can distribute the heavy load to larger area, though water is not
retained in this pavement structure.
A conventional ballast track system, which consists of rail, crossties, ballast, subballast,
and subgrade soils, is another common example and a successful case of permeable surface
layer applications for heavy load. Also note that the surface layer is permeable, but the heavy
load of a train is still able to be distributed to larger area. In addition, water is not retained
in this ballast track structure. As shown in Figure 5, after several years of service or improper
designs, pumping might still be a problem. Thus, in order to reduce the maintenance frequen-
cies, a slab track (or ballastless track) system has gradually gained popularity (Huang, 2004;
Yen and Lee, 2007).
1196
2 JW ECO-TECHNOLOGY PAVEMENT
1197
1. Impervious Portland cement concrete (PCC) surface layer embedded with plastic air-
circulating aqueduct frames: The the impervious PCC layer can distribute heavy load to
larger area. After removing surface strips to reveal holes of aqueducts, which consist of
two types of tubes to provide high permeability and high breathability. In which, one type
of tubes with wide upper ends and narrow bottom ends is for water permeation and the
other type of tubes with wide bottom ends and narrow upper ends is for air circulation.
2. Aggregate (crushed stone or gravel) base layer: The aggregate base layer is used for water
retention or storm water storage and can also enrich the sub-layer ecosystem.
3. Compacted fill subbase and subgrade soil.
The main features of JW eco-technology pavement including flood control, environmental pro-
tection and enriching the ecological system, durability and safety are briefly discussed as follows:
1198
block pavements. Evidences indicated that the JW pavement can be successfully used in road-
ways subject to light traffic under normal geological conditions.
The structural design of JW pavement may include the determination of concrete slab thick-
ness and aggregate base/subbase thickness. Through proper designs, the structured JW pave-
ment can possess adequate load bearing capacities as conventional concrete pavements do.
Traditional concrete pavement thickness design approaches are proposed to be used as the
basic guidelines for determining the required slab thickness. General guidelines for determin-
ing the slab thickness of pervious concrete material pavements are also available from the
National Ready Mix Concrete Association (NRMCA, 2008) and the Portland Cement Asso-
ciation (Leming, 2007) and the American Concrete Pavement Association (ACPA, 2011). In
addition, the thickness of aggregate base layer can be determined as the thicker of the two
results based on the layer’s structural bearing and detention reservoir design requirements.
The following primary design factors are proposed in a rational structural design method
for load-bearing JW permeable pavements:
1. Traffic Loading: The total anticipated traffic loads during the design period may be
expressed as single, tandem, and tridem axle load spectra or 18-kip (80kN) equivalent
single axle loads (or ESALs).
2. Structural Layer Properties: Currently, there exist no standard test methods for character-
izing the compressive strength and flexural strength of pervious concrete slab. Thus, as an
interim design procedure, these design inputs are assumed to be comparable to those used
in the fatigue equations for conventional concrete pavement designs (ACPA, 2011). The
elastic moduli of the base and subbase materials may be considered accordingly.
3. Subgrade Soil Strength: The subgrade strength is normally expressed as the modulus of
subgrade reactions (k) for rigid pavements. Empirical correlations between k and other
tests including the soaked California Bearing Ratio (CBR), R-value, or resilient modulus
(Mr) are available (AASHTO, 1993). An effective or composite modulus of subgrade reac-
tion (composite k) may be used to determine the support provided by the subgrade based
on roadbed soil resilient modulus, subbase elastic modulus, and subbase thickness.
4. Other Factors: Other environmental factors including freezing climates and extended sat-
uration of the soil subgrade should also be considered.
JW structural permeable pavements are much stronger than traditional porous material
pavements and block pavements. JW concrete slab can be as strong as traditional concrete
material, with compression strength of 510 kgf/cm2 (50.0 MPa) and tensile strength of 74 kgf/
cm2 (7.3 MPa) (Li, 2004). Traditional rigid pavement design approaches can be used as basic
design guidelines. As compared to JW eco-technology pavements, porous material pavements
and blocked pavements may require much thicker surface and base/subbase thicknesses.
Full-scale testings of the structural behavior of permeable pavements on open-graded bases
are yet to be investigated to better understand the relations between load and deformation.
Conservative values (i.e., AASHTO layer coefficients) are suggested for open-graded base
and subbase aggregates in permeable pavement design. The adverse effect of rigid permeable
pavement due to the saturation of subgrade soil is not as severe as that of flexible permeable
pavement, which is particularly true for JW eco-technbology pavement. Nevertheless, the
soil underneath the aggregate (or crushed stone) layer needs further investigation, especially
when it is suspicious to settlement or the loss of shear strength due to wetting.
The stability of subgrade soil under traffic load should be carefully reviewed. The low-
est saturated soil strength is recommended for its structural design for vehicular applica-
tions, i.e., generally with a minimum CBR = 4%, R-value = 9, Mr = 6,500 psi (45 MPa), or
k-value = 100 pci (27.1 kPa/mm). Besides, the subgrade soil needs to be compacted to at least
95% of standard Proctor density. If the underlying soil has a low CBR value of less than 4%,
special treatments and relatively thicker base/subbase thicknesses may be needed. Soil com-
1199
paction required to achieve the minimum strength may reduce the permeability of the soil.
Thus, the infiltration rate of soil should be assessed at the required compaction density.
This eco-technology pavement has also been accepted by our permeable pavement con-
struction specifications for non to light traffic conditions. However, there are still challenges
ahead such as to better understand its failure mechanisms, structural capabilities in terms
of allowable load repetitions, and other long term performance indices if the structured JW
pavement should be used for normal or heavier traffic loading conditions.
A series of 3-D finite element analyses were performed using the PLAXIS program. The
results indicated that with relative thick concrete surface and aggregate base, the immediate
deformation of this pavement under a truck load is very low (1-mm or less). Research efforts
are still underway to develop a systematic thickness design approach for JW eco-technology
pavement.
5 CONCLUSIONS
To mitigate and adapt to global warming, heat island effects, and other extreme climate
changes, the construction of green buildings and water-permeable pavements has become
very popular recently. However, pervious pavements are generally not structurally sound,
weak in material strength, and can be easily clogged. Thus, they were only used in sidewalks,
squares, and parking lots, but not for vehicular loads. An innovative JW eco-technology
pavement recently developed in Taiwan is introduced in this paper to alleviate such problems.
The JW pavement uses air-circulated aqueduct frames and impervious Portland cement con-
crete on top of an aggregate (crushed stone or gravel base) layer to form a structured perme-
able concrete pavement, which has the following features:
1. Flood Control: The aggregate (crushed stone or gravel) layer can store up to 50% of the
volume of pervious materials. The pervious base/subbase layer may serve as a detention
reservoir for regular rainfalls, which may also be connectted to other detention systems
(e.g., ditches, ponds, or water storage systems) to increase flood control capabilities.
2. Environmental Protection and Enriching the Ecological System: The stored water may
lower pavement surface temperature in summer and help to reduce heat island effects.
The air-circulating aqueducts can cool the ambient air temperature through releasing the
underneath cold air and vaporing the underground water in summer. Literature found
that exhausts from vehicles may be absorbed by the pavement system and become the
nutrients of an underground ecological system. Rainwater can flow into while being fil-
tered by the sub-layers, which may help to activate soil and create an underground wetland
ecological system.
3. Durability and Safety: The JW pavement has higher durability and hence greatly reduces
future maintenance activities and costs. It can also prevent water accumulation on roads
or sidewalks and reduce the formation of fog when raining to improve drive safety or
pedestrian safety.
The material strength of JW concrete pavement is as strong as that of traditional concrete
material. Traditional rigid pavement design procedures may be used as interim design guide-
lines. Research efforts are still underway to develop a systematic thickness design approach
for JW eco-technology pavement.
JW pavements generally outperform many other pervious pavements or block pavements.
Evidences indicated that the JW pavement can be successfully used in roadways subject to
light traffic under normal geological conditions. This eco-technology pavement has also
been accepted by our permeable pavement construction specifications for non to light traffic
conditions. However, there are still challenges ahead such as to better understand its failure
mechanisms, structural capabilities in terms of allowable load repetitions, and other long
term performance indices if the structured JW pavement should be used for normal or heav-
ier traffic loading conditions.
1200
REFERENCES
AASHTO. 1993. AASHTO guide for design of pavement structures. American Association of State
Highway and Transportation Officials, Washington, D.C.
ACPA. 2011. PerviousPave: background, purose, assumptions, and equations. American Concrete Pave-
ment Association. http://www.acpa.org/perviouspave
Chen, J.W. 2004. Maintenance and improvement in the ecological environment of the island: The design
and application in high bearing structure pervious pavement. The International Conference on the
Islands of the World VIII, Kinmen Island, Taiwan, November 1–7.
Fan, L., Wang, S., Chen, C., Hsieh, H., Chen, J., Chen, T. & Chao, W. 2014. Microbial community struc-
ture and activity under various pervious pavements. J. Environ. Eng., 140(3), 04013012.
FHWA. 2003. Distress identification manual for the long-term pavement performance program. Publica-
tion No. FHWA-RD-03-031, U.S. Department of Transportation, Federal Highway Administration,
Washington, D.C.
Huang, Y.H. 2004. Pavement analysis and design. Second Edition, Pearson Education, Inc.,
New Jersey.
IPCC. 2012. Summary for policymakers. In: Field, C.B., Barros, V., Stocker, T.F., Qin, D., Dokken,
D.J., Ebi, K.L., Mastrandrea, M.D., Mach, K.J., Plattner, G.K., Allen, S.K., Tignor, M. & Midgley,
P.M. (eds) Managing the risks of extreme events and disasters to advance climate change adaptation. A
special report of working groups I and II of the Intergovernmental Panel on Climate Change (IPCC).
Cambridge University Press, Cambridge, UK/New York, 1–19.
Lee, Y.H., Chou, N.S. & Chen, J.W. 2015. Applications and challenges of an innovative load-bearing
permeable pavement developed in Taiwan. 5th Annual Workshop of the International Association of
Chinese Infrastructure Professionals (IACIP) in Conjunction with the 94th Transportation Research
Board Annual Meeting, Washington, D.C., January 11–15. (Invited Speech)
Leming, M.L., Malcom, H.R. & Tennis, P.D. 2007. Hydrologic design of pervious concrete. Portland
Cement Association, Skokie, IL.
Li, W.F. 2004. Research project for JW precautionary air-cycle aqueduct pavement method. Taiwan
Construction Research Institute. (in Chinese)
Liu, C.M., Chen, J.W., Tsai, J.H., Lin, W.S., Yen, M.T. & Chen, T.H. 2012. Experimental studies of the
dilution of vehicle exhaust pollutants by environment-protecting pervious pavement. Journal of Air
and Water Waste Management, 62:1, 92–102.
Liu, C.M., & Hsieh, Y.S. 2013. Build sponge Taiwan initiative. http://eqpf.myweb.hinet.net/download/
2.pdf (in Chinese)
Liu, C.M., Chen, J.W., Hsieh, Y.S., Liou, M.L. & Chen, T.H. 2015. Build sponge eco-cities to adapt
hydroclimatic hazards. Handbook of Climate Change Adaptation, Eds. Walter Leal Filho, Springer-
Verlag Berlin Heidelberg, 1997–2009.
NRMCA. 2008. Structural design considerations (of pervious concrete pavements). National Ready
Mixed Concrete Association. Retrieved from http://rmc-foundation.org
Titi, H., Rasoulian, M., Martinez, M., Becnel, B. & Keel, G. 2003. Long-Term Performance of Stone
Interlayer Pavement. Journal of Transportation Engineering, ASCE, 129(2), 118–126.
USGBC. 2010. LEED 2009 for new construction and major renovations rating system. U.S. Green Build-
ing Council (USGBC) member approved November 2008 and updated October 2010. U.S. Green
Building Council: Washington, D.C.
Yen, S.T. & Lee, Y.H. 2007. Parameter identification and analysis of a slab track system using 3D
ABAQUS program. Journal of Transportation Engineering, 133(5), ASCE, 288–297.
Wang, D.C., Wang, L.C., Cheng, K.Y. & Lin, J.D. 2010. Benefit analysis of permeable pavement on
sidewalks. International Journal of Pavement Research and Technology, 3, 207–215.
Wikipedia. 2014. Crawlerway – Wikipedia, the free encyclopedia. Retrieved on 2014/02/14 from http://
en.wikipedia.org/wiki/Crawlerway.
Wu, C.S. 2005. Analysis of the environmental benefits of the permeable pavements. Master Thesis, Gradu-
ate Institute of Engineering, National Central University, Taiwan. (in Chinese)
1201
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Yongsheng Guan
Jiangsu Sinoroad Engineering Research Institute Co. Ltd., Nanjing, China
Tongji University, Shanghai, China
ABSTRACT: This paper adopted the compound modification method by using multi-chain
polyolefin and SBS to improve the workability and extend the construction temperature range of
ultra-thin wearing course materials. Therefore the dense graded Ultra-thin Pavement (U-PAVE)
mixture was designed with the nominal maximum aggregate size 10 mm. Based on laboratory
results of volumetric properties and pavement performance, design parameters and evaluation
indexes of U-PAVE10 were proposed. Meanwhile, the attenuation law of texture depth was com-
paratively studied via modified wheel tracking test. Besides, the Computed Tomography (CT)
scanning method was employed to investigate the void distribution law inside U-PAVE10 sam-
ples cored from test roads, and bonding strength with the lower layer was measured by pull out
test. The U-PAVE10 exhibited excellent pavement performance during laboratory and field tests.
Compared with traditional ultra-thin wearing course materials, the attenuation speed of texture
depth for U-PAVE10 was much slower and its bonding effect was better.
1 INTRODUCTION
Ultra-thin paving technique was originated from France in the late 1970s to recover the slid-
ing resistance of pavement, and could be also applied in the construction of newly built roads.
This technique had got widely used in European countries since emergence, and in China
researchers started to study ultra-thin paving technique at 1990s. At present, ultra-thin paving
materials are faced with several problems during the application process, including low degree
of compaction, moisture damage and rapid attenuation of sliding resistance at a later stage.
Those problems could be attributed to the fact that, the thickness of ultra-thin paving materi-
als is usually from 2 to 2.5 centimeters, thus the temperature cools too quickly to ensure degree
of compaction. When it comes to open-graded ultra-thin paving material, it was not suit-
able for milling and repaving method, because the void ratio was increased for open-graded
materials. Thus the water could easily permeate to lower the bonding strength between layers,
causing raveling and reducing the durability of ultra-thin paving materials. Besides, traditional
ultra-thin paving materials only possess general rutting resistance, and its texture depth and
sliding resistance decayed rapidly, resulting potential risks for vehicles’ safety.
In view of the existed problems of current ultra-thin technique, a dense-graded ultra-
thin pavement, namely U-PAVE10, was developed in this study. The U-PAVE10 mixture
was modified by compound modification method with multi-chained polyolefin modifier
and SBS, extending the temperature range for construction, improving the workability and
bonding effect between layers and improving the integrity of pavement structures as well.
Furthermore, through modified wheel tracking test, the attenuation mechanism for the
surface of U-PAVE10 at high temperature was investigated. Finally, samples of U-PAVE10
and SMA-10 mixtures cored from practical engineering projects were comparatively stud-
ied to evaluate the bonding strength by CT scanning method and pull-out tests.
1203
2 DESIGN OF U-PAVE10 MIXTURE
2.1 Materials
To improve the performance of asphalt mixture, despite PG-70 SBS modified asphalt, basalt
aggregate and limestone powders, U-PAVE10 material was added with compound synergistic
modifier. The properties of each material were listed in Table 1 and Table 2.
Sieve size (mm) 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075
Gradation envolope (%) 100 90∼100 40∼54 20∼36 16∼30 10∼24 7∼20 6∼12 4∼8
Designed gradation (%) 100 99.6 42.0 31.2 21.8 17.0 11.6 8.9 7.0
1204
could be cooled down from 160°C to 80°C in 5∼10 minutes, while the cooling rate was slowed
down with paving thickness increased. Therefore, the construction temperature should be
varied according to the thickness of paving course, and the compaction temperature was not
constant in laboratory mixture design process.
The design method for U-PAVE10 mixture was followed by those procedures:
1. The design void ratio was set between 5∼7%
2. Considering the thickness as well as the temperature T1 at the bottom of ultra-thin wear-
ing course after first compaction, and the heating temperature T2 for SBS modified
asphalt, the compaction temperature T3 was computed by equation T3 = (T1+T2)/2
3. Based on designed compaction temperature and void ratio, the estimated asphalt content
was calculated accordingly
4. The dense graded U-PAVE10 mixture was designed according to Marshall Design Method.
Finally, based on the 2.5 cm thickness for ultra-thin wearing course, the compaction tem-
perature was determined as 145°C in laboratory for this study, and asphalt content was 5.1%.
Asphalt Maximum
content theoretical Bulk gravity Stability Flow value
(%) density g/cm3 g/cm3 VV (%) VMA (%) VFA (%) (KN) (0.1 mm)
1205
graded asphalt mixture since opening to traffic. During the wheel tracking test, the relative
location for aggregates and asphalt mortar would change. Thus the TD at the surface of rut-
ting position would decrease if unstable asphalt mortar rose above. In this paper, the attenu-
ation mechanism for ultra-thin U-PAVE10 wearing course was investigated by observing the
TD variation at rutting position.
The principle of this test was that, the TD of each track plate was measured before wheel
tracking test. Then strip specimens were cut along with tracking positions and its TD was
obtained afterwards. By comparing the TD before and after wheel tracking test, the effects
caused by floating of asphalt mortar under traffic loads was studied.
In Chinese specification (T0719–2011) for wheel tacking test for dynamic stability, the
duration was 60 minutes. However, when the test time was too short, it was hard to observe
the attenuation for TD on the surface of mixture. Hence, the test duration was extended to
180 minutes, and test procedures were introduced as follows:
1. The board specimen for wheel tracking test was roll-compacted with dimensions of
300 mm × 300 mm × 50 mm, and the original TD K0 on the surface was tested with sand-
laying method, as shown in Figure 2.
2. The silica sands on the surface were washed away by water, and then wheel tracking test
was conducted in accordance with specification T0719–2011 under 0.7 MPa traffic loads,
but the test duration was extended to 180 minutes.
3. The strip specimen was cut from board specimen for wheel tracking test along with load-
ing positions.
4. The mass m1 of strip specimen was weighted after drying, as shown in Fig. 3.
5. Sands were laid on the surface of strip specimen as shown in Fig. 4 to ensure that the
whole surface texture was fully filled with sands, and extra sands were cleaned away with
a tiny brush.
6. The mass m2 for sand and specimen in total was weighted as shown in Fig. 5.
7. The density for standard sands was measured as ρG .
8. The area s of the strip specimen was calculated after measuring its length and width.
Using Equation 1 the TD for loading positions after wheel tracking test was determined
as K1:
Figure 2. Measure the TD K0 of wheel tracking Figure 3. Measure the mass m1 for
board specimen. strip specimen.
Figure 4. Lay sand on the surface of strip Figure 5. Weigh the total mass m2 for both
specimen and strip specimen. sand and strip specimen.
1206
m2 m1
K1 = (1)
ρG ⋅ s
By comparing K0 and K1, the degree of for TD on the surface was obtained after wheel
loads for 3 hours.
Compound Dynamic
modifier Asphalt stability Rutting depth Rutting depth
Mixture’s type content (%) content (%) (times/mm) after 1 hour (mm) after 3 hours (mm)
1207
Figure 6. The attenuation for TD of different mixtures.
Figure 7. Relationship between rutting depth and attenuation ratio for TD.
Thus it was obvious that adding multi-chained polyolefin modifier was an effective method
to slow down the attenuation rate for TD of U-PAVE10 mixture.
For the purpose of investigating the effects of high temperature performance of asphalt
mixture on attenuation for TD, correlation analysis was conducted between rutting depth
after 3 hours and attenuation ratio for TD, and the relationship was depicted in Figure 7.
From Figure 7, it could be observed that the attenuation ratio decreased with the increase
of rutting resistance, indicating that the attenuation rate for mixture could be slowed down to
some extent through improving the high temperature stability. Moreover, the range for filler
to bitumen ratio could be enlarged for dense graded asphalt mixture by improving its high
temperature stability.
3D-XY-Pixel Size,
Type X axis/mm Y axis /mm Z axis /mm 3D-Z-Pixel Size/mm
Figure 10. The scanning image for the interface betweenU-PAVE10 and original pavement.
X-ray after transmission was gathered by detector to rebuild 2D or 3D grey level image. The
scanned specimen was standard cylindrical sample cored from pavement and was cut flat at
each side by cutting machine. The test condition was that Voltage: 200 kV: Current: 0.43 mA:
Integration Time: 300 ms.
The cored U-PAVE10 and SMA10 samples were put up and down overlapped and scanned
by CT, and the 3D rebuilt image was shown in Figure 9.
The CT scanning result for U-PAVE10 was shown in Figure 10.
From the top view of the interface of 2 mixtures, the size and distribution of the voids
for U-PAVE10 mixture was relatively uniform. While for SMA10 mixture, its void size and
distribution was inhomogeneous. Using CT scanning technique and image analysis method,
the distribution mechanism of aggregates, voids, and cracks inside asphalt mixture could
1209
Figure 11. The scanning image for the interface between SMA10 and original pavement.
be precisely located. Through grey level processing of the image and using Origin software
to calculate the void ratio on the interface based on grey level analysis, the grey level versus
probability density distribution curve was depicted in Figure 12. The grey level value for void
was usually less than 30, therefore void ratio was calculated by integral from 0 to 30. The
relationship between air voids and height from up to down of those 2 paving structures was
depicted in Figure 13. The void ratio for interior structure and interface layer of 2 mixtures
as listed in Table 9 and Figure 14.
From Figure 13, it could be easily seen that the air void for the surface layer of U-PAVE10 was
smaller than that of SMA10 mixture. The thickness of SMA13 and U-PAVE10 was 26 mm and
28 mm respectively. From Table 9 and Figure 14, the void ratio in the interface for U-PAVE10 mix-
ture was smaller than SMA13 by 1.2%. This could be attributed to the fact that modified emulsi-
fied asphalt tack coat layer was sprayed between U-PAVE10 layer and original pavement, and
then the emulsified asphalt was reheated and softened when paving U-PAVE10, thus to make the
contact interface more compact between U-PAVE10 layer and original pavement.
On the other hand, because of the properties of dense gradation and compound modifier,
U-PAVE10 mixture could be easily compacted and without separation. Besides, the void
ratio inside asphalt mixture for U-PAVE10 was smaller than SMA10 for 0.8%. Consider-
ing the gap gradation for SMA10 mixture, the content for coarse aggregates was relatively
higher than U-PAVE10, thus the interior structure was formed by point to point contact by
aggregates. While for U-PAVE10 mixture, the aggregates could form a compact integrate by
forming an interlock structure.
1210
Figure 13. The relationship between air voids and height for those 2 paving structures.
Table 9. The void ratio for interior structure and interface layer of 2 mixtures.
Figure 14. The void ratio for interior structure and interface layer of 2 mixtures.
Figure 15. Field pull out test for U-PAVE10 and SMA10 mixture.
Mixture type Pull out force (kN) Pull out stress (Mpa) Temperature (°C)
1211
ers, were smaller than SMA10 mixture, proving its better bonding strength and compaction.
In all, compared with traditional gap-graded SMA10 mixture, the designed U-PAVE10 mix-
ture exhibited superior rutting resistance, impermeability and structural integrity.
5 CONCLUSIONS
REFERENCES
Zeng M, Peng L, Wu C, et al. 2012. Experimental Study of the Performance of Ultrathin Asphalt Fric-
tion Course. Journal of Wuhan University of Technology 04: 27–31.
Xiao Y, Wang Y, Wu S, et al. 2015. Assessment of bonding behaviours between ultrathin surface layer
and asphalt mixture layer using modified pull test. Journal of Adhesion Science and Technology
29(14): 1508–1521.
Pereira, D.D.S., Balbo, J.T., Khazanovich L. 2006. Theoretical and field evaluation of interaction
between ultra-thin whitetopping and existing asphalt pavement. International Journal of Pavement
Engineering 7 (4): 251–260.
Satoshi T, Jun O, Masayuki K. 2014. A study on characteristics evaluation to control quality of asphalt
mixture using X-ray CT. Road Materials and Pavement Design 15 (4):892–910.
Yang Y, Shen Y, Gao X. 2014. Analysis Interlayer Shear Stress of Ultra-Thin Wearing Course Consider-
ing Temperature and Different Interlayer Contact Conditions. Applied Mechanics and Materials 505:
102–105.
Xiao J, Xia A, Jiang W, et al. 2013. Air Void Features of Cement Asphalt Emulsion Mixtures from
Microcosmic-test. Journal of Tongji University (Natural Science) 09: 1398–1403.
Tan Y, Yao L, Wang H, et al. 2012. Performance evaluation indicator of ultra-thin wearing course
asphalt mixture. Journal of Harbin Institute of Technology 12: 73–77.
Rahaman F, Musty H Y, Hossain M. 2012. Evaluation of Recycled Asphalt Pavement Materials from
Ultra-Thin Bonded Bituminous Surface. Proceedings of Geo Congress: 25–29.
Zhang Y, Verwaal W, van de Ven M.F.C, et al. 2015. Using high-resolution industrial CT scan to detect
the distribution of rejuvenation products in porous asphalt concrete. Construction and Building
Materials 100: 1–10.
Liu T, Zhang X, Li Z, et al. 2014. Research on the homogeneity of asphalt pavement quality using X-ray
computed tomography (CT) and fractal theory. Construction and Building Materials 68: 587–598.
Satoshi T, Keiichiro O, Jun O, et al. 2013. A study on quality evaluation for bituminous mixture using
X-ray CT. Frontiers of Structural and Civil Engineering 7 (2): 89–101.
Yang X, Ling J, Wu C, et al. 2013. Evaluation of Performance of Novachip Ultrathin Asphalt Friction
Course. Highway Engineering 01: 9–13.
Praticò F G, Vaiana R. 2015. A study on the relationship between mean texture depth and mean profile
depth of asphalt pavements. Construction and Building Materials 101: 72–79.
1212
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
G. van Bochove
Heijmans, Rosmalen, The Netherlands
1 INTRODUCTION
Asphalt concrete is the most widely used structural material in the transportation industry.
It is chosen for highways, runways and parking lot constructions based on the special needs
of each structure in order to provide the desired mechanical and driving performance. Apart
from the mechanical performance, nowadays emphasis is giving on developing infrastructure
materials with new functionalities reducing also the environmental footprint. In this context,
a lot of researches has been done to develop electrically conductive asphalt concrete (CAC)
mixes for various purposes. Deicing of pavements (Minsk, 1968), acceleration of asphalt
healing with induction energy (Liu et al. 2010), (Garcia et al. 2013), (Liu et al. 2012) self-
sensing of pavement (Liu & Wu, 2009), (Wu et al. 2006) or reinforcing asphalt mixes (Serin
et al. 2012), (Wu et al. 2005), (Kandhal & Mallick, 2001), (Lee et al. 2005), (Huang & White,
1996) are typical examples. However, behavior under aggressive environmental conditions
like the impact of chloride on the mechanical performance of CAC is unknown, while pos-
sibly (especially for mixes with higher voids content) the electrochemical reactions of metallic
particles within the mixes can cause deterioration to the whole mix.
The concept of asphalt healing through induction heating has been developed in order
to restore the mechanical properties of asphalt concrete mixes, while in service. The induc-
tion heated asphalt mixes are produced by mixing in conductive particles. The asphalt
mortar in these mixes can be heated under a time-variable magnetic field, when micro-
cracks are expected to occur between the stone. According to Faraday’s law, this magnetic
field induces currents (eddy currents) in the conductive particles within the mix and they
are heated based on the principles of Joule’s law, Figure 1. The generated heat in the par-
ticles increases locally the temperature of the mortar including the bitumen, the micro-
cracks are healed through bitumen diffusion and flow and the mechanical properties are
recovered.
1213
Figure 1. Infrared image during induction heating of an asphalt pavement (A58 near Vlissingen,
the Netherlands, top figures) and the schematic of induction healing of asphalt concrete mix (bottom
figures).
Many experimental studies were conducted in the past to evaluate the effect of differ-
ent conductive particles on the effective electrical conductivity of asphalt mixes. The electri-
cal conductivity which is the most important factor in controlling the heating efficiency of
asphalt concrete mixes using induction power depends on the volumetric concentration and
the type of mixed conductive particles within the asphalt concrete. These particles are nor-
mally divided into categories according to their size such as filler-sized (e.g., graphite, carbon
black, iron powder) (Wu et al. 2005), (Huang et al. 2006), (Park et al. 2014), stone-sized (e.g.,
steel slag) (Ahmedzade & Sengoz, 2009) particles and fiber-type (e.g., steel and carbon fibers)
(Wu et al. 2005), (Garcia et al. 2009).
In addition, focus has been given on developing cost efficient ways to increase the conduc-
tivity of asphalt mixes without degrading the mechanical properties. Previous research has
been reported that the addition of fiber-type particles in asphalt mixes is the most effective
way to enhance the electrical conductivity of mixes improving simultaneously their dura-
bility (Mahrez et al. 2003). However, mixing fiber-type particles affects the distribution of
these particles and the other constituents of asphalt mixes, the porosity and subsequently
mixes mechanical performance. When fiber particles are inhomogeneously distributed within
asphalt concrete mixes, clusters of fibers are formed leading to higher air void contents of
mixes. The most crucial material parameters of conductive fiber-type particles in order to
acquire homogeneous distribution of particles and good dispersion of all mix components
in the asphalt mixes are ranked in: (i) the amount, (ii) the diameter and (iii) the initial length
of fibers (Garcia et al. 2014).
In contrast to the above studies, the performance of CAC exposed to very aggressive envi-
ronments is unexplored, apart from some efforts in the field of studying the sensitivity to
moisture damage (Ahmedzade & Sengoz, 2009). It is believed that bitumen can fully coat
aggregate and conductive particles to prevent metallic corrosion and the possible negative
effects of corrosion on asphalt pavement structures (Jones, 1996). The chloride damage of
1214
steel fibers in heterogeneous mixes like asphalt starts mainly from the external surface of steel
fibers with oxidation of the exposed areas. The oxidation process is called anodic reaction
and the mechanical degradation of asphalt mixes can be translated to the formation of non-
metallic areas in steel fibers.
In this paper, the possible chloride induced damage on a stone mastic asphalt (SMA)
type CAC with conductive particles is presented. The mixes type SMA are gap-graded
Hot-Mixed Asphalts (HMAs) developed to maximize rutting and cracking resistance and
through the years has been approved cost-effective pavement surface material with wide
applications in Europe and the USA (European Asphalt Pavement Association, 1998)
(Rockliff, 1996) (National Asphalt Pavement Association, 2002). Herein, a corrosion con-
ditioning protocol was utilized to evaluate the chloride corrosion damage and the impact
of chloride on the induction heating efficiency of SMA-CACs. The Non-Steady-State
Chloride Migration (NSSC) test method was used to accelerate the migration of chlo-
ride ions into the conductive particles within the asphalt mix (NT BUILD 492, 1999),
(Stanish et al. 2000). The chloride induced damage was determined via performing the
indirect tensile strength after different chloride conditioning exposure times. The influence
of chloride on the induction heating efficiency of different conductive asphalt mixes at
different conditioning levels was also investigated. The experimental process of damaging
and measuring the structural and non-structural performance of an SMA-CAC mix will
be demonstrated.
Based on the migration capability of chloride ions, an external electric potential is applied
axially across the CAC mix to accelerate the migration of ions into the material from out-
side. The velocity of ions transport is proportional to the strength of the electric field, the
charge and the size of the ions. The analysis of chloride migration is based on the Nernst-
Planck equation for a chloride mono-ion system assuming, a semi-infinite homogeneous
medium (Lizarazo-Marriaga et al. 2012). In the case of a chemical potential or concentra-
tion gradient and electric field, the flux for ionic transport can be expressed according to
formula (1):
∂c zF ∂E
J D − Dc (1)
∂x RT ∂x
where D is the diffusion coefficient of species, z is the electric charge, F is the Faraday con-
stant (9.65×104 Coulomb/mol), R is the gas constant (8.31 J/mol/oK), T is the absolute tem-
perature (K), E is the electric voltage (V), c is the ionic concentration of the species in the
pore fluid (mol/m3), and × is the distance from the surface of the sample (m).
As electro-migration tests they are carried out in laboratory facilities and the studied time
is limited to a few days; the chloride migration test assumes that the diffusion term of the
Nernst-Planck equation does not provide an important contribution to the total flux and can
be considered to be negligible. Thus:
zi F ∂E
JM = Di c(x,t ) (2)
RT ∂x
It is important to note that the temperature has more impact on chloride migration than
the diffusion mechanism. From the above relationship the ion movement under an electric
field and a concentration gradient is described as:
RTu
T i
Di = (3)
zi F
1215
Table 1. Composition of the SMA mix (AC) and the same mixes with particles
Composition (%)
In this study, two SMA-CAC mixes were produced and tested. The mixes differed in terms of
the combination of conductive particles. The aggregates consisted of crushed stone (density
2734 kg/m3), crushed sand (size between 0.063 mm and 2 mm and density 2697 kg/m3) and
mineral fillers; weak limestone (WL) and produced limestone (PR) with density 2780 kg/m3
and 2698 kg/m3. Cariphalte XS, an SBS modified bitumen with density 1030 kg/m3 was used
in the SMA mix with 6% air voids.
Table 1 shows the composition of the SMA mix without particles (CA) and the mixes
(respectively CAC.a and CAC.b) in which steel fibers (SF) and iron powder (IP) were added.
For CAC.a steel fibers (4% by total volume of bitumen) with density 7656 kg/m3, initial length
2.5 mm and diameter 0.083 mm were mixed with the asphalt mixes. For CAC.b, both steel
fibers and iron powder (15% by total volume of bitumen) with density 7507 kg/m3 were added
as the iron powder was considered a filler-sized conductive particle and replaced part of the
mineral filler. The added part of iron powder was 15% by total volume of bitumen. Figure 2
presents the different filler-sized particles and steel fibers which are used in this study.
Cylindrical samples, 150 mm in diameter and 110 mm in height, were produced using the
Superpave Gyratory Compactor (SGC). Samples with a height of 50 mm were cut from the
centre of the gyratory specimens and conditioned in the non-steady state chloride migration
test (NSSC Migration), after which the indirect tensile test (ITT) and the induction test were
done. In total 30 samples were prepared of which 12 samples were for CAC.a with 4% of steel
fibers, 12 samples for CAC.b with 4% of steel fibers and 15% of iron powder and 6 samples
for AC without conductive particles. The amount of aggregates and bitumen maintained
constant for all samples.
Chloride migration tests were performed after different exposure times for CAC mixes with
different combinations of conductive particles. The test method of Non-Steady-State Chlo-
ride Migration (NSSC) was used to accelerate the migration of chloride ions into the conduc-
tive particles within the asphalt mix, see Figure 3. After different chloride conditioning time
intervals, the chloride induced damage was determined via performing the indirect tensile
strength test. The impact of chloride migration on the induction heating performance of
CAC was determined as well.
The NSSC method requires a migration cell with cylindrical specimens of 100 mm diameter
and thickness of 50 mm. The surfaces of specimens should be dry and cleaned from any burrs.
For preconditioning of specimens before the migration test, they should be placed in the vacuum
1216
Figure 2. SEM images of filler-sized particles and fibers: mineral fillers WL & PR and IP & SF.
container for vacuum treatment, where a pressure of 10–50 bars (1–5 kPa) is applied for three
hours. At the end of preconditioning, the specimens are kept in the solution for 18 ± 2 hours.
Then, the samples are placed in a catholyte solution of 10% NaCl by mass in tap water (100 g
NaCl in 900 g of water). The concentration of NaCl in water was chosen following the process
recommendations of NSSC method. The anolyte solution was of the same NaCl proportion
with the catholyte solution and specimens and solutions were stored at 20 °C during the test.
According to the specification of the NSSC method (NT BUILD 492, 1999), the accelera-
tion of chloride migration is achieved by applying an external electric potential. The potential
is applied across the sample in order to force the chloride ions to ingress into the samples
and it can be adjusted to certain levels in order to reach the desired migration. The level of
migration of chloride is also related with the duration of NSSC test. Particularly, the recom-
mended duration and the electric potential of this test for cement concrete range from 6 to
96 hours and 10 to 60 V, respectively. However, the selection of the proper combination of
time and potential is dependent on the recorded initial current through the samples after
applying voltage of 30 V. Considering the fact that the chloride damage is unknown for
CAC mixes and subsequently of conductive asphalt pavements, constant voltage of 30 V was
used for all the samples and the analysis performed at two different time durations of 5 and
10 days. The set-up of the migration test is shown in Figure 3.
The indirect tensile strength test was used to get an indication of the effect of chloride
induced corrosion of conductive particles on the mechanical performance of asphalt mixes.
In this research, this test method was selected for the mechanical characterization of condi-
tioned mixes taking into account the previous successful experience of utilizing the same test
for assessing the impact of moisture on the performance and durability of asphalt concrete
mixes (Varveri et al. 2014). The indirect tensile strength test (ASTM D6931, 2012) was per-
formed at a test temperature of 20 °C in a displacement controlled mode at a rate of 50 mm
per minute. The indirect tensile strength of the asphalt samples was obtained using:
2F
T =
ITS (11)
π ttd
where F is the applied load, d and t are the diameter and thickness of the specimen
respectively.
1217
Figure 3. Schematic depiction of chloride migration test (middle) with the CAC mixes before (left) and
after (right) the NSSC test.
Finally, the CAC samples were heated up with a 550 V RF-generator 50/100 at a maximum
frequency 63.5 kHz and the induction heating efficiency of CACs. The maximum tempera-
ture at the surface of the CACs after 120 seconds of induction heating was determined before
and after the NSSC test using an infrared thermometer.
5 RESULTS
1218
Figure 4. Plot of load versus vertical displacement in indirect tensile test for different asphalt mixes
at 20 °C.
Figure 5. Effect of the chloride conditioning duration on the indirect tensile strength of two types of
CAC mixes.
1219
Figure 6. Effect of the chloride conditioning duration on the induction heating efficiency of two types
of CAC mixes.
simultaneously a new network of conductive. The new pathways connect isolated conductive
particles of the mix with iron oxides on the mix surface allowing electrons to flow more easily.
This phenomenon is called a ‘surface bridge effect’ because of the iron oxides.
The chloride-induced damage and effect of chloride on the mechanical performance of a con-
ductive SMA mix and the induction heating efficiency were studied by utilization of the accel-
erated chloride migration test in stone mastic asphalt mixes for different exposure times.
Mechanical analysis with indirect tensile strength indicates that the damage resulted by chlo-
ride migration is not significant after 10 days of the migration test. In order to distinguish the
factors causing the strength reduction in the conductive asphalt mixes, it is very important
to investigate also the moisture effect on the same mix type under the same temperature and
time duration conditions. Moreover, mixing corroded particles within the asphalt mixes can
be used as an alternative way to compare the performance of these asphalt mixes with the
uncorroded particles with the same characteristics (e.g. length, diameter). The results of chlo-
ride-induced damage after the chloride migration test should be correlated to the developed
corrosion damage in the field pavement. Among the recommendations for future studies are
also the experimental and numerical analysis of chloride and moisture damage of mixes with
similar or different porosity and aggregate structure, such as porous asphalt (PA) and dense
asphalt (DA) mixes.
Finally, the rate of induction heating efficiency appears to increase slightly after 5 and
10 days in the migration test. Because the heating efficiency under induction is related to the
effective electrical conductivity of the mix, this slight increase of the rate of induction heating
efficiency is probably caused by the formation of iron oxide conductive pathways on the mix
surface after steel fibers corrosion.
ACKNOWLEDGMENTS
The authors would like to thank Heijmans Integrale Projecten for the financial support and
Dr. Branko Savija for his guidance with the experiments.
1220
REFERENCES
ASTM D6931. 2012, Standard Test Method for Indirect Tensile (IDT) Strength of Bituminous Mixs.
Ahmedzade P., & Sengoz, B. 2009. Evaluation of Steel Slag Coarse Aggregate in Hot Mix Asphalt Con-
crete. Journal of Hazardous Materials, 165, 300–305.
European Asphalt Pavement Association. 1998. Heavy Duty Surfaces: The Argument for SMA. EAPA.
The Netherlands.
Garcia A., Norambuena-Contreras, J. & Partl, M.N. 2013. Experimental Evaluation of Dense Asphalt
Concrete Properties for Induction Heating Purposes. Construction and Building Materials, 46, 48–54.
Garcia, A., Norambuena-Contreras, J., Bueno, M., Partl, M.N. 2014. Influence of Steel Wool Fibers on
the Mechanical, Thermal and Healing Properties of Dense Asphalt Concrete. Journal of Testing and
Evaluation, 42, pp. 1107–1118.
Garcia, A., Schlangen, E., van de Ven, M. & Liu, Q. 2009. Electrical Conductivity of Asphalt Mastic Con-
taining Conductive Fibers and Fillers. Construction and Building Materials, Vol. 23, 3175–3181.
Huang B., Cao, J., Chen, Z., Shu, X., & He, W. 2006. Laboratory Investigation into Electrically Conductive
HMA Mixtures. Journal of the Association of Asphalt Paving Technologists, Vol. 75, 1235–1253.
Huang H. & White, T.D. 1996. Dynamic Properties of Fiber-Modified Overlay Mix. Transportation
Research Record, 1545, 98–104.
Jones, D.A. 1996. Principles and Prevention of Corrosion, 2nd ed. Prentice Hall, N.J.
Kandhal P.S., & Mallick, R.B. 2001. Effect of Mix Gradation on Rutting Potential of Dense-Graded
Asphalt Mixs. Transportation Research Board. Transportation Research Record, 1767, 146–157.
Lee S.J., Rust, J.P., Hamouda, H., Kim, Y.R., & Borden, R. 2005. Fatigue Cracking Resistance of Fiber
Reinforced Asphalt Concrete. Textile Research Journal, 75, 123–128.
Liu Q., Schlangen, E., & van de Ven, M. 2012. Induction Healing of Porous Asphalt Concrete. In Trans-
portation Research Record: Journal of the Transportation Research Board, No 2305, Transportation
Research Board of the National Academies, Washington, D.C., 95–101.
Liu Q., Schlangen, E., van de Ven, M. & Garcia, A. 2010. Induction Heating of Electrically Conductive
Porous Asphalt Concrete. Construction and Building Materials, Vol. 7, 1207–1213.
Liu X., & Wu, S. 2009. Research on the Conductive Asphalt Concrete’s Piezoresistivity Effect and Its
Mechanism. Construction and Building Materials, 23 (8), 2752–2756.
Lizarazo-Marriaga, J., Gonzalez, J. & P. Claisse. 2012. Simulation of the Concrete Chloride NT Build-
492 Migration Test. Numerical Modeling SSCS.
Mahrez, A., Karim, M. & Katman, H. 2003. Prospect of using Glass Fiber Reinforced Bituminous
Mixes. J. East Asia Soc. Trans. Studies, 5, pp. 794–807.
Minsk L.D., 1968. Electrically Conductive Asphalt for Control of Snow and Ice Accumulation. High-
way Research Record, 227, 57–63.
NT BUILD 492, Nordtest Method. 1999. Concrete, Mortar and Cement-based Repair Materials: Chlo-
ride Migration Coefficient from Non-Steady-State Migration Experiments.
National Asphalt Pavement Association. 2002. Designing and Construction SMA Mixtures – State-of-
the-Practice. NAPA, QIP 122.
Park, P., Rew, Y., & Baranikumar, A. 2014. Controlling Conductivity of Asphalt Concrete with Graph-
ite. Texas A&M Transportation Institute College Station, Report No SWUTC/14/600451–00025–1.
Rockliff, D. 1996. Stone Mastic Asphalt Pavement and its Effect on Highway Traffic Noise Levels.
Transportation Research Board.
Serin S., Morova, N., Saltan, M., & Terzi, S. 2012. Investigation of Usability of Steel Fibers in Asphalt
Concrete Mixs. Construction and Building Materials, 36, 238–244.
Stanish, K.D., Hooton, R.D. & Thomas M.D.A. 2000. Testing the Chloride Penetration Resistance
of Concrete: A Literature Review. FHWA Contract DTFH61–97-R−00022. University of Toronto,
Ontario, Canada.
Varveri, A., Avgerinopoulos, S. & Scarpas, T. 2014. Experimental Evaluation of Long- and Short-Term
Moisture Damage Characteristics of Asphalt Mixtures. Road Materials and Pavement Design, 17
(1), pp. 168–186.
Wu S., Liu, X., Ye, Q., & Li, N. 2006. Self-Monitoring Electrically Conductive Asphalt-Based Com-
posite Containing Carbon Fibers. Transactions of Nonferrous Metals Society of China, Vol. 16,
512–516.
Wu S., Mo, L., Shui, Z., & Chen, Z. 2005. Investigation of the Conductivity of Asphalt Concrete Con-
taining Conductive Fillers. Carbon, 43(7), 1358–1363.
1221
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
Asphalt mixture pavements are widely used in roads, urban roads, airports, and other
transport facilities due to its flatness, driving comfort, low noise, and so on(Wang, 2010).
Furthermore, asphalt mixtures can be classified in terms of the manufacturing temperature.
For example, a cold mix asphalt is manufactured at a temperature lower than 60°C, a half-
warm mix asphalt is manufactured at less than 100°C (normally 70∼95°C), and a warm mix
asphalt is manufactured at temperatures of 110∼140°C.
Previously, the hot asphalt mixture was widely used in road construction projects, but its
production and construction process require a lot of heat with significant carbon emissions.
Thus, new sustainable development policies have led to the road engineering technologies
that significantly reduce the manufacturing and application temperatures of hot asphalt
mixtures(Rubio et al., 2012; D’Angelo et al., 2008).
The half-warm asphalt mixture is a viable solution. Its manufacturing process considerably
reduces combustion gases, ranging 58% for CO2 and 99.9% for SO2. Further, less particles are
emitted into the atmosphere, except for VOCs and PAHs (María et al., 2013).
Some research has been conducted on the manufacturing of the half-warm asphalt mix-
ture and its properties. Punith et al. (2012; 2013) manufactured a half-warm asphalt mixture
with foamed asphalt and evaluated its moisture susceptibility and rutting resistance. Has-
san et al. (2012) produced a half-warm asphalt mixture based on superabsorbent polymers
(SAP); as a result, it possessed a good compaction ability and workability at lower tempera-
tures for increased haul distances and construction in colder climates. Ai et al. (2014) studied
the strength formed condition and influencing factors of a half-warm asphalt mixture and
proposed that the compaction could be appropriately increased to improve its strength.
This paper used a high-speed shear apparatus to shear matrix asphalt and a half-warm
asphalt additive at a ratio of 100:10 in order to prepare the half-warm asphalt. Various per-
formance aspects were evaluated. The half-warm asphalt mixture was preliminarily studied in
order to determine its mixing temperature, optimum asphalt content, and curing condition.
The results provide a theoretical basis for further study.
1223
2 EXPERIMENTAL
Figure 2 displays the TG curve of the half-warm asphalt. When the temperature was 25°C
to 188.5°C, the half-warm asphalt retained its quality, the half-warm asphalt did not decom-
pose, and the thermal stability was good. However, from 188.5°C to 665.5°C, the half-warm
asphalt gradually decomposed, and more mass was gradually lost. When the temperature
reached 665.5 °C, there was no quality change, and all of the half-warm asphalt was decom-
posed carbonization. Based on the above analysis, the asphalt temperature and the asphalt
mixture mixing temperature should be less than 188.5°C when the half-warm asphalt is a
binder-mixed asphalt mixture.
3.4 The mixing temperature and curing condition of the half-warm asphalt mixture
AC-13 was used for the mix design of the half-warm asphalt mixture. The preliminary asphalt
content was 5.0%, and the mixture underwent double-sided compaction 75 times. The mixing
temperatures were 75°C, 85°C, 95°C, and 105°C. Marshall specimens were formed, and the
void fractions of the specimens were tested. The results are given in Table 3.
In accordance with the specification requirements, the recommended void fraction range
for AC-13 asphalt mixtures is 3% to 6%, and the mixing temperature of a half-warm asphalt
mixture should be less than 100°C; thus, the mixing temperature of the half-warm asphalt
mixture was 95°C.
Marshall specimens were formed using the same preliminary asphalt content (5.0%) and
a mixing temperature of 95°C. The specimens were kept at room temperature (23°C∼27°C),
and the Marshall stability was tested every two days (see Table 4).
As seen from Table 4, the strength of the half-warm mixture continuously improved as
time elapsed. The Marshall stability of the half-warm asphalt mixture stabilized after curing
for 10 days. Thus, the curing condition of the half-warm asphalt mixture was 10 days at room
temperature (23°C∼27°C).
1226
Table 3. Void fraction results.
75 7.62
85 5.39
95 3.74
105 3.28
2 days 5.52
4 days 5.70
6 days 5.79
8 days 5.82
10 days 5.97
12 days 5.98
14 days 5.98
4 CONCLUSONS
This paper prepared a half-warm asphalt using a high-speed shear apparatus with a 100:10
ratio of half-warm asphalt additive to asphalt. The half-warm asphalt’s performance was
evaluated, and a preliminary study of the half-warm asphalt mixture was conducted. The
results revealed the following conclusions.
a. No new material appeared after mixing the half-warm asphalt additives and asphalt. Only
a physical mixing occurred; thus, they had good chemical compatibility.
b. When the temperature was lower than 188.5°C, the half-warm asphalt had good thermal
stability.
c. The PG grade upper temperature of the half-warm asphalt was 58°C. The elastic recov-
ery of the half-warm asphalt that aged in accordance with the rolling thin-film over test
(RTFOT) was improved, and the unrecoverable deformation resistance was increased.
d. The mixing temperature of the half-warm asphalt was identified as 95°C, the curing con-
dition was 10 days at room temperature (23°C−27°C), and the optimum asphalt content
1227
Figure 3. Asphalt content and physical and mechanical indexes.
1228
Table 5. Asphalt content and physical and mechanical indexes.
Asphalt content Void fraction Stability Flow value Theoretical density Saturation
was 4.8%. Half-warm asphalt has great application potential because it effectively reduced
energy consumption during the asphalt pavement paving process.
REFERENCES
Ai, C.F., Li, B.X., Peng, H., Qiu, Y.J. 2014. Study of Strength Forming Mechanism and Influencing Fac-
tors of Half-Warm Mix Asphalt. Journal of Highway and Transportation Research and Development
(English Edition) 8:1–6.
D’Angelo, J., Harm, E., Bartoszek, J. et al. 2008. Warm-mix Asphalt: European Practice. Report no. Fuel
Consumption 2:72.
Hassan, B., Mohsen, E., Emmanuel, V. et al. 2012 Warm and half-warm asphalt mixes with super
absorbent polymers: An innovative technology. European Roads Review 3: 56–60.
María del, C.R., Fernando, M., María, J.M., Germán, M. et al. 2013.Comparative analysis of emissions
from the manufacture and use of hot and half-warm mix asphalt Original Research Article. Journal
of Cleaner Production, 41:1–6.
Punith, V.S., Xiao, F., Wingard, D. et al. 2012. Moisture susceptibility and rutting resistance evaluation
of foamed Half Warm Asphalt mixture. Transportation Research Boar. USA: Washington, D.C.
Punith, V.S., Xiao, F., Wingard, D. 2013. Performance Characterization of Half Warm Mix Asphalt
Using Foaming Technology. Journal of Materials in Civil Engineering 3:382–392.
Rubio, M.C., Martínez, G., Baena, L., Moreno F. 2012. Warm mix asphalt: an overview. Journal of
Cleaner Production 24:76–78.
Wang Shuchao. 2010. The modified mechanism of WMA and research of Evaluation. Xi’an,
Chang’an.
1229
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Xin Yu, Fan Gu, Fuqiang Dong, Gongying Ding & Tianyu Wang
College of Civil and Transportation Engineering Hohai University, China
ABSTRACT: In this paper, the glass transition temperature (Tg) was used to evaluate the
low temperature performance of foamed warm asphalt, ductility test and dynamic shear
rheological test were used as well. Then the bending beam rheometer experiment was carried
out to verify the results which were obtained frorm Tg and ductility test which shows that the
low temperature performance of foamed SBS modified asphalt first decreases then increases
with the increasing of water content. It could be found that the low temperature performance
of foamed SBS modified asphalt is better than that of non-foamed SBS modified asphalt
when water content is 3wt%. Meanwhile, the foamed crumb rubber modified asphalt has the
similar low temperature performance. According to the comparison the results of BBR test,
it could be concluded that it is suitable for Tg to evaluate the low temperature performance
of warm mix asphalt.
1 INTRODUCTION
With the rapid development of the society, the asphalt pavement has been widely used with
the excellent performance and the comfortable driving experience. However, the property
of high viscosity of asphalt itself consumes lots of energy as the production of the asphalt
requiring high temperature. To solve this problem, foamed warm mix modified asphalt has
therefore been made. Condition of the foaming asphalt was tested by Castedo-Franco point-
ing out that any asphalt can be foamed under certain conditions (Castedo-Franco et al. 1983).
The performance of foamed asphalt mixture are determined by the aggregates and asphalt
types proved by Munir Nazzal. A comparative study indicating the properties of the foamed
asphalt was carried out to show that the indexes of foamed asphalt mixture are in accordance
with the requirements of ordinary hot mix asphalt mixture has been done by Guanhua Qiao.
Evaluation of low temperature properties of foamed asphalt is determined by low tempera-
ture ductility, bending beam creep test, direct tensile test, indirect tensile test, bending test,
bending creep test and splitting creep test, which are not widely recognized.
The low temperature performance of asphalt pavement is of great importance. Different
types and dosages of modifier have been made into the foamed SBS modified asphalt and
foamed rubber modified asphalt so that the special research should be implemented to evalu-
ate the low temperature performance of foamed asphalt. (Hurley et al. 2005). The low tem-
perature performance of SBS modified asphalt and rubber asphalt is difficult to evaluate by
using low temperature ductility as the incorporation of modified agent (Xu Bo et al. 2014).
An index with the clear physical meaning is therefore needed to evaluate the low temperature
performance of foamed warm mix asphalt.
To solve the problem, this paper uses the glass transition temperature Tg to evaluate the
low temperature performance of the foam warm asphalt (Cao LP et al. 2006). The glass tran-
sition temperature has been used by Dr Yin to test the hot mixed asphalt’s low temperature
performance which shows it’s own value (Yin YM et al. 2010).
In this paper, the low temperature performance of SBS modified asphalt was evaluated by
ductility value and glass transition temperature Tg which is verified by bending beam rheol-
1231
ogy experiment. Moreover, according to the paper, the physical significance and feasibility
of the glass transition temperature as the standard for the evaluation of the low temperature
performance of foamed warm mix asphalt is analyzed.
2 EXPERIMENTAL MATERIAL
3 EXPERIMENTAL METHOD
Asphalt type
Water consumption/%
Asphalt type 0 1 2 3
1232
causes a large strain emerging in the direction of the stretch. So the rubber asphalt’s ductility
is not high causing the specimen being easily pulled off. Therefore, the ductility index is not
suitable for evaluating the low temperature properties of foamed rubber asphalt.
Analysis can be concluded from Table 2 and Table 3 that the low temperature anti-cracking per-
formance of SBS modified asphalt is opposite to the foam rubber asphalt studied by using the
ductility value and the glass transition temperature. The low temperature anti-cracking perform-
ance of the foam rubber asphalt is better than that of the SBS modified asphalt at 5 degree centi-
grade in the practical engineering application as it is shown in the glass transition temperature.
Water consumption/%
Asphalt type 0 1 2 3
1233
Table 4. Creep stiffness S and m-value of foamed asphalt.
the quality of asphalt. The ideal state is that Tg can be lower than the lowest service tempera-
ture of asphalt so that the asphalt has good deformation ability to reduce the production of
low temperature cracks.
In order to test the results of the ductility experiment and the glass transition temperature
experiment, bending beam rheological experiment is used for inspection.
The material used to pave roads is asphalt mixture while the low temperature performance of
asphalt which is tested by using the bending beam combines to 80% of the asphalt mixture’s
low temperature performance (Zhang ZQ et al. 2005). So the accuracy and sensitivity of the
low temperature performance of rubber asphalt mixture and SBS modified asphalt mixture
should be evaluated by the application of Tg.
The low temperature performance test of the mixed material of the original sample and three
kinds of different water content (1%, 2%, 3%) which were marked B0,B1,B2,B3,S0,S1,S2,S3.
A lot of methods have been used to evaluate the low temperature performance of modified
asphalt mixture. However, there was no generally accepted method of evaluation. In low
temperature condition, to obtain good technical performance of asphalt mixture, the mixture
needed to have high strength and large deformation which means in order to evaluate the low
temperature anti-cracking performance of asphalt mixture, the two indexes of strength and
deformation must be considered. The compressive strain energy density test was the one to
1234
Figure 1. Relation of stress and strain of compression strain energy density.
Figure 2. Tg and compression strain energy density of different kinds of modified asphalts.
evaluate the low temperature performance of the mixture, so it was used to evaluate the rub-
ber asphalt mixture and the SBS modified asphalt mixture.
Uniaxial compression test method for asphalt mixture (T0714–1993) of Standard Test
Methods of Bitumen and Bitumen Mixture for Highway Engineering (JT J 052–2000) was
used at test temperature of 0°C and at the loading rate of 5 mm/min−1. The test equipment
was MTS whose stress-strain curve is shown in Figure 1 and the area under the curve is the
compression strain energy density. The greater the compression strain energy density is, the
greater the energy required for the occurrence of cracks in the mixture is which means the low
temperature performance of the mixture is better.
The compressive strain energy density and glass transition temperature of the rubber asphalt
mixture and SBS modified asphalt mixture were measured as the same as the original sample
and three kinds of different water content (1%,2%,3%). It can be seen from Figure 2 there is a
difference between rubber asphalt mixture and SBS modified asphalt mixture as the bigger the
Tg is, the worse the low temperature performance is. Therefore, sorting from 8 kinds of asphalt
low temperature performance are B3, B1, B2, S3, B0, S0, S2, S1. For the mixed material, the
smaller compression strain energy density is, the worse the value of the low temperature per-
formance is. Therefore, the low temperature performance of mixture from sort of B3, B1, B2,
S3, B0, S0, S2, S1. It can be seen that the results of the evaluation of modified asphalt by Tg
were consistent with the results of the evaluation of the asphalt mixture.
In summary, the low temperature performance of SBS modified asphalt by Tg is reasonable,
accurate and sensitive, it not only has a definite physical meaning, but also can characterize the dif-
ference of the low temperature performance of different modified asphalt which means the results
of the evaluation of the low temperature performance of the mixture have good correlation.
6 CONCLUSION
1. The low temperature performance of the foamed rubber asphalt is slightly worse than
that of the rubber asphalt before at first. With the addition of the water, the low tempera-
1235
ture performance of foamed warm mix asphalt has an increasing trend. When the water
consumption goes to 3%, the foam asphalt’s low temperature performance gets improved
which is better than that not foamed one.
2. The ductility values of the foam rubber asphalt at 5°C is less than the foamed SBS modi-
fied asphalt. But in fact, the low temperature performance of foamed rubber asphalt is
better than that of SBS modified asphalt which indicated that the ductility index is not
suitable for evaluating the low temperature properties of foam rubber asphalt.
3. The glass transition temperature Tg has clear physical meaning with the experimental veri-
fication of the bending beam rheometer showing that the glass transition temperature Tg is
suitable for evaluating the low temperature performance compared to the ductility values.
4. The low temperature performance of two kinds of modified asphalt mixture is tested by
using the strain energy density test which shows that the Tg can be used to evaluate the
performance of modified asphalt at low temperature.
5. In conclusion, using the glass transition temperature Tg to evaluate the low temperature
performance of foamed asphalt has a certain feasibility.
ACKNOWLEDGMENTS
This work was supported by the National Natural Science Foundation of China (51278173), the
Natural Science Foundation of Jiangsu Province for Distingished Young Scholar (BK20150038),
and the Fundamental Research Funds for the Central Universities (2015B21614).
REFERENCES
1236
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Asphalt mortar can be regarded as weak position in asphalt mixture, the
deformation property of asphalt mortar affect the damage evolution process, however, it is
hard to get the strain field of asphalt mortar by traditional method. Digital Image Correlation
(DIC) method has been adopted to obtain the strain field of asphalt mortar under indirect
tension test in this study. It is found that the proportion of high strain area to total asphalt
mortar area is very low, by contrast, the theoretical distribution is more uniform, the strain
variance increase rapidly at the later stage due to the strain localization; The modulus of
asphalt mortar can be calculated by the result of strain of different stress state, the result
show the modulus falls with time and the damage parameter increases with time for both
AC-16 and SMA-16, the modulus of AC-16 is larger than that of SMA-16.
1 BACKGROUND
1237
of aggregate and asphalt binder, therefore, studying the strain localization phenomena is an
approach to reveal the failure mechanism of asphalt mixture. Traditional strain test method,
such as strain gage and Linear Variable Differential Transformer (LVDT) can only get the
average strain (Roque& Buttlar, 1992; Kim, Seo, King & Momen, 2004), it is not be suit-
able for strain localization phenomena measurement, because of the limitation of the test
method, it is hard to study the damage evolution of asphalt mixture in depth.
Technology of optical measurement have been proposed for local deformation measure-
ment, Digital Image Correlation (DIC) as one of optical measurements was proposed in the
1980’s (Chu, Ranson & Sutton, 1985; Bruck, McNeill, Sutton & Peters, 1989), it is a direct
displacement and strain fields measurement method with advantages of full-field, real-time,
non-contact, flexibility and so on. (Seo, Kim, Schapery, Witczak & Bonaquist, 2004) first
proposed the use of a DIC technique as a possible displacement/strain measurement method
for asphalt mixture. (Tan, Zhang, Guo & Shan, 2012; Tan, Hou, L.Zhang & K.Zhang, 2014)
compared DIC method with LVDT and determine the minimum calculation gauge length of
the IDT test, according to the strain field obtained by DIC, it is found that the Poisson ratio
is not a constant. (Song, Paulino & Buttlar, 2006; Montepara, Romeo, Birgisson & Tebaldi,
2010; Birgisson, Montepara, Romeo & Tebaldi, 2011; Aragão & Kim, 2011) investigated
fracture phenomena in asphalt materials based on DIC method.
In previous studies, the DIC technique can accurately capture the strain field of asphalt
mixture and it is an effective method for investigating the fracture phenomenon, however,
there are little experimental studies combining the damage evolution and full strain field
information, especially on the strain field of asphalt mortar which can be considered as the
weak position of asphalt concrete. In this article, DIC method combined with digital image
processing method is introduced to determinate the strain field produced by IDT to investi-
gate the deformation properties of asphalt mortar, the failure process and evolution of dam-
age fields were characterized by the strain fields.
1238
Table 2. Basic properties of SBS polymer-modified asphalt.
ture, cylindrical specimens are obtained by compacting the mixtures 120 times using a gyra-
tory compactor. In order to observe the failure position, each cylindrical specimen is sawn
to obtain two effective plates discarding the top and the bottom plates for reducing density
gradient effects, the thickness of the final test specimen is 63.5 mm, The specimens are used
to perform the IDT test at 25°C as the Chinese specification JTG E-20–2011.
2 P ⎛ si 2 θ1 θ1 si 2 θ 2 θ 2 ⎞ 2P
σx = + ⎟⎠ − π d (1)
π ⎝ r1 r2
2P ⎛ cos3 θ1 3
θ 2 ⎞ 2P
σy = + − (2)
π ⎝ r1 r2 ⎟⎠ π d
2 P ⎛ cos 2 θ1 i θ1 cos 2 θ 2 i θ 2 ⎞
τ xy = + ⎟⎠ (3)
π ⎝ r1 r2
Figure 3. Distribution of material and strain field (a) binary image of asphalt mixture; (b) strain field
of asphalt mortar.
1240
Figure 4. Stress field calculation (a) stress field calculation method; (b) theoretical strain field of
asphalt mortar.
P is the load, d is the diameter of the disk, σx is the stress in the X direction, σy is the
stress in the Y direction, τxy is the shear stress, other symbols is showed in Fig. 4(a). For the
IDT test, the deformed destruction of the disc specimen is led by tensile stress in horizontal
direction, so the study of this article focused on the strain and stress in X direction. Accord-
ing to the stress distribution, the strain field can be calculated by assuming the modulus of
asphalt mixture, the strain of asphalt mortar was show in Fig. 4 (b).
show the strain mean increase with time, the slope of the curve is low in the earlier stage and
high in the later stage, it illustrate the deformation velocity acceleration due to damage or
softening of asphalt mortar which cause the overall deformation resistance decline.
The variance of a set of strain of pixels is the average of the squares of the deviations of
the observations from their mean, it reflect the level of dispersion. Fig. 7 (b) demonstrates
the strain variance increase with time, the curve present non-linear regularity, the slope of
the curve is low in the earlier stage and high in the later stage like the strain mean-time curve,
however, in the strain variance-time curve, its slope in the later stage increase faster and
faster, it shows a high level of dispersion. Comparing two curves, it shows the increasing of
strain mean and the steady of strain variance at the earlier stage, as time goes by, the strain
variance increase rapidly while the strain mean maintain a steady growth. The reason for
that is the strain distribution present local characteristic due to the progressive damage in the
asphalt mortar, the strain increase rapidly in the localized areas which will be the failure posi-
tion and the difference between the highest and the lowest strain values gradually widened, it
shows a serious localization phenomena before the failure of asphalt mixture.
4.2 Experimental analysis and description for damage evolution of asphalt mixture
Deformation resistance continues to decrease in the test processing because of the damage of
material, accurate damage evaluation method is the foundation of bearing capacity prediction
and precise simulation of failure processing. Modulus loss calculated by strain and stress can
be used as damage variable, in this research, real strain of each pixel can be obtained by DIC
method, but it is hard to get the real stress. Because there is no appreciable macroscopic crack in
the material before load reached the peak pressure, the specimen have good integrity, for the con-
venience of research, the stress field in this research is obtained by analytical method. The modu-
lus of each pixel of the surface can be calculated by the stress and strain, there are large amounts
of pixels in the tested surface, for the convenience of observing the relationship between stress
and strain clearly, average the strain of pixels in the same stress level and the stress-strain curve
of key loading points is showed in Fig. 8 based on the average strain of different stress level.
As shown in Fig. 8, overall trend of the curves is the linearity of the curve improve with
time, at the early stage, there is not significant linear relationship between strain and stress,
because of reconsolidation between aggregate and asphalt mortar and the calculation error
which is large for the strain in the early low strain stage, there is no evident regularity of
the relationship of stress and strain in Fig. 8 (a). As time goes on, the curves in Fig. 8 (b)
approach to be linear with a large fluctuation in the foreside, this may be because the low
strain zone is still in the reconsolidation processing, Fig. 8 (c)−Fig. 8 (f) show a better linear
trend, strain grows linearly as the stress increasing.
To explain those diagrams above clearly, the strain field of A, C, E are showed as Fig. 9. In the
earlier stage of loading, the strain field shows the overall increasing trend, there is no significant
1242
Figure 6. Strain distribution of asphalt mortar (a) Strain distribution of point B; (b) Strain distribu-
tion of point D; (c) Strain distribution of point F; (d) Theoretical strain distribution.
Figure 7. Strain distribution of asphalt mortar (a) Strain distribution of point B; (b) Strain distribu-
tion of point D;.
variation between the different positions, some high strain area with the red color randomly dis-
tribute, it is not consistent with the result based on the continuous homogeneous hypothesis that
high strain area is in the middle of the sample, this phenomenon presents heterogeneous mate-
rial such as asphalt mixture has a reconsolidation and dislocation motion in the earlier stage of
loading. After this stage, the high strain areas are still random distribution, but the overall trend
is the high strain areas concentrate in its middle of the sample around the aggregate, it match our
above conclusions well which show the strain is higher in the higher stress areas.
After the earlier stage, the stress-strain curve almost agrees with linear rule, the slope of the
curve is modulus and the modulus loss might be regarded as damage variable by the princi-
ples of damage mechanic. The least square method is applied to get linear curve fitting, the
slope of the fit line for stress and strain is the modulus of asphalt mortar. Finally, the modu-
lus of AC-16 and SMA-16 at different time point are showed in Fig. 10, it can be seen that
as load progresses toward the peak load, the modulus of asphalt mortar continues to fall, the
damage in the asphalt mortar induce the decrease of deformation resistant ability.
1243
Figure 8. Strain of different stress state in the IDT test process (a) point A; (b) point B; (c) point C;
(d) point D; (e) point E; (f) point F.
Figure 9. Strain field in the IDT test process (a) strain field of point A; (b) strain field of point C; (c)
strain field of point E.
E
D = 1− (4)
E0
1244
Figure 10. Modulus-time curve of different Figure 11. Damage parameter D—time curve
asphalt mixture. of different asphalt mixture.
where: E0, the elastic modulus of the un-damaged material; E, the elastic modulus of the
damage material at time t.
For heterogeneous materials, during the elastic stage, the strain localization and damage in
the asphalt mortar is not obvious, as show in the Fig. 8 (a), no marked relationship between the
stress and corresponding strain, therefore the damage parameter D was calculated from point
B and the modulus at point B was assumed as the elastic modulus of the un-damaged material,
based on that assumptions, the damage parameter time characteristic was shown in Fig. 11.
As shown in Fig. 11, the damage parameter increases at a near-liner trend as a whole,
however, at the end of the curve, it indicates a slower rate of growth. The damage parameter
D is used to characterize the modulus reduction which is related to the deformation property,
in the load-displacement curve (Fig. 5), the stage from point B to E is a process of crack
initiation and propagation, the strain of all the locations in the asphalt mortar increase, the
crack propagation result in the decrease of the modulus. After point E, the obvious strain
localization band appear and the strain growth mainly increased near the band, the strain
of other positions increase relatively slowly, therefore the overall growth of the strain field is
slow which will lead to the slow decrease of the modulus.
Comparing Figure 10 and Figure 11, the damage process of SMA-16 is closed to that of AC-16,
it illustrate the damage process of different asphalt mixtures are roughly the same, but the modulus
of AC-16 is larger than that of SMA-16, because the asphalt mortar of SMA-16 with SBS polymer-
modified asphalt can produce a larger interior deformation before failure for the IDT test.
5 CONCLUSION
In this work, indirect tension test have been conducted to investigate the damage evolution of
asphalt mortar combined with DIC technique, it can be concluded that:
1. Digital image correlation is an effective method to observe the evolution of strain fields
in asphalt concrete, the strain of asphalt mortar can be extracted by image processing
method combined DIC method. It can well describe the whole process of damage accu-
mulation and the strain localization phenomena asphalt mixture under the load.
2. The proportion of high strain area to total asphalt mortar area is very low, the proportion
present a descending trend with the strain value increase, comparing with the theoretical
strain value, the theoretical distribution is more uniform. The strain variance increase
rapidly at the later stage due to the strain localization.
3. According to the strain of each pixel in the asphalt mortar, the strain of different stress
state are obtained, there is not significant linear relationship between strain and stress at
the early stage due to the reconsolidation between aggregate and asphalt mortar and the
calculation error, after the early stage, the stress strain curve show a better linear trend.
4. Based on the stress strain curve, the modulus and the damage parameter D can be calculated,
the result show the modulus falls with time and the damage parameter D increases with time
for both AC-16 and SMA-16, the modulus of AC-16 is larger than that of SMA-16.
1245
ACKNOWLEDGEMENTS
The authors are grateful for the financial support of the National Science Fund for Dis-
tinguished Young Scholars of China (51225803) and Fundamental Research Funds for the
Central Universities (HIT.NSRIF.201662).
REFERENCES
Aragão, F.T.S., & Kim, Y.R. 2011. Characterization of fracture properties of asphalt mixtures based
on cohesive zone modeling and digital image correlation technique. Transportation Research Board,
Washington, DC.
Birgisson, B., Montepara, A., Romeo, E., & Tebaldi, G. 2011. Characterization of asphalt mixture
cracking behavior using the three-point bending beam test. International Journal of Pavement Engi-
neering 12(6): 569–578.
Bruck, H.A., McNeill, S.R., Sutton, M.A., & Peters Iii, W.H. 1989. Digital image correlation using New-
ton-Raphson method of partial differential correction. Experimental Mechanics 29(3): 261–267.
Chu, T.C., Ranson, W.F., & Sutton, M.A. 1985. Applications of digital-image-correlation techniques to
experimental mechanics. Experimental mechanics 25(3): 232–244.
Guddati, M.N., Feng, Z., & Kim, Y.R. 2002. Toward a micromechanics-based procedure to character-
ize fatigue performance of asphalt concrete. Journal of the Transportation Research Board 1789(1):
121–128.
Kim, Y.R., Allen, D.H., & Little, D.N. 2005. Damage-induced modeling of asphalt mixtures through
computational micromechanics and cohesive zone fracture. Journal of Materials in Civil Engineer-
ing 17(5): 477–484.
Kim, Y.R., Seo, Y., King, M., & Momen, M. 2004. Dynamic modulus testing of asphalt concrete in
indirect tension mode. Journal of the Transportation Research Board 1891(1): 163–173.
Lee, H.J., Daniel, J.S., & Kim, Y.R. 2000. Continuum damage mechanics-based fatigue model of asphalt
concrete. Journal of Materials in Civil Engineering 12(2): 105–112.
Masad, E., & Somadevan, N. 2002. Microstructural finite-element analysis of influence of localized
strain distribution on asphalt mix properties. Journal of Engineering Mechanics 128(10): 1105–1114.
Masad, E., Muhunthan, B., Shashidhar, N., & Harman, T. 1999. Internal structure characterization of
asphalt concrete using image analysis. Journal of computing in civil engineering 13(2): 88–95.
Masad, E., Somadevan, N., Bahia, H.U., & Kose, S. 2001. Modeling and experimental measurements of
strain distribution in asphalt mixes. Journal of Transportation Engineering 127(6): 477–485.
Montepara, A., Romeo, E., Birgisson, B., & Tebaldi, G. 2010. Strain localization and damage dis-
tribution in SBS polymer modified asphalt mixtures. Road Materials and Pavement Design 11(4):
899–915.
Park, S.W., Kim, Y.R., & Schapery, R.A. 1996. A viscoelastic continuum damage model and its applica-
tion to uniaxial behavior of asphalt concrete. Mechanics of Materials 24(4): 241–255.
Roque, R., & Buttlar, W.G. 1992. The development of a measurement and analysis system to accurately
determine asphalt concrete properties using the indirect tensile mode (with discussion). Journal of the
Association of Asphalt Paving Technologists, 61.
Seo, Y., Kim, Y.R., Schapery, R.A., Witczak, M.W., & Bonaquist, R. 2004. A study of crack-tip defor-
mation and crack growth in asphalt concrete using fracture mechanics. Journal of the Association of
Asphalt Paving Technologists, 73.
Song, S.H., Paulino, G.H., & Buttlar, W.G. 2006. Simulation of crack propagation in asphalt concrete
using an intrinsic cohesive zone model. Journal of Engineering Mechanics 132(11): 1215–1223.
Tan, Y., Hou, M., Zhang, L., & Zhang, K. 2014. Studying the strain field distribution of asphalt mixture
with the digital speckle correlation method. Road Materials and Pavement Design 15(1): 90–101.
Yi-qiu, T., Lei, Z., Meng, G., & Li-yan, S. 2012. Investigation of the deformation properties of asphalt
mixtures with DIC technique. Construction and Building Materials 37: 581–590.
1246
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
Soil stabilization is the process through which the mechanical properties of weak soil are
improved. It can be simply performed by excavating the weak soil, mixing it with the neces-
sary stabilizers and placing the improved soil in its initial position. However, the situation
differs radically when the subgrade under existing pavements is considered due to the fact
that the existing pavement needs to be removed at the start and then repaved at the end.
This process is associated with high costs, severe traffic disruptions and dissatisfied public.
Traditional soil stabilization also comes with high levels of dust, noise and pollution. There-
fore, the method examined in this paper, soil stabilization based on electrokinetic principles,
seems an attractive alternative as it does not maintain the drawbacks of the traditional soil
stabilization.
The examined technique introduces the stabilizer, an anode, and a cathode at the sides of
the pavement and leaves the original pavement intact. The stabilizer travels from one side of
the pavement to the other through the weak soil due to the electric field that is created by the
electrodes and the potential difference applied externally. The mechanisms that drive the sta-
bilizer in the soil matrix when direct current is applied are diffusion, migration and electroos-
motic flow. Diffusion is the result of the presence of a concentration gradient. Migration is
the movement of charged ions under the presence of the electric field. The electroosmotic
flow develops under the presence of the electric field due to the charge of the soil particle
surface. This charge is normally negative and it results from imperfection in the mineral lat-
tice (Ng et al 2014). The concept of in situ soil stabilization based on electrokinetic princi-
ples is illustrated in Figure 1, where ΔV is the potential difference applied between the two
electrodes.
Electrokinetic phenomena have been used in the past for soil remediation purposes, where
electrokinetic principles were utilized to remove pollutants existing in soil (Lageman 1993,
Shapiro and Probstein 1993). Electrokinetic phenomena have also been considered within the
oil recovery field, where electrokinetic principles were considered for enhanced oil recovery
(Ghazanfari et al. 2012). The same principles are employed in the proposed in situ soil sta-
1247
Figure 1. Concept of in situ soil stabilization based on electrokinetic principles.
bilization technique, but with a major difference: where previous applications were aiming
at removing a substance that already existed inside the soil (pollution, heavy metals, oil, etc.)
because it was either harmful or lucrative, the proposed method aims at adding a substance
(stabilizer) in the soil matrix that does not initially exist in the soil and it will be beneficial
when added.
2 GOVERNING EQUATIONS
This section develops the framework for electrokinetic soil stabilization in fine soils under
the presence of direct current. The charge conservation law is used to describe the electric
potential distribution and mass transport in porous media principles are used to describe the
“injection” of the stabilizer in the soil.
∂ ρc
+ ∇⋅J = 0
∂t (1)
J k ∇V
(2)
Assuming constant electrical conductivity and charge density (steady current), equations 1
and 2 can be simplified to equation 3, where ∇ 2 is the Laplacian.
∇2 = 0 (3)
∂c
= −∇ ⋅ N
tot
(4)
∂t
The total mass flux Ntot is made up of three different mass fluxes according to equation 5
(Acar and Alshawabkeh 1993), where Nd is the diffusive mass flux, Nm is the migrational flux
and Ne is the mass flux due to the electroosmotic flow.
N +N
tot d m e
N N (5)
The diffusive mass flux Nd can be described by equation 6, where D efff is the effective dif-
fusion coefficient accounting for the fact that diffusion takes place in soil. The effective dif-
fusion coefficient D efff can be calculated by equation 7 (Moldrup et al. 2011), where D 0 is the
diffusion coefficient in water, θ is the volumetric soil water content and τ is the tortuosity as
defined by (Ball 1981).
D efff ∇c
d
N
(6)
D 0θ
D efff = 2
τ (7)
The migrational flux Nm can be described by equation 8 (Acar and Alshawabkeh 1993),
where zc is the valence, F is the Faraday’s constant, D efff is the effective diffusion coefficient,
R is the universal gas constant and T is the absolute temperature. In equation 8, substituting
the effective diffusion coefficient as defined in equation 7 yields equation 9 for the migra-
tional flux.
D efff
c∇V
m
N zc F
RT (8)
θ D0
c∇V
m
N zc F 2
τ RT (9)
e
The mass flux due to the electroosmotic flow N can be described by equation 10, where u
is the interstitial velocity. The interstitial velocity is related to the superficial velocity by equa-
tion 11, where us is the superficial velocity and n is the porosity; us can be calculated by equa-
tion 12 (Scales and Tait 2004), where εf is the permittivity of the fluid, ζ is the zeta potential
and η is the dynamic viscosity. Combining equation 11 and equation 12 with equation 10
yields the mass flax due to the electroosmotic flow Ne in equation 13.
e
N cu
(10)
1
u = us
n (11)
n ε fζ
us = ∇V
τ η (12)
ε fζ
N = c∇V
e
τη (13)
By combining equations 4, 5, 6, 7, 9 and 13 the governing equation for the mass transport
can be expanded to equation 14.
1249
Figure 2. Main dimensions of the 2D model.
∂c D 0θ 2 θ D0 ε fζ
= 2 ∇ c+ ∇V ⋅ ∇c − ∇V ⋅ ∇c (14)
∂t τ τ RT τη
c 2
3 NUMERICAL MODEL
In this section, the development of a 2D model is discussed. The governing equations are
equations 3 and 14. The software that was used to develop the model is COMSOL Multi-
physics 5.2.
3.1 Geometry
The main dimensions of the model are presented in Figure 2, where the meshing of the weak
soil is also visible. The pavement is not meshed as it is not actively participating in the model;
it is only presented for illustrative purposes. The electrodes are located at the side of the road;
the anode is located at the right side and the cathode at the left side.
3.3 Parameters
The variables that will be solved for are the molar concentration c and the electric potential
V. Besides the physical constants, all the other items that appear in this paper are parameters
1250
Figure 3. Boundaries defined as anode and cathode.
of the problem and typical values based on engineering judgement are assumed. A full list of
the parameters, their values and a brief description is presented in Table 1.
By observing Table 1 the importance of modeling becomes clear as measuring all the
involved parameters and their variation with respect to time and space can be from challeng-
ing to impossible.
4 RESULTS
In this section the results of the simulation are presented. The two variables V and c are
solved for and a time period of four months is assumed (1 month = 30 days). The elec-
tric potential V remains constant in time and its distribution is presented in Figure 4. The
molar concentration c changes over time and its distribution is plotted for different times in
Figures 5, 6, 7, 8 and 9.
Figures 5 to 9 illustrate how the stabilizer is transferred inside the soil over time. The
stabilizer is located mostly in the area that we are interested in improving, namely the soil
located under the pavement, and between the anode and cathode. This is promising for real-
life applications as it suggests that the proposed method can be effective in improving the
targeted areas and it is not losing its effectiveness due to diffusion in areas that are out of
interest. Whether the examined volume is large enough or not can be checked by observing
Figure 9. Since there is no concentration accumulation at the outer boundaries at the end of
the simulation, these boundaries do not affect the system and the volume selected is deemed
sufficient. The zero initial condition for the concentration can be validated from Figure 5,
1251
Figure 4. Electric potential, units in [V].
1252
Figure 6. Concentration of the stabilizer at t = 1 month, units in [mol/m3].
where concentration is zero everywhere at t = 0 except at the location of the anode. This is
due to the fact that the boundary condition at the anode for the concentration is Canode. The
small area with smeared concentration next to the anode is due to the finite size of the mesh
elements and the interpolation between points of zero concentration and points with con-
1253
Figure 8. Concentration of the stabilizer at t = 3 months, units in [mol/m3].
centration equal to Canode. This reasoning is in line with the thickening of the smeared area
close to the top boundary, as from Figure 2 it can be seen that the mesh elements become
larger in that area.
1254
Figure 10. Experimental setup in development.
5 EXPERIMENTAL SETUP
In Figure 10 the medium scale experimental setup that is in development is presented. The
length of the container is ∼1 meter, the width ∼0.7 meters and the height ∼0.2 meters. The
area between the red sieved plates will be filled with clay. The red plates are sieved in order
to allow for water and the stabilizer to enter and leave the soil matrix. They are covered with
cloth so that the clay is held at its position. The gap between the red plates and the walls of
the container form the anodic and cathodic chamber, where water and the potential stabiliz-
ers will be added. The anode and cathode will be placed against the walls of the containers
and small PVC blocks will be glued to act as guides. A circuit will be created by connecting
the anode and the cathode to an apparatus that provides a potential difference between the
anode and the cathode. The thin rods that are distributed in the container (in the area that
will be filled with soil, in the anodic chamber, and in the cathodic chamber) will act as passive
electrodes. The passive electrodes will be used for monitoring the potential distribution over
time. An appropriate pumping system has been designed to maintain a constant level in the
two chambers. The water level in each container is determined by a sensor with adjustable
height.
In this paper, the framework for indirect soil stabilization of weak fine soil based on electrok-
inetic principles is presented. Starting from the governing equations a 2D model is developed
that allows for estimating the transport of the stabilizer inside the soil that is located under
the pavement.
The results show an effective transport of the stabilizer in the targeted area. After 4 months
almost the entire width of the road is covered, and losses of the stabilizer to the surrounding
areas that are not of interest are small. The height of the stabilized soil stripe is larger than
the height of the electrodes, suggesting that shorter electrodes than the height of the targeted
soil can be used allowing for further optimization.
Some topics that are recommended for future research are:
− Development of a 3D model to assess the impact of the spacing between the electrodes
along the pavement. The one dimension is fixed and it is determined by the pavement
width, but the other dimension should be the result of an optimization process.
1255
− Conducting experiments to determine whether there is any potential drop at the area of
electrodes. Different electrode configurations might result in different effective potential
applied to the system.
− Examining whether reversing the anode and cathode (still direct current, no alternating
current) at some point during the experiment results in higher efficacy.
− Establishing relations about how the introduced stabilizer reacts and provides soil with
improved properties.
− Performing sensitivity analysis to determine which parameters of table 1 are the most
important, as the amount of the parameters and the associated complexity of measuring
and monitoring all of them varies from challenging to impossible.
− Exploring which stabilizers may yield the best outcome. Smaller scale experimental setup
could be used for speeding up the process.
REFERENCES
Acar, Y.B. & Alshawabkeh, A.N. 1993. Principles of electrokinetic remediation. Environmental Science
& Technology, 27, 13, 2638–2647.
Ball, B.C. 1981. Modelling of soil pores as tubes using gas permeabilities, gas diffusivities and water
release. Journal of Soil Science, 32, 465–481.
Ghazanfari, E., Shrestha, R.A., Miroshnik, A. & Pamukcu, S. 2012. Electrically assisted liquid hydro-
carbon transport in porous media. Electrochimica Acta, 86, 185–191.
Kestin, J., Sokolov, M. & Wakeham, W.A. 1978. Viscosity of liquid water in the range -8 oC to 150 oC.
Journal of Physical and Chemical Reference Data, 7, 3, 941–946.
Lageman, R. 1993. Electroreclamation, applications in The Netherlands. Environmental Science & Tech-
nology, 27, 13, 2648–2650.
Malmberg, C.G. & Maryott, A.A. 1956. Dielectric constant of water from 0 oC to 100 oC. Journal of
Research of the National Bureau of Standards, 56, 1.
Moldrup, P., Olesen, T., Komatsu, T., Shcjonning, P. & Rolston, D.E. 2001. Tortuosity, diffusivity
and permeability in the soil liquid and gaseous phases. Soil Science Society of America Journal, 65,
613–623.
Ng, Y.S., Sen Gupta, B. & Hashim, M.A. 2014. Stability and performance enhancements of electrok-
inetic-fenton soil remediation. Reviews in Environmental Science and Bio/Technology, 13, 3, 251–263.
Scales, N. & Tait, N. 2004. Modelling electroosmotic flow in porous media for microfluidic applications.
Proceedings of the 2004 International Conference on MEMS, NANO and Smart Systems.
Shapiro, A.P. & Probstein R.F. 1993. Removal of contaminants from saturated clay by electroosmosis.
Environmental Science & Technology, 27, 2, 283–291.
1256
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Urban express corridor in China are characterised by intensive traffic loads,
extreme climatic conditions, short allowable construction time and therefore high difficulty
of construction and maintenance. By mechanical analysis, the essential requirements on pave-
ment structure of bridges and tunnels on urban expressway were identified. Typical pavement
structures and materials were optimised regarding the special operational and construction
conditions. Warm mix and flame retardant asphalt for city tunnel were introduced. A coloured
anti-skid and warning topping was developed for the pavement at the inlet and outlet of tunnel
in order to alleviate the high safety risk in that section. Typical solutions for steel bridge deck
were compared. Cold mix resin asphalt and SMA pavement structures were suggested for use
on urban viaduct. Some practical application of the introduced structures on a viaduct project
proved the feasibility of the studied structures. Long-term performances need to be observed.
1 INTRODUCTION
*AC stands for asphalt concrete, SMA for stone mastics asphalt, OGFC for open-graded friction
course, Sup for the asphalt mixture which is designed according to the American Superpave method
and is supposed to have good skeleton structure, GA for Gussasphalt.
Figure 1. Typical concrete box girder and maximal pavement response under different loading
positions.
The stress level in the pavement structure under the most critical position was considered
(right side of Fig. 1). In the longitudinal direction maximum tensile stress occurs when the
wheel load lies on the middle of the span. Transversely, on Positions 1, 6 and 17, the tensile
stress is most critical, when one of the tyres lies on top of the middle of the box gird.
Thermal stress resulting from the temperature gradient in the heating and cooling process
is also considered, as shown in Fig. 2. This stress was added up to the stress from wheel axe
load to form the overall stress.
1259
Figure 3. Influence of the E-Modulus of lower layer on maximal stress.
According to the results, the overall maximal tensile stress reaches 0.796 MPa and the
overall maximal shear stress reaches 0.430 MPa. The maximal tensile stress and the shear
stress are approaching the tensile strength of normal asphalt concrete and bitumen-based
tack coat material, especially at high temperature (60°C). For the stability of the structure,
the structure and material need to be strengthened to alleviate the overall stress level.
With increasing E-Modulus of the lower layer, the maximal tensile stress decreases by
more than 1/3 (see Fig. 3). This means a high-modulus lower layer can reduce the maximal
stress level of the pavement structure. As a result, the fatigue life of the structure will be
prolonged.
Figure 5. Operating of the sandblasting machine and the appearance of the surface after
sandblasting.
Figure 6. Spraying waterborne epoxy binder manually (left) or with intelligent asphalt sprayer
(right).
simultaneously. With minor blowing, the surface will be cleaned, as shown on the right side
of Figure 5. With little dust emission, this solution excels the fine grinding and sweeping.
Alternatively, the cement concrete surface can be exposed in fresh condition by applying
retarders and brushing in time.
As the tack coat material, waterborne epoxy binder is applied. It is a gel composed of
epoxy resin, coal tar and waterborne solvent. It can be spread either manually (Fig. 6 left)
or with the intelligent asphalt sprayer, which is used for spraying bitumen emulsion (Fig. 6
right).
1261
Figure 7. Pullout test (left) and shear test (right) to determine the bonding strength of the tack coat.
The tack coat hardens after one day at room temperature (25°C) and develops sufficient
bonding strength. The bonding strength of the tack coat can be determined by the pullout
test and the direct shear test on a composite test specimen with both parts of cement concrete
substrate and asphalt layer. For the pullout test, the specimen is bored from the asphalt side
to the bonding interface. A steel plate will be firmly glued to the surface of the asphalt. After
hardening the plate will be pulled out with increasing load, and the tensile stress level when
the asphalt was removed from the concrete substrate will be recorded as the pullout strength.
Direct shear test is performed on the same composite specimen, the shear stress till sliding on
the interface is taken as the shear strength.
In comparison with the conventional tack coat solutions such as the hot SBS bitumen or
crumb rubber modified bitumen with chippings, waterborne epoxy binder maintains rela-
tively higher bonding strength (0.86 MPa and 1.18 MPa) at the working temperature (about
45°C) as shown in Table 2. The waterborne epoxy binder tack coat can well accommodate
the maximal tensile (0.796 MPa) and shear stress (0.430 MPa) according to the mechanical
analysis in Section 2.2, while the other two solutions based on hot sprayed bitumen cannot.
3 TUNNEL PAVEMENT
1263
Figure 8. Surfactant and the compaction behaviour of wax and surfactant warm mix asphalt.
Figure 9. Test specimen and the limiting oxygen index test of flame retardant bitumen.
Table 4. Marshall test results of warm and hot mix asphalt with and without flame retardant.
Properties ρ (g/cm3) VV (%) VMA (%) VFA (%) VCAMix (%) MS (kN)
Hot mix no flame retardant 2.520 3.7 16.7 77.7 40.6 10.26
Hot mix flame retardant 2.509 3.8 16.7 77.1 40.9 16.94
Warm mix no flame retardant 2.532 3.2 16.2 80.1 40.3 9.38
Warm mix flame retardant 2.506 3.9 16.8 76.6 40.9 11.12
density decreases from 80% to 74%. The asphalt binder tends to be less flammable and less
smoke is generated when a fire emerges. The physical properties of the flame retardant are
comparable with those of normal limestone fillers and are therefore deemed as fillers in the
asphalt mix composition.
The flame retardant and surfactant in asphalt are compatible. The compaction behaviour
of warm-mix flame retardant asphalt is studied by a cross test, in which the surfactant and
flame retardant are applied separately as well as together. Results of the density and volumet-
ric indexes show that warm mix surfactant has no interaction with flame retardant (Table 4).
The flame retardant improves the Marshall stability (MS) while warm mix asphalt has a
relatively lower stability than the corresponding hot mix asphalt. Warm mix asphalt tends to
be more easily compactable.
3.3 Skid resistant and warning topping in inlet and outlet of tunnel
Statistics of accident around tunnel sections show that most accidents happen within the
200 m range near the inlet and outlet (Fig. 10 left). Aside from the lighting issue, the reason
1264
Figure 10. High accident occurrence rate in tunnel and layout of skid resistant warning topping.
Figure 11. Structure and the visual effect of the multi-colour skid resistant warning topping.
lies in the insufficient skid resistance and the subjective reasons from the drives, such as the
not timely deceleration when driving in or too early acceleration when driving out.
In view of this, a scheme of skid resistant warning topping is proposed, as shown on the
right side of Figure 10. The feathers of this scheme are several stripes of resin chipping top-
pings with the colour of red or yellow and the length of 5 m or 10 m. The colour is red out-
side and yellow inside, for yellow reflects the light better and red colour warns significantly.
The length is 10 m outside and 5 m inside the tunnel, with first narrower and then 10 m wide
spacing.
The topping consists of a layer of reaction resin binder on top of the existing asphalt or
cement concrete (sandblasting or cleaning if necessary) surface and a spread layer of stones
(Fig. 11 left). The paving of this topping can be completed within a day. The curing time for
the reaction resin is one day. Loose stones will be swept away before opening to traffic. The
bonding strength is essential for the safety and durability. With no time pressure, extra resin
can be spread on top to ensure the bonding strength. An extra day is needed for curing. The
visual effect of the topping is shown in Figure 11. Long-term effect of the topping on the
reduction of traffic accident is however not clear for the moment.
Figure 12. Cold mix resin asphalt and SMA composite structure for steel bridge deck on viaduct (left:
scheme of the pavement structure; middle: completion of EBCL; right: finished RA with spreading).
material and failure of the tack coat are the two major problems, which often lead to erosion
of the steel plate and endangers the structure safety and driving comfort.
The representative form of steel bridge deck pavement is mastics asphalt (Britain), Gussas-
phalt (Germany), epoxy asphalt (US and Japan) and composite structures using SMA as the
surface and resin asphalt as bearing layer. In Table 5, the using performances and construc-
tion issues of typical types of steel bridge pavement are compared. The typical American
double-layer epoxy asphalt has strict requirement on the temperature and time control. On
urban expressway, these conditions can seldom be fulfilled, only resulting in poor quality.
Due to the fine composition, gussasphalt or mastics asphalt is susceptible to the high tem-
perature. The high rut potential makes it unsuitable for the climatic conditions in China.
1266
performance and durability. This pavement structure can be opened to traffic in three days
and was proven to be durable in scores of highway projects (Pan et al. 2012).
5 CONCLUSIONS
This paper introduces some pavement structures and techniques for the cement concrete and
steel bridge deck pavement on urban viaduct as well as for city tunnel pavement. The solu-
tions can meet the requirements regarding the using performance and construction condi-
tions on urban expressway. The solutions enable good using performance and convenience
during construction. The long-term performances of the solutions are yet to be observed and
studied.
REFERENCES
ASTM D1074, Standard test method for compressive strength of bituminous mixtures.
ASTM D3497, Standard test method for dynamic modulus of asphalt mixtures.
ASTM D2843, Standard test method for density of smoke from the burning or decomposition of
plastics.
ISO 4589–2:1996. Plastics—Determination of burning behaviour by oxygen index—Part 2: Ambient-
temperature test.
The RST Working Group “Design of bituminous mixtures”, Laboratoire Central des Ponts et Chaussées,
2007. LPC Bituminous Mixtures Design Guide. Paris.
Jiangsu Technical Quality Controlling Bureau. 2012. Standard Specification for Waterborne Epoxy-
Binder Waterproof Coating for Concrete Bridges (DB32/T 2285–2012). Nanjing.
Pan Youqiang, Zhang Zhixiang, Cao Rongji. Research on ERS steel deck pavement techniques. Journal
of Testing and Evaluation. 40(7):20120152, December 2012.
1267
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
The seasonal variation of surface temperatures is known to strongly affect the structural
resistance and reliability of pavements, as high or low peaks may soften or embrittle the
asphalt, respectively. As a result, during hot periods pavements are susceptible to rutting
(Bobes-Jesus et al. 2013) and acceleration of the ageing process (Mollenhauer et al. 2012),
while during cold periods ice is a hazard for traffic (Wang et al. 2008, Dai et al. 2012) and the
risk of fatigue failure of asphalt is increased (Wang et al. 2013).
A number of methods have been studied to fight overheating or freezing issues separately.
In the case of high pavement temperatures, the methods considered are mostly related to
changes in the paving material properties, e.g., thermal conductivity, specific heat capacity,
albedo, or emissivity (see, e.g., Golden et al. 2006, Sarat et al. 2012, Santamouris 2013), how-
ever, energy harvesting pavements have been recently considered as an alternative (Chiarelli
et al. 2015a, b). In the case of freezing problems, the most common solutions are the use of
chemical substances (Wang et al. 2008, Dai et al. 2012) or the use of hot fluids circulated
under the pavement surface (Lund 2000, Lund et al. 2011).
The mitigation of both overheating and freezing problems can be pursued by allowing air
to flow in purposely built channels under the asphalt wearing course, thus, making absorbing
or releasing heat. The excess heat, if any, can be discharged in the environment through a
chimney or used for a chosen application, as done in convection powered energy harvesting
pavements (García & Partl 2014, Chiarelli et al. 2015a, b). If the temperature of air entering
the channels installed under the pavement surface is controlled, the pavement temperature
can be influenced in both summer and winter conditions.
In this paper, an experimental setup meant to simulate a geothermal heat source is used to
feed air at a realistic reservoir temperature to a pavement prototype.
The aim of the paper is to evaluate this equipment to see whether it provides a competent
evaluation of the opportunities for heating and cooling pavements by convective air flow.
Tests were conducted in the environment at the University of Nottingham, UK, in order to
1269
Figure 1. Scheme of the ground source heat simulator; the letters refer to Table 3.
assess whether the pavement temperature can be effectively controlled or not with this kind
of setup and to study the effects of weather on the behavior of the system.
2 METHODOLOGY
1270
Figure 2. Photograph of the ground source heat simulator at the University of Nottingham.
1271
Table 1. Parameters measured in the experimental campaign.
The data analyzed with the Pearson’s correlation coefficient was tested for statistical sig-
nificance with a 2-tailed test and Pearson’s coefficients between +0.8 and +1 or −0.8 and −1
were considered as signs of a strong positive or negative correlation, respectively (Kent et al.
2015). In addition, values between +0.5 and +0.8 and −0.5 and −0.8 were considered as a sign
of moderately positive or negative correlations, respectively.
In the analysis of data, rainfall was neglected as it was extremely scarce during the period
considered and, thus, could not cause any measurable effect.
In Table 2 and Table 3, the numerical results obtained from the statistical analysis are shown.
In Table 2, the minimum, maximum, and average values for all the parameters measured are
shown along with the standard deviation of each. It can be observed that the experimen-
tal campaign was performed with environmental temperatures between 7.30ºC and 25.80ºC,
relative humidity between 33% and 98%, and wind speed between 0 m/s and 2 m/s. These
conditions slightly differ from the averages for the month of August in the Nottingham area,
where environmental temperatures ranging between 12.0 ºC and 21.0 ºC are reported by
the Met Office for the years 1981–2010 (Met Office, n.d.). The average relative humidity of
72.5% is in line with values in the area (UK Climate Projections, n.d.).
A first observation of the data suggests that the ground source heat simulator did provide
a measurable effect on the surface temperature of the pavement prototype. In fact, if the
surface temperature of the prototype in Table 2 is compared to that of the control slab it is
possible to notice that it has a higher minimum value, a lower maximum value, and a lower
mean value. This means that the use of the simulated geothermal heat source was able to
reduce the oscillation of the pavement temperature during the tests. In particular, the air
absorbed heat when the prototype temperature was higher than 15ºC and it released heat
when the prototype temperature was below 15ºC. A confirmation of the reduction of the
temperature oscillations is also provided by the standard deviation, which is 6.46ºC for the
pavement prototype and 7.85ºC for the control slab.
The data shown in Table 2 also shows that the temperature in the inlet air box was not con-
stant during the tests. The average value of 16ºC is similar enough to the value set in the ther-
mostat in the experimental setup (i.e., 15 ºC). This inexact correspondence is due to the fact
that the inlet air box is in direct contact with the prototype pavement, thus, its temperature is
influenced by the temperatures of asphalt and aggregates. It is also worth considering that in
the inlet air box the air flows by buoyancy, thus, its temperature is constantly varying due to
the variable conditions present at the bottom of the inlet pipe providing air to the prototype
pavement. Finally, the non-constant temperature in the air box may be related to not perfect
thermal insulation or thermal bridges in the volume under analysis.
1272
Table 2. Descriptive statistics for the data gathered.
In Table 3, the Pearson’s correlation coefficient between the weather conditions and the
temperatures analyzed is shown. To begin with, it is relevant to mention that a level of signifi-
cance of 0.01 was achieved for all the correlations, thanks to the high amount of data points
collected. Therefore, the results obtained can be considered as reliable measures of the linear
correlation between the parameters under investigation.
The analysis of the correlation coefficient shows that among all the parameters consid-
ered the absolute pressure seems not to have any influence on the temperatures measured.
The reason why this happened is simply related to the fact that the pressure does not vary
significantly, thus, it cannot strongly influence the results obtained. In addition, pressure
changes would involve both ends of the system, thus, the differential effect would be small
or negligible.
All the other parameters, however, show moderate to strong positive or negative correla-
tions with the temperatures in the domain. The outdoor temperature and wind chill both
have strong correlations with the surface and top layer temperatures of both the prototype
and the control slab.
This result is fairly easy to understand, since the pavement temperatures are always highly
dependent on the environmental temperature. However, it is interesting to notice that the out-
door temperature and the wind chill have almost the same correlation coefficients with the
parameter under analysis. This behavior is seen because the values of outdoor temperature
and wind chill measured were extremely close to each other, as reported in Table 2 (the vari-
ation is smaller than the digits shown here), therefore, the correlation between the datasets
must be very similar.
1273
In addition, since the wind chill is defined as the decrease in the air temperature felt by
the human body due to air flow (Bluestein 2015), the similarity of the wind chill corrected
temperature to the actual outdoor temperature can be explained by the rather low values of
wind speed measured at the time of the experiments (see Table 2).
Furthermore, the relative humidity has a strong negative correlation with the temperatures
measured in the experimental setup. This behavior is mostly related to meteorology, as low
values of relative humidity are generally reported when the temperature is higher, while high
values of relative humidity are found when the temperature is lower. Therefore, when the
measured outdoor temperature increases, the surface temperature of the prototype increases.
In a very similar way, when the relative humidity decreases, the surface temperature increases.
The relationship between outdoor temperature, relative humidity, and surface temperature is
shown in Figure 3 to allow a clearer understanding of the phenomena at work. In Figure 3,
the data is sorted by increasing values of temperature, so that the relationship between the
parameters discussed is easily seen.
The Pearson’s correlation coefficient between wind speed and gust and the measured temper-
atures is moderate and not very noticeable. The reason for this is again that during the experi-
ments low speeds were measured, thus, this parameter could not highly affect the results.
It is, however, expected that the presence of stronger wind will cause a decrease in the
measured temperatures due to an increased thermal convection heat flux on the surfaces of
the prototype and the control slab (Liu et al. 2013). It may also beneficially increase air flows
through the system by inducing a suction over the chimney, further cooling the prototype
pavement containing the air conduits.
It is also interesting to notice that the weather conditions don’t have any noticeable effect on the
temperature at the bottom of the prototype, probably due to the high amount of thermal insula-
tion used in the experimental setup and possibly due to the contact with the ground source heat
simulator. The ground source heat simulator was highly insulated with extruded polystyrene foam
and aluminum bubble foil insulation, however, heat losses cannot be completely ruled out.
The main drawback of the analysis of the Pearson’s correlation coefficient is that it does
not provide any information about the actual effectiveness of the experimental setup being
considered.
This information can be extracted directly from the experimental data gathered by plotting
the surface temperature variation with time for both the pavement prototype and the control
slab, as seen in Figure 4. The curves shown in Figure 4 represent the most important result of
this paper, i.e., they are the proof of the effectiveness of the experimental setup tested. In fact,
the oscillations of the surface temperature of the prototype are always smaller than those
seen in the control slab. This means that when the control slab is warm the prototype pave-
ment is always cooler, while when the control slab is cold the prototype pavement is warmer.
Figure 3. Trends in the datasets showing the relationship between outdoor temperature and pressure
with the surface temperature of the prototype (data sorted by increasing surface temperature).
1274
This result provides confirmation of the comments made after the preliminary observation
of the descriptive statistics about the data gathered. The maximum and minimum tempera-
ture differences obtained in the experiments can be quantified as +6°C and −6°C.
Moreover, in Figure 4 it can be observed that the maximum and minimum daily and
nightly temperatures (corresponding to the high and low peaks in the data) change every day.
It is interesting to notice that the temperature reduction or increase is achieved consistently
every day, however high or low the maximum and minimum temperatures get. This aspect
is of high importance, because it is a proof that the system designed behaved reliably over a
fairly long period of time.
The raw data obtained from the datalogging equipment can also be used to assess the effec-
tiveness of the specific configuration chosen for the ground source heat simulator. In par-
ticular, it is possible to plot histograms of the inlet air box temperature and of the outdoor
temperature, as seen in Figure 5. The histograms represented show that the most frequent
inlet temperature obtained is around 15°C, thus, confirming that the experimental setup is
effective for the simulation of a constant soil temperature. The effectiveness of the designed
setup is also shown by the fact that the outdoor temperature goes well below 10°C, while the
inlet temperature is always higher.
The data in Figure 5 is able to show a weakness of the system developed, i.e., the impossi-
bility to simulate a ground temperature when ambient air is above 15°C. As a result, the inlet
air box can reach higher temperatures that may not be representative of a ground heat source.
Therefore, future studies may focus on a more complex system made of heating elements and
a cooler, so that the air box temperature can be kept constant or controlled at all times.
Figure 4. Surface temperature of the prototype vs. Surface temperature of the control slab.
Figure 5. Distribution of the inlet temperatures and of the outdoor temperatures measured.
1275
The experimental setup considered in this paper is currently being tested in the environ-
ment for a longer and more detailed analysis with a higher variety of weather conditions. As
a result, it is expected that the knowledge gathered in this pilot study will be improved in the
next year with further publications by the authors.
4 CONCLUSIONS
In this paper, the use of air convection to mitigate the oscillation in pavement temperatures
was presented and analyzed. The results gathered have a number of consequences for the
infrastructure system. First of all, there is a direct effect, which is an increase of the durabil-
ity of the material, thanks to the reduced thermal stress on the asphalt mixture. In addition,
a number of secondary positive effects come with the use of the system described, e.g.,
(i) the minimum temperatures during winter may be increased, thus, reducing the risk of
freezing and the consequent reliability and safety issues; (ii) the maximum pavement tem-
peratures may be reduced, which would cause a reduction in the thermal energy radiated by
the pavement and of the urban heat island effect; (iii) the reduced pavement temperature
during warm or hot periods would imply lower energy consumption for air conditioning
and, therefore, a lower primary energy consumption in buildings; and (iv) the heat extracted
from the pavement in warm periods could be used for possible applications such as heating
a chosen fluid.
The following main conclusions can be drawn from the analysis of the data gathered:
− It is possible to couple a set of pipes buried under a pavement with a constant tempera-
ture heat source simulating a geothermal reservoir in order to obtain a buoyancy powered
flow.
− The oscillations in the surface temperature can be reduced by ±6ºC with the experimental
setup considered when compared to a control slab that has no air conduits.
− The temperature reduction or increase when compared to a control slab is consistently
achieved no matter what are the maximum and minimum daily temperatures.
− The environmental conditions have a strong linear correlation with the surface tempera-
ture of the pavement studied and with the inlet air box temperature. However, the absence
of strong wind was not studied here due to the low values of air speed recorded in the
period when the study took place.
− The system successfully delivered the required inlet temperature when ambient air was
colder than the design temperature of 15ºC, thereby showing that it is possible to heat up
a pavement in winter by the means of ground source heat.
REFERENCES
Bluestein, M. 2015. Reference Module in Earth Systems and Environmental Sciences, Encyclopedia of
Atmospheric Sciences (Second Edition), Basic Atmospheric Structure and Concepts | Wind Chill.
Elsevier B.V.
Bobes-Jesus, V., Pascual-Muñoz, P., Castro-Fresno, D. & Rodriguez-Hernandez, J. 2013. Asphalt solar
collectors: A literature review. Applied Energy 102: 962–970.
Chiarelli, A., García, A. & Dawson, A. 2015a. Analysis of the performance of an air powered energy
harvesting pavement - in press. Transportation Research Record: Journal of the Transportation
Research Board.
Chiarelli, A., García, A. & Dawson, A. 2015b. Parametric analysis of energy harvesting pavements
operated by air convection. Applied Energy 154: 951–958.
Dai, H., Zhang, K., Xu, X. & Yu, H. 2012. Evaluation of the effect of deicing chemicals on soil and
water environment. Procedia Environmental Sciences 13: 2122–2130.
Ferroplast. FERROTERM Geothermal System. Web: http://www.ferroplast.es/pdf/Tech_Cat_Geo-
thermal_System_FERROTERM_ENG.pdf. Accessed 29/02/2016.
García, A. & Partl, M. 2014. How to transform an asphalt concrete pavement into a solar turbine.
Applied Energy 119: 431–437.
1276
García, A., Hassn, A., Chiarelli, A. & Dawson A. 2015. Multivariable analysis of potential evaporation
from moist asphalt mixture. Construction and Building Materials 98: 80–88.
Golden, J.S. & Kaloush, K.E. 2006. Mesoscale and microscale evaluation of surface pavement impacts
on the urban heat island effects. International Journal of Pavement Engineering 7: 37–52.
Kent, M.G., Altomonte, S., Tregenza, P.R. & Wilson, R. 2015. Temporal variables and personal factors
in glare sensation. Lighting Research and Technology.
Liu, Y. & Harris, D.J. 2013. Measurements of wind speed and convective coefficient on the external
surface of a low-rise building. International Journal of Ambient Energy.
Lund, J.W. 2000. Pavement snow melting. Web: http://www.oit.edu/docs/default-source/geoheat-center-
documents/publications/snow-melting/tp108.pdf ?sfvrsn = 2. Accessed 14/10/2015.
Lund, J.W., Freeston, D.H. & Boyd, T.L. 2011. Direct utilization of geothermal energy 2010 worldwide
review. Geothermics 40: 159–180.
Met Office. Web: http://www.metoffice.gov.uk/public/weather/climate/gcrje93b8. Accessed 26/10/2015.
Mollenhauer, K., Mouillet, V., Pierard, N., Tusar, M. & Gabet, T. 2012. Laboratory aging of asphalt
mixtures: simulation of reclaimed asphalt and application as test method for durability. 5th Euras-
phalt & Eurobitume Congress: hal-00850875.
Santamouris, M. 2013. Using cool pavements as a mitigation strategy to fight urban heat island - a
review of the actual developments. Renewable and Sustainable Energy Reviews 26: 224–240.
Sarat, A.A. & Eusuf, M.A. 2012. An experimental study on observed heating characteristics of urban
pavement. Journal of Surveying, Construction and Property 3: 1–12.
UK climate projections. Web: http://ukclimateprojections.metoffice.gov.uk/media.jsp?mediaid = 87923.
Accessed 06/10/2015.
Wang, D., Linbing, W., Christian, D. & Zhou, G. 2013. Fatigue Properties of Asphalt Materials at Low
In-Service Temperatures. Journal of Materials in Civil Engineering 25(9): 1220–1227.
Wang, H., Zhao, J. & Chen, Z. 2008. Experimental investigation of ice and snow melting process on
pavement utilizing geothermal tail water. Energy Conversion and Management 49: 1538–1546.
1277
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: In this paper, the use of silane, an inorganic compound from semiconductor
industry, is evaluated with the objective of improving the production and the performance of
asphalt mixtures. Two types of asphalt mixtures commonly used in Germany, one for binder
layer and one for wearing course, were prepared with and without silane. Mixing, compaction
effort and volumetric properties were recorded, while low temperature tests were to deter-
mine the evolution the material thermal stress. The results suggest that, for comparable power
demand, particles coating occurs 20% faster when silane is used. The compactability does not
show any difference for both mixture types with and without silane. Similar bulk density was
obtained for all asphalt mixtures. The low temperature behavior was not affected from silane
addition, indicating that this additive can be potentially used for pavement applications.
1 INTRODUCTION
Road pavements in Germany are commonly made of hot asphalt mixture (Kern 2005).
Conventionally, compaction temperature is in the range of 135°C to 160°C. A decrease in
compaction temperature would lead to significant technical, economic and environmental
advantages:
• The construction process would be less dependent on weather and season conditions;
• The time window for compaction would increase with potential benefits in terms of higher
bulk density and lower air voids content resulting in a longer life time;
• A lower energy demand associated to lower natural resources consumption and
exploitation.
One technique to achieve this is the use of wax based additives to modify the viscosity of
bitumen. The use of these additives was attempted in the recent past. For example, the Ger-
man Federal Highway Research Institute (BASt) evaluated and monitored a number of tests
sections where wax additives were used (Radenberg et al. 2006), (BASt 2010). Good work-
ability was observed while the mixing temperature was reduced of 20 °C. Performance tests
showed good results, except in terms of low temperature behavior.
In a research project at the Institut für Straßenwesen (ISBS) at Technische Universität
Braunschweig different wax based additives were used for rolled asphalt to validate the
results from BASt (Renken 2012). The 20 °C temperature reduction could not be replicated
showing that this parameter doesn’t seem to be characteristic for wax based additives. Nev-
ertheless, the time window for compaction could be extended. The worse performance at low
temperature observed in the field sections were confirmed by the laboratory tests. According
to the BASt specification and recommendations, the use of these additives is required for
mastic asphalt (BASt 2010) (TL Asphalt—StB 07/13 2014). However, for health and safety
measures, the mixing temperature was reduced from 250°C to 230°C.
An alternative way to reduce mixing temperature is the use of Zeolites, a crystalline
hydrated aluminum silicate with a water content of ∼20%. During the mixing process water
vaporizes and, by remaining as foam in the bitumen, helps reducing the viscosity of the mix-
1279
ture for some hours. So the mixing and compaction temperature can be reduced of ∼30°C
(Barthel et al. 2004). Additional details can be found elsewhere (D’Angelo 2008).
A different approach is based on the Swedish KGO III technique by Karl Gunnar Ohlson
(Viman et al. 2004): bitumen and coarse aggregate are first mixed together and only, in a sec-
ond stage, filler is added. Using this technique a reduction of mixing temperature of ∼30°C
can be achieved. Nevertheless, this technique is not used in Germany.
Organic silane, commonly used in semiconductor industry as coupling agents to facili-
tate the adhesion between bio-inert materials and specific polymer matrices, represents an
attractive solution for improving the compaction process. This additive creates a stronger
bonding effect of bitumen and aggregates (Cui et al. 2014), (Ranka et al. 2014) based on
an irreversible chemical reaction between silane and aggregate associated to the evapora-
tion of a small amount of water. A known effect of the increased bonding properties is an
improved water resistance as water cannot dissolve the saline-aggregate interaction (Taylor
et al. 2011).
In this paper the temperature reduction effect during mixing and compaction process of
a commercially available silane additive is evaluated. The improved bonding effects and the
minimum remaining amount of water potentially lead to a consistent reduction of the viscos-
ity in the hot asphalt mixtures during production. So a decrease in temperature for mixing
and the following compaction processes should be possible. Due to the experience of poorer
low temperature performance when using wax based additives, thermal stress restrained spec-
imen tests (TSRST) and uniaxial tension stress tests (UTST) were conducted to evaluate the
behavior of mixture containing silane.
In order to evaluate the compaction resistance, an asphalt mixture (AC 16) for binder course
as well as a mixture (AC 11) for wearing course were produced in the ISBS pavement labora-
tory. Both types of asphalt mixtures were prepared according to mix design conventionally
used in Germany (TL Asphalt). The two mixtures were designed with Gabbro aggregate,
limestone filler and with two asphalt bitumens: one plain bitumen having penetration range
50/70 (EN 1426 2007) and one polymer modified bitumen with penetration range 25/55 (EN
1426 2007) and softening temperature of 55ºC (EN 1427 2007), named 25/55-55. Silane addi-
tive was used in 3 out of 4 prepared mixtures. Table 1 presents the mix design for the asphalt
mixtures used in the present research.
1280
The addition of Silane was carried out suing a special mixer with a Cowles agitator for
producing emulsions (see Figure 1). Silane was mixed together with the bitumen and heated
up to 160°C for at least 10 minutes. The newly obtained bitumen was then added to the pre-
heated aggregates in the mixer and the mixing process performed.
Both asphalt mixture were prepared with and without silane compound so that the last
plain mixtures could be used as control. In addition, different compaction temperatures were
selected. Table 1 provides a summary of the experimental plan.
The same compaction method (load, number of roll passes, etc.) was used for each mixture type;
only compaction temperatures were different. Four asphalt slabs of sizes 300 × 260 × 40 mm3
and 300 × 260 × 50 mm3 were produced for each mixture AC 11 and AC 16.
3 EXPERIMENTAL PROCEDURES
The mixing process was performed according to the current German Standard TP Asphalt-
StB, part 35 (TP Asphalt-StB, Teil 35 2009) which based on the European standard EN
12697-35 (EN 12697-35 2004). During the two minute mixing, all eight asphalt mixture
types were filmed and then the time for a specified coating percentage of aggregate was
visually estimated. For evaluation, four degrees of coating were selected: 50%, 75%, 90%
and 100%.
During the mixing process, the laboratory mixing device was used to record the torque of
the mixing agitator attached to the driveshaft and of the mixing drum and the data converted
into power [W]. The accumulated power energy [Ws] = [J] from the agitator and drum are
specified in a diagram over a two minute mixing duration.
The compactability T was determined on the basis of the change in thickness of the Mar-
shall sample during the compaction procedure according to German Standard TP Asphalt-
StB, part 10 (TP Asphalt-StB, Teil 10 2014) which is derived from EN 12697-10 (EN 12697-10
2002). On the basis of the reducing sample thickness, due to the number of impacts, an
exponential function can be determined whose parameter T provides an estimation of the
material compactability. Small values of T are typical of materials easy to compact, whereas
larger values are representative of asphalt mixtures that require significant effort for achiev-
ing the desired compaction. In addition, bulk density, ρb,SSD (TP Asphalt-StB, Teil 6 2012),
(EN 12697-6 2012), maximum density, ρm (TP Asphalt-StB, Teil 5 2013), (EN 12697-5 2009),
and air voids content V (TP Asphalt-StB, Teil 8 2012), (EN 12697-8 2003) were measured for
each asphalt mixture slab produced.
Figure 1. Cowles agitator (left) and mixer used (right) for blending bitumen and silane.
1281
Table 2. Experimental plan.
Asphalt Compaction
ID Mixture Bitumen Silane Temperature
1a AC 16 50/70 no 135°C
1b Yes 135°C
1c Yes 115°C
1d Yes 95°C
2a AC 11 25/55-55 A no 145°C
2b yes 145°C
2c yes 125°C
2d yes 105°C
1a 1b 1c 1d 2a 2b 2c 2d
50% 25 s 15 s 19 s 19 s 37 s 22 s 30 s 30 s
75% 35 s 22 s 30 s 29 s 45 s 29 s 38 s 39 s
90% 50 s 32 s 40 s 43 s 55 s 42 s 44 s 46 s
100% 67 s 47 s 50 s 55 s 80 s 56 s 66 s 61 s
Two slabs of each asphalt mixture were cut into prismatic specimen with the dimensions
of 40 × 40 × 160 mm3 for AC 11 and 50 × 50 × 160 mm3 for AC 16. The specimens were sub-
jected to low temperature tests according to EN 12697-46: TSRST and UTST. In TSRST the
specimen is restrained and the two ends of the sample are kept at the same distance, while
temperature is reduced with a rate of 10 K/h. As thermal shrinkage is prevented, a kryogenic
stress develops until failure occurs. By recording the kryogenic stress, a curve of stress versus
temperature, the failure temperature and failure stress can be specified. In UTST specimen is
cooled down to test temperature avoiding any stress to build up. Then the specimen is pulled
until failure. The selected UTST temperatures were +20 °C, +5 °C, -10 °C and -25 °C; hence,
a minimum number of 10 specimens were needed. For each temperature the maximum stress
is recorded as tensile strength and all results are then used to generate the regression curve of
strength as function of temperature.
4 RESULTS
1282
Figure 2. Accumulated power demand [W] during the mixing processes for mixtures AC 16 (left) and
AC 11 (right).
Asphalt Compaction
ID mixture temperature Compactability T
1a AC 16 135 °C 41,6 Nm
1b 135 °C 43,5 Nm
1c 115 °C 42,8 Nm
1d 95 °C 41,3 Nm
2a AC 11 145 °C 37,3 Nm
2b 145 °C 34,2 Nm
2c 125 °C 36,5 Nm
2d 105 °C 36,5 Nm
Asphalt mixture AC 16 AC 11
ID 1a 1b 1c 1d 2a 2b 2c 2d
Compaction Temperature °C 135 135 115 95 145 145 125 105
Bulk Density g/cm3 2,511 2,508 2,506 2,513 2,508 2,500 2,491 2,483
Max. Density g/cm3 2,684 2,711 2,713 2,688 2,618 2,615 2,627 2,622
Air Voids Content Vol.% 6,4 7,5 7,6 6,5 4,2 4,4 5,2 5,3
Mean Max. Density g/cm3 2,699 2,621
Mean Air Voids Content Vol.% 7,0 7,1 7,2 6,9 4,3 4,6 5,0 5,3
Figure 3. Results of UTST and TSRST: Tensile strength and cryogenic stress as function of temperature
for mixtures AC 11 (left) and AC 16 (right).
shows that the tensile strength of AC 16 is not as high as the one exhibited by AC 11 with
slightly shifted maximum toward higher temperatures.
With respect to the tensile strength, both plots indicate that there are no significant differ-
ences between asphalt mixtures with or without silane. However, thermal stresses show small
differences. The asphalt mixture with silane presents a clear trend toward lower stresses at
lower temperatures with respect to the asphalt mixture without silane (solid lines in figure 3).
As there’s no distinct trend for failure temperature or failure stress, only the trend of lower
thermal stresses can be noted.
In this research the influence of the silane binding agent on the mixing process and to com-
paction properties of asphalt mixture and their impact on low temperature behavior was
investigated. For this purpose, two typical asphalt mixtures used in Germany, one for wear-
ing course and one for binder course, were prepared. During the mixing processes, the power
consumption of the laboratory mixer was recorded and the time that the degrees of coating
of 50%, 75%, 90% and 100% were reached was recorded. In addition compactability and
air voids content was measured. Prismatic specimens were cut of the slabs and TSRST and
UTST were conducted.
The following conclusions can be drawn:
• Coating requires, on average, 20% less time when using silane.
• Less power consumption was observed for binder layer mixture when adding silane to the
mix design.
1284
• The compactability does not show any difference for both asphalt mixture types with and
without binding agent.
• The maximum density of the asphalt mixtures produced can be considered equivalent, while
the bulk densities of the roller compacted mixture slabs indicate minimal differences.
• An effect of compaction temperature on air voids content was observed only for the wear-
ing course mixture with increasing values for lower compaction temperatures. Neverthe-
less, the variation of air voids content was not very remarkable.
• The low temperature behavior was not affected from silane addition. In case TSRST
slightly better results were observed.
Overall, when including silane in the mix design, aggregate coating time is reduced as
well as compaction temperature resulting in air voids contents comparable to those obtained
without binding agents. The negative impact of wax additives to low temperature behavior
can be avoided. The results of this research, although at the preliminary stage, are very prom-
ising, opening the opportunity for a follow up study.
REFERENCES
Barthel, W., Marchand, J., Von Devivere, M. 2004,“Warm Mix Asphalt by Adding a Synthetic Zeolite,”
Proceedings of the Third Eurasphalt and Eurobitume Conference, Book 1, Foundation Eurasphalt,
Breukelen, The Netherlands, 2004, 1241–1249.
BASt 2010. Erfahrungssammlung über die Verwendung von Fertigprodukten und Zusätzen zur Temper-
aturabsenkung von Asphalt, Bundesanstalt für Straßenwesen (BASt), Bergisch-Gladbach, Novem-
ber 2010.
Cui, Blackman, Kinloch, Taylor. Durability of asphalt mixtures: Effect of aggregate type and adhesion
promoters. International Journal of Adhesion and Adhesives, Volume 54, p. 100–111, Oct. 2014.
D’Angelo, Harm, Bartoszek, Baumgardner, Corrigan, Cowsert, Harman, Jamshidi, Jones, Newcomb,
Prowell, Sines and Yeaton. Warm-Mix Asphalt: European Practice. Federal Highway Administration,
U.S. Department of Transportation, American Association of State Highway and Transportation
Officials, National Cooperative Highway Research Program, FHWA-PL-08-007, February 2008.
EN 1426 2007, Bitumen and Bituminous Binders - Determination of Needle Penetration, European
Committee for Standardization. 2007.
EN 1427 2007, Bitumen and Bituminous Binders - Determination of the Softening Point—Ring and
Ball Method, European Committee for Standardization. 2007.
EN 12697-5 2009. Bituminous Mixtures. Test Methods for Hot Mix Asphalt. Determination of the
Maximum Density, European Committee for Standardization. 2009.
EN 12697-6 2003. Bituminous Mixtures. Test Methods for Hot Mix Asphalt. Determination of Void
Characteristics of Bituminous Specimens, European Committee for Standardization. 2003.
EN 12697-8 2012. Bituminous Mixtures. Test methods for Hot Mix Asphalt. Determination of Bulk
Desnity of Bituminous Specimens, European Committee for Standardization. 2012.
EN 12697-10 2002. Bituminous Mixtures. Test Methods for Hot Mix Asphalt. Compactibility,
European Committee for Standardization. 2002.
EN 12697-35 2004. Bituminous mixtures. Test methods. Part 35: Laboratory mixing, European
Committee for Standardization. 2004.
EN 12697-46 2012. Bituminous Mixtures. Test Methods for Hot Mix Asphalt. Low Temperature Crack-
ing and Properties by Uniaxial Tension Tests, European Committee for Standardization. 2012
Kern 2005. Faszination Strassenbau, Motorbuch Verlag, ISBN 3-613-02499-3, Stuttgart.
Radenberg, Bielenberg, Damm 2006. Nachuntersuchungen an ausgesuchten Streckenabschnitten zur
Beurteilung des Langzeitverhaltens von TA-Asphalten. Abschlussbericht zum Forschungsprojekt
FE 89.0161/2005/AP im Auftrag der Bundesanstalt für Straßenwesen (BASt), Wahlstedt, August
2006.
Ranka, Mehta 2014. AAPT/ISAP International Forum: Nanotechnology Organosilane Compounds for
Chemical Bonding in Road Construction. Asphalt Paving Technology 2014, Volume 83, The Associa-
tion of Asphalt Paving Technologists, (AAPT), Lancaster, 2014.
Renken, P. 2012. Walzasphalte mit viskositätsabsenkenden Additiven—Entwicklung und Optimierung
der Eignungs—und Kontrollprüfungsverfahren und Bestimmung der Einflüsse auf die performance-
orientierten Asphalteigenschaften. AiF-Forschungsvorhaben Nr. 15589 N, Technische Universität
Braunschweig, Institut für Straßenwesen (ISBS), April 2012.
1285
Taylor, Willis 2011. Effects of Nanotac Additive on bond strength and moisture resistance of tack
coats. Final report, National Center for Asphalt Technology, Auburn University, Auburn, Alabama,
November 2011.
TL Asphalt-StB 07/13, Technische Lieferbedingungen für Asphaltmischgut für den Bau von Verkehrs-
flächenbefestigungen. FGSV-Nr. 797, FGSV-Verlag, Köln, January 2014.
TP Asphalt-StB, Teil 5 2013. Rohdichte von Asphalt. FGSV-Nr. 756/5, 2013, FGSV-Verlag, Köln.
TP Asphalt-StB, Teil 6 2012. Raumdichte von Asphalt-Probekörpern. FGSV-Nr. 756/6, August 2012,
FGSV-Verlag, Köln.
TP Asphalt-StB, Teil 8 2012. Volumetrische Kennwerte von Asphalt-Probekörpern und Verdichtungs-
grad. FGSV-Nr. 756/8, August 2012, FGSV-Verlag, Köln.
TP Asphalt-StB, Teil 10 2014. Verdichtungswiderstand mit Hilfe des Marshall-Vedichtungsgerätes.
Verfahren B: Änderung der Probekörperdicke. FGSV-Nr. 756/10B, Mai 2014, FGSV-Verlag, Köln.
TP Asphalt-StB, Teil 35 2009. Asphaltmischgutherstellung im Laboratorium. FGSV-Nr. 756/35,
February 2009, FGSV-Verlag, Köln.
Viman, Wendel, Said 2004. Long-Term Performance of flow mixing technique. proceedings of 3rd
Eurasphalt & Eurobitume Congress, 13–14 May 2004, Vienna.
1286
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Peiqing Wang
Highway Administration of Taizhou City, Taizhou, Zhejiang, China
Fujian Wang
Institute of Transportation Engineering, Zhejiang University, Hangzhou, Zhejiang, China
Mingqiang Ke
Highway Administration of Taizhou City, Taizhou, Zhejiang, China
Zhigang Lu
Zhejiang Scientific Research Institute of Transport, Hangzhou, Zhejiang, China
Houquan Zhang
Highway Administration of Taizhou City, Taizhou, Zhejiang, China
Zhiyuan Zeng
Ji Zhun Fang Zhong Architecture Design Co. Ltd., Chengdu, Sichuan, China
1 INTRODUCTION
Asphalt pavement is widely used in advanced road construction due to its excellent perform-
ance. Adding fiber to asphalt mixture is a significant way to improve its pavement perform-
ance. Compared with other kinds of fiber, basalt fiber has unique advantages such as good
mechanical property, surface wettability, chemical stability, aging resistance, water stability,
remarkable high and low temperature resistance, and hygroscopicity, etc. (Guo et al., 2010;
Rabinovich et al., 2001). It is necessary to study on basalt fiber reinforced asphalt mixture to
improve the performance and service life of asphalt pavement.
Researchers throughout the world have studied a lot on basalt fiber. Li et al analyzed
temperature field and temperature stress of cement concrete pavement under periodic vari-
able low temperature (Li et al., 2014); Liu et al studied the high temperature stability, water
stability and low-temperature crack resistance of AC-16 reclaiming asphalt mixture with dif-
ferent proportion of old material (Liu et al., 2014); Garcia et al proved the good behavior of
1287
fiber reinforce plastic with basalt fiber when applied to timber beams (Garcia et al., 2013);
Subagia et al showed that the interplay hybrid composite with carbon fiber at the compres-
sive side exhibited higher flexural strength and modulus than when basalt fabric was placed
at the compressive side (Subagia et al., 2014); Jiang’s experiment indicated that the concrete
containing basalt fiber presents higher porosity(Jiang et al., 2014).
Nowadays investigation on basalt fiber reinforced asphalt concrete concentrates on labo-
ratory test which includes basalt fiber reinforced asphalt cement and pavement performance
of basalt fiber reinforced asphalt mixture (Guo et al., 2007). There are few investigations
on structural analysis of basalt fiber reinforced asphalt pavement, especially under moving
loads. This paper built finite element model on basalt fiber reinforced asphalt pavement in
ABAQUS based on completed wheel tracking test and according numerical simulation. And
then mechanical characteristics and time-history responses of basalt fiber reinforced pave-
ment under moving loads were analyzed.
Laboratory test on basalt fiber reinforced asphalt mixture had been completed, which inves-
tigated the mixture performance and were used for pavement structural analysis. Viscoelastic-
ity parameter was obtained through generalized Maxwell model which is connected parallelly
by several Maxwell models and the Maxwell model is a series connection of Young model
and viscosity coefficient. Rutting numerical simulation was also conducted in ABAQUS and
the result was compared to that in laboratory test.
1288
Table 1. Immersion Marshall test result on AC-13C.
Results in Table 1 are averages of four Marshall specimens. One can see the AC-13C
immersion residual stability of both conditions satisfies requirements. When basalt fiber dos-
age is 0.3%, the immersion residual stability increases by 2.2%, which shows basalt fiber can
indeed improve asphalt mixture’s water stability.
∫ G (t − )J ( ) dτ = t
t
(1)
0
where G(t) is relaxation elasticity modulus, J(t) creep compliance, t time and τ relaxation time.
According to time-temperature equivalence principle (Friedrich et al., 2000), for simple
rheological materials, one can use shift factor αT to perform time-temperature conversion for
relaxation modulus in different temperature. The conversion formula is equation (2):
G (T ,lgtt ) G (T , lg t − lg T ) (2)
From equation (2) we know if we move the relaxation modulus curve in reference tempera-
ture T0 along the logarithm time axis for a distance αT, we can get relaxation modulus curve
in any temperature T. This paper selected 25°C as a reference temperature and generalized
Maxwell model was used to describe viscoelasticity of basalt fiber reinforced asphalt mixture.
Equations (1) and (2) were used to fit relaxation modulus curve in 25°C reference tempera-
ture (Figure 2). The curve fits well to the calculation result in creep test and the correlation
coefficient reaches 0.946, which shows generalized Maxwell model can simulate viscoelastic-
ity of asphalt mixture satisfactorily.
1289
the “part” module of ABAQUS, with the size of 300 mm × 50 mm. The boundary condi-
tion of finite element model was fixing the bottom and restricting both sides’ displacement
in horizontal direction. The test temperature was 60°C. CPE4R (4-node bilinear, reduced
integration element) type of two-dimension-plane strain unit was used to divide the network
as shown in Figure 3. Wheel load was simplified with the peak value of 0.7 MPa (Figure 4).
The load moved from point O to D and we aimed to obtain the subsidence of point B. The
load was simplified as triangle shape and divided the vehicle’s moving length into four equal
parts with each part’s length of 57.5 mm.
For wheel tracking test, the vehicle rolled 42 times/min and moved the distance of 230 mm,
and the wheel is 50 mm wide. Then we calculated the moving velocity V = (42 × 0.23)/60 = 0.161
(m/s) and rolling period Tr = 60/42 = 1.429 (s), and the moving time for each phase is
TOA = TAB = TBC = TCD = Tr /4 = 0.35725 s.
The first 100 s local enlarged drawing of the vertical subsidence curve of the load’s central
acting point is shown as Figure 5; one can see vertical displacement fluctuates with time and
its variation form is the same as that of the load. The curve’s upper end ligature is the rutting
plate’s permanent deformation, while the lower end ligature is total deformation. In the early
loading, rutting increases dramatically, while as time goes by the deformation gradually tends
to stable, which corresponds to compression and flow phase of asphalt mixture respectively.
When loading is over, the permanent deformation of rutting plate is 1.44 mm, slightly smaller
than the laboratory test result. Table 2 lists comparison between laboratory tests and rutting
simulation results for 45 min and 60 min, respectively.
1290
Figure 5. Local enlarged drawing of vertical displacement at the loading center (Displacement: mm;
Time: s).
Time (min) Simulation Results (mm) Laboratory Test Results (mm) Difference (mm)
From Table 2 we find the rutting depth of simulation is slightly smaller than laboratory
test. The former is smaller by 7.3% for 45 min situation and only 0.7% for 60 min. Generally
speaking, finite element simulation and laboratory test have good consistency, which further
reveals that the fitted viscoelasticity parameters are appropriate.
Elastic layer system theory, which assumes that load on pavement is static and uniformly
distributed, is used for asphalt pavement design. However, the mechanical characteristics of
pavement are complicated. Pavements under loads from moving vehicles bear huge dynamic
effects. Stress and deformation vary along with the variation of load and temperature. There
is a great error to analyze the dynamic response of asphalt pavements by static loads. In
fact, the loads which vehicles impose to pavements include vertical forces and horizontal
forces, because vehicles are subjected to loads from both directions when braking. In this sec-
tion, 3D finite element model is built to analyze dynamic response of basalt fiber reinforced
asphalt pavement under moving loads in ABAQUS.
Ei
Rectangular load distribution which is 0.7 MPa was applied to simulate wheels and the
scale of loading area is 192 mm × 186 mm. The vehicle was moving at a speed of 90 km/h, and
its moving distance is 2976 mm and divided into 48 parts, on each of which the wheel takes
0.00248 s. This means each time step is 0.00248 s and the entire computation time is 0.114 s
by getting rid of first and final rectangle. The wheel accounts for three rectangles for one time
step as shown in Figure 7 and moves forward for one rectangle in a time step. For instance,
1292
the wheel takes rectangles 1, 2 and 3 in the first time step and then rectangles 2, 3 and 4 in the
second time step. The more rectangles are divided, the more accurate the result is.
Figure 8. Vertical displacement variation at different layers under the middle of loading area (P in
Figure 7).
1293
Figure 9. Vertical stress variation at different layers under the middle of loading area (P in Figure 7).
Figure 10. Horizontal stress variation at different layers under the middle of loading area (P in
Figure 7).
Figure 11. Horizontal shear stress variation at different layers under the middle of loading area (P in
Figure 7).
1294
Figure 12. Maximum shear stress variation at different layers under the middle of loading area (P in
Figure 7).
Figure 13. Horizontal shear stress variation at different layers under the middle of loading area (P in
Figure 7).
Figure 14. Maximum shear stress variation at different layers under the middle of loading area (P in
Figure 7).
Table 5. Maximum shear stress comparison under different load types (MPa).
1295
It can be seen from Figure 14 that there is a fluctuation on middle surface layer curve
and the maximum shear stress with the value of 0.349 MPa is at lower surface layer bottom.
Maximum shear stress at basement layer is much smaller than that at other layers. Maximum
shear stresses at upper and middle surface layers have increased more compared to that with-
out horizontal load. Comparison has been listed in Table 5, from which we can conclude that
horizontal load has a significant influence on maximum shear stress at asphalt surface layer.
4 CONCLUSIONS
Semi-rigid asphalt pavement 3D finite element model was built by ABAQUS based on
completed laboratory test of basalt fiber reinforced asphalt mixture and relevant numerical
simulation was conducted. Dynamic responses of asphalt pavement under moving loads con-
sidering rheological property in asphalt were analyzed. Main conclusions are as follows:
1. Vertical compression stress and shear stress are pretty large under moving loads. The alter-
nating of horizontal stress and shear stress are the main cause of fatigue failure in asphalt
pavement structure.
2. Horizontal moving load has less influence on vertical displacement, vertical and horizon-
tal stress, but greater influence on shear stress at surface layer than vertical moving load
does. Greater shear stress can generate at surface layer under both horizontal and vertical
moving loads which can result in shear failure.
3. It is necessary to mix basalt fiber into asphalt surface materials to improve load capacity
as stress at asphalt surface layer is much great under moving loads.
ACKNOWLEDGEMENT
This work was financially supported by a grant from the Department of transportation in
Zhejiang province (Grant Number: 2014H55).
REFERENCES
Fan, W., Kang, H. & Zheng, Y. 2010. Experimental study of pavement performance of basalt fiber-
modified asphalt mixture. Journal of Southeast University (English Edition) 26(4): 614–617.
Friedrich, M., Schulze, A. & Prosch, G. et al. 2000. Investigation of Chemically Treated Basalt and
Glass Fibers. Mickrochimica Acta 133(114): 171–174.
Guo, H., Ma, Y. & Chen, S. et al. 2010. Development and Application Prospect of Continuous Basalt
Fiber. China Fiber Inspection (5): 76–79. (in Chinese).
Guo, N. & zhao, Y. 2007. Viscoelastic Response of Asphalt Pavement under Moving loads. Journal of
Shenyang Jianzhu University 23(6): 922–925. (in Chinese).
Garcia, P.D., Escamilla, A.C. & Garcia, M.N.G. 2013. Bending Reinforcement of Timber Beams with
Composite Carbon Fiber and Basalt Fiber Materials. Composites Part B-Engineering 55: 528–536.
Jiang, C.H., Fan, K. & Wu, F. et al. 2014. Experimental Study on the Mechanical Properties and Micro-
structure of Chopped Basalt Fiber Reinforced Concrete. Materials & Design 58: 187–193.
Li, G., Chen, C. & Jiang, B. et al. 2014. Research on the Behavior of Rigid Pavement of Basalt Fiber
Reinforced Dowel Bar under the Condition of Variable Temperature. Applied Mechanics and Materi-
als 579: 1209–1215.
Liu, X. et al. 2014. Study on Pavement Performance of High Volume Reclaiming Asphalt Mixture.
Advanced Materials Research 1081: 260–264.
Park, S.W. & Kim, Y.R. 1999. Interconversion between Relaxation Modulus and Creep Compliance for
Viscoelastic Solids. Journal of Materials in Civil Engineering 11: 76–82.
Rabinovich, F.N., Zueva, V.N. & Makeeva, L.V. 2001. Stability of Basalt Fibers in a Medium of Hydrat-
ing Cement. Glass and Ceramics 58(11–12): 431–434.
Subagia, A., Kim, Y. & Tijing, L.D. et al. 2014. Effect of Stacking Sequence on the Flexural Properties of Hybrid
Composites Reinforced with Carbon and Basalt Fibers. Composites Part B-Engineering 58: 251–258.
Zhao, Y. & Zhong, Y. 2009. Dynamic viscoelastic analysis of asphalt pavements. Journal of Vibration
and Shock 28(9): 159–162. (in Chinese).
1296
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Zhigang Zhang
Southeast University, Nanjing, P.R. China
Qian Zhang
University of Louisiana, Lafayette, LA, USA
Shunzhi Qian
Nanyang Technological University, Singapore
Victor C. Li
Southeast University, Nanjing, P.R. China
University of Michigan, Ann Arbor, USA
ABSTRACT: In this study, new ECC mixture designs have been developed using a type of
low-cost PVA fiber without oil-coating. High fly ash content, fly ash cenosphere and crumb
rubber were incorporated into the mixture design to compensate for the micromechanical
change associated with using the low-cost PVA fiber. Compressive and tensile test were con-
ducted to characterize the performance of the newly developed ECC mixtures. ECC with
tensile strength of 5.6 MPa and tensile strain capacity of 1.9% has been successfully devel-
oped using such fiber. For applications requiring higher tensile ductility, by increasing fly ash
content and incorporating FAC or crumb rubber, the tensile ductility of the new ECCs can
be further enhanced with moderate decrease in compressive and tensile strength. By incorpo-
rating low-cost PVA fiber, the cost-effectiveness of ECC has been greatly improved, which is
expected to greatly promote the application of ECC in pavement engineering field in China.
1 INTRODUCTION
Normal concrete pavement undergoes repeated vehicle loading and environmental loading
during service, such as freezing-and-thawing cycles, restrained shrinkage, and temperature
changes. Cracking in concrete is inevitable due to its brittleness that results in the loss of
mechanical properties and corrosion of re-bar, thus shortening its service life. The frequent
maintenance and repair throughout the service life cause significant negative life-cycle eco-
nomic, social, and environmental impacts. In order to enhance the durability of concrete
pavement, reduce the maintenance work and ultimately extend its service life, a kind of
crack-free concrete material with high tensile ductility is desired.
Engineered Cementitious Composites (ECC) belong to the family of High-Performance
Fiber-Reinforced Cementitious Composites (HPFRCC), which have been developed over the
last two decades as a ductile alternative to conventional concrete. Unlike brittle concrete,
ECC features a metal-like pseudo strain-hardening behavior with high tensile strain capacity
up to 3 to 5%, which is 300∼500 times that of conventional concrete (Li 1998, Li 2001). In
contrast to the brittle fracture under tension commonly observed in normal concrete, multi-
ple fine cracks (typically less than 100 μm wide) develop in ECC under tension before final
fracture, therefore resulting in highly ductile behavior. With high tensile ductility and tight
crack width, ECC overcomes many challenges faced by concrete material associated with
1297
brittleness and cracking, greatly improving the structural and durability performance of the
infrastructure (Sahmaran et al. 2007, Qian et al. 2013, Li 2011)
Apart from performance, the cost of the material is also important for large-scale applica-
tions of ECC in the construction. The history of ECC development also reflects the con-
tinuous endeavor of finding more economical polymer fibers. ECC was originally developed
using Polyethylene (PE) fibers (Li 1998). However, PE fiber is almost prohibitively expensive
to use in large-scale construction applications. In search for a lower cost alternative, Polyvinyl
Alcohol (PVA) fiber was selected to produce PVA-ECC (Kanda & Li 1998, Li et al. 2002,
Lin et al. 1999). The price of PVA fiber in US is about 1/8 of that of PE fiber, thus greatly
reducing the cost of ECC material. To better adopt PVA fibers into ECC design, collaborat-
ing with a Japanese fiber manufacturer, Kuraray Co. Ltd., researchers at the University of
Michigan found that oil coating the surface (typically 1.2% oiling coating by fiber weight)
of PVA fiber lead to better tensile behavior (Li et al. 2002). This process reduces excessive
interfacial bond between the hydrophilic PVA fibers and cementitious matrix, which results
in desired tensile behavior of the PVA-ECC. The oil-coated PVA fibers have since been used
extensively in the development of ECC materials.
With recent rapid economic growth in China, China has seen a huge construction boom.
With such background, scholars and professionals have also been working on introducing
ECC technology to China in the past decade (Pan et al. 2015, Qian & Zhang 2012).
One of the major obstacles for widely adopting ECC in China is the high cost of Kura-
ray PVA fiber (in Chinese market). Kuraray PVA fiber is very expensive in China (around
140 RMB (23 US$)/kg), significantly higher than that in US (6.6 US$/Kg). Such price,
although considerably cheaper than PE fiber, significantly increases the initial material
cost of ECC (in China), given that the fiber typically makes up more than 50% of the total
material cost. Therefore the high material cost greatly limited the broader applications of
ECC in China.
To reduce ECC cost in China, a low-cost PVA fiber manufactured in China is used in
attempt to produce Low-Cost ECC (LC-ECC) in this study. In the present stage, the ten-
sile strength and elastic modulus of this PVA fiber are both notably lower than those of
Kuraray PVA fiber. Furthermore, unlike the Kuraray PVA fiber typically used in current
ECC mixtures, this PVA fiber has no oil-coating on the surface, which could potentially
lead to excessive chemical bond between the fiber and surrounding matrix. Both are consid-
ered undesirable according to micromechanics underlying ECC design (Li et al. 2002, Lin
et al. 1999). With these different fiber properties, re-tailoring of the ECC mixtures might be
needed to achieve mechanical behavior comparable to the Kuraray PVA-ECC.
In this paper, new ECC mixtures have been developed using the Chinese PVA fibers. High
fly ash content, crumb rubber and Fly Ash Cenosphere (FAC) were also incorporated to fur-
ther enhance the tensile ductility of the mixtures by reducing the matrix fracture toughness
and modifying the interface properties between fiber and matrix. The compressive and tensile
behaviors of the newly developed mixtures were experimentally investigated. The underlying
micromechanics were also discussed in this paper. All experimental procedures and findings
were documented in the following sections.
Unlike the typical trial-and-error material development methodology, ECC is designed based
on micromechanics. The micromechanics-based design theory links the microstructural and
micromechanical properties (matrix toughness, fiber property, matrix/fiber interfacial prop-
erty, etc.) to the macro-level material properties (tensile strength, tensile ductility, cracks
width, etc.). In this section, the ECC design theory is briefly reviewed. This will help the
readers to understand the discussion in the later sections.
The core of the ECC design theory is the strain-hardening conditions (Lin et al. 1999, Li
1992). Two conditions (energy and strength based criteria) need to be satisfied in order to
achieve tensile strain-hardening behavior. The two criteria are stated as follows:
1298
δ0
Energy based criterion:JJtiip δ ∫0
δ dδ ≡ J b′ (1)
Where σ0 is the maximum bridging stress corresponding to the crack opening δ0; σcs is the
cracking strength of the matrix; Jtip is the fracture energy of the matrix, which is approxi-
mately equal to Km2/Em, where Km is the matrix fracture toughness, and Em is the matrix
Young’s modulus. Equation1 describes the energy balance in the crack extension process.
Failure to meet this criterion results in the Griffith crack mode with unrestricted crack width
as the crack extends (Griffith 1921). The composite ingredients should be tailored to achieve
sufficient margin between complementary energy of fiber bridging J′b and crack tip tough-
ness Jtip to enable saturated multiple cracking and robust tensile ductility (Kanda & Li 1998,
Marshall & Cox 1988, Li & Leung 1992). This requires limiting the matrix fracture tough-
ness (that limits Jtip) and/or increasing the J′b by adjusting fiber properties (volume fraction,
length, diameter, tensile strength and modulus) and fiber/matrix interfacial properties (fiber/
matrix frictional and chemical bonds). Equation 2 describes the strength criterion of multiple
cracking, which requires the fiber bridging strength to be higher than the matrix cracking
strength that is a function of the matrix defect size. If either inequality (Equation 1 or 2) is
not satisfied, the composite fails with a single localized fracture, instead of multiple cracking,
and a typical tension-softening behavior of normal fiber reinforced concrete is observed.
In ECC design, the ratios PSHE = J′b/Jtip and PSHS = σ0/σcs are considered as multiple-
cracking potential indicators (Wang & Li 2007). Higher PSH means greater margin for satis-
fying the strain hardening criteria and thus usually leads to higher tensile ductility (Wang &
Li 2007, Kanda & Li 1999). For PVA-ECC, the PSHE is particularly of interest. PVA-ECC
tends to have narrower margin for satisfying the energy criterion because of lower J′b associ-
ated with excessive interfacial chemical bond and relatively low fiber strength. Compared
to Kuraray PVA fibers, Chinese PVA fibers have notably lower strength (by 22%). It is also
expected that the interfacial chemical bond will increase since the Chinese fibers were not oil-
coated. These might contribute to further lowered J′b, PSHE, and therefore, tensile ductility.
Based on above discussion, when incorporating the Chinese PVA fiber in ECC design, in
order to compensate the potential adverse effects, retailoring of the composition might be
needed to restore the J′b/Jtip and enhance the tensile ductility. For a given type of fiber, this
might involve lowering the chemical bond (Gd) through matrix modification and lowering the
matrix toughness (Jtip). Three approaches explored in the present study include incorpora-
tion of high fly ash content, Fly Ash Cenosphere (FAC) and crumb rubber into the mixture
design. All approaches were found to lower the interfacial chemical bond and /or the matrix
toughness according to previous researches (Wang & Li 2007, Zhang et al. 2015, Zhang et al.
2014, Huang et. al 2013), and therefore are plausible for the present study.
3 EXPERIMENTAL PROGRAMS
Figure 1. Constituent materials (a) crumb rubber and (b) fly ash cenosphere used for ECC mixtures.
Figure 2. Particle size distributions of fly ash cenospheres and crumb rubber.
Crumb
Mix Cement Fly ash Sand rubber FAC Water HRWR PVA
1300
The process of specimen preparation can be found in reference (Zhang 2014). All of speci-
mens were cured in air for 28 days before tests at room temperature of 20 ± 3°C and Relative
Humidity (RH) of 40 ± 5%.
1301
Figure 4. Measured compressive strengths of ECC mixtures.
rubber surface during mixing process according to previous researches on rubberized con-
crete (Zhang et al. 2015). As evidence, more air voids can be seen on the crossing section of
ECC specimen with crumb rubber as shown in Figure 5, compared with M2.2-C. Neverthe-
less, these mixtures (with compressive strengths above 30 MPa) are still considered suitable
for pavement or road repair applications.
1302
Figure 6. The representative tensile stress-strain curves of ECC mixtures.
Mix Ef (GPa) σfu (MPa) Gd (J/m2) τ0 (MPa) β σ0 (MPa) J’b (J/m2) PSHE
relationship developed by Yang et al. (2008) was used for the micromechanical analysis. The
inputs for the model, including the fiber properties and fiber/matrix interfacial properties
of M2.2-K and M2.2-C are listed in Table 4. In Table 4, Ef and σfu are the in-situ fiber elas-
tic modulus and tensile strength, which are lower than the nominal values listed in Table 1.
PVA fibers show reduced modulus and strength when pulled out from the matrix due to the
fibrillation of the fibers (Kanda & Li 1998). Therefore, when calculating the fiber bridging
behavior, the in-situ fiber elastic modulus and strength should be used instead of the nomi-
nal values. The listed Ef and σfu values for Kuraray PVA fibers are adopted from literature
(Yang et al. 2008). Due to lack of experimental data, these values for Chinese PVA fibers are
assumed based on the nominal values. Interfacial parameters Gd, τ0, and β of Kuraray PVA
fibers within M2.2-K matrix were measured using single fiber pull-out test by the authors
and listed in Table 4. The detailed procedure of the test can be found in references (Redon
et al. 2001). The interfacial properties between the Chinese PVA fibers and M2.2-C matrix
(same as M2.2-K) were assumed to be the same as those of the Kuraray PVA fibers due to
lack of experimental data, as listed in the table 4. When calculating the fiber bridging behav-
ior, random 2-D distribution of fibers within the matrix was assumed. The calculated fiber
bridging strength σ0, fiber bridging complementary energy J’b and PSHE are listed in Table
and the calculated σ(δ) curves are plotted in Figure 8.
From Table 4 and Figure 7, it can be found that σ0 and Jb’ of M2.2.-C are both lower than
those of M2.2-K due to the lower fiber strength and modulus of the Chinese PVA fibers.
Given the same matrix proportion and therefore the same crack tip toughness Jtip (measured
1303
Figure 7. Calculated fiber bridging σ(δ) relationships.
to be 16 J/m2), the PSHE index decreases from 3.4 (for M2.2-K) to 2.7 (for M2.2-C). Further-
more, the actual fiber bridging curve of M2.2-C should further shift leftwards (and slightly
upwards) if higher interfacial chemical bond due to the non-oil-coated fibers is accounted
for, leading to an even greater decrease in J′b and PSHE. The recommended PSHE values for
achieving robust multiple cracking behavior is above 3 (Kanda & Li 1998). Based on this
criterion, the PSHE of M2.2-C falls to the lower side. Besides that, the decrease of σ0 and
consequently the decrease of ultimate tensile strength of M2.2-C also cause the lower PSHS.
The lower PSHE and PSHS indicate smaller margin for satisfying the strain-hardening criteria
(Eqs.1 and 2), causing relatively low tensile ductility.
To improve the tensile ductility of M2.2-C, mixtures with higher fly ash content (M4.0-C),
FAC (M2.2FAC-C), and crumb rubber (M2.2CR-C) were developed. All the above mixtures
achieve higher tensile ductility by effectively lowering the matrix toughness. The measured
matrix toughness Km and Jtip of all mixes are listed in Table 5. The lower matrix toughness
also leads to lower first cracking strength as shown in Table 3. For these mixtures, both
PSHE and PSHS are effectively restored hence the tensile ductility are greatly improved over
unmodified M2.2-C. In particular, M2.2CR-C achieves an average tensile strain capacity of
5.2%. It is worth noticing that the cracking and ultimate tensile strengths of these mixtures
also notably decrease due to the weakening of the matrix and fiber/matrix interfacial bond.
1304
Table 5. Measured matrix toughness of ECCs.
Such mixtures are more suitable for applications that require high ductility and relatively low
strength, for example, pavement and road repairs.
previous research on the pavement overlay using different materials, Qian (2007) found that
ECC can double the service life while only using about 40% thickness compared with that of
normal concrete overlay, which can further decreases the cost of ECC construction. There-
fore, this study has demonstrated the feasibility and effectiveness of using Chinese PVA fibers
to develop lower cost ECC material while retaining desirable performance. The successful
development of such cost-effective ECC is expected to greatly promote the application of
ECC in pavement engineering field in China.
5 CONCLUSION
In the present paper, low-cost Chinese PVA fibers have been used to develop cost effective
ECC mixtures for civil infrastructure applications. Based on the investigation results, the fol-
lowing conclusions can be drawn:
1. ECC mixtures with tensile strength of 4.54∼5.67 MPa and tensile strain capacity of
1.9∼5.2% have been successfully developed using low-cost Chinese PVA fibers. By replac-
ing Kuraray fibers with Chinese ones, the unit cost of ECC material could be reduced by
as much as 60%, greatly promoting broader applications of ECC in China.
2. Directly replacing Kuraray PVA fibers with Chinese PVA fibers could result in decreased
tensile ductility due to lower tensile strength and higher interfacial chemical bond of Chi-
nese PVA fibers.
3. Higher fly ash content, incorporation of Fly Ash Cenosphere (FAC) and crumb rubber
could effectively enhance the tensile ductility of ECC with Chinese PVA fibers. However,
the compressive and tensile strength will slightly decrease.
4. ECC produced with Chinese PVA fibers exhibits tighter crack width than that of Kuraray
ECC, which is desirable for maintaining low transport properties under cracked stage and
enhancing durability performance.
ACKNOWLEDGMENTS
Zhigang Zhang is supported by a grant from the Chinese Scholarship Council (CSC) as a vis-
iting scholar at the University of Michigan. The authors would like to graciously thank the
National Natural Science Foundation of China (No.51278097) for the financial support for
this work. The co-author (Shunzhi Qian) also wants to acknowledge the support of startup
grant from Nanyang Technological University, Singapore under Grant No. M4081208.
REFERENCES
ASTM C109. 2012. Standard Test Method for Compressive Strength of Hydraulic Cement Mortars.
ASTM E399–12. 2012. Standard test method for linear-elastic plane-strain fracture toughness KIc of
metallic materials.
Griffith, A.A. 1921. The Phenomena of Rupture and Flow in Solids, Philosophical Transactions of the
Royal Society of London, 221: 163–198.
Huang, X., Ranade, R., Zhang, Q., Ni, W., & Li, V.C. 2013. Mechanical and thermal properties of green
lightweight engineered cementitious composites, J. Construction and Building Materials, 48:954–960.
1306
JSCE. 2008. Recommendations for design and construction of high performance fiber reinforced
cement composites with multiple fine cracks. Tokyo: Japan Soc. of Civil Engineers.
Kanda, T. & Li, V.C. 1998. Interface property and apparent strength of high-strength hydrophilic fiber
in cement matrix, Journal of Materials in Civil Engineering, ASCE, 10(1): 5–13.
Kanda, T. & Li, V.C. 1998. Multiple Cracking Sequence and Saturation in Fiber Reinforced Cementi-
tious Composites, JCI Concrete Research and Technology, 9(2):19–33.
Kanda, T. & Li, V.C. 1999. A New Micromechanics Design Theory for Pseudo Strain Hardening
Cementitious Composite, Journal of Engineering Mechanics, ASCE, 125(4): 373–381.
Lepech, M.D. & Li, V.C. 2009. Water permeability of engineered cementitious composites, Cement &
Concrete Composites, 31:744–753.
Li, V.C.1992. Post-Crack Scaling Relations for Fiber-Reinforced Cementitious Composites, Journal of
Materials in Civil Engineering, 4(1): 41–57.
Li, V.C. &Leung, C.K.Y. 1992. Theory of Steady State and Multiple Cracking of Random Discontinu-
ous Fiber Reinforced Brittle Matrix Composites, Journal of Engineering Mechanics, ASCE, 11(11):
2246–2264.
Li, V.C. 1998. Engineered cementitious composites-tailored composites through micromechanical mod-
elling, Fiber Reinforced Concrete: Present and the Future. Montreal, Canada: Canadian Society for
Civil Engineering, 64–97.
Li, V.C. 1998. Engineered Cementitious Composites—Tailored composites through micromechanical
modeling, Fiber-Reinforced Concrete: Present and the Future, Banthia N., Bentur A., and Mufri A.,
eds., Canadian Society for Civil Engineering, Montreal, Canada, 64–97.
Li, V.C, Wang, S. & Wu, C. 2001. Tensile Strain-hardening Behavior of PVA-ECC, ACI Materials Jour-
nal, 98(6): 483–492.
Li, V.C. Wu, C. Wang, S.X, Ogawa, A. & Saito T. 2002 Interface tailoring for strain-hardening Polyvinyl
Alcohol-Engineered Cementitious Composite (PVA-ECC), ACI Materials Journal; 99(5): 463–452.
Li, V.C. 2011. High-Ductility Concrete for Resilient Infrastructure, Journal of Advanced and High-Per-
formance Materials, 16–21.
Lin, Z., Kanda, T. & Li, V.C. 1999. On interface property characterization and performance of fiber
reinforced cementitious composites, Concrete Science and Engineering, RILEM, (1): 173–184.
Marshall, D.B. & Cox, B.N. 1988. A J-Integral Method for Calculating Steady-State Matrix Cracking
Stresses in Composites, Mechanics of Materials, 8:127–133.
Pan, Z.F., Wu, C., Liu, J.Z. et al. 2015. Study on Mechanical Properties of Cost-Effective Polyvinyl
Alcohol Engineered Cementitious Composites (PVA-ECC), Construction and Building Materials,
78:397–404.
Qian, S.Z. 2007. Influence of concrete material ductility on the behavior of high stress concentration
zones, PhD Thesis, Univ of Michigan.
Qian, S.Z. & Zhang, Z.G. 2012. Development of Engineered Cementitious Composites with local ingre-
dients, Journal of Southeast University (English Edition), 28(3): 327–330.
Qian, S.Z., Li, V.C., Zhang, H. & Keoleian, G.A. 2013. Life cycle analysis of pavement overlays made
with Engineered Cementitious Composites, Cement & Concrete Composites, 35:78–88.
Redon, C. Li, V.C., Wu, C., Hoshiro, H., Saito, T. & Ogawa, A. 2001. Measuring and Modifying Interface
Properties of PVA Fibers in ECC Matrix, ASCE J. Materials in Civil Engineering, 13(6): 399–406.
Sahmaran, M., Li M. & Li, V.C. 2007. Transport Properties of Engineered Cementitious Composites
under Chloride Exposure, ACI Materials Journal, 104(6):303–310.
Wang, K., Jansen, D.C., Shah, P.S., Karr, A.F. et al. 1997. Permeability study of cracked concrete. Cem
Concr Res, 27:381–93.
Wang, S.X. & Li, V.C. 2007. Engineered Cementitious Composites with High-Volume Fly Ash, ACI
Material Journal, 104(3): 233–241.
Yang, E.H., Wang, S.X., Yang, Y.Z. & Li, V.C. 2008. Fiber-bridging Constitutive Law of Engineered
Cementitious Composites. Journal of Advanced Concrete Technology, 6(1):181–193.
Zhang, Z.G., Qian S.Z. & Ma H. 2014. Investigating mechanical properties and self-healing behavior
of micro-cracked ECC with different volume of fly ash. Construction and Building Materials, 52(2):
17–23.
Zhang, Z.G. Ma, H. & Qian, S.Z. 2015. Investigation on properties of ECC incorporating crumb rubber
of different sizes, Journal of Advanced Concrete Technology, 31(5): 241–251.
1307
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
idea of Rollpave®, this study developed a modified asphalt specifically tailor for rollable asphalt
pavement. A comparison was conducted between this asphalt and other asphalt. The perform-
ance of asphalt pavement using this asphalt was evaluated. This studies will provide technical
support for the further development of pre-fabricated asphalt pavement.
2.1 Methodology
At room temperature or relatively low temperature like 0 to 30°C, to avoid asphalt cracking
during the rolling process, the asphalt mortar of asphalt mixture should be tough enough. The
idea of traditional asphalt binder modification is to add a high molecule modifier to the vir-
gin asphalt binder along with a stabilizing agent (Zhang 2008; Zhang Dommelen et al. 2009).
The stabilizing agent is used to disperse the high molecule modifier uniformly into the virgin
binder and blend into a stable phase. Under low temperature, the virgin binder became brittle,
and is low in flexibility and strength. To make the best of high molecule modifier, one way is
to increase the concentration of the modifier, thus enable the high molecule modifier to form a
continuous phase and have virgin binder disperse into this phase. The modified asphalt binder
obtained only in this manner can achieve enough flexibility, which enable asphalt mixture pro-
duced using this binder to roll like a blanket and not sacrifice its mechanical performance.
Performance and rheology tests were conducted on asphalt binder modified for the Rollpave,
the tests results were then compared with virgin binder and SBS modified asphalt binder.
1311
Figure 2. Asphalt binder sample demolding.
respectively. As the rutting resistant index increases with temperature increases. It is probable
that the failing temperature of the developed asphalt binder will exceed 100°C, which sug-
gested that it has a superior high temperature performance compared to the virgin binder
and SBS modified binder.
1313
Table 4. Low and high temperature performance test results.
Dynamic Stability/
Mixture type times ⋅ mm−1 Tensile strength/MPa Failure strain/10−6
Asphalt
Binder Maximum Deflection/ Flexural Flexural Flexural
type force/N mm strength/MPa strain/με stiffness/MPa
respectively. It can be observed that the developed asphalt binder enhanced the low tempera-
ture performance of the asphalt mixture to a noticeable degree.
1314
5 PRE-FABRICATED ROLLABLE ASPHALT PAVEMENT TEST
SECTION CONSTRUCTION
To validate the construction feasibility of the pre-fabricated rollable asphalt under low tempera-
ture environment, this study chose to construct the test section at the middle of January, the cold-
est month of Beijing. The paving temperature is −1 to 3°C. A section require immediate repair
was selected. The part observed serious alligator cracking of the existing pavement was milled
first, the milling depth was determined by depth of the pre-fabricated rollable pavement. After
milling, the milled surface was cleaned thoroughly to ensure the evenness of the new surface.
To secure a good bonding between the pre-fabricated asphalt pavement with the underly-
ing layer, the bottom of the pre-fabricated rollable asphalt pavement and tack coat (SBS
modified binder) was heated with fuel gas blowtorch. After the construction, the pavement
was compacted with a steel wheel compactor twice.
Table 6 showed the test results from the test section. As shown in Table 6, the skid resist-
ance of the rollable asphalt pavement meets the requirements of the specification well. There-
fore, rollable asphalt pavement could be applied to high class highway and municipal roads.
After one year of using, three measurements were made on the test section. No upheaval
or wave were found, only small amount of rutting was spotted, no new cracks appeared
around the joints, sealants worked well, the measurement are shown in Table 7.
Measurements Requirements
Date 1 2 3 1 2 3
1315
As indicated in Table 7, the rut depth increase with time, but in general, the number is
small. Due to the short gap between each measurement, though the BPN increased with the
time, the increment is still very small.
6 CONCLUSION
1. This study developed a modified asphalt binder (RMA) for rollable pre-fabricated asphalt
pavement, which enable asphalt mixture using this binder to bear large deformation. This
binder consist of virgin binder, special GTA particles, SBS particles, softening agent, and
auxiliary agent.
2. The results of the penetration, softening point, and ductility tests showed that the RMA
has far higher softening point and ductility than the virgin binder and SBS modified, but
its penetration was almost the same the virgin binder.
3. The performance of the RMA was further evaluated through DSR and BBR test, the
results were in line with regular tests. Compared to the virgin and SBS modified binder,
the RMA significant increased both the low and high temperature grade.
4. Asphalt binder type exerted noticeable influence on the performance of asphalt mixture.
Mixture prepared with RMA was superior in both low and high temperature performance
to those of virgin binder and SBS modified binder. Results of bending beam test con-
ducted at 10°C showed that RMA asphalt mixture could meet the deformation require-
ment of pre-fabricated rollable asphalt pavement.
5. The construction of test section using rollable pre-fabricated asphalt pavement was suc-
cessful, which indicated the feasibility of constructing rollable pre-fabricated asphalt
pavement at large scale and under low temperature. The performance measurements from
the test sections showed the skid resistance of the test section met the requirements of the
specification well. This suggest that the rollable pre-fabricated asphalt pavement could be
applied to high class highway and municipal roads.
REFERENCES
Dommelen A.E., J. van der Kooij, Houben L.J.M., et al. 2004. LinTrack APT research supports
accelerated implementation of innovative pavement concepts in the Netherlands // Minneapolis, USA:
www.mrr.dot.state.mn.us/research/MnROAD_Project/index_files/pdfs/VanDommelen.
Dong Yuanshuai, Liu Qingquan, Cao dongwei. 2013. Research of influencing factor on the bending
property of asphalt mixture based on grey correlation entropy. Applied Mechanics and Materials, v
405–408: 1923–1927.
Dong Yuan-shuai. 2015. Study on Material Properties and Construction Technologies of Rolla-
ble Prefabricated Asphalt Pavement. Nanjing: Transportaiton College, Southeast University. (in
Chinese).
Houben, L.J.M., J. van der Kooij, R.W.M. Naus., et al. 2004: APT Testing of Modular Pavement Struc-
ture ‘Rollpave’ and Comparison with conventional asphalt motorway structures//2nd International
Conference on Accelerated Pavement Testing. Minneapolis, USA. 1–24.
Ingram L.S., Herbold K.D., Baker T.E., et al. 2004. Superior Materials, Advanced Test Methods and
Specifications In Europe. Washington DC: Federal Highway Administration U.S. Department of
Transportation.
Molenaar J.M.M., Montfort J. 2011. Super StilleDeklaag Rubber Rollpave. Delft: Road and Hydraulic
Engineering Institute, the Netherlands. (in Dutch).
Zhang Bao-chang. 2008. Study on Compound Mineral Fillers Modified Asphalt and its Mixtures Paving
Performance. Shenyang: School of Science, Northeastern University. (in Chinese).
Zhang Jin-sheng, Zhang Yin-yan, Xia Xiao-yu, et al. 2009. Asphalt Materials. Beijing:Chemical Indus-
try Press.
1316
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: In this paper, the characteristics of semi flexible pavement base materials
were studied through experiments. Based on the experiments, the design method of emulsi-
fied asphalt and cement stabilized macadam mixture, performance test, mechanical analy-
sis of semi flexible pavement structure, construction technology were studied. The results
show that by using the framework of the skeleton block distribution, the mixture has good
physical, mechanical properties, high temperature stability, low temperature flexibility and
anti fatigue properties. The use of emulsified asphalt and cement stabilized crushed stone
base in place of semi-rigid base can effectively reduce the maximum vertical shear stress and
improve the crack resistance of pavement. Meanwhile, the trial road shows that the mixture
of emulsified asphalt and cement stabilized macadam mixture has good uniformity, and easy
compaction.
1 BACKGROUND
Semi-rigid base asphalt pavement are widely used in highway in China. Traditional semi-rigid
base course materials, due to the suspended dense structure, higher cement content, the traf-
fic load, environment, cement stabilized material dry shrinkage and temperature shrinkage
factors of multiple roles, more serious diseases of cracking and reflection to the pavement
appear, which caused pavement structure damage.[Ding, 2012] Based on the shortage of semi rigid
pavement, the semi flexible pavement technology has been widely concerned and applied in
the near future. [Du, 2012] Semi-flexible pavement technology can effectively improve the pave-
ment crack resistance, mechanical properties, reduce the asphalt pavement thickness, and
improve the durability of asphalt pavement.[Hou, 2010] The application of semi-flexible pave-
ment technology is of great significance to enrich and improve the structure type of asphalt
pavement.[Jiang, 2013]
Semi-flexible pavement base usually adopts the emulsified asphalt, foamed asphalt,
cement and other adhesive material, mixing water, rolling and molding.[Li, 2008] The per-
formance of foamed asphalt stabilized material is greatly influenced by asphalt foaming
equipment, so the semi-flexible base of emulsified asphalt cement stabilized crushed stone
is adopted in this paper.[Wang, 2013] On the basis of this, several parts have been studied, such
as design of emulsified asphalt cement stabilized crushed stone mixture, performance test,
mechanical analysis of semi-rigid base pavement structure and construction of semi flex-
ible pavement base. Cement and emulsified asphalt stabilized macadam mixture pavement
technology, reduce reflection crack in asphalt pavement and maintenance frequency, and
reduce the traffic delay caused by maintenance, ensure the quality of road driving. Fur-
ther more, through the application of this technology the level of highway construction
in Jiangsu Province has been promoted, which has a very good direct and indirect social
benefits.[Yan, 2011]
1317
2 DESIGN OF EMULSIFIED ASPHALT CEMENT STABILIZED CRUSHED
STONE MIXTURE
Numerical Index
Test item test requirements
Sieve pore Through the sieve (square mesh sieve, mm) percentage (%)
31.5 26.5 19.0 16.0 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075
1# sample 100 90.7 32.4 14.1 2.4 0.4 0.4 0.4 0.3 0.3 0.2 0.2 0.1
2# sample 100 100 100 100 99.5 64.8 1.2 0.3 0.3 0.3 0.2 0.2 0.1
3# sample 100 100 100 100 100 100 97.0 66.5 46.4 30.5 18.1 13.0 9.6
1318
Table 4. Performance indexes of common emulsified asphalt.
Table 5. Requirements of design gradation range of emulsified asphalt cement stabilized crushed
stone mixture.
Sieve pore (mm) 31.5 19.0 9.5 4.75 2.36 0.6 0.075
Grading limit (%) 100 68∼86 38∼58 22∼32 16∼28 8∼15 0∼5
Gradation 100 78.2 49.8 29.3 18.6 9.3 2.8
optimum moisture content. Emulsified asphalt content of test prepared is 3.0%. The water
contents are 3.0%, 4.0%, 5.0%, 6.0% and 7.0%. Test results are plotted of dry density and the
actual water content relationship curve. Figure in maximum dry density corresponding to
the water content is the optimum moisture content OWC. Vibration compaction test results
are shown in Table 6, From the test results, vibration compaction method to determine the
optimum water content of OWC was 4.4%.
1319
Table 6. Determination of optimum water content by vibrating compaction method.
Emulsified asphalt content (%) Actual water content (%) Dry density of specimen (g/cm3)
1320
Table 8. Test results of optimum emulsified asphalt content and optimum water content.
Optimum Maximum
emulsified Gross relative Dry Wet
asphalt Optimum volume density of splitting splitting
content water relative measured Void strength strength
Mixture type (%) content (%) density theory volume (%) (MPa) (MPa)
strength of the representative values are greater than 3.5 MPa, meet the requirements of
strength of the pavement.
Stiffness Failure
Emulsified asphalt Maximum Mid span Flexural tensile modulus strain
content (%) load (kN) deflection (mm) strength (MPa) (MPa) (με)
Table 14. Results of fatigue test of emulsified asphalt cement stabilized crushed stone mixture.
The emulsified asphalt and cement stabilized crushed stone mixture has excellent high
temperature stability, while retaining the flexibility of the asphalt mixture, the maximum
bending strain still reached 3057.6με. Cement was added to increase the brittleness of cement
and emulsified asphalt stabilized macadam mixture, but the cement as cementing material
improves the strength. Both have excellent high temperature performance, while retaining
good low temperature flexibility, this is an important feature of semi flexible base.
1322
Figure 2. Calculation results of vertical shear stress of pavement structure.
Specimens were placed in the box more than 6 hours, so that each part reaches a constant
temperature. The small beam fatigue test is carried out at a temperature of 20°C.
Under asphalt content 3%, cement content 2%, and 200 με conditions, the fatigue life
of the emulsified asphalt and cement stabilized macadam mixture reached 32300 times.
The emulsified asphalt and cement stabilized macadam mixtures at low strain levels, has
good anti fatigue performance. Strain levels increase, fatigue life decreases rapidly.
4.2 The influence of semi flexible base modulus change on the stress state
of pavement structure
Two layers of semi flexible base thickness was 16cm respectively. Though Semi flexible base
modulus of 700MPa, 800MPa, 900MPa, 1000MPa and 1100MPa, calculate the pavement
1323
Figure 4. The change of deflection with the thickness of the base layer.
Figure 5. Opened to traffic about one year, the test road of emulsified asphalt and cement stabilized
macadam base.
deflection, bottom layer tensile stress and vertical shear force. Results show that pavement
deflection value decreases with the increase of the modulus of base. Base bottom tensile
stress increases with the increase of modulus. The shear stress increases with the modulus
increases. Therefore, reduce the modulus of pavement base can prevent reflection crack, and
improve fatigue performance of pavement structure. But the modulus is too small, it would
cause that the pavement bearing capacity is insufficient.
4.3 The influence of the thickness change of the semi flexible base layer on the stress state
of pavement structure
Changing the thickness of semi flexible base, study the changes of semi flexible base layer
thickness on the structure stress. The thickness of semi flexible base is 6cm, 10cm, 15cm and
20cm, the calculation results are shown in Figure 4. The results showed that the road surface
deflection, base layer tensile stress at the bottom, and the vertical shear stress decreased with
the increase of the semi flexible base thickness. So appropriately increases semi flexible base
thickness, it can improve the ability to resist failure load.
In November 2013, relying on highway 237 Gaoyou period of renovation and expansion
project, semi flexible pavement base test paving of emulsified asphalt and cement stabilized
macadam was carried out.
The field core tests of 5 days in the test section show that: emulsified asphalt and cement
stabilized macadam base was formed well, the core sample was complete, and unconfined
1324
compressive strength was 4.4 MPa. One year after test road opened to traffic, emulsified
asphalt and cement stabilized macadam base test road with good appearance, no distresses.
6 CONCLUSION
1. The characteristic of emulsified asphalt and cement stabilized crushed stone mixture are
as follows. The maximum nominal particle size of aggregate is 31.5mm. Synthetic gra-
dation is skeleton interlocking gradation. Determine the optimum moisture content by
vibrating compaction method. Determine the optimum amount of emulsified asphalt by
dry and wet splitting strength. The optimum amount of emulsified asphalt is 3.2% and the
optimum moisture content is 4.4%. The strength and void rate of emulsified asphalt and
cement stabilized macadam mixture meet the design requirements.
2. The performance of emulsified asphalt and cement stabilized crushed stone mixture is
studied. Compressive strength by vibrating compaction method is greater than 3.5MPa,
meet the pavement strength requirements. Compressive modulus of 20°C is about 800MPa,
with a certain flexibility. Dynamic stability is more than 15000 times /mm, with excellent
high temperature stability. Low temperature bending test shows that the maximum bend-
ing strain reached 3057.6 με, which has better low temperature flexibility. At low strain
levels, it has good fatigue performance.
3. Instead of semi rigid base, emulsified asphalt and cement stabilized macadam base can
reduce the maximum vertical shear stress of pavement, which is beneficial to improve the
crack resistance of pavement. Pavement surface deflection, bottom layer tensile stress, and
vertical shear stress decrease with the increase of the semi flexible base thickness. Properly
increase the thickness of semi flexible base can improve the ability to resist failure load.
4. The application of semi flexible pavement base test shows that the emulsified asphalt and
cement stabilized crushed stone mixture has the characteristics of good uniformity and
easy compaction.
REFERENCES
Ding Wuyang et al. Research and application of the key technology of emulsified asphalt plant mix cold
recycling mixture [J]. China and Foreign Highway, 2012, 05, 216–219.
Du Shaowen. The cement and emulsified asphalt recycled cement stabilized macadam[J]. Highway,
2012, 7, 250–251.
Hou Yanming et al. Dry shrinkage performance of cement and emulsified asphalt stabilized macadam[J].
Highway, 2010, 10, 107–108.
Jiang Liwei. The performance test of emulsified Asphalt and cement stabilized graded crushed[J]. Build-
ing Science Research of Sichuan, 2013, 03, 222–223.
Li Zhigang et al. Shrinkage characteristics of cement and emulsified asphalt stabilized macadam[J].
Journal of PLA University of Science and Technology, 2008, 2, 177–179.
Technical Guideline: Bitumen Stabilised Materials, A Guideline for the Design and Construction of
Bitumen Emulsion and Foamed Bitumen Stabilised Materials. Asphalt Academy, 2009.
Twagira M E, Jenkins J K. Durability properties of materials and mixture composition of bitumen
stabilized materials. Journal of Wuhan University of Technology, 2010, 17.
Wang Hong et al. High temperature stability of emulsified cold recycled mixture[J]. Highway engineer-
ing, 2013, 04, 191–192.
Yan Jinhai et al. Indirect tensile fatigue performance of emulsified asphalt cold recycled mixture[J].
Journal of building materials, 2011, 01, 58–61.
Yu Shaoqiu. The water stability of cement and emulsified asphalt cold recycled mixture[J]. China and
Foreign Highway, 2013, 01, 249–250.
1325
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Lei Wang
Chang’an University, Xi’an, China
Research Institute of Highway, Ministry of Transport, Beijing, China
Mingxin Li
Chang’an University, Xi’an, China
Delft University of Technology, Delft, The Netherlands
ABSTRACT: Through establishing discrete element models of rubber asphalt and HMA,
the authors elaborated the meso-damage mechanism of each mixture. With the adhesion
index of asphalt and fine aggregate and a revised evaluation method of durability of rubber
asphalt, the improvement of durability with additives of TAFPACK-Super (TPS) modifier
was evaluated. An optimized scheme of TPS modified rubber asphalt pavement was devel-
oped based on lab tests. To study the influence of various factors on elastic modulus of
rubber asphalt, a meso-mechanical model to predict microscopic performance of composite
mixture was established. With the FE analysis of ice layer and its state of stress and strain,
the relevance between de-icing efficiency and internal and external factors such as environ-
ment temperatures, ice thickness, elastic modulus of mixtures, structures combination. The
results revealed ice removal mechanism of rubber asphalt and elaborated the effectiveness of
environmental conditions, working status and optimum structural arrangements.
1 INTRODUCTION
In seasonal frozen inland areas, winter can be both cold and dry while last a long period.
The snow and ice on pavement surface are not easily melted. It is prone to traffic accidents.
There is an approach, that makes crumb rubber granular as part of fine aggregates in asphalt
mixture, to build safer pavement with better deicing property. The crumb rubber granular
is produced from waste rubber tires and grinded into particular shapes and sizes. Rubber
particles in the asphalt mixture will effectively improve the elastic deformation capacity of
the pavement (Shimomura, et al., 2009). As a consequence, the repeat loads of traffic can
effectively suppress the accumulation of snow or formation of ice since adhesive condition
between snow/ice and pavement is changed.
Even though researches on crumb rubber granular asphalt mixture are popular for many
years, a number of problems still remain with the application of rubber asphalt as deic-
ing pavement. Analysis of rubber asphalt on durability is relatively scarce. Accordingly, the
evaluation indicators and methods of rubber asphalt durability also had many deficiencies.
For its deicing property, Zhou (2006) analyzed the stress condition in ice layer and the effects
of various factors on deicing. However, comprehensive research on either damage states of
ice layer or relationship between rubber asphalt elastic modulus and deicing capacity is not
performed thoroughly, yet. Therefore, further study is demanded before this deicing technol-
ogy being promoted on a wider range.
1327
2 MESO-STRUCTURE ANALYSIS WITH DISCRETE ELEMENT METHOD
To build the mesoscopic numerical model which is able to simulate internal mesostructure of
rubber asphalt mixture, random sampling principal was employed. 2-D numerical model was
used to analyze stress and strain states, deformation characteristics and failure mechanism of
rubber asphalt mixtures under repeated wheel loads to simplify computation.
Considering the typical operating conditions of pavements in China, traffic load was
adopted as standard axle load BZZ-100 with 100 kN of axle load, 0.7 MPa of wheel pres-
sure, and 32 cm of wheel center spacing. The tire footprint is simplified as 18.9 × 18.9 cm2.
The size of 2-D model is 36 cm by length and 4 cm by depth. The model is symmetric to the
center plane of two wheels. In the direction of depth, only upper surface that rubber asphalt
mixture was used was taken into account. Parameters used in discrete element analysis were
adopted from works of Zhang (2012) and Wang (2013). A gap-graded rubber asphalt mixture
was designed as study object. By observing particle displacement and cohesive and adhesive
contact conditions in mesostructures, the author tried to obtain the damage mechanism of
rubber asphalt pavement under traffic loads.
1328
Table 1. Average values of strain rates in circular measurement domains.
Circular measurement
domain ε y × 10 −2 γ xy × 10 −2 γ xy × 10 −2 Note
Figure 2. Enlarged scheme of internal bonding failures of gap-graded rubber asphalt mixture.
crack initiation were identified with short red lines. From the scheme, it can be recognized
that bonding failures first occurred around the rubber particles. That is to say, interfaces of
rubber particles and between rubber particles and aggregates are relatively weak.
Results of discrete element analysis indicated that flexible pavement with rubber asphalt
had the characteristic of relatively large local deformations with rubber particles distribution.
Within the area where rubber particles existed, bonding state was dominant by adhesion
between aggregates and rubber particles. Consequently, the repeated traffic loads will lead
to lower bonding strength in rubber asphalt mixtures than that in asphalt mixtures without
rubber particles. Weak stone-rubber interfaces resulted in peeling problems which would be
reflected as reveling and poor durability for pavement performance.
TAFPACK-Super (TPS) modified asphalt was originally invented for the use of porous asphalt
pavement. The main ingredient of this modifier was thermoplastic rubber. Other components
such as adhesive resins and plasticizers were added. Bitumen was modified with this modifier
by mechanical mixing to obtain high viscosity. According to former research findings, the
dosage of TPS modifier should be in a range of 12 15% of total bitumen amount. The author
performed tests on aggregate-binder adhesion and water stability of asphalt mixtures with
TPS dosage of 12%, 13.5% and 15%, respectively, to determine the optimum content.
1330
Figure 4. 30 min water immersion test results.
were recorded. The peeling rates of coating fine aggregates to coarse aggregates were calcu-
lated and analyzed.
However, this approach could be only used to evaluate overall adhesion of fine aggregates,
including rubber particles, to coarse aggregates. It was not able to investigate binder-rubber
adhesion. Moreover, this test was employed for qualitative comparison rather than quantita-
tive analysis.
Due to the high viscosity of TPS asphalt, it was recommended to increase 15_ mixing tem-
peratures for samples with TPS modifier involved. Test results were shown in Figure 5.
From the results above, it could be seen that adhesion between SK-70 asphalt and fine
aggregates was the poorest since it had the highest peeling rate. Adhesion of SBS asphalt
samples was better than that of SK-70 samples. TPS asphalt samples had the lowest
peeling rates which were all less than 45% for 175°C or higher mixing temperatures. It
could be inferred that TPS modifier would improve adhesion between asphalt and fine
aggregates.
As seen from the results, adhesion decreased with higher content of TPS modifier. It could
be explained that higher viscosity of modified asphalt would make the mixing more difficult
so that fine aggregates were more prone peeling off. By increasing mixing temperature to
185°C, the mixing uniformity improved so that adhesion increased accordingly. The peeling
rates of TPS samples were all less than 35% while also increasing with higher TPS dosage
used.
1331
Figure 5. Adhesion test results of fine aggregates.
By analyzing results in Figure 6, it could be seen that water stabilities of SBS and TPS
modified asphalt samples were similar and much better than that of SK-70 asphalt samples
with 170_ mixing temperature. As proven before, if mixing temperature was increased to
185_, the water stability would grow significantly. Meanwhile, water stabilities decreased with
more TPS modifier used.
Figure 7 illustrated the results of TSR tests. The results were very consistent with immer-
sion stability test results. For all the samples with modified asphalt involved, the TSRs were
higher than 85% which was required by Chinese code for HMA.
As shown in Figure 8, TPS modifier could significantly reduce the loss rates of the tests.
For example, the standard scattering loss rate of 12% TPS modified asphalt samples was 12%
lower than that of SK-70 samples and 6% lower than that of SBS modified samples. For the
flooding scattering tests, the differences could be more remarkable. This could be interpreted
1332
Figure 7. Tensile strength ratio test results.
1333
4 DEICING PERFORMANCE EVALUATION OF RUBBER ASPHALT PAVEMENT
1334
As seen in Table 3, for ice thickness is 2, 4, 6 and 8 mm, the discriminant function would
be less than 0 at the mixture modulus of 2600, 2000, 2000 and 1800 MPa, accordingly. When
ice thickness reached 10 mm or thicker, it was unable for deicing.
Assuming the ambient temperature as –10°C, the FE calculation was conducted at elastic
modulus of 2600 MPa and 2700 MPa, at ice thicknesses of 2 mm and 4 mm. The results were
shown as following in Table 4.
Only at 2600 MPa elastic modulus and 2 mm ice thickness was the ice found to be elim-
inated effectively. Either Higher elastic modulus of asphalt pavement or thicker ice layer
would result in poor deicing performance.
Through finite element analysis of combinations of various operating conditions, the rela-
tionship between deicing performance of rubber asphalt pavement and other key factors was
presented. Deicing ability of rubber asphalt pavement got less effective with increasing elastic
modulus of rubber asphalt mixture, increasing thickness of ice and decreasing temperature.
4.2 Mechanical analysis in compound structures of ice and rubber asphalt pavement
Excellent deicing performance required a reasonable design of pavement structure. There-
fore, numerical simulation of compound structures consisting of ice layer, rubber asphalt
mixture and HMA layers was performed in order to get stress and stain states inside the
ice. The assumptions in this study were low temperature, repeated traffic loading and elastic
multilayer system. The pavement 3-D model consisted of ice, rubber asphalt, HMA, base
layer and subgrade, vertically. After comparing with typical pavement structures in China
and error analysis of model sizes, the actual size of model was determined as 3.8 m in length
and width, while 6 m in depth. Mechanical parameters of all materials at 0°C were presented
in Table 5.
The traffic load used in FE analysis was the same with that of discrete element analysis.
For different compound structure scenarios, the thicknesses of ice, base layer, subbase layer
and subgrade remained the same. Thickness of ice layer was 6 mm. Four different structure
combinations were simulated and the maximum equivalent stress in ice was calculated for
each structure.
As the results shown in Table 6, the second structure combination had the highest equiva-
lent stress in ice. In another word, it had the best deicing performance among the four struc-
tures that investigated in the research.
1335
Table 6. Max. equivalent stress of ice layer for various pavement structure.
Comparison of Combination 1 and 2 revealed that thicker surface layers would result in
better deicing performance since surface asphalt layers were more potential to deform. Differ-
ences between Combination 1 and 3 showed that materials of bottom layer could also affect
deicing ability of pavements. The poorer deicing performance of Combination 3 could be
attributed to its relatively high equivalent elastic modulus. When the rubber asphalt layer was
allocated at middle layer instead of top layer, deicing performance of the pavement decreased
significantly. However, it was believed that rutting and skid resistance of the pavement would
be enhanced in Combination 4.
5 CONCLUSIONS
1. With the aid of discrete element approach, failure mechanism of crumb rubber granular
asphalt mixtures was studied. It was found that rubber particles in the mixture would lead
to a relatively large deformation under repeating loads. It was prone to adhesion failures.
2. The existing durability evaluation methods were improved to meet the requirements and
features of rubber asphalt.
3. The optimum content and recommended mixing temperature of TPS modified asphalt
were put forward by conducting a series of tests involving adhesion and water stability.
4. A numerical model was established in order to predict the deicing performance of rubber
asphalt pavement in certain operating conditions.
REFERENCES
Li, J., Wang, X. & Zhang, C. 2013. Study on durability of granulated crumb rubber elastic pavement
based on TPS high viscosity asphalt. Road Machinery & Construction Mechanization, 1, 32−36.
Ma, X., Wang, X. & Yao, L. 2013. Evaluation of Deicing Effect of Granulated Crumb Rubber Elastic
Pavement. Road Machinery & Construction Mechanization, 1, 37–40.
Shimomura, S., Motofuji, T. & Takahashi, T. 1995. Freezing controlled pavement with rubber matted
concrete flat plate utilizing waste tire. Nippon Doro Kaigi Ronbunshu, 21.
Wang, X. & Lu, K. 2013. Technology and development of snow melting and deicing of highway pave-
ment. Road Machinery & Construction Mechanization, 1, 26–31.
Zhang, C. 2012. Study on durability of granulated crumb rubber asphalt pavement. Master dissertation,
Chang' an University, Xi'an, China.
Zhou, C. 2006. Study on granulated crumb rubber asphalt mixture technology in frost region. Doc-
toral dissertation, Harbin Institute of Technology, Harbin, China.
1336
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Mingxin Li
Chang’an University, Xi’an, China
Delft University of Technology, Delft, The Netherlands
1 INTRODUCTION
In the cold regions pavements covered with snow and ice put road users in significant safety
risks. These layers of snow/ice on the top of the pavement surfaces significantly reduce the
skid resistance accidents (Wang et al 2013). Deicing i.e. getting rid of unwanted snow/ice on
these pavement surfaces is still challenging to the road authorities and researchers around the
world. Recently researchers have found that compared to the traditional deicing methods, the
use of microwave technology has certain advantages. To the best of authors’ knowledge not
much attention has been paid to properly evaluate the effectiveness of these the microwave
technologies in this direction.
Microwaves are electromagnetic waves with frequencies (wavelengths) in the range from
300 MHz (λ = 1 m) up to 300 GHz (λ = 1 mm) (Vollmer 2004). The principle of microwave
deicing technology is related to the heating mechanism of these microwaves. In general, the
complex permittivity of a semiconductor material is given by (1) as follows:
σ
ε ε 0 (ε r ε r″ − j ) = ε 0ε r (1 − j tanδ ) (1)
ωεε 0
where is the effective dielectric loss tangent of the semiconductor given by (2) as follows:
σ
tanδ tanδ d + (2)
ωεε 0ε r
where ε 0 is the permittivity of the vacuum, ε r is the relative real permittivity of the semicon-
ductor, ω is the angular frequency, σ is conductivity, and tanδ d is the dielectric loss tan-
1337
Figure 1. Energy transfer process of microwave deicing.
gent associated with pure dielectric loss mechanisms (e.g., electronic and ionic polarization)
(Krupka 2006). It is noted here that the dielectric loss tangent (tanδ) of ice is low, the micro-
waves would reach the pavement surface rather being absorbed by ice (Hopstock 2003).
In order to study the effectiveness of the microwave technology, dedicated test section
surface samples were prepared in laboratory. The microwave absorbing capabilities of test
surfaces were enhanced by replacing some fraction of the filler material with microwave
absorbing materials such as hydroxyl iron powders and ferriferous oxide (also known as mag-
netite or Fe3O4). Henceforth, the mixes prepared from the microwave absorbing materials are
referred as MAMs. The effectiveness of the these absorbing materials were studied by com-
paring the performance of such materials against a section with conventional mix (CMs).
The result shows that a section with traditional materials performs poorly under usage of
microwave technology and results in unnecessary waste of energy. As rightly pointed out by
several past studies, the implementation of such technology is practically infeasible due to
such huge waste of energy. (Jiao, Tang & Gao 2008 and Wang & Lu 2013).
There was an alternative way to take advantage of microwave absorption materials in pave-
ment deicing rather than reconstruction of microwave absorbing surface. The use of micro-
wave absorption coating would significantly save the construction budget.
1338
oxide, were selected for the tests. Henceforth, the two hydroxyl iron powders used in the study
will be referred as YW1 and RW whereas ferriferous oxide will be referred as Fe3O4. Param-
eters of YW1 and RW powders are presented in Table 1.
It could be seen from Table 1 that the two most significant differences between YW1 and
RW are the iron contents and the average particle sizes. Iron content is believed as the main
difference that will dictate the microwave absorption capacities.
Experiments were conducted to compare the deicing performance of YW1, RW and FX.
For this experiment, samples were cut into quarters from the standard plate size sample
(300 mm × 300 mm × 50 mm). This results in 50 mm thick samples with 150 mm length/width.
Out of the whole thickness of the sample, top 20 mm was replaced by MAMs. The content of
MAMs used 3% and 5% of microwave absorbing materials respectively. Samples of the same
MAMs were given the same number. Sample No. 1 was CMs used for comparison.
Surface temperatures of all samples were recorded at the beginning and at the end of the
operation. To measure the surface temperature, a sheathed thermocouple was closely adhered
with silicone to the specimen surface. During the test, the temperature at specimen surface
could be controlled by the microwave operating time. The thermocouple’s wires were tightly
wrapped with an aluminum foil tape in order to prevent it from the microwave damage.
Content Content
# of sample Material (by mass) # of sample Material (by mass)
1 No absorption material / 9 RW 7%
2 RW 3% 10 RW 9%
3 RW 5% 11 YW1 1%
4 YW1 3% 12 YW1 7%
5 YW1 5% 13 YW1 9%
6 Fe3O4 3% 14 Fe3O4 1%
7 Fe3O4 5% 15 Fe3O4 7%
8 RW 1% 16 Fe3O4 9%
1339
50 mm, 60 mm, 90 mm, 120 mm and 250 mm, respectively, according to the wavelength of
used microwave with 2.45 GHz. Power of microwave was 800 W for these experiments while
microwave operation time for each sample is 60 s. For ease of presentation, a constant initial
surface temperature of 0°C is used and surface temperature increases are recorded.
1340
higher frequency and closer distance would make the samples more effective in microwave
absorption.
4. Samples 6 and 7 were less sensitive to frequency and distance changes.
Some of the findings were found in good accordance with other studies as explained later
in the paper.
3.3 Influence of distance from waveguide outlet to pavement surface on deicing efficiency
when using microwave to heat pavement surface, the amount of energy absorption presents an
exponential decay with depth. Therefore, a depth indicator was put forward by other research-
ers to relate the heating efficiency with pavement depth. The heating depth indicator referred to
the depth where attenuation of the microwave power reached to 0.3679 (1/e) times of the initial
value at material surface. The equation of heating depth indicator D is presented as following:
1341
λ0
D= (3)
π 2ε ( + − )
where, λ0 is free-space wavelength of microwave. tanδ is the dielectric loss tangent of pave-
ment material. ε ′ is the relative dielectric constant (Jiao, S., Tang, X. & Gao, Z 2008).
tanδ and ε ′ of asphalt pavement are both constants while tanδ of ice is so small that could
be ignored. Therefore, the depth indicator is a positive linear function of microwave length.
In another word, it is possible to evaluate deicing efficiency of pavement by investigating
distance between waveguide outlet to pavement surface. Since the wavelength of microwave
λ is the function of light speed c divided by frequency f, 122.45 mm of wavelength yields for
2.45 GHz frequency.
The results of deicing efficiencies with various distances from waveguide outlet to sample
surface are presented below in Figure 7.
1342
Results showed that the highest surface temperature increase happened when the distance
between the waveguide outlet and the sample surface was at a quarter distance of the wavelength
of the used microwave, i.e. in this case a distance of 30 mm. Whereas, the lowest surface tem-
perature increase occurred when distance from waveguide outlet to sample surface was 90 mm,
which was about 3/4 of the wavelength. Moreover, temperature increase at sample surface had
a reducing tendency with an increasing operation distance. It could also be inferred that MAMs
had better deicing efficiencies when the depth of ice layer on top is thinner than 40 mm.
Figure 10. 40 mm ice thickness test. Figure 11. 50 mm ice thickness test.
1343
From results in Figure 13, the following findings could be inferred.
1. CMs had the least temperature rise.
2. The microwave absorbing capabilities of Fe3O4, RW and YW1 were in a decreasing order.
For the two types of hydroxyl iron powders, microwave absorption of RW samples were
slightly better than that of YW1 samples. Microwave absorptions of RW and YW1 were
1.21∼2.19 times and 1.20∼2.12 times to that of the reference set, respectively.
3. Samples with ferriferous oxide additives showed the best heating performance among all.
4. With an increasing content of microwave absorbing material, MAMs samples had greater
temperature increases under identical microwave operations.
Figure 13. Temperature effect of each type and content of absorption material.
1344
Figure 14. Heating efficiency of MAC. Figure 15. Deicing efficiency of MAC.
coatings above the surface. For the set of tests without ice coatings, the temperature increase
was 17°C, 38°C, 51°C and 91.5°C for reference sample, RW sample, Fe3O4 sample and EG
sample, respectively, after 50 s of microwave operation. For the other set of tests, microwave
started operate at the surface temperatures of −16°C. It took 197 s, 124 s, 108 s and 23 s,
respectively, for the surface temperature of reference sample, RW sample, Fe3O4 sample and
EG sample to reach 0°C. It could be inferred that the EG coating had the highest heating
and deicing efficiency followed by Fe3O4 and RW. As coating material, Fe3O4 was observed to
be slightly better than RW. It was also clear that the CMs samples took much more time and
energy to heat up to the same level as samples treated by microwave absorbing coatings.
4 CONCLUSIONS
This paper describes the input requirements such as frequency, power and distance from
microwave source on the deicing efficiency. For a typical road network of China, the required
deicing microwave energy was analyzed at different temperatures and thicknesses of ice. Two
different approaches to add microwave absorbing materials on the pavement surface are pre-
sented: a) by partial replacement of filler material and b) by using them as surface coat-
ing treatment. Microwave heating tests were carried out to evaluate the response of both
approaches under the influence of environmental factors. Overall, the results showed that the
sample “EG” is most suitable as coating materials whereas sample “FX” is more suitable for
microwave absorption asphalt pavement. The following findings can be presented briefly:
a. for partial replacement of filler material technique
1. On deicing efficiencies both internal & external factors showed significant influence. Internal
factors are associated with types and contents of microwave absorption materials, power and
frequency of microwave, and distance from waveguide outlet to pavement surface. External
factors includes parameters such as thickness and initial temperature of ice layer.
2. Heating efficiency tests showed that sample “FX” containing FE3O4 has the highest microwave
absorbing ability followed by samples “RW” and “YW1” respectively. Increase of “microwave
absorption material” in MAMs enhances the deicing performance of these mixes.
b. for surface coating treatment
3. Among those coating materials, expandable graphite (EG) was the most effective in
heating and deicing under microwave operation followed by samples “FX” and “RW”
respectively.
REFERENCES
Hopstock, D.M., 2003. Microwave-absorbing road construction and repair material. Final Report to
NRRI on Idea Evaluation Subcontract (June 11, 2003). Duluth: University of Minnesota Duluth.
1345
Jiang, S. 2012. Study on an environmental ice-deicing and snow-melting coat for asphalt pavement.
Master dissertation, Chang' an University, Xi'an, China.
Jiao, S., Tang, X. & Gao, Z. 2008. Study of key technology on microwave deicing efficiency. China
Journal of Highway and Transport, 06,121–126.
Krupka, J., Breeze, J., Centeno, A., Alford, N., Claussen, T. & Jensen, L. 2006. Measurements of Per-
mittivity, Dielectric Loss Tangent, and Resistivity of Float-Zone Silicon at Microwave Frequencies.
IEEE Transactions on Microwave Theory and Techniques, Volume:54, Issue: 11, 3995–4001.
Li, X., Xu, Y. & Liu, F. 2003. Application of microwave heating for ice removal on streets. Journal of
Harbin Institute of Technology, 11,1342–1343.
Vollmer, M. 2004. Physics of the microwave oven. Physics Education, 39, 74.
Wang, X. & Lu, K. 2013. Technology and development of snow melting and deicing of highway pave-
ment. Road Machinery & Construction Mechanization,1, 26–31.
Wen, J. 2009. Research Progress of Microwave Radiation Damage and Protection. Bulletin of the Acad-
emy of Military Medical Sciences, 04.
Wright, M. 2013. A study of the effects of adding ice retardant additives to pavement surface course
material. PhD dissertation, The University of Nottingham, Nottingham, United Kingdom.
Yang, G., Kang, Y. & Meng, Q. 2010. Progress in development of microwave absorbing materials.
Applied Chemical Industry, 4, 584–589.
1346
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
The uniformity of the asphalt mixture has an important role on asphalt pavement. The con-
figuration of the coarse aggregate directly affects the uniformity of the mixture. In general, is
the coarse aggregate configuration is related to the construction technique, but some research
show it is also related to the mixture gradation, forming method and other factors (Sefid-
mazgi & Bahia, 2014; Peng & Sun, 2007). At present, research focused on coarse aggre-
gate configuration and number without considering the influence of coarse aggregate shape
irregularity. Meanwhile the changes of coarse aggregate configuration in the forming process
were known very little (Hassan, Airey, Hainin, 2014; Bruno, Parla & Celauro, 2012).
Since the 1990s, digital image processing has been applied in the micro-structure study
of bituminous mixture (Masad et al., 1999). There are now substantial researches focused
on digital image processing from two-dimensional (2-D) as well as three-dimensional (3-D)
methods (Wang LB, Wang XR, Mohammad & Abadie, 2005; Liu & You, 2009). The former
approach has been proven to be more cost-effective (Al-Rousan, Masad, Tutumluer & Pan,
2007). However, the features captured and quantified parameters used to evaluate the angu-
larity and surface texture of coarse aggregates using 2-D imaging techniques studied relatively
simple aggregate structures; only a few have been applied to studying complex structures
(Liu, Zhang, Li & Chen, 2014). In this paper, under different compaction numbers, dense
gradation asphalt mixture AC-20 specimens were molded by SGC, then a cross-section was
obtained by cutting.
A typical asphalt pavement consists of 86% aggregates, 10% asphalt cement and 4% air-
void by volume. The configuration characteristics of coarse aggregate largely reflect the uni-
formity of mixture. Most of existing studies focus on the mixture images partition (Moon
& Wistuba, 2015). This paper offers a synthesis center of mass of the coarse aggregate as
parameter.
1347
In the 2-D images, the coordinates of synthesis of center of mass is (X,Y). Coarse aggre-
gate centroid is (xi, yi). Coarse aggregate area is si .
The mathematical expectation is given in Equations 1 and 2, which can be simplified as
Equation 3.
∑ s (X
n
i i )=0 (1)
∑ s (Y − y ) = 0
n
i i (2)
⎛ ∑ si xi ∑ si yi ⎞
( )=⎜ n , n ⎟ (3)
⎜ ∑ si ∑ si ⎟
⎝ n n ⎠
D ( X − X 0 )2 (Y − Y0 )2 (4)
According to the AC-20 mixture design, the specimens have taken shape under 60 and 80
times gyratory compaction, which represent compaction effort. The cross-section images
were obtained. At the same time, in order to get the best optical effect, the dolomite was used
to replace the common limestone ore material. AH-70 bitumen was used as well.
1348
The mixture of the specimens is shown below.
Figure 1. shows how to the cross-section images were obtained.
First, the images were pre-processed with gray processing and tailoring. Then, the median
filter algorithm was deployed to process the droplet images. The dolomite which employed
as coarse aggregate provided good conditions for image segmentation. The image resolution
was adjusted to 564 * 564.
We adopted the OTSU threshold method to confirm the threshold of binary image. This
method has been extensively applied in image segmentation for being simple and intuitive.
The binary images separated the aggregate with open operation. Then, these are processed
by hole-filling. The images were next processed by morphological operations to filter differ-
ent particle sizes of coarse aggregates. Finally, the diameters of the circles which were at the
same area of 2-D coarse aggregates were obtained.
1349
4 RESULTS AND DISCUSSION
Using the algorithm, synthetic centroid coordinates were obtained and plotted on the
graph.
Figure 4 shows the position and diameters of different sizes of particles in the synthetic.
A, B, C, D represent >4.75 mm, >9.5 mm, >13.2 mm, >16 mm, respectively. In this picture,
we didn’t find any 2-D aggregate whose diameter was above 19 mm. It illustrates this obvious
trend in the specimen synthetic centroid.
They are closer to the picture center as the sieve became smaller. Synthetic centroid coor-
dinates may or may not be on the white aggregate.
1350
Figure 5. Synthesis of center distance—sieve.
601 4 6 8 18 72 108
602 3 6 9 26 65 109
603 0 3 8 27 67 105
801 2 5 14 18 73 112
802 2 4 10 27 75 118
803 2 6 9 22 69 108
ference in the synthetic centroid distance below 13.2 mm under 80 times is smaller. However,
the synthetic centroid distance above 13.2 mm under 80 times indicate fluctuations.
Compaction also makes the coarse aggregate arrangement more reasonable. In order to
further clarify this phenomenon, the average value under different compaction numbers are
illustrated below.
Figure 6 shows that coarse aggregate total configuration does not increase significantly,
but the larger particles become smaller with the increase in compaction number. Therefore,
large particle configuration has a uniform movement trend by compaction.
1351
4.2 Different size range
Further calculation of the synthesis of center distance is done based on different size
ranges. The 2D coarse aggregates are shown below.
Because these numbers obtained from different cross-sections are close, it implies that the
cross-sections are not segregate and that the above analysis is reliable.
Different size ranges of coarse aggregate were also listed. Figure 7 shows synthesis of
center distance change for these ranges.
Obviously, the synthesis of center distance of coarse aggregate below 13.2 mm changed
little whether it was 60 times or 80 times. The synthesis of center distance of coarse aggregate
above 16 mm are unstable. Because the number of coarse aggregates above 16 mm is less,
they have no statistical value. Overall, the synthesis of center distance of different size ranges
have greater variability. This suggests that different size ranges of coarse aggregates exist as
an interference characteristic.
The average values of synthesis of center distance were calculated and plotted as follows.
Figure 8 shows the synthesis of center distances of different size ranges changed signifi-
cantly from 60 times to 80 times. The curve of synthesis of center distances under 60 times
is smooth, unlike the curve of synthesis of center distances of different size ranges under 80
times. This phenomenon shows that gyratory compactor has a kneading action. Considering
AC-20 density under 60 times, the knead effect is a main factor at the end of the compaction
stage.
1352
5 CONCLUSIONS
The synthesis of center distance was built to statistically investigate the forming character-
istics of asphalt mixture during the gyratory compaction process. We have compared them
obtained from AC-20 model and that got directly from the AC-20 cross-section images in the
lab, at the same time the different size range were also studied in this paper. The following
conclusions could be drawn:
1. The synthesis of center distances of coarse aggregate are smaller as the sieve pore becomes
smaller.
2. Gyratory compaction may not improve the uniformity of the coarse aggregate but larger
coarse aggregate configurations are improved.
3. At the end of the gyratory compaction stage, the knead effect is a main factor.
ACKNOWLEDGMENT
The authors would like to acknowledge the financial support provided by the National Sci-
ence Foundation through the project 51378123. The funding supported the generated speci-
mens and image processing.
REFERENCES
Al-Rousan, T., Masad, E., Tutumluer, E. & Pan, T. (2007). Evaluation of image analysis techniques for
quantifying aggregate shape characteristics. Construction and Building Materials. 21(5):978–90.
Bruno, L., Parla, G. & Celauro, C. (2012). Image analysis for detecting aggregate gradation in asphalt
mixture from planar images. 3(28):21–30.
Hassan, N.A., Airey, G.D., & Hainin, M.R. (2014). Characterization of micro-structural damage in
asphalt mixtures using image analysis. 3(54):27–38.
Liu, T., Zhang, X., Li, Z., & Chen, Z. (2014). Research on the homogeneity of asphalt pavement qual-
ity using X-ray computed tomography (CT) and fractal theory. Construction and Building Materials.
10(68):587–598.
Liu, Y. & You, Z.P. (2009).Visualization and simulation of asphalt concrete with randomly generated
three-dimensional models[J].Journal of computing in civil Engineering. 23(6):340–347.
Masad, E. et al. (1999).Quantifying laboratory compaction effects on the internal structure of
asphalt concrete.Transportation Research Record: Journal of the transportation Research Board.,
1691:179–185.
Moon, K.H., & Wistuba, M.P. (2015). Analyzing aggregate size distribution of asphalt mixtures
using simple 2D digital image processing techniques. Arabian Journal for Science and Engineering.
40(2):1309–26.
Peng, Y. & Sun, L. (2007). Effects of Experimental Method and Aggregate Type on Index of Asphalt
Mixture Homogeneity. Journal of Tongji University. 3(35):346–350
Sefidmazgi, N.R. & Bahia, H.U. (2014). Effect of compaction conditions on aggregate packing using
2-dimensional image analysis and the relation. Materials and Structures. 8(47):1313–1324.
Wang, L.B., Wang, X.R., Mohammad, L. & Abadie, C. (2005). Unified method to quantify aggre-
gate shape, angularity and texture using Fourier analysis. Journal of Materials in Civil Engineering
17(5):498–504.
1353
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The objective of this paper is to propose a rutting prediction method of nano-
modified asphalt concrete. The nano-modifier consists of nano-organobentonite, nano-SiO2
and SBS. A viscoelastic-viscoplastic damage constitutive model is utilized to characterize the
mechanical behaviour of the modified asphalt concrete. Finite element simulations of rutting
tests are conducted and the results are compared with laboratory wheel tracking tests to verify
the validity of the rutting prediction method. Then real pavement structures are modelled in
ABAQUS software and rutting is simulated using the proposed prediction method. The devel-
opment of rutting is studied and the rut depth of an asphalt pavement under different loading
conditions and temperatures is calculated. The FEM simulation results exhibit a good agree-
ment with laboratory experiments. The results of the rutting prediction indicate that both the
loading level and temperature have a significant effect on the development of rutting.
1 INTRODUCTION
2 CONSTITUTIVE MODEL
where εijve = the viscoelastic strain rate while εijvpd = the viscoplastic strain rate.
2.2 Viscoelastic
The Generalized Maxwell model, which is widely used to characterize viscoelastic mechani-
cal behavior, is chosen as the viscoelastic part.
The relaxation modulus E(t ) is formulated in the form of a Prony series:
k
E (t ) = E∞ + ∑ Ei e −t / ρi (2)
i =1
where E∞ = equilibrium modulus; Ei and ρi represent the elastic modulus and relaxation time
of the ith Maxwell element, respectively; k = the number of the Maxwell element.
2.3 Viscoplastic
The Mohr-Coulomb yield criterion is adopted. In past studies it was verified that this crite-
rion is suitable for asphalt concrete. It is given below:
1356
⎧ I1
⎪ f Rmc 3J 2 − 3 tan φ − c = 0
⎪ 1 π 1 π
⎪Rmc = sin(θ + ) − cos( + ) tan φ
⎪ 3 cos φ 3 3 3
⎪ ⎛ ⎞
3
⎪ r
⎨cos( ) = ⎜ ⎟ (3)
⎪ ⎝ q⎠
⎪ 27
⎪r = 3 J3
⎪ 2
⎪ q = 3J 2
⎪⎩
∂ff
εij λ (4)
∂σ ij
g J 2 − α I1 (5)
∂g
εij λ (6)
∂σ ij
where F f ) = the function of yield criterion; Γ and N are material constants. x represents
the Macaulay brackets:
⎧ x, x ≥ 0
x =⎨ (8)
⎩ 0, x < 0
∂g
εijvp = Γ F f )
N
(9)
∂σ ij
2.4 Damage
According to the strain equivalence principle, the effective stress is defined as σ ij σ ij ( − D ).
Correspondingly, the yield criterion with damage can be derived:
1357
I1
Rmc 3J 2 − tan φ
I 3
F f ) = Rmc 3J 2 − 1 tan φ − c = −c (10)
3 1− D
where I1 I1 (1−
1 D ) and J 2 1 D )2 .
J 2 (1−
c = c0 + a [ b
bp ] (11)
D Aexp(
A np )(1 − D )m (12)
Modifier A 1% 5% /
Modifier B 1% 5% 5%
1358
Figure 1. Gradation of asphalt mixture.
where D0 = glassy compliance; D j and τ j represent the creep compliance and retardation time
of the jth element, respectively; m = number of elements.
The processed experiment data of the viscoelastic strain is fitted based on the least squares
method in the MATLAB software. The fitted curve of asphalt concrete with modifier A at
40°C is shown in Figure 2 and the viscoelastic parameters are summarized in Table 2.
1359
Figure 2. Viscoelastic strain fitted curve.
A 40 4.85 2.44 0.40 0.14 29.01 349.37 71.06 5.54 4.01 8.60
A 60 4.70 2.90 0.33 0.20 31.26 124.66 111.95 6.10 3.75 7.48
B 40 4.64 2.80 0.30 0.34 40.85 235.98 119.81 1.38 5.41 9.98
B 60 6.39 3.37 0.29 0.40 62.36 214.69 20.51 2.39 1.96 5.07
*T = Temperature.
1360
⎧k ⎫
min ( , N ,α ,φ , c0 , a, b, A, n,, ) = ⎨∑ ( vpdc vpde )2 ⎬ (14)
⎩ i =1 ⎭
where ε vpdc = calculated viscoplastic damage strain; ε vpde = laboratory test data; k = the number
of experiment data.
The fitted curve of asphalt concrete with modifier A at 40°C is shown in Figure 3 and the
viscoplastic damage parameters are summarized in Table 3.
1361
Figure 5. FEM model validation.
The equivalent loading principle is adopted to calculate the loading time in the FEM simu-
lation. The loading time of the laboratory wheel tracking tests is 60 min, which is equivalent
to 340 s.
Figure 5 presents the results of the validation in terms of rut depth. In general the qualita-
tive trend of the curves of the simulation is similar to that of the laboratory tests for both
mixtures. It shows that the mixture with modifier B has a better rutting resistance than that
with modifier A. The results indicate that the proposed FEM model is reasonably practicable
to simulate the development of rutting of nanomaterial modified asphalt concrete.
tom. The similar seed distribution is used to reduce the mesh density outside the load area
and the same boundary conditions are implemented. The tire contact area is rectangular. The
model is shown in Figure 6.
Asphalt concrete with modifier B is used to characterize the asphalt layer. The properties
of the other materials are given in Table 4.
Rutting simulations under different temperature and different load levels are conducted.
The results are presented in Figure 7a, b. It indicates that both the temperature and load level
have significant effects on the development of rutting.
5 CONCLUSIONS
REFERENCES
Atef, Saleeb, Robert Y. Liang, Husam Al Qablan, & Dave Powers. 2005. Numerical simulation tech-
niques for HMA rutting under loaded wheel tester. International Journal of Pavement Engineering,
6(1):57–66.
Fischer, H.R., & Bos, B. 2010. Bitumen composition with improved ageing resistance. US, US7803222.
1363
Ghile, D.B. 2006. Effects of nanoclay modification on rheology of bitumen and on performance of
asphalt mixtures. Delft University of Technology. Delft, The Netherlands.
Hua J. 2000. Finite element modeling and analysis of accelerated pavement testing devices and rutting
phenomenon. Purdue University. West Lafayette, USA.
Huang, X., Zhao, Y., & Gao, Y. 2006. Highway asphalt pavement design theory and method, Beijing:
China Communications Press.
Hunter, A.E., Airey, G.D., Harireche, O., 2007. Numerical modeling of asphalt mixtures wheel tracking
experiments. International Journal of Pavement Engineering and Asphalt Technology, 8:52–71.
Jahromi, S.G., & Khodaii, A. 2009. Effects of nanoclay on rheological properties of bitumen binder.
Construction & Building Materials, 23(8): 2894–2904.
Kettil, P., Lenhof, B., Runesson, K., & Wiberg, N.E. 2007. Simulation of inelastic deformation in road
structures due to cyclic mechanical and thermal loads. Computers & Structures, 85(12): 59–70.
Lu, Y., & Wright, P.J. 1998. Numerical approach of visco-elastoplastic analysis for asphalt mixtures.
Computers & Structures, 69(2):139–147.
Ma, F. 2004. Research on performance of pavement and modification mechanism of Nano-CaCO3 mod-
ified asphalt. Chang’an University. Xi’an, China.
Peng, M. Xu, Z. 2006. Nonlinear constitutive model of asphalt pavement permanent deformation. SCI-
ENCE IN CHINA, 36(4):415–426.
Rashid K. Abu Al-Rub, Masoud K. Darabi, Chien-Wei Huang, Eyad A. Masad, & Dallas N. Lit-
tle. 2012. Comparing finite element and constitutive modelling techniques for predicting rutting of
asphalt pavements. International Journal of Pavement Engineering, 13(4):322–338.
Sun, L., Xin, X., & Wang, H. 2012b. Comprehensive comparison of nano-materials for asphalt modifi-
cation. Journal of Transportation Engineering & Information, 10(2):1–5.
Sun, L., Xin, X., Wang, H., & Gu, W. 2012a. Performance of nanomaterial modified asphalt as paving
materials. Journal of the Chinese Ceramic Society, 40(8):1095–1101.
Ye, C., & Chen, H. 2009b. Study on road performance of Nano-SiO2 and Nano-TiO2 modified asphalt.
New Building Materials, 36(6):82–84.
Ye, C., Chen, H. & Li, J. 2009a. The effect of Nano-TiO2 on modified asphalt’s high temperature proper-
ties. Journal of Shandong Jiaotong University. 17(03):63–66.
Zhang, J. et al. 2004. The effect of Fe3O4 nanometer particles on the three main indexes of modified
asphalt. Journal of Shandong Jiaotong University. 12(4):10–14.
Zhang, J., Huang, X., & Wang, X. 2007. Analysis of asphalt pavement rut based on elastic-viscoplastic
theory. Journal Of Highway And Transportation Research And Development, 24(10):20–24.
Zhu, H., & Sun, L. 2013. A viscoelastic–viscoplastic damage constitutive model for asphalt mixtures
based on thermodynamics. International Journal of Plasticity, 40(1):81–100.
Zhu, Y. 2010. Study on the mechanistic behavior of asphalt pavement based on viscoelastic-viscoplastic
theory. Southeast University. Nanjing, China.
1364
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
1365
Finally, data which refer to weather conditions, air pollution, noise pollution and other envi-
ronment-related pieces of information can minimise the environmental impact and improve
sustainability and make the transportation infrastructure more bearable and equitable.
Furthermore, better knowing load spectrum, weather conditions, materials quality, infra-
structural health conditions (strains, stresses, deflections) reduces process design variance
and improves asset management over time (maintenance and rehabilitation options, etc.),
therefore lowering agency costs. From the above factors, it turns out that sensor-based moni-
toring (close or/and remote sensing, i.e., platforms which gather electromagnetic radiation
emitted or reflected from the object or area of interest) can benefit road maintenance and
rehabilitation design. Sensors embedded in the pavement or in the tyre can provide many of
the above pieces of information.
For example, sensors embedded in the tyre can be capable of detecting normal pressure,
deflection, and/or longitudinal strain, indicating also tractive force (Armstrong et al, 2015).
Another issue is the transition from many indicators (provided by sensors) to real informa-
tion. Understanding how the integration of models and sensors can contribute to harnessing
Big Data and, more importantly, make the vital step from ‘Big Data’ to ‘Big Information’ is
vital to the overall process (Reis et al, 2015).
The same concept of sensor is almost unlimited and encompasses a wide array of exam-
ples. For example, fiber Bragg grating (FBG) is a type of distributed Bragg reflector con-
structed in a short segment of optical fiber that reflects particular wavelengths of light and
transmits all others. It can be used for quality control of asphalt pavement compaction (Yiqiu
et al, 2014).
On the other hand, sensor operations and data transmission need energy and energy har-
vesting can help supply this required energy. Not only may energy harvesting provide energy
in order to increase the self-supporting energy capability of roads, but also it may provide
electricity to areas that are remote from main electricity lines (Kim et al., 2015).
To this end it is noted that there the following elements require energy in a roads (Cooper
and Mitchell, 2008):
• Street lights (Traffic routes have lamps with about 150-watt bulbs, while town centres have
the strongest bulbs of between 250 and 400 watts.)
• Traffic bollards (9*11 W)
• Electroluminescent signs (15–25 W)
• Traffic lights (Typical incandescent traffic lights use 100-watt or 150-watt bulbs; LED
arrays consume 12–20 watts).
In Italy the cost of 1 kWh varies from 0.04€ to 0.17€. It turns out that in eight hours, the
cost of one out of the above devices varies from 0.003 € up to 0.50€, as a function of technol-
ogy and hourly cost. Furthermore, it is noted that solar-powering is growing in importance
and in this case energy supply isn’t required (Cooper and Mitchell, 2008). Can energy har-
vesting be useful in the aim of providing road and road users energy, so improving safety and
asset management? The energy that can be recovered from a piezoelectric system depends
on a number of variables. For example, a bending element bolted to a rigid surface that
undergoes 80-gram force to its tip at a frequency of 60 Hz can produce a power output of
3.6 mW.
A small piezoelectric switch can generate 0.5 mJ per push and has a cost around 20 €. Now,
given 2 pushes/sec (people typically walk at about 2 paces/second, therefore their frequency
of excitation is around 1 Hz), this gives about 1 mW (1 mJ/s) of power. With 30 minutes of
continuous walking, this generates 30 minutes * 60 sec/minute * 1 mJ/sec = 1800 mJ. Tyre-
road interaction is very complex. This notwithstanding, it is possible to observe the wheel-
base (spacing between axles, e.g., bicycle: 1.1 m, motorcycle: 1.5 m; light vehicle: 2.8 m; truck:
4.5 m) has an outstanding effect on dominant frequencies. From the standpoint of an object
embedded into the friction course, this implies that frequencies around 3 Hz (bicycles), 11 Hz
(motorcycles), 8.0 (light vehicles), 3 Hz (heavy vehicles) may be reasonable (see also Cebon,
2000). Under the abovementioned hypotheses, this would correspond to 1–6 mW, i.e., to 2–10
1366
Joule in one hour. Of course the above numbers provide only a ballpark figure because many
parameters are involved (including well path wandering).
In contrast, solar cells, based on photovoltaic modules, can supply about 3–5 kW/m2 (when
working), with a cost around 500–1500€/m2.
It is clear that technologies are growing in performance, they have a different level of
development, more applications are emerging, which makes comparisons unreliable. This
notwithstanding, if we compare the two systems, it may be observed that the cost of one Watt
of solar energy is around 10–30 cents (of euro) versus 3–10 k€ in the case of the piezoelectric
system.
For thermoelectric materials, they show the thermoelectric effect (i.e., conversion of tem-
perature differences to electric voltage) and this allows harvesting energy. Furthermore, note
that this implies the external active mechanical cooling of the pavement, which points towards
cooler pavements. This seems to have the potential to help address the problem of urban heat
islands and global warming (Hasebel et al, 2006; EPA, 2008).
2 OBJECTIVES
In the light of the above, the goal of the study described in this paper is to investigate the fea-
sibility of a smart device, part of a larger Wireless Sensor Network (WSN), which integrates
different energy harvesting devices in the pursuit of optimising pavement maintenance and
rehabilitation (Figure 1).
Figure 1 illustrates the graphical abstract of the paper. It focuses on three different har-
vester systems:
• The Piezoelectric Energy Harvesting System (PEHS);
• The solar cell;
• The thermoelectric generator.
The above harvester systems are linked to an energy harvesting management circuit, and
to a battery. This allows transmitting data to a laptop (located on a vehicle). Data are meas-
ured by a Wireless Sensor Network (WSN) and transmitted by means of a transmitter.
1367
The next section focuses on how better monitoring can positively affect the lifecycle of
pavements and, on a more general note, road asset management. Afterwards, design of
experiments and preliminary experiments are discussed. Finally, conclusions and references
are reported.
1368
Importantly, fewer, more sustainable and better planned interventions increase the safety
for the workers. To this end, the quantification of the impact can be derived based on prob-
ability analysis and based on the same algorithms used to quantify agency and user costs for
a period of analysis (even infinite, see Weed, 2001, Pratico’ et al, 2012).
The conceptual framework of the model here proposed to quantify sensing effects can be
formulated and applied based on the following steps:
i. estimation of the expected life of the flexible pavement for a given pavement design (see
AASHTO Guide for Design of Pavement Structures, 1993; MEPDG, 2007);
ii. estimation of the present value of agency costs for each solution;
iii. estimation of the corresponding present value of user costs;
iv. estimation of the present value of externalities;
v. derivation of the overall present value (eq.16);
vi. estimation of the resilience (which sum up the consequences with respect to the control
case);
In pursuit of the estimation of the overall gain for each solution, note that Agency Costs
(AC), user costs (UC) and externality costs (EX) contribute to the overall Present Value, PV
given as:
PV = PV
VAC + PV
VUC + PV
VEX (1)
In order to model resiliency, it may be observed that it refers to the ability to recover
quickly from negative distress conditions and it entails three interrelated aspects: i) lower
probabilities of failure, ii) less-severe negative consequences when failures do occur, and iii)
faster recovery from failures.
As for the resiliency, Re, defined in (0,1), the following tentative algorithm can be set up:
Re = ΔPV ⋅ C −1 + 1 (2)
ΔPV = PV
VA − PV
VB (3)
where C = C0 is the construction cost, PVA and PVB are the present value of costs before and
after the event. Note that difference ΔPV/C is defined in (-1, 0). By referring to the above
algorithms in the context of resiliency studies, it may be observed that: i) agency costs should
be considered; ii) the concept of robustness (as a minimum reduction of functionality) refers
to low values of ΔPV; iii) the concept of redundancy (i.e., presence of alternative options)
is not explicitly included in equations 12–13; iv) the rapidity (rate of recovery) is indirectly
related to ΔPV; v) the resourcefulness of the system (availability of tools, device to activate
the recovery) is not included; vi) single or multiple hazards can be considered by using the
above algorithms.
4 EXPERIMENTS
Overall, the main tasks of the designed experiment include what follows: i) Theoretical analy-
sis of piezoelectric devices, deformations, stresses, and strains (static scheme, loads, defor-
mations, stress and strain, computer-aided simulation); ii) 3D Computer-Aided Design of
the experimental set up; iii) Material testing through the experimental set up (controlled
variables: temperature; Inputs: deformation, with a given frequency, a given load history;
Outputs: energy, number of cycles to failure, fatigue life evaluation, S-N curve); iv) Theoreti-
cal analysis, set up design and material testing for the Thermoelectric Generator (TEG); v)
Theoretical analysis, set up design and material testing for the solar cell for pavement appli-
cations; vi) Theoretical analysis, set up design and material testing for the abovementioned
three devices. This section refers to the preliminary simulations carried out.
1369
Converting vibrational or mechanical energy into usable electrical energy for powering
wireless sensors is an attractive approach that can be pursued in terms of piezoelectric effect.
Depending on the mechanical energy source, for example a moving human body or a vibrat-
ing structure, the excitation frequency varies from less than 10 Hz (human motion) to over
30 Hz (machinery and equipment vibrations).
Piezoelectric energy harvesters are also known as vibration-based power generators since,
in simple terms, when a piezoelectric materials is mechanically strained, electric charges are
collected at the electrodes located on its surface (Telba and Wahied, 2012). The equivalent
electric circuit is a parallel resonant circuit.
With regard to MIDE V21BL (see Figure 2, MID, 2015), the cantilevered-beam piezoelec-
tric energy harvester used in this work, each piezo beam can be thought of as a small current
source in parallel with a capacitor and a parasitic resistance (LTC, 2015). Typical values of
parallel capacitance and resistance are on the order of 10nF and >40MΩ, respectively, where
nF stands for nanofarad MΩ stands for Megaohm.
Nowadays, computer simulations are an essential part of the design process in electronics.
The possibility to predict the behaviour of a system well before its development paves the way
to its fine-tuning and reduces the developing time of the prototype.
To achieve this result, a model of the real system should be readily available or customized
to take into account physical and electronic effects that could affect the simulations (LTC,
2015). The analysis performed within this work is based on a SPICE simulator, a standard
simulation program for electrical and electronic circuits, for which plenty of models of elec-
tronic components and integrated circuits are available.
As above mentioned, two devices were selected to carry out experiments, namely a MIDE
V21BL and an LTC3588–1. Preliminary results from investigation, modelling and simulation
of the MIDE V21BL Piezoelectric energy harvester and an LTC3588–1, an energy harvesting
power supply chip developed by Linear Technology, are reported hereafter. In particular, the
LTC3588–1 is used to obtain a stable DC supply voltage of 3.3V starting from the AC ouput
of the MIDE V21BL.
Figure 3 illustrates a typical application schematic of the above mentioned energy harvest-
ing circuit. The energy harvester and energy management circuit are shown. Sensors, trans-
mitter and microcontroller will be connected to the Vout terminal.
1370
The internal structure of LTC3588–1 is shown in Errore. L’origine riferimento non è stata
trovata.. The internal low-loss full-wave bridge rectifier rectifies the alternating current (AC)
signal from MIDEV21BL. The ultralow quiescent current under voltage lockout (UVLO)
block enables operation of buck converter in presence of enough power to be transferred from
input to the output, otherwise accumulating it on an input capacitor. The buck converter,
activated when the voltage level on the input capacitor is above the threshold, is formed by
one P-type metal-oxide-semiconductor (PMOS) and one N-type metal-oxide-semiconductor
(NMOS) transistor, a control block and an external inductor and capacitor and uses hyster-
etic voltage algorithm to control the output through internal feedback from Vout pin (Errore.
L’origine riferimento non è stata trovata., pin 6). In regulation mode, the sleep state maintains
both input and output quiescent currents minimal. Note that in Errore. L’origine riferimento
non è stata trovata., the internal block diagram of LTC3588–1 is depicted, as per its datasheet
(LTC, 2015).
As anticipated, the equivalent electric circuit of a Piezoelectric energy harvester is a paral-
lel resonance circuit.
Considering a transient time domain analysis, the resonance circuit could be simplified
with a sine wave generator working at a single frequency (41 Hz, as reported in LTC, 2015),
with an Rseries of 119kΩ. The AC amplitude has been empirically defined considering the
force applied on the system. The LTC 3588–1 chip requires only a few external components
to set the output voltage at the requested value of 3.3V. The used schematic is shown in
Figure 4. Internal block description (LTC3588–1) Figure 5. Simulations have been carried
out using LTSpice a SPICE based simulation software for electronic circuits developed by
Linear Technology (LIN, 2015).
In Figure 4. Internal block description (LTC3588–1) Figure 5, the V1 source is the piezo-
electric energy harvester and the yellow box is the LTC device described above. The capacitor
C2 is an energy storing element introduced to allow the operation of a load when the piezo-
electric transducer is not producing energy, although for limited time intervals
Figures 6 and 7 show the output voltage and current settling from start, derived through the
above mentioned simulations. In particular, a 5 s observation time was set, showing that the
output voltage achieves 3.3V with a continuous current of 65uA, delivering 200uW of power
to the output, after 1.5 s of transient. By using duty-cycled operations, the output storage
capacitor C2 could allow providing higher output current bursts. The “power good” signal,
shown in Errore. L’origine riferimento non è stata trovata., is set by the energy management
circuit when the output voltage has reached the requested suitable value. It is then transferred
1371
Figure 5. Simulation of the system.
to the microcontroller that overviews the whole operation of the systems, allowing the right
energy balance between energy storage, sensor’s data acquisition and data transmission.
Based on the above it is possible to conclude that the system is able to provide a sufficient
amount of energy to fulfil the energy requirements for data acquisition and transmission
using low energy RF standards (eg. Bluetooth Smart) or, in conjunction with a standard
RFID reader, to operate autonomously within the reading range thus establishing a bidirec-
tional communication channel for remote sensing application (Merenda et al., 2014).
1372
5 DISCUSSION AND CONCLUSIONS
REFERENCES
1373
Peterson, D.E., Good Roads Cost Less—Report No. UDOT-MR-77–8, Utah DOT, Salt Lake City, UT,
1987.
Praticò F.G., Casciano A., and Tramontana D. The Influence of Dispersion and Location on Pay
Adjustment in Construction Engineering, Journal of Construction Engineering and Management
doi:http://dx.doi.org/10.1061/(ASCE)CO.1943–7862.0000540. ISSN: 0733–9364 (print). 1943–7862
(online), Vol. 138, No. 10, October 1, 2012.
Praticò F.G. & Vaiana R., Improving infrastructure sustainability in suburban and urban areas: is Porous
asphalt the right answer? And how?, Urban Transport: Urban Transport and the Environment in the
21st Century, Edited by: C.A. Brebbia, J.W.S. Longhurst, WIT Press, ISBN: 9781845645809, WIT
Transactions on Ecology and the Environment, 2012, pp.673–684.
Reis, S., Seto, E., Northcross, A., Quinn, N.W.T., Convertino, M., Jones, R.L., Maier, H.R., Schlink, U.,
Steinle, S., Vieno, M., Wimberly, M.C., 2015. Integrating modelling and smart sensors for environ-
mental and human health Environmental Modelling and Software, 238–246, Volume 74.
Schnebele, E., Tanyu,, B.F., Cervone, G., Waters, N., 2015. Review of remote sensing methodologies for
pavement management and assessment, European Transport Research Review, June 2015, 7:7.
Smadi, O, Quantifying the Benefits of Pavement Management, 6th International Conference on Man-
aging Pavements, 2004.
Telba, A., Wahied, G.A., 2012. Modeling and Simulation of Piezoelectric Energy Harvesting, Proceed-
ings of the World Congress on Engineering 2012 Vol II, WCE 2012, July 4–6, 2012, London, U.K.
Walls, J. and Smith, M.R. Life-Cycle Cost Analysis in Pavement Design. FHWA report FHWA-SA-
98–079. 1998. Federal Highway Administration. Washington, D.C.
Weed, R.M. (2001). “Derivation of equation for cost of premature pavement failure.” Transp. Res. Rec.,
1761(1), 93–96.
Wilde, W.J., S. Waalkes, and R. Harrison. Life Cycle Cost Analysis of Portland Cement Concrete Pave-
ments, Report No. FHWA/TX-00/0-1739-1, U.S. Department of Transportation, 1999.
Yiqiu, T., Wang, H., Shaojun, M., Huining, X., 2014. Quality control of asphalt pavement compaction
using fibre Bragg grating sensing technology, Construction and Building Materials 54:53–59 ⋅ March
2014.
Yun et al., (2014) Smart wireless sensing and assessment for civil infrastructure, Structure and Infrastruc-
ture Engineering: Maintenance, Management, Life-Cycle Design and Performance, 10:4, 534–550.
M. Merenda, F.G. Della Corte, C. Felini, (2014) An Autonomous and Energy Efficient Smart Sensor
Platform, IEEE Sensors 2014
1374
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1 INTRODUCTION
Structure materials, such as concrete, age and deteriorate during their service life due to load-
ing, aggressive environment, difficulties in proper inspection methods, and so on. Therefore,
Structure Health Monitoring (SHM) is critical to prevent structure failure. One of the key
components of SHM is data acquisition. Currently, sensors are used to collect strain, tem-
perature, stress, etc. However, these sensors have disadvantages of high cost, loss of mechani-
cal properties in structure, and poor durability. Different from sensors, structural self-sensing
materials, which are structural materials themselves and able to sense their own strain and
damage, could overcome the drawbacks of sensors. Therefore self-sensing materials are ideal
for data collection.
Self-sensing materials were first studied by Chen & Chung (1996). They found that cemen-
titious materials with carbon fiber have self-sensing ability. After that the piezoresistivity of
cement/carbon fiber material had been extensively studied (Banthia et al. 1992, Chung 2002,
Sun et al. 2000, Wen & Chung 1999, Wen et al. 2000, Xie et al. 1996). Results indicated that
cement-fiber composite had higher electrical conductivity than cement paste and was able to
sense the strain change. There was a good correlation between the fractional change in resis-
tivity and loading. However, the piezoresistivity was not reversible when the strain was more
than 0.2% (Wen & Chung 1999).
Different from carbon fiber, Research found that the piezoresistive characteristics of Car-
bon Nanotubes (CNTs) were highly reversible even for a huge strain of 3.4% (Tombler 2000).
Therefore, the CNTs are more suitable than carbon fiber for producing cement-based self-
sensing materials. Li et al. (2007) first studied the mechanical properties and piezoresistiv-
ity of CNT/cement composite. After that, a few studies had been carried out to investigate
the mechanical properties, electrical resistance, and piezoresistivity (Han et al. 2009, 2012,
Konsta-Gdoutos et al. 2010, Saafi 2009, Yu & Kwon 2009). These studies showed promising
results and indicated that the CNT could increase the mechanical properties. Also, CNT/
cement composite had reversible piezoresistivity under repeated or impulsive loading. How-
ever, most of the research focused on the effect of doping level, surface treatment, and disper-
sion on properties of CNT/cement composite. Limited studies were carried out to investigate
the effects of water content, environment, and fatigue. Han et al. (2010) found that the water
content significantly influenced the piezoresistivity of the CNT/cement composite. Ge et al.
1375
(2015) investigate the temperature and water content on the electrical resistance and piezore-
sistivity of CNT/cement composite. The objective of this paper is to study the influence of
temperature and fatigue on the resistivity and piezoresistivity of CNT/cement composite.
2 EXPERIMENTAL PROCEDURE
2.1 Materials
The Ordinary Portland cement P.O. 42.5 with specific surface area of 343 m2/kg was used.
Table 1 shows cement chemical compositions. The initial and final setting times are 173 and
253 minutes, respectively. The multi-walled carbon nanotubes (MWCNTs) were from Shen-
zhen Nano Port Co. Ltd., China. Its basic physical parameters are listed in Table 2. The
sodium dodecylbenzenesulfonate (NaDDBS), a pale yellow powder produced by Shanghai
Connaught Chemical Technology Co. Ltd, was employed as surfactant to disperse MWC-
NTs. Tributyl phosphate (TBP) was added as defoaming agent.
Compositions (%) CaO SiO2 Al2O3 Fe2O3 MgO SO3 Na2Oeq f-CaO
nm um % % m2.g-1
1376
3 EXPERIMENTAL RESULTS
1377
3.2 Effect of fatigue
During the service life, the repeated loading could cause fatigue damage inside material and,
therefore, change the electrical behavior. In this case, it is important to study the effect of
fatigue on the electrical properties of CNT/cement composite. As shown in Figure 4, the
resistivity continuously increased with fatigue stress cycle number. One possible reason is
that micro-crack could be induced and developed under cycle fatigue loading. The cracks
inside specimen breaks the connections and increases the distance among nanotubes, there-
fore, increases the electrical resistance. When subjected to fatigue loading, the change in
resistivity of specimen first increased quickly and then became kind of stable as the cycle
number increased (Figure 5). This phenomena could be caused by the micro-cracks induced
by fatigue. When the specimen was under loading, the micro-cracks will be closes and result
1378
in larger change in resistivity. As the cycle number increased, more cracks could be formed
inside and, therefore, increased the piezoresistivity.
4 CONCLUSION
The CNT/cement composite had good piezoresistivity and could be used for structural health
monitoring. Both resistivity and piezoresistivity were significantly influenced by temperature
and fatigue cycles. Specimens at higher temperature had lower resistivity and higher piezore-
sistivity. As temperature increased from 20 ºC to 50ºC, the fractional change in resistivity was
almost doubled under the same stress. The fatigue loading could cause micro cracks inside
specimen and, therefore, increased the resistivity. As the cycle number increased, the frac-
tional change in resistivity also increased.
ACKNOWLEDGEMENTS
This study is supported by National Natural Science Foundation of China (Grant No. 51108247)
and Scientific Research Foundation for the Returned Overseas Chinese Scholars (20131792).
REFERENCES
Banthia, N. Djeridane, S. & Pigeon, M. 1992. Electrical resistivity of carbon and steel micro-fiber rein-
forced cements. Cement and Concrete Research 22:804–814.
Chen, P. & Chung, D.D.L. 1996. Concrete as a new strain/stress sensor. Composites Part B: Engineering
27(1):11–23.
Chung, D.D.L. 2002. Piezoresistive cement-based materials for strain sensing. Journal of Intelligent
Material Systems and Structures 13:599–609.
Ge, Z. Wang, D. Sun, R. Oeser, M. 2015. Carbon nanotube-cement composite for self-sensing concrete
pavement. Bautechnik. 92:189–195.
Han, B. Yu, X. Kwon, E. 2009. A self-sensing carbon nanotube/cement composite for traffic monitor-
ing. Nanotechnology. 20:445501.
Han, B. Yu, X. Ou, J.P. 2010. Effect of water content on the piezoresistivity of MWCNT/cement com-
posites. Journal of Material Science. 45:3714–3719.
Han, B. Zhang, K. Yu, X. Kwon, E. & Ou, J. 2012. Fabrication of piezoresistive CNT/CNF cementitious
composites with superplasticizer as dispersant. Journal of Materials in Civil Engineering. 24(6):658–65.
Konsta-Gdoutos, M.S. Metaxa, Z.S. & Shah, S.P. 2010. Multi-scale mechanical and fracture character-
istics and early-age strain capacity of high performance carbon nanotube/cement nanocomposites.
Cement and Concrete Composite. 32(2):110–5.
Li, G.Y. Wang, P.M. & Zhao, X. 2007. Pressure-sensitive properties and microstructure of carbon nano-
tube reinforced cement composites. Cement and Concrete Composites 29:377–382.
Luo, J. Duan, Z. Zhao, T. Li, Q. 2011. Effect of compressive strain on electrical resistivity of carbon
nanotube cement-based composites. Key Engineering Materials. 483:579–83.
Saafi, M. 2009. Wireless and embedded carbon nanotube networks for damage detection in concrete
structures. Nanotechnology. 20:395502.
Sun, M. Liu, Q. Li, Z. & Hu, Y. 2000. A study of piezoelectric properties of carbon fiber reinforced con-
crete and plain cement paste during dynamic loading. Cement and Concrete Research 30:1593–1595.
Tombler, T.W. et al. 2000. Reversible electromechanical characteristics of carbon nanotubes under local-
probe manipulation. Nature 405:769–772.
Wen, S. & Chung, D.D.L. 1999. Piezoresistivity in continuous carbon fiber cement-matrix composite.
Cement and Concrete Research 29:445–449.
Wen, S. Wang, S. & Chung, D.D.L. 2000. Piezoresistivity in continuous carbon fiber polymer-matrix
and cement-matrix composite. Journal of Materials Science 35:3669–3675.
Xie, P. Gu, P. & Beaudoin, J.J. 1996. Electrical percolation phenomena in cement composites containing
conductive fibers. Journal of Material Science 31(15): 4093–4097.
Yu, X. & Kwon, E.A. 2009. Carbon nanotube/cement composite with piezoresistive properties. Smart
Materials and Structures. 18(5):055010
1379
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
1381
1.3 Vehicle-Pavement Interaction
V-PI can be defined as “the system in which the vehicle and the pavement exert mutual forces
on each other”. This system should be able to describe the pavement and the vehicle their
components as well as the way in which these components influence each other (Steyn and
Monismith, 2010).
There are thee frequency ranges that should be noted during vehicle-pavement interaction
(Jarimopas et al., 2005):
0.1 to 5 Hz. (Range representing body bounce)
5 to 20 Hz. (Range representing axle hop response)
>20 Hz. (Responses from structure, road roughness and drive train)
1.5 Accelerations due to truck type and the position on the truck
Numerous types of vehicles are used for the transportation component of the supply chain.
For the transportation of fresh produce in South Africa there are three main vehicle types
that are being used, including two-axle rigid trucks, articulated trucks with a tandem drive
axles and interlinks with seven axles (Steyn and Monismith, 2010). These different trucks
with different suspensions give rise to different measured accelerations. Steel-spring and air-
ride suspensions were analysed by Hinsch et al. (1993). Vibrations were measured at the cen-
tre of the truck and near the rear. This study indicated that the accelerations measured at the
centre were 36% and 59% of accelerations measured at the rear of the truck for steel-spring
and air-ride suspensions respectively.
1382
1.7 Effect of different road conditions
Pretorius and Steyn (2012) analysed the accelerations generated by different road conditions.
The roads included a well maintained national road, a provincial road and a gravel road. The
national and provincial roads had good IRI values (between 0.79 m/km and 2.54 m/km) and
the gravel roads were in a poor condition with an IRI value of 8.06 m/km.
Acceleration distributions for the different roads were compared. The gravel road had the
widest distribution followed by the provincial road and lastly the national road.
The energy absorbed by the system was also calculated and it was found that the energy
for the gravel roads was significantly higher than for the provincial and the national roads.
Although the section of gravel road was much shorter than the section of provincial and
national road more damage to produce could occur in a short distance of road in a poor
condition.
The literature review supports the notion that road conditions and vehicle operating condi-
tions affect the response of vehicles traveling on roads, and also the condition of the trans-
ported cargo. Based on this background, an experiment was devised to evaluate the effect of
road condition specifically for transported tomatoes in South Africa. Three tomato farms,
in the Limpopo province, owned by the ZZ2 group were selected for this study. The farms
include Esmefour, Pondrift and Steve Mohale.
Trucks travel from the farms via a gravel road followed by a provincial road and a section
of national road to the central packaging house at Mooketsi. From the packaging house
trucks travel on the national road via Gauteng until they reach the market in Pietermaritz-
burg. The location of the ZZ2 farms and the fresh produce market is shown in Figure 1. The
average distance travelled from the farms to the fresh produce market is roughly 1 000 km.
The maximum speed that the trucks are allowed to travel is 80 km/h. It therefore takes them
between 12 and 15 hours to complete the trip. Trucks leave the central packaging house in the
afternoon around 17h00 and they reach the market at around 6h30 the next morning.
Figure 1. Location of ZZ2 farms in Limpopo and fresh produce market in Pietermaritzburg (Image
adapted from Wikipedia).
1384
Figure 2. USB Accelerometer Model X6-1A (GC data concepts, 2010).
the accelerations that were logged by the accelerometers to different positions on the route
travelled by comparing the GPS time log to the accelerometer time log. The accelerations
measured can thus be compared for different locations with different IRI values.
One kilometre road sections of good, average and poor quality were selected for the gravel
and provincial road. A Fast Fourier Transform (FFT) analysis was done for each acceler-
ometer placed on the truck, boxes and half bins for the one kilometre section to determine
the dominant frequencies. The dominant frequencies of the different sections and container
types were compared.
Parameters such as the temperature and humidity were also monitored during the experi-
ment, and these data will be used in a complimentary study by food technologists to evaluate
the specific effect of these parameters on the tomato deterioration (this part falls outside of
the scope of this paper).
1385
3 RESULTS AND DISCUSSION
To compare the effect that different road conditions has on the transportation of toma-
toes, in two different container types, one kilometre sections of good, average and poor road
condition were selected based on IRI data. Figure 4 shows a representation of IRI values,
travelling speed and processed acceleration measurements for the gravel road section between
Esmefour and the central packaging house at Mooketsi. Three one kilometre sections identi-
fied as good, average and poor can also be recognized on the graph. The mean IRI values for
the good, average and poor sections of the gravel road is 3.14 m/km, 4.54 m/km and 5.89 m/
km respectively.
Accelerometer data for the sections were assessed using a FFT analysis to determine the
dominant frequencies for the sections.
Figure 5 displays the Power Spectral Density (PSD) plot for the accelerometer placed on
the truck for a good portion of the farm road. On the graph two peaks can be observed, one
at a frequency of 3 Hz and one at a frequency of 14 Hz. These frequencies represent body
bounce and axle hop respectively as mentioned in Section 1.3.
The frequencies that were most dominant for each accelerometer-IRI pair were compared
and is shown in Figure 6.
Figure 4. Acceleration Distributions, Speed and IRI measurements for the gravel road portion from
Esmefour.
1386
Figure 5. PSD plot for the accelerometer on the truck on a good farm road section.
The 3D plot in Figure 6 shows that the small boxes experience the high frequencies from
axle hop as dominant. For the frequency and IRI value ranges of the small boxes the Z-axis
power value is relatively high. The power value is a representation of the energy that the sys-
tem absorbs, or put differently, a higher power value indicates that less energy is required for
the system to accelerate vertically.
These findings are in line with previous studies done by O’Brien (1965), Jarimopas et al
(2005), Berardinelli et al (2004) and Pretorius and Steyn (2012) indicating that although pal-
lets unify a large amount of boxes, the boxes still respond individually to vertical accelera-
tions (especially the top boxes of the pallet).
The half bins experience both body bounce and axle hop frequencies as dominant, although
the power value is low. The half bins weigh significantly more than the small boxes and more
energy is required to vertically accelerate the larger mass.
The accelerometers placed on the truck also experience body bounce and axle hop fre-
quencies. Dependent on the speed and the position of the accelerometer, more or less energy
is transferred to the system.
Although more energy is required to accelerate the half bins and the vertical movement
of the bins are limited there are still internal pressures between the tomatoes in the different
packaging that plays an important role during the development of a damage model.
The bottom layer of the half bins has at least six to eight layers of tomatoes on top of it,
whereas the small boxes only has two layers. It is expected that the tomatoes in the half bins
would experience higher contact stresses than those in the boxes. Further studies in this field
is recommended to enhance the development of a damage model for in-transit damage to
tomatoes.
The focus of this paper is on the relationship between the road condition, the dominant
frequencies and the PSD data for the two types of packaging. A detailed statistical analyses
of the data obtained for the total length of the routes are outside the scope of the current
paper, and will be dealt with in a subsequent paper.
1387
Figure 6. Comparison of IRI, Power and Frequency for different one kilometre road sections.
Dominant frequencies for one kilometre road sections with different IRI values were calcu-
lated for accelerometers placed on the trucks used for transportation, half bins and small
boxes. Two main groups of dominant frequencies were apparent from the evaluation. These
groups are body bounce, with a frequency range between 0.1 Hz and 5.0 Hz, and axle hop,
with a frequency range between 5,0 Hz and 20.0 Hz.
When considering the small boxes it was observed that they mostly experience axle hop
frequencies and absorb a large amount of energy. The half bins in comparison experience
both axle hop and body bounce frequencies but because of the mass of the unit less energy
is absorbed by the system.
There is however not conclusive evidence to imply that half bins are a better form of
packaging for transportation. The bottom layer of the half bins has at least six to eight
layers of tomatoes on top of it, whereas the small boxes only has two layers. Although the
frequency of vertical acceleration is higher for the small boxes the contact stress and contact
time between the tomatoes are expected to be much less than the contact stress and time in
the half bins. Considering the studies by Van Linden (2006) it is expected that the impact time
would affect the mechanically induced damage.
Fruit with different maturities has different factors that affect bruise susceptibility. To
develop a holistic damage model these factors should also be considered.
For future studies and the development of an in-transit damage model for tomatoes it is
recommended that laboratory simulations should be conducted where in-transit frequencies
are simulated on a vibration table for both half bins and small boxes. The simulations should
be carried out over varying time-periods. The field contact stresses should be compared to
the simulation data to ensure accuracy of the experiment. Damage induced during the exper-
iment should be note. The tomatoes should be stored and monitored together with a control
set to determine the average time by which the shelf life is reduce.
1388
This study could serve as the basis for the development of damage models for other fruits
and vegetables. With this knowledge growers can consider different route options to optimize
and balance financial and time inputs for the best solution.
REFERENCES
Beraedinelli, A., Donati, V., Giunchi, A., Guarnieri, A., Ragni, L. 2004. Damage to pears caused by
simulated transport. Journal of Food Engineering. Vol. 66, March, pp. 219–226.
Chonhenchob, V., Sittipod, S., Swasdee, D., Rachtanapun, P., Singh, S.P., Sing, J. 2009. Effect of truck
vibrations during transport on damage to fresh produce shipment in Thailand. Journal of Applied
Packaging Research, Vol 3, No1, January, pp. 27–38.
Havenga, J.H., Simpson, Z.P., De Bod, A. & Viljoen, N.M., 2014, ‘South Africa’s rising logistics costs:
An uncertain future’, Journal of Transport and Supply Chain Management 8(1), Art. #155, 7 pages.
http://dx.doi. org/10.4102/jtscm.v8i1.155.
Hinsch, R.T., Slaughter, D.C., Craig, W.L., Thompson, J.F. 1993. Vibration of fresh fruits and vegetables
during refrigerated truck transport. Transactions of the ASAE. Vol 36, No 4, July, pp. 1039–1042.
Jarimopas, B., Singh, S.P., Saengnil, W. 2005. Measurement and analysis of truck transport vibration
levels and damage to packaged tangerines during transit. Packaging Technology and Science, Vol 18,
March, pp. 179–188.
Lurie, S., Handros, A., Fallik, E., Shapira, R. 1996. Reversible Inhibition of Tomato Fruit Gene expres-
sion at high temperature. Plant Pathology. Vol 110, pp. 1207–1214.
O’Brien, M., Centry, P.J., Gibson, R.C. 1965. Vibrating Characteristics of fruit as related to in-transit
injury. Transactions of the ASAE. Vol 8, pp. 241–243.
Pretorius, C.J., Steyn, W.J.vdM. 2012. The influence of road roughness on the transportation of fresh
produce. South African Transport Conference. Pretoria:SATC.
SAICE (South African Institute of Civil Engineering) 2011. SAICE Infrastructure Report Card. Mid-
rand: SAICE. Available at: www.saice.co.za [accessed on 26 November 2015].
SOL (State of Logistics) 2013. Tenth Annual State of Logistics Survey for South Africa. Pretoria:
CSIR.
Sayers, M.W., Gillespie, T.D., and Queiroz, C.A.V. (1986) “The international road roughness experi-
ment. Establishing correlation and a calibration standard for measurements.” Washington, D.C.,
USA: World Bank. (World Bank Technical Paper Number 45).
Steyn, W.J.vdM., Monismith, C.L. 2010. Analysis of Vehicle-Pavement Interaction effects on vehicular
loads and transported cargo. University California, Davis. UCPRC-SR-2010-xx. Davis, California.
Van Linden, V., De Ketelaere, B., Desmet, M., De Baerdemaeker, J. 2006. Determination of bruise
susceptibility of tomato fruit by means of an instrumented pendulum. Postharvest Biology and Tech-
nology. Vol 40, pp. 7–14.
Van Linden, V., Scheerlinck, N., Desmet, M., De Baerdemaeker, J. 2006. Factors that affect tomato
bruise development as a result of mechanical impact. Postharvest Biology and Technology. Vol 42,
pp. 260–270.
Wright, R.C., Pentzer, W.T., Whiteman, T.M., Rose, D.H. 1931. Effect of various temperatures on stor-
age and ripening of tomatoes. United States Department of Agriculture. Washington DC.
1389
Rigid Pavements
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Hanwen Deng
Southeast University, Nanjing, P.R. China
Shunzhi Qian
Southeast University, Nanjing, P.R. China
Nanyang Technological University, Singapore
ABSTRACT: Due to the lack of durability, the rate of deterioration for existing con-
crete infrastructures is alarming. The demand for fast and durable concrete repair materi-
als is urgent. This paper reports a fast and durable materials, entitled High-Early-Strength
Engineered Cementitious Composites (HES-ECC). This material can be applied to large scale
repair or retrofit applications, such as long span steel bridge deck, where the time require-
ment has to be finished in limited time. In this study the PE fiber is used in ECC to com-
pletely replace PVA fiber, thus reduce the volume of fiber. Two particle size of sand (40 μm
and 100 μm in average) is used in this research. Experiments of HES-ECC were conducted on
compressive strength, crack width and tensile properties. In this paper, HES-ECC achieves a
compressive strength of 23.4 MPa and tensile stress of about 2.51 MPa after six hours with
strain capacity over 2%. Moreover, HES-ECC exhibits high strain capacity under uniaxial
tension along with a self-controlled crack width of below 60 μm. HES-ECC is a potential
candidate material for fast and durable concrete repairs.
1 INTRODUCTION
The transportation infrastructures, such as expressway, bridge and airport etc., are vital to
the quality of life and economic wellbeing of its citizens. In the US, the nationwide net-
work of roads and bridges alone transports over 75% of goods across the US. While the
importance of the transportation infrastructure is clear, the rate of deterioration is alarming.
Concrete structure deteriorate over time under the combined environment and mechanical
loading, such as freezing and thawing cycles, corrosion and fatigue etc. Concrete damage is
caused by earthquakes and overloading that result in cracking, spalling, or complete failure
[1]. Although numerous materials and techniques are in practice to meet the demand for
rapid, inexpensive, and durable concrete repairs, few address the cracking and debonding
from concrete due to the relatively brittle repair material’s shrinkage [2]. It has been estimated
that the half of all concrete repairs prematurely fail, leading to frequent maintenance and
repair, which result in significant economic and environmental impacts [3].
To reduce the environmental impacts, there is an urgent demand for fast and durable repairs
of concrete structures. This requires that repair materials can rapidly get strength during early
age and remain durable during service. Highway transportation authorities often claim the
repair job to be finished in 6 to 8 hours at night so the pavement can be reopened to traffic in
the following morning [4]. Although high-early-strength and fast-setting concrete materials
have been successfully developed in the past decades, the research on engineered cementitious
composite (ECC) mixed polyethylene (PE) fiber is limited. ECC is a unique class of the new
generation high-performance fiber-reinforced cementitious composites (HPFRCC) featuring
high ductility and medium fiber content, which designed based on micromechanics theory
1393
by Victor Li at 1990s. Tensile strain capacity at a range of 3 to 5% has been demonstrated in
ECC materials using polyethylene fibers and polyvinyl alcohol (PVA) fibers with fiber vol-
ume fraction no greater than 2% [5,6]. Figure 1 shows a typical tensile stress-strain curve of
ECC and tight crack width maintained with increasing strain [5]. The large strain capacity
in ECC is contributed by sequential development of multiple cracks, instead of continuous
widening of only localized crack in concrete. The associated high fracture toughness and
controlled crack width (typically below 100 μm) make ECCs to be an ideal repair material to
improve the serviceability and durability of civil infrastructures.
In this paper, it is attempted to do the research about fast and durable concrete repair
material on High-Early-Strength Engineered Cementitious Composites (HES-ECC). Uniax-
ial tensile and compressive tests were performed to investigate the effect of a low volume
of different fiber on properties of HES-ECC. Moreover, the influence on crack width and
density were also highlighted.
2 EXPERIMENTAL PROGRAM
2.1 Materials
The mixtures are shown in Table 1, the HES-ECC mixture includes the sulfoaluminate cement
(SAC), silica fume (SF), ground silica, silica sand, PVA fiber, and PE fiber. Totally there are
Figure 1. Tensile stress-strain curve and tight crack width control of ECC.
Table 1. Mix proportion of HES-ECCs with different, micro-fibers (Mix by weight (fiber by
volume)).
1394
two particle sizes of sand, including ground silica with 40 μm and silica sand with 100 μm in
average. There are four mixtures which include different dosages (PVA fiber with the content
of 2%, PE fiber with the content of 0.5% and 0.8%). The dimensions of the fibers that were
used for the four different mixture are given in Table 2.
1395
Figure 3. Tensile behaviors of: (a) ECC1; (b) ECC2; (c) ECC3 at different ages.
1396
Figure 4. Comparison of tensile properties of different mixtures at different ages (a) tensile stress and
(b) tensile strain.
compressive strength between ECC2 and ECC3 with the different content of PE fiber. It can
be explained by the mechanism that compressive strength of ECC in controlled not only by
the toughness of matrix but also the fiber bridging restrain the horizontal deformation of
specimen, consequently, increasing compressive strength. It demonstrates that the content of
PE fiber with 0.5% and 0.8% cannot change the property of compressive strength.
1397
Figure 5. Multiple cracking along the length of the specimen.
The age-dependent development of HES-ECC’s tensile strain capacity, as shown in Fig.4 (b),
the tensile strain of ECC1 is only 0.7 MPa at 6 hours and 1.5 MPa at 28 days, far below that of
ECC2 and ECC3. It can be explained by the hydrophobic nature of PE fiber, which eliminates
fiber/matrix chemical bond and enhances the complimentary energy of fiber bridging [8]. For
the ECC2 mixtures, exceeds 4% at 6 hours, increase to 5% at 24 hours, and decreases to 3% at
28 days. The change in tensile strain capacity with age can be explained by the combined effects
of changes in matrix toughness and matrix/fiber interfacial properties. Regardless, the whole
age tensile strain capacities of HES-ECC are several hundred times larger than the normal
concrete materials (only 0.01%). The phenomenon can be reflected by that the tensile strain
capacity of HES-ECC mixtures with low volume of PE fiber is better than that of PVA fiber.
4 CONCLUSIONS
In this paper, it is attempted to develop ECC incorporating a lower volume of fiber with char-
acteristics of high early strength, high deformability and high ductility for rapid and durable
repair of infrastructures. The influences of PE fiber and PVA fiber on the mechanical behavior
of HES-ECC were studied. The following specific conclusions can be drawn from this study:
1. The influence on compressive strength of is negligible when the content of PE fiber is
0.5% and 0.8% which due to the fiber bridging restrain the horizontal deformation of
specimen. While the compressive strength value of ECC2 and ECC3 is higher than ECC1
because of the higher strength of PE fiber.
2. The tensile strain capacity gets improved when the selected fiber is PE fiber, while it does
not change a lot when the content of PE fiber with 0.5% or 0.8%. And, the first cracking
strength increases with time, denoting the toughness of HES-ECC increases. To some
degree, the high fiber strength of PE fiber increase the fiber-bridging capacity relative to
the high matrix cracking strength of HES-ECC and make matrix not easy to crack.
3. The addition of PE fiber decreases the interface chemical bond and enhances the energy
of fiber bridging which result in the fiber pullout which is contributed to the multiple
cracking and smaller tight crack width, consequently, improve HES-ECC’s durability.
ACKNOWLEDGEMENTS
The authors would like to graciously thank the National Natural Science Foundation of
China (No.51008071) and Nanyang Technological University for providing start-up grant
(M4081208).
REFERENCES
Delatte, N.J., Concrete Pavement Design, Construction, and Performance, E&FN Spon, New York,
2007, 392 pp.
Emmons, P.H., and Sordyl, D.J., The State of the Concrete Repair Industry, and a Vision for its Future,.
Concrete Repair Bulletin, July-Aug. 2006, pp. 7–14.
Li V C, Wu C, Wang S, et al. Interface tailoring for strain-hardening polyvinyl alcohol-engineered
cementitious composites (PVA-ECC). ACI Materials Journal, 2002, 99(5):463–472.
Li V C. Engineered cementitious composite-tailored composites through micromechanical modeling
Fiber Reinforced Concrete: Present and the Future. Montreal, Canadian Society for Civil Engineer-
ing, 1998:64–97.
Li, M., and Li, V.C. High-Early-Strength Engineered Cementitious Composites for Fast, Durable Concrete
Repair—Material Properties. ACI Materials Journal, V. 108, No. 1, January-February 2011, pp. 3–12.
Li, M. Multi-Scale Design for Durable Repair of Concrete Structures. PhD dissertation, University of
Michigan, Ann Arbor, MI, 2009, 425 pp.
Peerapong Suthiwarapirak, Takashi Matsumoto, and Tetsushi Kanda.Multiple Cracking and Fiber
Bridging Characteristics of Engineered Cementitious Composites under Fatigue Flexure. Journal of
Materials in Civil Engineering, Vol. 16, No. 5, October 1, 2004, pp. 433–443.
Ravi Ranade, Victor C. Li, Michael D. Stults, William F. Heard, and Todd S. Rushing. Composite Prop-
erties of High-Strength, High-Ductility Concrete. ACI Materials Journal, V. 110, No. 4, July-August
2013, pp. 413–422.
Vaysburd, A.M., Brown, C.D., Bissonnette, B., and Emmons, P.H. ‘Realcrete’ versus ‘Labcrete’ Con-
crete International, V. 26, No. 2, Feb. 2004, pp. 90–94.
1399
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Xu Wang
Xi’an Eurasia University, Xi’an, Shaanxi, China
1 INTRODUCTION
As a new type of pavement, the Exposed-aggregate Concrete Pavement (EACCP) has excel-
lent performance because the surface mortar has being removed and the aggregates are
exposed. Firstly, with rich macro and micro structures, EACCP has good skid-resistance,
especially on rainy days. Secondly, the air between tire tread and pavement is easily given off,
which can reduce air-pumping noise caused by air compression, so EACCP is considered as
an excellent low-noise pavement. Thirdly, compared with Portland cement concrete pave-
ment, EACCP has lower reflection coefficient and its anti-glare performance is good(Xu,
O.M. et al. 2009).
Until now most research(Gibbs, D. et al. 2005, Hall, K. et al. 2007, Han, S. et al. 2005) has
focused on the performance of EACCP. But there is not a suitable method to evaluate the dis-
tribution condition of exposed-aggregates. Exposed-aggregate degree, which was the number
of exposed-aggregates greater than 5 mm in the specific measuring area, was used for evalu-
ation of surface texture. Exposed-aggregate degree was usually obtained by visual way, so
that the result was not accurate. Meanwhile, the measuring area was too small to reflect the
distribution condition of exposed-aggregates. As shown in Figure 1,there is a great difference
between district A and B, the area of which are the same of 10 cm*10 cm.In this study, in
order to evaluate distribution condition of exposed-aggregates quickly and accurately,digital
image technology was introduced as a new method and it was expected to overcome the dis-
advantage of traditional method.
1401
Figure 1. Distribution condition of exposed-aggregates in different measuring areas.
2.1 Materials
Table 1 shows the properties of Qin-ling cement(CEM I 42.5R).The aggregates were basalt.
Strength(3d)(MPa)
Fineness Normal
(Vacuum sieve) Consistency Initial setting Final setting Compressive Flexural
(%) (%) time(min) time(min) Soundness Strength Strength
6.8 24.6 142 222 Qualified 26.4 4.5
4 MORPHOLOGICAL THEORY
Any image of object has its own unique morphological characteristics, that is its geometry.
This is mainly because of the fact that different objects are composed of different materials,
which have different light absorption and reflection properties(Vincent, L. 1993). Morpho-
logical characteristics reflect the difference between the object and its background. The basic
principle of morphology theory is to extract the geometric characteristics of the object based
on the useful brightness information (or gray scale information) of the image, then to calcu-
late the corresponding evaluating indicators on a computer (Elad, M. et al. 2005).
1403
As is well known, the skid-resistance of pavement depends on macro-texture(pavement
surface) and micro-texture(aggregate surface). And usually the micro-texture will not change
if the aggregate is chosen. So the skid resistance of EACCP is closely related with the macro-
texture. In this study the exposed-aggregate degree, which is the number and the area of
the exposed-aggregates, was employed for the evaluation of the macro-texture and the skid-
resistance of EACCP.
In this research, the MATLAB software was applied for the extraction of exposed-aggre-
gates in the processed image of EACCP. And then exposed-aggregate degree was calculated.
During the calculation, it was assumed that 1) each aggregate was a circular particle and its
diameter should be calculated, 2) if only partial ones were contained in the image, aggregates
were complete for analysis.
The exposed-aggregate degree was calculated according to equation 1.
E= N×A (1)
1404
Figure 3. The image after smoothing processing.
Table 3. Calculated result for the number and the area of exposed-aggregates in EACCP.
Table 4. Test results of exposed-aggregate degree, texture depth and friction coefficient.
Exposed-aggregate degree 45.2 47.2 31.1 32.2 30.3 34.3 23.6 28.3 20.4 21.1
Texture depth(mm) 0.95 1.03 1.03 1.13 1.1 1.08 0.98 1.23 1.05 1.08
Friction coefficient 95 97 87 94 93 92 79 82 75 75
Figure 7. Correlation between exposed-aggregate degree and texture depth, friction coefficient.
1406
The results show that exposed-aggregate degree has good correlation with friction coeffi-
cient, while the correlation coefficient between exposed-aggregate and texture depth is small.
It is mainly because of that, texture depth of EACCP is contributed by three indicators: the
number and area of Exposed-aggregate, and exposed-aggregate depth, so exposed-aggregate
degree may not show an overall representation for the texture depth of pavement. Though
the Portable Skid Resistance Tester(SRT) has some disadvantages, it is applied widely in the
world. The correlation analysis with friction coefficient shows that the digital image method
is feasible and can give an accurate and quick evaluation of macro-texture of EACCP.
6 CONCLUSION
REFERENCES
Elad, M. et al. 2005. Simultaneous cartoon and texture image inpainting using morphological compo-
nent analysis (MCA). Applied and Computational Harmonic Analysis 19(3): 340–358.
Gibbs, D. et al. 2005. Quiet Pavement Systems in Europe. Transportation research board: 05–011.
Hall, K. et al. 2007. Long-Life Concrete Pavements in Europe and Canada. Transportation research
board: 07–027.
Han, C.Q. et al. 2008. Temporal noise and spatial noise of CCD camera at different illuminance. Journal
of Applied Optics 29(2):207–210.
Han, S. et al. 2004. Key technology of exposed-aggregate cement concrete pavement. China Journal of
Highway and Transport 17(4):17–20.
Han, S. et al. 2005. Noise reduction performance of exposed-aggregate cement concrete pavement. Jour-
nal of Traffic and Transportation Engineering 5(2): 32–34.
Mitra, S.K. et al. 1991. A new class of nonlinear filters for image enhancement. Acoustics, Speech, Sig-
nal Processing: 2525–2528.
Russo, F. 2000. Noise removal from image data using recursive neurofuzzy filters. IEEE Transactions on
Instrumentation and Measurement 49(2): 307–314.
Russo, F.2002. An image enhancement technique combining sharpening and noise reduction. IEEE
Transactions on Instrumentation and Measurement 51(4): 824–828.
Tsai, W.H. 1985. Moment-prserving thresholding: a new approach. Computer Vision, Graphics, and
Image Processing 29: 377–393.
Vincent, L. 1993. Morphological grayscale reconstruction in image analysis: applications and efficient
algorithms. IEEE Transactions on Image Processing 2(2): 176–201.
Wang, J.W. 2009. Graph and image processing based on MATLAB7.6. Beijing: National defense industry
press
Wu, W.L. et al. 2008. Study on rutting test of asphalt mixture based on digital image processing tech-
nique. Road Machinery & Construction Mechanization 25(3):55–57.
Xu, O.M. et al. 2009. Effect of inter-layer interface on structural performance of concrete bridge deck
pavement.. Journal of Chang’an University 29(5):17–20.
1407
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Jian Ouyang
School of Transportation and Logistics, Dalian University of Technology, Dalian, China
Yiqiu Tan
School of Transportation Science and Engineering, Harbin Institute of Technology, Harbin, China
1 INTRODUCTION
Cement asphalt emulsion (hereinafter abbreviated as CA) composite materials has been
found wide application in road (Schmidt et al. 1973; Brown et al. 2000; Bocci et al. 2011;
Oruc et al. 2007) and building (Li et al. 1998; Song et al. 2006) construction due to their
well properties combined with the merits of cement with high strength and asphalt with
good flexibility. Recently, cement asphalt emulsion mortar served as elastic cushion layer has
played an important role in the non-ballast track structure of high-speed railway (Wang et al.
2008; Tan et al. 2013). The properties of CA mortar are gained more and more concerns due
to the mass construction of high-speed railway in the world.
In placement, CA mortar is grouted into a narrow pre-adjusted chamber with the size of
6450 × 2550 × 30 mm between the concrete roadbed and track slab. The grouting method
requires that CA mortar should possess a good workability. The initial workability not only
governs the construction result of CA mortar in slab track, but also affects the demulsification
rate of asphalt emulsion and asphalt membrane forming process to some extent. However,
there is a challenge to design the workability of CA mortar. In CA mortar, cement is rigid
material and asphalt is flexible material. It can be expected that the mechanical properties
1409
and durability are governed by asphalt to cement mass ratio (A/C) (Wang et al. 2012; Wang
et al. 2011; Wang et al. 2010). It is reasonable that CA mortar and other CA materials can be
adjusted by A/C to get different properties according to the requirements of different struc-
tures. However, the mix proportions of CA mortar differ greatly when A/C is changed, which
definitely affect the design of workability. For pure cement mortar or paste, the workability
and mechanical properties are normally designed in the basis of water-cement ratio (W/C),
and a constant W/C is preferred in the study of different cement mortars or pastes. However,
fresh CA paste is a very complicated suspension combined with cement grains and asphalt
droplets, a constant W/C to study CA paste with different A/C would ignore the effect of
asphalt droplets on the workability of CA paste. The W/C and the sensitivity of W/C on
workability should be different if A/C is changed. Therefore, the W/C of CA mortar with
different A/C should be concerned if studying its workability.
The workability of CA mortar is studied in many previous papers (Cheng & Zheng, 2011;
Xu et al. 2010). The results show that CA mortar with constant W/C has different flow
time and spreading diameter at different A/C. However, the effect of A/C is not paid much
attention to researchers. Besides, flow time of a cone flow test and spreading diameter are
engineering experience indicators to evaluate workability, which cannot describe workabil-
ity from mechanics standpoint. Fortunately, the workability of grouting CA mortar can be
interpreted through the previous studies on the viscosity of cement paste. Numerical simula-
tion and field test show that the flow time of a cone flow test is mainly related with plastic
viscosity (Cremonesi et al. 2010; Ferrara et al. 2012), and the dynamic segregation resistance
is related with viscosity (Saak et al. 2001; Tregger et al. 2012). CA mortar is mainly composed
of cement, asphalt emulsion, water, and sand, in which CA paste take up above 2/3 volume
fraction. Thus, the workability of CA mortar is mainly governed by CA paste. Therefore, the
workability of CA mortar can be interpreted through viscosity study of fresh CA paste.
The rheological properties of fresh CA paste were studied in recent years (Zhang et al.
2012; Peng et al. 2012; Tan et al. 2014). Results from most papers indicate that the rheologi-
cal properties of CA paste is dependent on A/C, W/C, and emulsion types. However, seldom
papers focused on study the W/C characteristic of CA paste with different emulsion types
at different A/C. To study the W/C characteristic of CA paste with different A/C through
viscosity, CA paste should be taken as a suspension which CA particles are suspended in
water. Some viscosity models indicate that the viscosity of suspension (η) is dependent on
particle volume fraction (φ) (Liu 2000). The authors introduced a suspension viscosity model
to quantitatively study the rheology of CA paste with different A/C (Ouyang & Tan 2015a).
The results show that the viscosity of CA paste with different A/C can be predicted by the
employed viscosity model.
In the paper, the relation between viscosity and particle volume fraction of fresh CA paste
with different A/C was studied. Through the φ-η relation, a suitable viscosity range for con-
struction was chosen to discuss the effect of A/C and emulsion types (anionic emulsion and
cationic emulsion) on the W/C characteristic of CA paste. The W/C of CA paste with different
A/C in the same viscosity range was discussed, and the sensitivity of W/C change on viscosity
was studied. At last, the effect of W/C on the flow time of two typical CA mortar was studied.
This work can well reveal the effect of W/C on the viscosity of CA pastes with different A/C
and also reflect the characteristic of W/C for construction of CA mortar to some extent.
1410
respectively. The corresponding A/C was 0, 0.12, 0.24, 0.36, and 0.6. In order to establish the
relation between viscosity and particle volume fraction of CA pastes, four different particle
volume fractions were required for CA pastes with a constant A/C. For the relation between
particle volume fraction and viscosity of different CA pastes could vary greatly, the four
volume fractions for different CA pastes were not fixed but determined according to the real
viscosity test of CA pastes, as shown in the Table 1. Besides, a polycarboxylate-based super-
plasticizer (SP) was used, and its mass ratio to cement was 0.5%. This SP dosage was chosen
because SP has little effect on the viscosity in CA paste when SP/C is above 0.4% (Tan et al.
2014). Furthermore, superplasticizer can retard the demulsification process of asphalt emul-
sion in CA paste (Ouyang et al. 2015b).
All CA pastes were prepared by hand-stirring. For each CA0, cement, superplasticizer,
and water were stirred for four minutes. For other CA pastes, cement, superplasticizer, and
water were stirred firstly for two minutes, and asphalt emulsion was added and continually
stirred for two minutes. All pastes were prepared twice for rheology test.
The mixed CA mortars were employed to study the effect of W/C on the flow-ability of
CA mortars. Two typical CA mortar were fabricated. The basic mix proportions are shown
in Table 2. One is called as high-strength CA mortar, in which A/C is 0.24 and sand to cement
mass ratio (S/C) is 1.4. The CA mortar with this proportion can be employed in CRTS-II
slab track in China, and anionic emulsion is used. The other is named as low-strength CA
mortar, in which A/C is 0.84 and S/C is 2.0. This CA mortar is used well in CRTS-I slab track
in China, and cationic emulsion is employed in it.
φC+φA φC+φA
Mix (%) C(g) AEa(g) W(g) SP(g) (%) C(g) AE(g) W(g) SP(g)
1411
Table 2. Basic mix proportions of CA mortars.
Figure 1. Cone size for flow time test (mm): (a) high-strength CA mortar, (b) low-strength CA mortar.
between rotor and cylinder is 0.85 mm. After loaded, samples were sheared immediately with a
linear increasing shear rate from 0 to 100 s−1 in 1 minute, and then with a linear decreasing shear
rate to 0 s−1 in 1 minute. After the first hysteresis, the samples were sheared again with a gradual
increasing shear rate from 0 to 100 s−1 in 1 minute. The data in the last step was utilized to ana-
lyze. This was performed for each volume fraction twice each on nominally identical mixes.
where ηr is the relative viscosity defined as a ratio of the suspension viscosity (η) to the
viscosity of suspending medium (η0), φ is the particle volume fraction, φm is the maximum
packing density and a is an adjusting factor.
1412
The apparent viscosity of CA paste decreases firstly and then increases slightly with shear
rate (Ouyang & Tan 2015a). Therefore, CA paste has its minimum apparent viscosity. In this
paper, the minimum apparent viscosity of CA pastes was studied by Eq.(1).
The parameters of Eq.(1) are easy to fit from experimental data after Eq.(1) is transformed
to Eq.(2). As can be seen from Eq.(2), 1-ηr−1/2 has linear relationship with φ. Through plotting
the experimental data of 1-ηr-1/2 at four different φ, the parameters φm and a can be fitted by
Eq.(2), as shown in Fig. 2. In the basis of the fitted values φm and a, the minimum appar-
ent viscosity of CA pastes can be predicted through Eq.2 (the viscosity of water in 23°C is
0.000936 Pa.s), as shown in Fig. 3.
1 ηr 1/ 2 11− φm − aφ (2)
ρw ( φ )
W /C = (3)
ρC φC
where ρW and ρC are density of water and cement, respectively. φC is cement particle volume
fraction, which can be calculated by A/C (mA/mC), asphalt density (ρC), cement density, and
Figure 2. Determination of φm and a from the 1-ηr−1/2 versus φ curve (CA6 with anionic emulsion).
Figure 3. The predicted viscosity of CA paste as a function of particle volume fraction (CA6 with
anionic emulsion).
1413
Figure 4. Viscosity versus W/C curves of CA paste (CA6 with anionic emulsion).
total particle volume fraction, as deduced in Eq.(4). Through Eq.(1), Eq.(3), and Eq.(4), the
relation of W/C and viscosity of CA paste can be established, as shown in Fig. 4.
φC φ φ
φC = φ= = (4)
φC φA 1 + φA / φC 1 + A / ρA
m
mC / ρC
A successful construction requires that binder materials should have suitable viscosity
range. A high viscosity of the binder can lead to a loose structure for concrete or mortar;
and also CA mortar with high viscosity cannot be utilized for grout. A low viscosity of the
binder definitely causes bleeding and segregation in construction. For asphalt mixtures, the
rotary viscosity of asphalt is recommended in 0.17 ± 0.02 Pa.s for mix and 0.28 ± 0.03 Pa.s for
compaction in the specification (Ministry of Communications of PR China 2011). For CA
mortar, there is no clear rule about the viscosity of CA paste in the specification, however,
the viscosity of CA paste should be in a suitable range to guarantee the successful grout-
ing. From the reported references (Peng et al. 2014; Wang et al. 2008; Hu et al. 2009), the
minimum viscosity of CA paste for grouting were discussed about between 0.15 and 0.4 Pa.s.
Therefore, the viscosity of CA paste between 0.15 and 0.3 Pažs was studied to discuss the
W/C range required in CA paste for construction in this paper, which is shown in Fig.4.
1414
Table 3. Regression parameters to predict the viscosity of CA pastes.
Mix a φm R2 a φm R2
CSS-1 have lower φm at low A/C below 0.24 when compared to those with anionic emulsion,
but differs little when A/C is upper to 0.36.
Besides, CA pastes with cationic emulsion have greater W/C than those with anionic emul-
sion at the same A/C. The results are similar to the above analysis. It is proved again that
CA paste with different A/C cannot be designed with the same W/C for workability, and CA
paste with cationic emulsion requires more water than CA paste with anionic emulsion.
In actual construction, the workability sensitivity of CA mortar about W/C is also an
important indication for grouting. The differences between upper and lower W/C are shown
in Fig. 9. As can been seen that the differences of W/C for CA pastes with anionic emulsion
1416
Figure 9. Sensitivity of W/C for all CA pastes in construction.
reduce when A/C is below 0.36 compared to cement paste, and then rise a lot for A/C at 0.6.
For CA pastes with cationic emulsion, the differences keep growing with the increasing A/C.
In contrast of CA pastes with the two different asphalt emulsion, the W/C of CA pastes with
anionic emulsion on viscosity is more sensitive, especially when A/C is below 0.6, the differ-
ence of W/C is no more than 0.04, which means that a little change in W/C can bring about
a big difference about workability. Therefore, we should pay much attention to control W/C
for the construction of CA paste with anionic emulsion.
1417
5 CONCLUSIONS
The characteristics of W/C for construction of fresh CA paste was investigated through
studying viscosity of CA paste and flow time of CA mortar. The following conclusions can
be drawn from the results.
1. The required W/C of CA pastes changes with A/C and emulsion types. The required W/C
of CA pastes with anionic emulsion changes little at low A/C below 0.12, and then increases
stably with the increasing A/C. For CA pastes with cationic emulsion, the required W/C
keeps growing with A/C, and also is larger than CA pastes with anionic emulsion at the
same A/C.
2. The effect of W/C change in CA pastes with anionic emulsion on viscosity is more sensi-
tive than cement paste when A/C is below 0.36, while it is less sensitive for A/C at 0.6.
However, the W/C sensitivity on viscosity reduces with the increasing A/C for CA pastes
with cationic asphalt emulsion. The W/C sensitivity on the viscosity of CA pastes with
cationic emulsion is less than those with anionic emulsion. Adjusting the viscosity of CA
pastes with anionic emulsion should be careful to control W/C.
3. The W/C is low for high-strength CA mortar, and its change is very sensitive to the flow-
ability of high-strength CA mortar. However, the W/C is high for low-strength CA mor-
tar, and its change is not sensitive to flow-ability compared to high-strength CA mortar.
ACKNOWLEDGEMENTS
The first author acknowledges the support of fundamental research fund from Dalian Uni-
versity of Technology (DUT15RC(3)083) and the suggestion of Prof. Surendra P. Shah in
Northwestern University.
REFERENCES
Brown, S.F. & Needham D. 2000. A study of cement modified bitumen emulsion mixtures. Proc
AAPT 69.
Bocci, M., Grilli, A., Cardone, F., et al. 2011. A study on the mechanical behaviour of cement–bitumen
treated materials. Construction and Building Materials 25(2): 773–8.
Cheng, Y. & Zheng, Y. 2011. Research on the influencing factors on flowability of cement emulsified
asphalt mortar. Experiment and Research 3: 20–22. [Chinese]
Chinese Railway Specification. 2008a. Technical specification of CA mortar for CRTS I non–ballast
slab track. China: China Railway Publishing House. [Chinese]
Chinese Railway Specification. 2008b. Technical specification of CA mortar for CRTS II non–ballast
slab track. China: China Railway Publishing House. [Chinese]
Cremonesi, M., Ferrara, L., Frangi, A., et al. 2010. Simulation of the flow of fresh cement suspensions
by a Lagrangian finite element approach, Journal of Non-Newtonian Fluid Mechanics 165(23–24):
1555–63.
Ferrara, L., Cremonesi, M., Tregger, N., et al. 2012. On the identification of rheological prperties of
cement suspensions: Rheometry, Computational Fluid Dynamics Modeling and field test measure-
ments. Cement Concrete Research 42(8): 1134–46.
Hu, S.G., Wang, T., Wang, F.Z., et al. 2009. Adsorption behaviour between cement and asphalt emul-
sion in CA mortar. Advances in Cement Research 21: 11–14.
Hu, S., Zhang, Y. & Wang, F. 2012. Effect of temperature and pressure on the degradation of cement
asphalt mortar exposed to water. Construction and Building Materials 34: 570–574
Li, G., Zhao, Y., Pang, S., et al. 1998. Experimental study of cement-asphalt emulsion composite.
Cement and Concrete Research 28(5): 635–41.
Liu, D.M. 2000. Particle packing and rheological property of highly-concentrated ceramic suspensions:
Φm determination and viscosity prediction. Journal of materials science 35(21): 5503–5507.
Ministry of Communications of PR China. 2011. Standard test methods of bitumen and bituminous
mixtures for highway engineering. Beijing. [in Chinese]
1418
Oruc, S., Celik, F. & Akpinar M.V. 2007. Effect of cement on emulsified asphalt mixtures. Journal of
Materials Engineering and Performance16(5): 578–583.
Ouyang, J. & Tan, Y. 2015a. Rheology of fresh cement asphalt emulsion pastes. Construction and Build-
ing Materials 80: 236–243.
Ouyang, J., Tan, Y., Li, Y., et al. 2015b. Demulsification process of asphalt emulsion in fresh cement-
asphalt emulsion paste. Materials and Structures 48(12): 3875–3883.
Peng, J., Deng, D., Yuan, Q., et al. 2014. Study of the rheological behavior of fresh cement emulsified
asphalt paste. Construction and Building Materials 66: 348–355.
Saak, A.W., Jennings, H.M. & Shah, S.P. 2001. New Methodology for designing self-compacting con-
crete. ACI Materials Journal 98(6): 429–39.
Schmidt, R.J., Santucci, L.E. & Coyne LD. 1973. Performance characteristics of cement modified
asphalt emulsion mixes. Proc AAPT 42: 300–19.
Song, H., Do, J & Soh, Y. 2006. Feasibility study of asphalt-modified mortars using asphalt emulsion.
Construction and Building Materials 20(5): 332–337.
Tan, Y., Ouyang, J., Lv, J., et al. 2013. Effect of emulsifier on cement hydration in cement asphalt mor-
tar. Construction and Building Materials 47: 159–164.
Tan, Y., Ouyang, J. & Li, Y. 2014. Factors influencing rheological properties of fresh cement asphalt
emulsion paste. Construction and Building Materials 68: 611–617.
Tregger, N., Gregori, A., Ferrara, L. & Shah, S.P. 2012. Correlating dynamic segregation of self-consol-
idating concrete to the slump-flow test. Construction and Building Materials 28(1): 499–505.
Wang, F., Liu, Z., Wang, T. & Hu, S. 2008. A novel method to evaluate the setting process of cement and
asphalt emulsion in CA mortar. Materials and Structures 41(4): 643–7.
Wang, F., Liu, Z., Wang, T. & Hu, S. 2010. Temperature stability of compressive strength of cement
asphalt mortar. ACI Materials Journal 107(1):27–30.
Wang, Q., Yan, P., et al. 2011. Strength Mechanism of Cement-Asphalt Mortar. Journal of Materials in
Civil Engineering 23(9): 1353–1359.
Xu, J., Hong, J., Liu, J., et al. 2010. Research on factors influencing bleeding of CA mortar. Journal of
Southeast University 40(Sup(II)): 56–60. [Chinese]
Zhang, Y., Kong, X. & Hou, S. 2012. Study on the rheological properties of fresh cement asphalt paste.
Construction and Building Materials 27(1): 534–544.
Wang, F., Wang, T., Hu, S., et al. 2008. Rheological behavior of cement asphalt mortar. Engineering
Journal of Wuhan University 41(4): 69–72. [in Chinese]
1419
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Rigid pavements are predominantly subjected to flexural stresses due to traf-
fic and thermal loading throughout their design life. In conventional concrete pavements,
there is a considerable contribution of thermal stresses in total. This may not be true for
thinner short paneled rigid pavement and consequently economical concrete slab can be
constructed. In this paper, a comparative analysis has been presented for maximum verti-
cal deflections in short paneled rigid pavements due to traffic loading and combined traffic
with temperature loading, based on 3-Dimensional (3-D) Finite Element (FE) technique. It
is observed from the results that percentage increase in deflection due to consideration of
temperature loading showed an increasing trend with increase in panel size. Results also sug-
gest that the percentage increase in deflections due to consideration of temperature loading
increases with slab thickness. Percentage increase in maximum vertical deflection is increas-
ing with the value of effective subgrade modulus, when both loadings are considered.
1 INTRODUCTION
Bituminous and concrete pavements are two major type of pavements designed and con-
struted over the world. In recent years, concrete pavements are being adopted in many high
volume road projects, but still on the low volume roads concrete pavements are not preferred.
Concrete pavements are having many advantages over bituminous pavement in view of their
lesser maintenance requirements, smoother riding surface, better performance in adverse
moisture conditions and longer services lives. But on the other hand capital cost of concrete
pavements is much higher than bituminous pavements, which restricts its usage particularly
over low volume roads. This drawback leads to a new type of thinner concrete pavement with
shorter panel size similar to the thin white topping over bituminous pavements. This can be
used in the construction of concrete pavements for low traffic volume roads, parking lanes
and shoulder construction because of low flexural stresses caused by reduced panel sizes.
Such pavements may be termed as short-paneled concrete pavements.
It has been identified through various research projects related to deformation of con-
crete pavements that there are several critical parameters that individually or in combination
affect the vertical deflections in concrete pavements. Some important parameters include slab
dimensions, thickness of slab, expected axle loads, subgrade support, strength of concrete
and environmental conditions (Gandhi et al., 2015).
Slab dimensions and geometry is an important factor related to the performance of con-
crete pavements (AASHTO 1993, Smith et al. 1998, ARA 2007). Traditional Jointed Plain
Concrete Pavements (JPCP) with size of the order 4500 mm × 3500 mm have also produced
satisfactory service lives (Smith et al. 1998, Yu et al. 1998). However, with these slab sizes,
the thicknesses generally range from 15 to 36 cm. As slab sizes become larger, the vulner-
ability of the pavement system to temperature, moisture and built-in curling increases. With
the presence of large built-in curling, mechanistic-empirical designs with conventional slab
sizes can result in larger required thicknesses (ARA 2007, Hiller 2007). On reducing the
1421
panel sizes, curling stresses might be less as it depends on the horizontal dimensions of the
slab (Pandey, 2005). Hence, for heavy temperature differential and excessive built-in curling,
conventional JPCP may not be the most economical pavement system to construct.
Thickness of concrete slab is another important parameter that varies the deflection in con-
crete pavements. Undoubtedly, deflections in concrete pavements will reduce with the increase
in thickness of slab. However, quantitative analysis about percentage reduction in deflection
with the increase in slab thickness, particularly for short-paneled concrete pavement has not
been studied. Strength of foundation, symbolized with effective modulus of subgrade reaction
(k-value) is also a key factor to variation of deflections. To optimise the pavement design, there
is an obvious need to do a sensitivity analysis with considering all these parameters.
Conventional concrete pavements allows full standard axle in single panel, while in paneled
concrete pavement, panel-size of slab was shorten to allow only one wheel load per slab to reduce
the curling stresses (Covarrubias and Covarrubias, 2008). Temperature loading in addition to
traffic loading is also a matter of investigation for short-paneled concrete pavements. Chand
and Pandey (2013) simply neglected the thermal stresses in 3-D FEA on short-paneled concrete
pavements. Hence, it is imperative that the comparative study of these short-paneled concrete
pavements should be carried out for both traffic loading; and traffic with temperature loading.
Distribution of temperature differential across the slab is also important. Several researchers
have emphasized the importance of nonlinear temperature distribution (Thomplinson, 1940;
Choubane and Tia, 1995; and Harik et al., 1994). However, most of current analyses are still
limited to linear thermal distribution (Huang, 2004). Thus, it is crucial to find out the percentage
variation in stresses and deflections with both linear and nonlinear thermal gradient.
Limited literature is available for the short-paneled concrete pavements over granular or
concrete bases. Most literatures deal with the concrete pavements with short slabs laid as over-
lays over milled bituminous pavements, commonly termed as thin or ultra-thin white top-
pings. IRC:SP:76 (2008) provides tentative guidelines for design of white and ultra-thin white
topping. Jundhare et al. (2011) presented details of edge stresses and deflections in unbonded
conventional white topping overlay. Kumara et al. (2003) developed a 3-D finite element model
for stress analysis of ultra-thin white topping pavement under critical loading conditions. The
model was also used to perform a parametric analysis to determine the effects of various
parameters on the maximum stresses in ultra thin white topping under typical conditions of
Florida, USA. Li and Vandenbossche (2013) developed a 3-D finite-element model subjected
to environmental and wheel loads to verify the failure mechanism of thin and ultra-thin white
topping. Chand and Pandey (2013) performed a 3-D finite element analysis for short concrete
panels using ANSYS. The study gives an analytical evaluation of stresses in paneled concrete
pavements subjected to wheel loads. There is still ample gap in the knowledge about the values
of maximum flexural stresses and maximum vertical deflections, developed in the short panel
concrete slab due to wheel load. Therefore, further research is required for comprehensive
understanding of the behavior of these short-paneled concrete pavement systems.
2 METHODOLOGY
Maximum vertical deflections in concrete pavements have always been a major concern and
are used as a criterion for pavement design. In this paper, an effort has been made for eval-
uating the maximum vertical deflections in short-paneled concrete pavement using Finite
Element (FE) technique. Practically, concrete pavement comprises of several layers like
subgrade, sub-base, base course and concrete slab. But in present analysis, short-paneled
concrete pavement system is idealized as two layered structure. Two types of loading cases
were considered in the analysis:
• traffic loading, and
• combined traffic and temperature loading
Initially, a comprehensive 3-D FE model of concrete pavement is generated in ANSYS
14.0 (2010), using upgraded versions of elements used by previous researchers. Two-layered
1422
single-panel concrete pavement system is used for the analysis. Concrete slab was assumed
to be resting over dense-liquid (Winkler’s) foundation. A comparative parametric study was
performed using 3-D FEM considering afore-mentioned parameters and their responses on
maximum defections for wheel loading; and wheel with temperature loading. Results of both
the studies were plotted and comparative analysis was performed.
Figure 1. Two layered single panel short concrete pavement considered in the analysis.
Foundation
Parameters PCC Slab (Base + Subbase + Subgrade)
1423
roads, design load of 50 kN on dual wheel having a spacing of 310 mm centre-to-centre is
considered in present paper as recommended by IRC:SP:62 (2014). In both FE models, load-
ing is applied as uniform pressure obtained by dividing the wheel load with contact area. The
dual wheel loads are placed with the outer wheel tangential to the longitudinal edge of the
slab as suggested in IRC:58 (2011). Figure 3 gives the details of the geometrical arrangement
of the dual wheel loads on the concrete slab. In 3-D FE modelling of pavement system, wheel
load was applied on the surface of concrete slab in the form of uniform pressure. Figure 4 is
showing the pavement slab with uniform wheel load.
1424
Table 2. Temperature differential for different slab depth.
while mid-nodes values were obtained from Eq. (1) and Eq. (2) for linear and non-linear
temperature variation respectively. In 3-D FE modelling of pavement system temperature
differential was applied as a body load. Figure 9 is representing application of temperature
on concrete slab in 3-D FE model.
1
Tmid op + Tbotto
(Tttop m)
bottom (1)
2
1
Tmid Ttop − (Ttop Tbottom)
bottom (2)
3
3 DISCUSSION
Figure 7 represents the variation in maximum vertical stresses with panel-size for different
thickness and subgrade modulus. Three dimensional FEM was used for the current analysis.
1425
Figure 7. Variation in maximum vertical deflection with slab thickness for different panel-size and
subgrade modulus.
Values of deflections were compared for two cases. In first case pavement was subjected to only
wheel-loading while for second case temperature loading was also considered along with wheel-
loading. It can be clearly seen from the Figure 7, that deflection in short-paneled concrete slabs
are increased slightly with the consideration of thermal gradient. For a subgrade modulus of
0.15 MPa/mm and slab thickness 150 mm, deflections are increased by 0–50% with the con-
sideration of thermal gradient, for all panel sizes. This percentage increase in deflection due
to consideration of temperature loading along with traffic loading is increasing with increase
in panel size of the slab. Results suggested almost negligible increment in deflection for 0.5 m
panel size to 67% increment in deflection for 2.0 m panel size, for a slab thickness of 150 mm.
Almost similar trends of deflections has been observed for other thicknesses also.
Figure 8 exemplify the variation of maximum vertical deflection in short-paneled concrete
slabs with change in thickness of the slabs for both loading cases. Results suggested the
percentage increment in deflection due to consideration of temperature loading is increases
1426
Figure 8. Variation in maximum vertical deflection with slab thickness for different panel-size and
subgrade modulus.
with slab thickness. Vertical deflections are coming 0–70% higher and 0–115% higher for slab
thickness of 50 mm and 200 mm respectively, when temperature loading is also considered
along with wheel loading.
As observed from the Figure 7 and Figure 8, percentage increase in maximum vertical
deflection due to consideration of temperature loading along with traffic loading is increas-
ing with the value of effective modulus of subgrade reaction. Increase in vertical deflec-
tion due to inclusion of thermal gradient for a subgrade modulus of 0.08 MPa/mm and
0.30 MPa/mm is 5–45% and 8–115% respectively.
4 CONCLUSIONS
In the present research wok, sensitivity analysis has been performed for various parameters
including slab panel size, slab thickness, effective subgrade modulus and thermal gradient.
Effects due to both traffic and temperature loading were considered. The major conclusions
drawn from the present work are summarized in this section.
1427
1. Percentage increase in deflection due to consideration of temperature loading along with
traffic loading showed an increasing trend with increase in panel size of the slab.
2. Results suggest that the percentage increase deflections due to consideration of tempera-
ture loading increases with slab thickness. Vertical deflections were found 0–70% higher
and 0–115% higher for slab thickness of 50 mm and 200 mm respectively.
3. When thermal loading was considered along with traffic loading, it was observed that for
slab thickness 150 mm the stresses for lower k-values up to 0.15 MPa/mm increase steeply
while it tends to attain plateau condition for k-values more than 0.15 MPa/mm for panel
sizes exceeding 1500 mm. Also, for slab thickness 200 mm, it was observed that stresses
for lower k-values up to 0.15 MPa/mm attained plateau condition while for k-values more
than 0.15 MPa/mm, it increases steeply, for panel sizes exceeding 1500 mm.
4. Percentage increase in maximum vertical deflection was increasing with the value of effec-
tive modulus of subgrade reaction, when both loadings were considered.
Along with wheel and temperature loading, analysis can also be done with taken care of the
self-weight and frictional force to better understand the behaviour of these short concrete
pavements. Various loading position, axle loads, temperature gradient can be taken for the
analysis to find out the critical combinations. Dynamic loading can also be included for the
analysis for more realistic approach. In order to optimise the design thickness, a comprehen-
sive study on the reduction of stresses due to load transfer through aggregate interlocking
can be done. To estimate the design life and durability of these short-paneled pavement sys-
tems, fatigue and damage analysis due to repetition of wheel loads and temperature cycles
can be carried out for better understanding.
REFERENCES
American Association of State Highway and Transportation Officials, USA, 1993. AASHTO Guide for
Design of Pavement Structures.
APDL, A.M., (2010). Mechanical applications Theory reference. ANSYS Release, No. 13.
ARA, Inc, 2007. Interim mechanistic–empirical pavement design guide manual of practice. Final Draft.
National Cooperative Highway Research Program Project 1–37A.
Chand, M.V., and Pandey, B.B. (2013). Analytical design of short panelled concrete pavements. Journal
of the Indian Road Congress, Oct-Dec, 322–328.
Choubane, B., & Tia, M. (1995). Analysis and verification of thermal-gradient effects on concrete pave-
ment. Journal of Transportation Engineering, 121(1), 75–81.
Covarrubias, J.P.T. and Covarrubias, J.P.V., 2008. TCP design for thin concrete pavements. In: 9th Inter-
national conference on concrete pavements, August 17–21, 2008, San Francisco.
Gandhi, T., Goyal, R., Gupta, A. and Kumar, B. (2015). “Parametric Study on Short Panelled Rigid
Pavements using Finite Element Method”, Proc. of 3rd International Conference of Transportation
Research Group, CTRG-2015, Kolkata, India, Paper Id – 496.
Harik, I.E., Jianping, P., Southgate, H., & Allen, D. (1994). Temperature effects on concrete pavements.
Journal of transportation engineering, 120(1), 127–143.
Hiller, J.E., 2007. Development of mechanistic–empirical principles for jointed plain concrete pavement
fatigue design. Thesis (PhD). University of Illinois Urbana–Champaign.
Huang, Y.H., (2004). Pavement Analysis and Design, Pearson Prentice Hall Inc.
IRC publication, No. 58 (2011).Guidelines for the design of plain-jointed concrete pavements for high-
ways. The Indian Roads Congress, New Delhi.
IRC special publication, No. 62 (2014). Guidelines for Design and Construction of Cement Concrete
Pavements for low volume Roads. The Indian Roads Congress, New Delhi.
IRC special publication, No. 76 (2008). Tentative Guidelines for Conventional, Thin and Ultra-Thin
Whitetopping. The Indian Roads Congress, New Delhi.
Jundhare, D.R., Khare, K.C., and Jain, R.K., (2011). Edge stresses and deflections of unbonded con-
ventional whitetopping overlay. Indian Concrete Journal, Vol. 85(3), pp.35–44.
1428
Kumara, W., Tia, M., Wu, C.L., and Choubane, B., (2003). Evaluation of applicability of ultrathin
whitetopping in Florida,. Transportation Research Record: Journal of the Transportation Research
Board, Vol. 1823(1), pp.39–46.
Li, Z., and Vandenbossche, J., (2013). Redefining the Failure Mode for Thin and Ultrathin Whitetop-
ping with a 1.8-x1. 8-m (6-x 6-ft) Joint Spacing. 92nd Transportation Research Board Annul Meeting.
Mahboub, K.C., Liu, Y., and Allen, D.L. (2004). Evaluation of temperature responses in concrete pave-
ment. Journal of transportation engineering, 130(3), 395–401.
Maitra, S.R., Reddy, K.S., and Ramachandra, L.S. (2013). Estimation of Critical Stress in Jointed Con-
crete Pavement. Procedia-Social and Behavioral Sciences, 104, 208–217.
Masad, E., Taha, R., and Muhunthan, B. (1996). Finite-element analysis of temperature effects on
plain-jointed concrete pavements. Journal of Transportation Engineering, 122(5), 388–398.
Pandey, B.B. (2005). Warping Streses in Concrete Pavements- A Re-examination. HRB No. 73, Indian
Roads Congress, pp. 49–58.
Smith, K.D., et al., 1998. Performance of concrete pavements. Volume II–evaluation of in-service con-
crete pavements. FHWA-RD-95–110, Washington, DC.
Thomlinson, J., Temperature variations and consequent stresses produced by daily and seasonal tem-
perature cycles in concrete slabs. Concr. Constr. Eng., 1940, 36(6), 298–307.
Yu, T., et al., 1998. Performance of concrete pavements. Volume III – Improving concrete pavement
performance. FHWARD- 95–111, Washington, DC, 302 pp.
1429
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Josef Stryk
CDV—Transport Research Centre, Brno, Czech Republic
Zdenek Nevosad
Specialist of Building Materials and Constructions, Brno, Czech Republic
Miroslava Gregerova
Department of Geological Sciences, Faculty of Science, Masaryk University, Brno, Czech Republic
ABSTRACT: Over the recent years, hairline cracks in cement concrete pavements appear
often very early, within a few months or a couple of years after the concrete is laid. Visible
cracks arise primarily around joints at the edges and in the corners of concrete slabs; from
there, they gradually spread over the whole slab. This fault does not only occur in the Czech
Republic, but currently also in other European countries, the USA, etc. As the cracks occur
in pavements with heavy traffic as well as without traffic, the reason for the occurrence of
cracks seems to be grounded in technology. The cause is the expansive reaction in concrete.
Despite considerable effort of specialists on international level, no generally recognized meth-
odology leading to successful diagnostics has been discovered. This paper presents one of
new concepts of geologic diagnostic research of concrete and its components using optical
microscopy, scanning electron microscopy and chemical microanalyses.
1 INTRODUCTION
The increasing occurrence of hairline cracks in cement concrete pavements in the recent
years is attributed to material properties of produced concrete. They are probably caused
by long-term expansive reactions generally called Alkali-Aggregate Reaction (AAR). They
include expansive reactions which occur in concrete structures. AAR is further divided into
Alkali-Silicate Reactions (ASR) and Alkali-Carbonate Reactions (ACR). The Alkali-silicate
Reactions (ASR) can be either ASR—Alkali-silica, with reactive SiO2; or ASR—Alkali-
silicate. Together they create a set of physical and chemical reactions between the particles of
reactive SiO2 and alkaline solutions in concrete.
The road concretes, which are installed in the outside environment, are intensively affected
by climatic and other factors of each specific site. The factors include temperature, mois-
ture, temperature changes, moisture changes, pollution of water and air, migrating chemical
defrosting agents, aggressive substances entering concrete with water and moisture, etc.
The joint effect of hydration reactions and effects of migrating substances from the
environment and non-constant climate leads to occurrence of different forms in concrete.
Those are forms building up from alkali-silica gel. The forms, which increase their volume
by absorbing a large amount of water, induce high pressures that are able to tear concrete
structure (Figure 1). High tensile, bending and friction forces occur at these spots which
are sufficient to disrupt the reacting clast as well as the adjacent concrete binder. In case
alkali solutions from the environment enter the concrete, the process of ASR expansion in
1431
Figure 1. Example of a core of examined cement concrete pavement.
Figure 2. Thin cracks in motorway pavement, risky area of corners has a dense net of easily visible
cracks.
1432
Figure 3. Airport slab cross section, deterioration of concrete pavement above subbase, horizontal
crack in concrete.
1433
Figure 5. Glassy view of expansive gels. Their gradual drying is accompanied by an occurrence of a
dense network of cracks. Photo A. Frybort.
In the first stage, the formed alkali-silica gel fills the free pores in cement concrete. The
amount of gel depends on the amount and type of SiO2 and on the sources of alkalis. After
filling the pores, the reactions between aggregates and alkali hydroxides can continue along
the intergranulars (space with a certain inconsistence between crystals, i.e. intercrystalline
space) and in weakened, discontinuous zones among aggregate grains. Reaction rims around
the aggregates may gradually occur and white products on cracks appear (drying gels, see
Figure 5). The cracks are often filled with transparent silica gel in layers (Figure 6). Alkali
silicate gels may, particularly depending on moisture, get activated in months or even in sev-
eral years (16 years in a known case).
The presence of gels in hydrating concrete is necessary but not every gel necessarily means
an indication of a destructive effect of ASR. What is important is the availability of a suf-
ficient amount of calcium for the formation of calcium silicate hydrates, which tends to be
limited. Most of the calcium is bound in stable phases of hydration intermediate products
and hydration products.
A long-term surplus of water in concrete causes expansion of alkali silicate gel, which
contributes to mechanical damage of concrete. Over time the migration of water through the
structure leads to binding of calcium into these gels, therefore the resultant product is more
and more similar to calcium silicate hydrate. Unfortunately, it is too late for the structure
damaged by cracks. The major damage in concrete had been done.
The solubility of amorphous and other reactive forms of SiO2 generally increases with
high pH of solutions. The formed alkali silicate gel is strongly hygroscopic, absorbs water and
thus increases its volume.
1434
Figure 6. Microstructure of concrete: Gradual spreading of younger gels along cracks in concrete.
Photo A. Frybort.
Mechanisms of the reaction were described by a number of authors, e.g. French (1986),
Broekmans (1996, 1999), Hou et al. (2004). Jensen (1993) presented an extensive overview of
literature at the end of the last century.
If aggregates in concrete contain limestone or other calcareous rocks, the following reac-
tions were observed:
− ACR of dolomitic limestones is a result of dedolomitization;
− ACR of limestones without dolomites leads to an occurrence of reaction rims;
− ASR are in progress in other limestones if they contain reactive SiO2.
ACR is a rare and usually wrongly understood form of AAR. The mechanism of reactions
is completely different from ASR even though the final result is very similar. It is generally
assumed that alkali solutions react with dolomite to form brucite, calcite, and sodium car-
bonate. Even though dedolomitization occurs, its damaging effect during expanding have not
been proven. Katayama (2004) proved it microscopically. Alkali carbonate reactions do not
produce gel.
The experience in practice showed that cracks occurring in young concrete by shrinking
and drying may speed up the expansion in concrete structure due to ASR. This is probably
caused by capillary phenomena in cracks, which speed up the migration of water in concrete
and swelling of reaction products.
The mentioned overview makes it obvious that this issue does not only concern the cement
concrete pavements, but also all exterior concrete constructions, including those that are part
of roads (bridges, culverts, concrete crash barriers, etc.). Detecting and considering real risks
1435
of occurrence of the mentioned degradation processes induced by expansive reactions may
lead to conclusions that can help to increase life span of concrete constructions.
The original faults – micro cracks in concrete structure are not visible by human eye. Later
they gradually extend and interconnect due to continued expansive reactions. They usually
spread from the wettest spots, i.e. edges (joints) and corners to the centres of slabs (struc-
tures) and gradually spread until they can be diagnosed by human eye.
In the final stages of this type of degradation, fragments or individual aggregate grains chip
off. In case any of these cracks gain any active function, such as a joint (contracting or dilatation),
even a partial one, a risk of the so-called edge chipping may occur. In case expansive reactions
occur in the most exposed road engineering structures, the first visible hairline cracks are known
to have appeared as early as within several months up to ten months of the age of the concrete.
The presented summary points out to rather warning prognoses:
− We usually have no idea about the existence of initial stages of expansive reactions due to
the fact the micro cracks cannot be recognized by human eye.
− Experience in practice showed that cracks occurring in young concrete (due to shrinking)
may speed up excessive expansion due to ASR.
− The existing diagnoses of prognoses for expansive reactions in concrete are insufficient, it
is not technically finalized.
− The occurrence of micro cracks cannot be currently eliminated by any known method; all
known types of reinforcement/fibres fail to deal with such narrow cracks.
− If there are conditions for more types of expansive reactions on a given spot in concrete
structure, they all occur concurrently in the defined relative representation and influence
the disintegration of concrete together.
− Once the expansive reaction begins, they cannot be usually stopped, just slowed down.
The issue of expansive reactions in concrete is one of the basic issues dealt with by many
countries at national and higher levels. It is still without general conformity and there are
several opinion groups. Each region has different opinions as well as raw materials, environ-
ment, conditions and traditions.
It is believed that no known test, including its unambiguous evaluation, for expansive
reactions has been sufficiently satisfactory. However, renowned specialists with long standing
experience, who have their own, usually non-standard methodologies, are often respected.
They are usually successful in their own areas; the extrapolations to other areas have not been
apparently very successful so far.
The common method of evaluation is based on petrographical examination of aggregates
and dilatometric measurements on prepared samples (mortar or concrete bars and prisms).
In Europe the latest findings concerning AAR test methods were published in State-of-the-art
report of the RILEM technical committee 219-ACS in 2016. It USA the different methods
are covered by stThe different approach is used for testing a potential for ASR and ACR.
According to RILEM recommendation the first step is petrographical examination (AAR-
1.1) and classification of aggregates into 3 classes (from I - very unlikely to be alkali-reactive, to
III - very likely to be alkali-reactive) with indication whether aggregates are mainly siliceous, or
carbonates. Next step is a rapid 16 day long screening test on mortar bars (AAR-2), which is in
case of carbonate aggregate combined with a test on concrete microbars (AAR-5). In case that
the tests indicate reactive or potentially reactive aggregates the further step is expansion testing on
concrete prisms, AAR-3 (38°C test method) requires a lengthy period, up to 12 months or more,
for reliable results to be obtained and AAR-4.1 (60°C test method) requires up to 4 months.
In USA are used the following tests. Determining potential for ACR include a rock cylin-
der expansion test (ASTM C586) and an expansion test on concrete prisms (ASTM C1105).
In case of determining potential for ASR there is more methods. Petrografic evaluation of
aggregates according to ASTM C295 is recommended but must be combined with other
methods. The expansion test using concrete prisms according ASTM C1293 is basic test
1436
method but it requires one year period. More common is an accelerated mortar bar expan-
sion test according to ASTM C1260 which provides the result in 16 days. Other option is to
use a long term mortar bar expansion test according to ASTM C227.
2.2 Experience with our microscopic diagnostics for expansive reactions in concrete
We have been gaining experience since 1998 with microscopic diagnostics of concrete (cores,
chippings) from airport and motorway pavements which were suspected for faults caused by
expansive reactions. When using optical microscopy, electron microscopy and microanalysis,
X-rays and other methodologies, at first only available in Masaryk University in Brno, we
combined experience and testing from geology and civil engineering.
It is conceptually completely different way of diagnostics and absolutely non-transform-
able evaluation of results in comparison to national regulations and standards. The above
mentioned methods were used for the diagnostics of concretes, or aggregates and cement
stone more specifically, collected by sampling from the existing structures.
Measurement and evaluation of concrete in the Czech Republic has so far led to the fol-
lowing findings:
− The expansive reactions of concrete exist and predominantly occur in the capillaries of the
cement stone.
− The cracks by the expansive reactions in the concrete structure affect all components, i.e.
cement stone, they are in the contact of aggregates and cement stone, and less frequently
in aggregates.
− The petrographic analyses provide a unique source of information on potential expansive
reactions in concrete. Regarding the concrete samples coming from Czech regions, the risk
of sulphate expansion is the most common both separately and in combination with ASR
gels (Figure 7).
− When concretes from the same motorway segments were repeatedly tested after time peri-
ods, different relative representation of types of expansive reactions and concrete deg-
radation were found over time. Regarding cement concrete pavements, younger concrete
showed dominant sulphate expansion; in approx. 7 years (with more intensive concrete
degradation) the relative representation of sulphate expansion and ASR was comparable.
− Regarding concrete with detected distinct alkali-silicate expansion (ASR), no aggregates
themselves contained the sufficient amount of alkalis for the detected high concentrations
of alkalis.
− Additional filling of some capillaries in the cement stone was found with older concrete.
1437
Figure 7. Microstructure of concrete. Occurrence of ettringite in pores. Once filled, an expansion
occurs in concrete together with microscopic cracks. Deficit of H2O leads to occurrence of contraction
cracks inside pores and creates space for reactions to continue. Photo A. Frybort.
3 CONCLUSIONS
Expansive reactions in concrete are much feared and an extremely quick form of a self-
destruction of cement concrete. Currently used testing and evaluation methods are not suf-
ficient to provide us with satisfactory results. Therefore, this issue has attracted considerable
attention for a long time.
In the paper we present one of new conceptions. The expansive reactions in concrete
structure are particularly determined on the basis of fields of specialized optical microscopy.
Regarding the conception, the solution is completely different from and non-transformable
towards the existing national regulations and standards. We believe that the special opti-
cal microscopy, including electron microscopy and microanalysis, have discovered new prov-
able findings. The findings come from testing of laid cement concrete pavements in airports
and motorways with concrete containing natural aggregates from quarries in the Czech
Republic.
We assume that the designed combination of methods is suitable for evaluating the risk
factors which initiate expansive reactions in cement concretes.
Brief summary of the existing experience:
− Special optical microscopy is a diagnostic method suitable and used for determining all
types of expansive reactions in concrete. It requires available, but expensive, equipment
and highly specialized training, or individual education of personnel, respectively.
1438
− Special optical microscopy is a diagnostics method suitable and used for bores and frag-
ments of concrete as well as for evaluation of quality and reactiveness of aggregates during
certification and control tests.
− So far, all tested concretes in the Czech Republic have contained aggregates susceptible to
sulphate, or calcareous (calcareous magnesium), expansion.
− It was found that the quantitative representation of individual types of expansive reactions
may considerably change with growing age of concrete.
− However, not in all cases it is possible to clarify the occurrence of increased contents of
alkalis necessary for an formation of ASR in the pores of cement stone. When Na+ is
increased, antropogenesis is most commonly considered. The presence of high concentra-
tions of K+ has not been clarified. It can be related to the presence of clays and clay rocks
or alkali feldspars. Their source may originate in the external environment, particularly
from the soils under the pavement.
− When using Portland cements CEM I, the occurrence of alkalis from cement is highly
improbable.
− A combination (or connection) of more types of expansive reactions was observed with all
tested concretes where any expansive reactions were found.
We assume that on the basis of the performed analyses of damaged concretes, occurrence
of expansive reactions in new cement concrete structures is more probable. These, even higher,
risks concern the complete as well as partial reuse of recycled aggregates from demolished
cement concrete pavements in new cement bound road construction layers. In our opinion, it
is necessary to continue to develop new forms of diagnostics which are able to detect expan-
sive reactions in time, in order to adopt timely and effective measures for their prevention.
ACKNOWLEDGEMENT
This paper was prepared within solving of a project of the Technology Agency of the Czech
Republic No. TE01020168: Centre for Effective and Sustainable Transport Infrastructure
(CESTI) with use of devices acquired as a part of project of Operational Programme Research
and Development for Innovation No. CZ.1.05/2.1.00/03.0064: Transport R&D Centre.
REFERENCES
Ahlstrom, G. 2008. FHWA Alkali-Silica Reactivity Development and Deployment Program, HPC
Bridge Wiews, Issue 51.
Broekmans, M.A.T.M. 1996. Pilot study three ASR-attacked viaducts in route A59 (in Dutch). Nebest
BV Open Report C1264 (24), 126 p.
Broekmans, M.A.T.M. 1999. Classification of the alkali-silica reaction in geochemical terms of silica
dissolution. Proceedings of the 7th Euroseminar on Microscopy Applied on Building Materials, Delft,
Netherlands: 155–170.
Broekmans, M.A.T.M. 2004. Structural properties of quartz and their potential role for ASR. Materials
Characterization, vol. 53: 129–140.
French J. 1986. A review of some reactive aggregates from the UK with reference to the mechanism of
reaction and deterioration. Concrete alkali-aggregate reactions: 226–230. Ottawa: Grattan-Bellew.
Hou X, Struble L.J. and Kirkpatrick R.J. 2004. Formation of ASR gel and the roles of C-S-H and port-
landite. Cement and Concrete Research, vol. 34: 1683–1696.
Jensen V. 1993. Alkali aggregate reaction in southern Norway. Doctor Technice- thesis Technical Uni-
versity of Trondheim, 262 p.
Katayama T. 2004. How to identify carbonate rock reactions in concrete. Materials Characterization,
vol. 53 (2–4): 85–104.
RILEM State-of-the-art report - Recommendations for the Prevention of Damage by Alkali- Aggregate
Reactions in New Concrete Structures, Technical Committee 219-ACS, 2016.
Stutzman, P.E. 1999. Deterioration of Iowa Highway Concrete Pavements, Part A – Petrographic Study,
NISTIR 6399, Gaithesburg, USA, 73 p.
1439
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Yi Xu
AECOM Infrastructure and Environment Ltd., Ashford, UK
Paul Phillips
Aggregate Industries Ltd., Ashbourne, Derbyshire, UK
1 INTRODUCTION
The development and usage of Steel Fibre Reinforced, Roller Compacted, Polymer Modified
Concrete (SFR-RC-PMC) material as an overlay to damaged pavements, can demonstrate
superior properties and can even be more economical and effective as it can speed-up cur-
rent slow construction processes and reduce long traffic disruptions. The new overlay can be
thinner, as well as providing adequate toughness, crack control, high flexural, shear and bond
strengths and good resistance to fatigue. This is Part I of a three-part research project on
Optimum Design for Sustainable, ‘Green’ Concrete Overlays. It emphasizes on the develop-
ment of a special mix design and the assessment and control of flexure, a common mode of
failure in pavements. Part II examines shear failure and Part III focusses on delamination.
The method outlined here introduces an innovative approach in determining the optimal
water content in RC concrete mixes used explicitly for pavement overlays. Two types of
mixes were developed in an effort to achieve good bond with existing concrete, branded by
their polymer content: SBR (Styrene Butadiene Rubber) and SBR-PVA (Polyvinyl Alcohol)
(hybrid) mixes.
1441
2.1 Constituents, properties, workability, design-criteria
The focal point of this mix design is the determination of the optimal water content. First, in
order to simulate conventional concrete overlays on old concrete pavements in the laboratory,
Ordinary Portland Cement Concrete (OPCC) composite cylinders and blocks were studied.
The OPCC mix proportion in Table 1 was similar to that used in conventional bonded
concrete overlays in successful, real pavement sites (Mokarem et al., 2007). Their tensile bond
strength was reported equal to 1.65 MPa compared to our splitting tensile bond strength of
2.17 MPa. Thus, the OPCC-to-OPCC bond strengths obtained, were selected to be the lower
strength boundary for SFR-RC-PMC.
The particular suitability of a RC-BCO (Bonded Concrete Overlay) mix was defined in
terms of paver ‘‘placeability’’ (workability to suit placing by asphalt pavers) and roller ‘‘com-
pactability’’ (workability suitable for roller compaction).
In general, mixes with appropriate water content for roller compacted, bonded con-
crete overlays should satisfy the following criteria simultaneously: (a) The mix should not
become lumpy during mixing and placing; rather, it should behave like a granular mate-
rial for paver placeability; (b) the water content should not be higher than that determined
by the S-P method for roller compactability; (c) the direct shear bond and splitting tensile
bond strengths of PMC-to-OPCC should not be less than that of the OPCC-to-OPCC bond
strength (4.09 MPa and 2.17 MPa, respectively); (d) compacted specimens with optimal water
content should have no voids.
OPCC-to-OPCC bond
Quant. for 1 m3 concrete (kg) strength
Flex Direct Split.
Cube tensile shear tensile Dir. shear Split. tensile
strength strength strength strength bond stre. bond stre.
Cement CA FA Water (MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
1442
Figure 1. Compacted samples (M9 and M11) by the M-L method reaching highest splitting tensile
strength and maximum dry density at the same water content.
2P
σc = (1)
πA
where: σ c is the splitting tensile strength (MPa); P is the maximum load (N); A is the area
undergoing shear (mm2).
When high bond strength was attained, nearly 50% of the failed sectional area in the PMC-
to-OPCC specimens, made of M5(M-L) and M7(M-L), was covered with bonded material.
In contrast, when the bond strength was low, the interface did not contain traces of the
bonded material. Several mixes were tested. The results are plotted in Figure 1.
Flexural strengths for conventional Steel Fibre Reinforced Concrete (SFRC) have been inves-
tigated since the early 1980s. Recently, Neocleous et al. (2011) investigated the flexural per-
formance of steel fibre-reinforced RC concrete for pavements with steel fibres recovered from
used tyres, whereas the mix was conventional RC concrete.
The mechanical properties of SFR-RC-PMCs have not been investigated to date. Steel
fibres in these types of mixes may exhibit a different behaviour to those in conventional
SFRCs as the former contains much less cement paste than the latter, plus that roller compac-
tion may result in deformation of steel fibres. Furthermore, the flexural performance of PVA
(Polyvinyl Alcohol) modified concrete has rarely been investigated.
This section aims to reveal the flexural performance, specifically the equivalent flexural
strengths of SFR-RC-PMC for overlay pavement design and the efficiency of fibres in RC
concrete. The mix proportion, the ingredient materials and the preparation of the specimens
were detailed in an article by Lin et al. (2013).
3.1 Flexural strength of PMC beams under 4PB and 3PB tests
Cube strengths obtained from tested saw-cut beams were of the order of 80 MPa to 105 MPa.
Figure 2 displays the flexural Strength ∨ CMOD relationships, while the interfacial fracture
toughness and splitting tensile bond strength of composite specimens are shown in Table 2.
Details for testing interfacial fracture toughness can be found in Lin and Karadelis (2015).
It was seen that:
a. Compared to conventional SFRC, SFR-RC-PMC exhibited very high flexural strengths,
which are desired for worn concrete pavement rehabilitation;
1443
Figure 2. Flexural strengths of six 20 mm-notched PMC beams under 3PB tests.
b. compared to the strengths measured under 4PB for the same mix, the obtained strengths
under 3PB were remarkably higher.
However, flexural strengths are not suitable to be used directly for overlay pavement design.
The design method for SFRC pavements proposed by Altoubat et al. (2008), requires the
flexural strengths to be converted into equivalent flexural strengths.
(Mix ID: SBRPMC1%-35 = Styrene Butadiene Rubber, Polymer Modified Concrete with
1% (vol) fibres in the mix and 35 mm length of fibre.
δ1 s1 ( 2 − )
= 0
(2)
δ2 s2 h1
(Parameters: δ1, δ2, s1, s2, h1, h2 and a0 . are detailed in the above reference).
After substituting for the known parameters, δ 2 δ1. In this case, the deflection limit,
δ 2 ., is 5 mm and its equivalent strength is fe,5. The equivalent flexural strengths, fe,5, are listed
in Table 3.
Also, Figure 2 shows that nearly all the flexural strength-CMOD curves for all mixes are
basically parallel to each other, except for mix PVAPMC1.5%-35. This indicathat these mixes
have the same scale factor for the equivalent strength conversion. In order to use the SFRC
pavement design method proposed by Altoubat et al. (2008), the fe,5 has to be converted twice to
obtain the standard equivalent flexural strength, fe,3. Table 3 above demonstrates the process.
In effect, first the flexural strength has to be converted from the 3PB to 4PB. Then, it has
to be converted from a 4PB test with the beam of 100 mm height to that of a standard beam
1444
Table 3. Standard equivalent flexural strengths, f3,e.
fP = max flexural strength, fe,5 = equivalent flexural strength, fe,3 = standard equivalent flexural strength.
Table 4. Standard equivalent flexural strength determined using experimental results of beams with
different notch length, a0, and beam depth, h.
of 150 mm height. Both conversion factors are attributed to the size effect (Karadelis and
Lin, 2015).
It is seen from Table 3 that mix PVAPMC1.5%-35 developed the lowest standard equivalent
flexural strength, fe,3, although it exhibited very high maximum flexural strength, fp. The stand-
ard equivalent flexural strength, fe,3 can be used for SFR-RC-PMC overlay pavement design.
Also, it is seen from Table 4 that experimental fe,3 results of beams with different notch
lengths and depths are almost identical. This confirms the method for calculating the stand-
ard equivalent flexural strength, fe,3, for overlay pavement design.
fp fe,3 GF
Mix ID (MPa) (MPa) (J/m2) It
Table 7. Fibre bridging law for Stage-II under 3BP [σ (MPa), w (mm)] (Lin and Karadelis, 2015).
Figure 3. (a) Fibre tensile stress after a hinge formation beneath the point load (a0 = notch depth).
(b) Plots of fibre bridging laws, Stage II, plotted from Table 7.
a given face opening displacement (Figure 3b) indicating that the superior properties of the
former are attributed to its lower water/cement ratio. In addition, the plot of fibre bridging
law of mix SBRPMC1.5%-50 lays above the plot of SBRPMC1.5%-35 at all crack face open-
ing displacements, implying that the former provided higher fibre traction.
The fibre bridging law can serve as an index to evaluate the fibre efficiency for the selection
of ingredients during the mix design process in practical (site) applications.
The relationship between the tensile stress developing in the fibres and the crack opening dis-
placement should fully characterize the contribution made by the fibre-matrix interaction.
1446
Three-point bending tests have been favourable to avoid multiple-cracking occurrence. The
simulation of load ∨ CMOD of SFRC beam in flexure can be classified into two categories:
one is based on the mechanics of materials, the other on fracture mechanics. Whatever the
choice of the method, both resort to the same relationship of fibre tensile stress and crack
opening displacement, usually named as fibre bridging law, or σ (w ) w law for convenience
(Lindhagen et al., 2000; Zhang and Li, 2004). Several shapes of σ (w ) w law have been pro-
posed based on the mechanics of materials method to simulate flexural performance (Foote
et al., 1986; Armelin and Banthia, 1997) but none of them depicted rigorously the actual path
of the curve. In an effort to portray the σ(w)-w law closely, the two distinct stages were con-
sidered and are outlined below. To the authors’ best knowledge no researcher has established
the σ(w)-w laws by employing fracture mechanics and mechanics of materials yet.
∂K IF
a
2
E ∫0
CMOD = K Ia da (3)
∂F
CMOD = CMOD
Da − CMODb . (4)
CMODa and CMODb are the crack mouth opening displacement induced by the applied
load and fibre tensile stress, respectively; and evaluated by eqn.5 and eqn.6, below (Lin, 2014)
4σ a ⎛ a ⎞
Da =
CMOD V1 ⎜ ⎟ . (5)
E ⎝ h⎠
⎡ ⎤
⎢ a G ⎛ x , a′ ⎞ G ⎛ 0 , a′ ⎞ ⎥
⎝ a ′ h ⎠ ⎝ a ′ h ⎠ σ (w ( x )) ⎥
a
8 ⎢
π E ∫0 a∫0 ⎛ a ′ ⎞ 3
Db =
CMOD dx ⎥ da ′ (6)
⎛ x⎞
2 a′
⎢ 1− 1− ⎥
⎢⎣ ⎝ h⎠ ⎝ a′ ⎠ ⎥⎦
Mf B ∫ I (w (x ))(h x ) dx
d +B ∫ σ (w (x ))(h − x ) dx
III d (7)
h x0 a0
1447
Figure 4. (a): Tensile stress of fibres at cracked section of SFRC beam. (b): Experimental load-CMOD
curves of two beams of mix SBRPMC1.5%-35 under 3PB.
Figure 6. Experimental load-CMOD curves of under 3PB test two beams of mix SBRPMC1.5%-35
under 3PB.
The fracture mechanics method was used until a hinge was formed just under the point
load. At this point the crack face opening was equal to 0.958 mm (w0 = 0.958 mm). After that
the mechanics of materials theory was employed.
Figure 7(a) shows the established fibre bridging law for a typical SBRPMC1.5%-35-L80
beam. Figure 7(b) enlarges and details the σ(w)-w relationship for Stage-I, where fracture
mechanics theory is used.
The above procedure was repeated for Con.SBRPMC1.5%-35-L80 and SBRPMC1.5%-50
beams. The general tendency of the fibre bridging law at Stage-I includes an ascending—
descending path very similar to that established experimentally by Zhang and Stang (1998).
After the hinge formation, the tendency is descending, to failure. The stress profile of the
σ(w)-w curve of mix SBRPMC1.5%-50 was higher than that of mix SBRPMC1.5%-35, indi-
cating that long fibres provide higher resistance to crack opening than short ones.
As a final check, three groups of mix SBRPMC1.5%-35, notched beams with different
ligaments, L, of 80, 60 and 40 mm were tested under 3PB to explore the size effect on the
flexural strength. The results obtained showed that the bridging law is affected by specimen
size.
1448
Figure 7. Calculated tensile stress σ(w) for a given crack face opening, w, and ‘best-fit’ for beams of
mix SBRPMC1.5%-35-L80, under 3PB. (a): Fracture Mechanics and Mechanics of Materials regions.
(b): Fracture Mechanics enlarged.
The interface of the overlay pavement system experiences both tension and shear under the
action of vehicular and environmental loads in normal service life. The overlay being fully
bonded with the existing concrete pavement is the key to success. A guiding FE analysis
revealed that the interface very near a crack of a composite beam under 4PB was in tension. In
contrast, the interface was in compression for a 3PB. This indicated that the 4PB test was suit-
able to investigate the interfacial bond of composite beams. The maximum bending moment,
M(OPCC+PMC), carried by the entire composite beam before cracking is determined by letting the
stress at the bottom of the OPCC base be equal to its maximum flexural strength, fOPCC.
( fOPCC fPMC ) ( E I + E I )
M (OPCC + PMC ) = (8)
E2 y2
After the OPCC beam has cracked, the maximum moment (load carrying capacity, MPMC),
is carried by the PMC overlay and is determined by letting the stress at the bottom of the
PMC overlay be equal to its maximum flexural strength, fPMC. Hence:
fPMC Bhh12
M (PMC ) = (9)
6
The first peak load, P1, can be determined via M(OPCC+PMC), with a given flexural OPCC
strength, fOPCC and a maximum SFR-PMC flexural overlay strength, fPMC, using eqn.8. After
failure of the OPCC base, only the overlay, SFR-PMC, carries the load, thus the second peak
load, P2, is evaluated through M(PMC) by eqn.9.
Three PVAPMC1.5%-35-on-OPCC-R (rough-interface) beams were tested (ASTM
C 1609/C 1609M-06 2006). All three beams cracked progressively from the OPCC bases,
through the PVA overly, to the top, until failure. No interfacial cracking was observed.
Three SBRPMC1.5%-35-on-OPCC-S (smooth-interface) beams were also tested. Close
examination revealed that cracking did extend into the interface. Theoretical and experimental
mid-span deflections were also compared and found to be in agreement. It is well known that
far higher correlation can be obtained by comparing strain than displacement in flexural tests.
The first peak load P1 is one of the indicators for verifying fully-bonded composite behav-
iour. It is dependent on both, the flexural strength of the OPCC base and the bond quality
at interface. The predicted (eqn.8) and experimental results are listed in Table 8 with good
agreement.
The splitting tests procedure complied with BS EN 123906:2009. The loading rate
was 1.4 kN/s. The splitting tensile bond strength was evaluated using eqn.1. It was found
that the strength of prisms of SBRPMC1.5%-on-OPCC-R is much higher than that of
1449
Table 8. Predicted (theoretical) peak loads and test results of composite beams under 4PB test.
Theoretical Experimental
B H1 H2
No. ID of beams Interface (mm) (mm) (mm) P1 (kN) P2 (kN) P1 (kN) P2 (kN)
1 SBRPMC1.5%-on-OPCC-R Rough
2
3 100 78 47.5 24.224 30.866 28.093 28.235
4 SBRPMC1.5%-on-OPCC-S Smooth 100 79.5 48.5 27.268 31.798 26.963 30.387
5 100 78.5 49.5
6 100 79.5 48.5
7 PVAPMC1.5%-on-OPCC-R Rough 100 75 50 25.989 29.773 24.73 26.153
8 100 74 48
9 100 71 50
SBRPMC1.5%-on-OPCC-S. This explains why the latter exhibited debonding under a 4PB
test although the former did not.
6 CLOSING REMARKS
1450
Figure 8. Typical load-midspan deflection curve exhibiting first peak and second peak loads.
Figure 9. Interfacial debonding of composite beams: (a): under 3PB test, (b): under shear test.
(c): Typical load vs. CMOD of composite beams with horizontal interface notch under 3PB.
1451
6.4 Flexural behaviour of a composite beam
It was confirmed by both, laboratory tests and theoretical analysis that no delamination
occurred in all SFRPMC-on-OPCC composite beams with roughened interface under 4PB
test, while at least one out of three composite beams with smooth interface experienced
debonding at the interface. It was found that the average splitting tensile bond strength of
composite blocks of SBRPMC1.5%-on-OPCC with roughened interface is 2.96 MPa, higher
than the criterion of 2.17 MPa, which was previously defined for mix design. It is worth
noting that the bond strength of SBRPMC1.5%-on-OPCC beams with smooth interface is
1.80 MPa, lower than the criterion of 2.17 MPa. This demonstrated how crucial a good bond
is for the composite beam. A poor bond results in interfacial delamination under bending.
This indicates that the criterion of splitting tensile bond strength of 2.17 MPa for bonded
overlay mix design is experimentally confirmed.
REFERENCES
Altoubat, S.A; Roesler, J.R.; Lange, D.A.; Rieder, K. A. (2008), Simplified method for concrete pave-
ment design with discrete structural fibres, Construction and Building Materials, 22, 384–393.
Armelin, H.S., Banthia, N. (1997), Predicting the flexural post-cracking performance of steel fibre rein-
forced concrete from the pullout of single fibres, ACI Materials Journal, V.94, No.1, pp.18–31.
ASTM C 1609/C 1609M-06, 2006, Standard Test Method for Flexural Performance of Fiber-Reinforced
Concrete (Using Beam with Third-Point Loading), ASTM International, USA.
BS EN 12390–7:2009, Testing hardened concrete Part 7: Density of hardened concrete, BSI, UK.
Foote, R.M.L., Mai, Y.W., Cotterell, B. (1986), Crack growth resistance curves in strain-softening mate-
rials, Journal of Mechanics and Physics of Solids, Vol. 34, No.6, pp.593–607.
Giaccio, G., Tobes, J. M., Zerbino, R. (2008) Use of small beams to obtain design parameters of fibre
reinforced concrete, Cement and Concrete Composites, v 30, n 4, 297–306.
JSCE-SF4. (1984) Methods of tests for flexural strength and flexural toughness of steel fibre reinforced
concrete, Concrete Library International, Japan Society of Civil Engineers, Part III-2, No.3, 58–61.
Karadelis, J.N., Lin, Y., (2015) Flexural Strengths and Fibre Efficiency of Steel Fibre Reinforced,
Roller Compacted, Polymer Modified, Concrete. Construction and Building Materials, DOI:1016/j.
conbuildmat.2015.04.059, 93, 498–505.
Karadelis, J.N., Lin, Y. (2016), Strain Energy Release Rate at Interface of Concrete Overlaid Pavements.
International Journal of Pavement Engineering, DOI: 10.1080/10298436.2016.1149833, 1–10.
Lin, Y. (2014) Optimum Design for Sustainable ‘Green’ Bonded Concrete Overlays: Controlling Flexu-
ral Failure. PhD Thesis, Coventry University, UK. (unpublished).
Lin, Y., Karadelis, J.N., Xu, Y.(2013) A New Mix Design Method for Roller Compacted Bonded Con-
crete Overlays, Journal of Construction and Building Materials, 48, 333–341.
Lin, Y., Karadelis, J.N. (2015) On Establishing the Fibre Bridging Law by an Inverse Analysis Approach.
ASCE Journal of Materials in Civil Engineering. DOI: 10.1061/(ASCE)MT.1943–5533.0001386 ISSN
(print): 0899–1561, ISSN (online): 1943–5533.
Lindhagen, J.E.; Gamstedt, E.K; Berglund L.A. (2000) Application of bridging-law concepts to short-
fibre composites Part 3: Bridging law derivation from experimental crack profiles, Composites Sci-
ence and Technology 60, 2883–2894.
Mokarem, D.W., Galal K.A., Sprinkel, M.M. (2007), Performance Evaluation of Bonded Concrete
Pavement Overlays after 11 Years, Transportation Research Record, Vol.2007 No.2005, pp3–10.
Neocleous, K., Angelakopoulos, H., Pilakoutas, K., Guadagnini, M. (2011), Fibre-reinforced roller-
compacted concrete transport pavements, Proceedings of the Institution of Civil Engineers, UK,
Transport 164, Issue TR2, pp 97–109.
Tada, H., Paris, P.C., Irwin, G.R. (2000), The stress analysis of cracks handbook, (third edition), Profes-
sional Engineering Publishing.
Zhang, J.; Li, V.C., 2004, Simulation of crack propagation in fibre-reinforced concrete by fracture
mechanics, Cement and Concrete Research, 34 (2004) 333–339.
Zhang, J., Stang, H. (1998) Applications of stress crack width relationship in predicting the flexural
behavior of fibre-reinforced concrete, Cement and Concrete Research, Vol. 28, No. 3, pp. 439–452.
1452
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Yi Xu
AECOM Infrastructure and Environment Ltd., UK
Paul Phillips
Aggregate Industries UK Ltd., Ashbourne, Deryshire, UK
1 INTRODUCTION
A Pavement Repair Management System (PRMS) has been developed at Coventry Univer-
sity offering a pioneering approach to concrete road rehabilitation. It aims to enhance the
structural and functional deficiencies and extend the life of the pavement and at the same
time introduce substantial benefits by bonding a layer of high strength concrete on top of the
existing, damaged pavement. The utilization of the residual structural potential of the exist-
ing worn pavement makes it more sustainable in both environmental and economic terms, as
an alternative to the wholesale demolition and reconstruction of the existing pavement.
Shear failure at cracks/joints is a major cause of degradation of concrete pavements. Not only
it leads to serviceability problems but also introduces reflective cracks and becomes an issue of
structural integrity, durability, riding quality and safety for the users. The optimized overlaid
material benefits from its high strength and stiffness. However, it raises some concerns regarding
its shear performance due to the potential brittleness and relatively smooth fracture surface. This
research aims to make a contribution in understanding the behaviour of a concrete pavement
overlay under shear loading, and to control and evaluate reflective cracking due to shear by
means of utilisation of steel fibres to provide adequate resistance to reflective cracking.
The shear capacity of the concrete mix was scrutinized experimentally in Section 2 employ-
ing the purposely developed single notch shear beam test. Not only it enabled the evaluation
1453
of the material performance under predominant shear mode but also allowed for analysis on
a progressive failure process (crack development), as presented in Section 3. The cohesive zone
model (CZM) was adopted in fracture simulations of the test to reflect the fibre bridging effect
and aggregate interlock at the crack interface. The multi-cracking feature of steel fibre reinforced
concrete (SFRC) overlay can provide a safe buffering zone and an effective crack control.
2 EXPERIMENTAL INVESTIGATION
1454
system to minimize the influence of horizontal friction. The displacement was measured at
the loading points at the front and the rear to eliminate eccentricity in the loading system. To
investigate the crack tip displacement, a LVDT was placed across the crack tip to measure the
crack tip opening displacement. Also, a clip gauge was measuring the crack sliding displace-
ment with the aid of two small attachments. The test was carried out on a 150 kN Denison
machine jointly with a Servo hydraulic close-loop system. Displacement controlled loading
was applied to capture a stable post-peak crack development.
Figure 1. Arrangements for monolithic beams and composite beams (left) shear force and bending
moment diagram (right).
1455
Figure 2. Test rig for monolithic beams.
The Steel Fibre Reinforced Concrete (SFRC) may be regarded as normal concrete with
enhanced mechanical properties. The onset crack resistance was measured by the onset shear
strength and the mode II stress intensity factor KIIonset. It was calculated based on Tada’s
formula (2000) in Eqn.1
(a b) + 0.37 (a b) + 0.28(a b)
2 3
where FCR1 = applied shear force at the notch at the onset of slow crack growth; a = crack
depth; b = height of the beam and B = width of the beam.
1456
Figure 3. Typical crack development stages.
1457
The addition of 2% steel fibre has enhanced the onset shear resistance up to 1.7 times that
of the plain concrete. Hence the effect of steel fibre on the onset shear resistance of SFRC
cannot be neglected. Considering an OPCC and SFRC of a similar size, a load which could
cause complete failure in OPCC specimens, may not even crack a SFRC specimens with as
little as 1% volume fraction of steel fibre.
The shear performance of SFRC cannot be described explicitly by the ultimate shear
strength as in OPCC. It may be noted that SFRC presents a very different strain hardening
and softening behaviour from conventional concrete. This can be explained by the Voo and
Foster’s (2003) diagram as in Figure 4. Tension softening happens in unreinforced concrete
following crack opening. In contrast, with the addition of a minor amount of fibre, the fibre
bridging effect enables the loss of load-carrying capability at a reduced rate, as presented in
the Figure 4(a). If the fibre bridging effect is enhanced due to increased fibre content, favour-
able fibre properties and/or effective fibre-matrix bond, the strain hardening may occur prior
to the final softening section, as demonstrated in Figure 4(b). If a substantial fibre bridging
effect is present, the ultimate stress may be achieved even after the crack opens.
The fracture energy is deemed as the energy dissipating per unit crack length. It was calcu-
lated as the total areas under P-LPD curves F1–δ1 and F2–δ2 divided by the measured area
of fracture plane. For the sake of brevity, an example of the F–δ curve in a SFRC beam is
plotted in Figure 5. It is noticed that the fracture energy for unreinforced concrete are in the
order of 10−4 kN/mm. The introduction of steel fibre by as little as 1% magnified the fracture
energy by ten times. The fracture energy could be amplified by increasing the nominal size of
coarse aggregate or the volume fraction of steel fibre. However, the addition of a low volume
of steel fibre has very significant effect on the fracture energy. SFRC continued to carry loads
long after the initial cracking occurred, in which case large fracture energy value implied
good post-cracking resistance.
This test was also performed on composite beams with Ordinary Portland Cement Con-
crete (OPCC) substrate overlaid by the SF-RC-PM-BCO and conventional OPCC overlay
to compare and examine the behaviour of the composite system under shear loading con-
ditions. Multiple cracking was observed in the SF-RC-PM-BCO system, as shown in Fig-
ure 6. Considerably more fracture energy was required for the crack to propagate through
the overlay section than the conventional OPCC overlay. In other words, it is proved that the
SF-RC-PM-BCO in the composite system acted as an effective crack arresting mechanism
and delayed appearance of the surface crack and crack propagation under shear loading
condition.
1458
3 FINITE ELEMENT ANALYSIS
1459
A predominant mode II was revealed at crack initiation, as shown in Figure 9. The stress
intensity factors (SIFs) are considerably reduced as a result of effective bridging and aggre-
gate interlocking on crack control. As crack further extends, mode II transforms into mode
I due to the bridging effect. It has been pointed out by many other researchers that the addi-
tion of steel fibre causes the transition from shear failure mode to flexural failure mode.
(Nguyen-Minh and Rovnak 2011).
1460
Figure 10. Pavement structure model in FEA.
simulating a local defect due to rocking and pumping effect. A typical inflated tyre pressure
of 0.7 MPa was applied immediately to the right of the crack on the pavement top distributed
over a length of 0.25 m.
The 6-node triangular element PLANE183 was deployed in this 2D fracture model. It is
assumed that any well-developed cracks appearing on top of an existing worn pavement will
be sealed or treated at the stage of site preparation. However, the underlying cracks could be
overlooked in the initial site investigation, which potentially could reflect upwards at a later
stage. Hence, the initial crack tip was assumed to be 50mm below the top of CRCP layer.
Contact elements with the interface properties derived experimentally as below (Xu, 2014)
were employed at the overlay—CRCP interface.
As the overlay thickness increases from 100 to 200, the shear mode stress intensity factor
KII and the maximum vertical displacement decreased. It is manifested that increasing over-
lay thickness can effectively reduce the susceptibility to shear failure and reflective cracking
and minimise the differential displacement at underlying joints/ cracks. On the other hand, KI
suffered a slight increase, as shown in Figure 11. Considering the ratio γ = o/d; where o is the
distance between vehicle loading and crack, and d is the remaining thickness above crack tip, as
γ decreases, the dominant failure mode transforms from shear to flexure. When a vehicle load
is located right above the crack (o = 0), the flexural failure mode dominants. In this simulation,
although o was fixed, the increasing overlay thickness led to a larger γ ratio, indicating the same
tendency of transforming from mode II to mode I. Hence, a slight increase in KI was evident.
1461
Comparing the performance between SFRC and OPCC overlay systems, shearing is the dom-
inant failure mode in both cases. Since the crack tip located in the OPCC substrate, the KIC and
KIIC values in OPCC were employed as the crack propagation criteria, as shown by the dash
lines in Figure 11. In this design case it dictates a safe thickness design of approx. 170 mm in
OPCC overlay system and 150 mm in SFRC overlay system to resist the reflective cracking in the
CRCP. SFRC showed little advantage of reducing the overlay thickness compared to the OPCC
overlay. The slight reduction in overlay thickness is probably attributed to the enhanced stiffness
in SFRC. However, it is believed that the advanced cracking resistance of SFRC is activated only
when the crack is present in the overlay itself. Hence the following simulation was conducted.
To examine the overlay performance after the crack has extended to the overlay, a develop-
ing crack of various lengths was introduced in the 150mm thick overlay. The cohesive zone
model was applied on the crack faces in the SFRC overlay to simulate the crack bridging
effect as per Section 3.1. Again, the relevant interface properties were applied for OPCC over-
lay and SFRC overlay respectively, as defined above. The crack deformation and the shear
stress contours are shown in Figure 12.
The SIFs and their critical values in OPCC and SFRC were plotted against crack depths
and ligament depth, as shown in Figure 13. As the crack was approaching the surface, the
rise in KII and crack tip shear stress was evident. The dotted lines represent the critical SIFs.
The SFRC managed to enhance both KIC and KIIC values, compared to OPCC, although the
fibre bridging effect appears more effective in reducing the KIC. As a matter of fact, the direct
result from fibre bridging is to prevent cracks from opening. The reduction in crack sliding
(mode II SIF) is a secondary and indirect result from steel fibre reinforcement.
If KII = KIIC is adopted as the failure criterion, the overlay fractures when the KII curve
meets the corresponding KIIC lines. In other words, the SFRC overlay system appeared to
reach the critical value when the crack was 90 mm deep into the overlay (ligament depth
of 60mm). In contrast, the OPCC overlay fractured when the crack was only 30mm into
the overlay. It is manifested that the fibre bridging effect in conjunction with the enhanced
mechanical properties in the SFRC have a significant advantage in controlling the deforma-
tion and the stress concentration at the crack tip.
Moreover, one single sharp crack was assumed in the SFRC overlay in order to calculate
the crack tip SIF. However, if the phenomenon of multiple cracking is taken into considera-
tion, the stress concentration at any crack tip could be deviated and considerably reduced.
Therefore, a safety margin is allowed in the above simulation. The results are valuable in
understanding the behaviour of the “green” overlay system under a typical shear loading con-
dition and provide an important baseline for future verifications in large-scale field trials.
Figure 11. SIF in OPCC /SFRC overlay system compared with critical SIF.
1462
Figure 12. Pavement model crack in overlay shear stress Sxy (deformation × 100).
The key to success lies on the development of a material mix with superior mechanical prop-
erties and the versatility to enable fast construction. A mix design method (M-L method) has
been developed to determine the optimal water content in SF-RC-PM-BCO. It was experi-
mentally verified to produce a satisfactory mix, meeting the criteria of “roller compactabil-
ity”, “paver placability” and adequate bond to underlying concrete. The mix developed with
the above method is suitable for use with fast construction technologies and can demonstrate
superior structural integrity in combination with the old pavement.
The shear capacity of the developed mix was scrutinized experimentally and was proven to
be higher than conventional concrete overlay materials. Explicit evaluation of shear capacity
is a highly complex task. The single notch shear beam test was developed specifically to eval-
uate the material performance under mixed mode conditions with shear being the predomi-
nant mode. Not only it enabled the acquisition of a number of key strengths and fracture
parameters but also allowed for analysis to be carried out on a progressive failure process.
The addition of steel fibre succeeded to enhance concrete strength and ductility. The load
bearing capacity of the steel fibre reinforced specimens did not drop rapidly after reaching
the peak load, as it would occur in conventional concrete. Instead, the specimens continued
to carry a large proportion of the peak load for a considerably long time before gradually
losing their load bearing capacity. The fracture energy of SFRC was found to be ten times
larger than that of plain concrete. It was evident that the concrete specimens could continue
to resist shear even after a crack opens.
1463
The appearance of first crack took place at a higher load than that of OPCC or plain PMC.
Therefore, the onset cracking resistance was also increased. It was shown that the superior
quality of the new overlay material provided an efficient resistance to reflective cracking, as
the initial cracking resistance was considerably higher than that of the conventional concrete.
Moreover, following the appearance of reflective cracks, the overlay can serve as an effective
crack control measure and maintain a suitable load carrying capacity.
The progressive failure process was successfully simulated using finite element techniques.
The stresses, deformations and stress intensity factors were obtained at various crack-tip
positions. The fibre bridging effect was replicated by the cohesive crack model, the highlight
of which was a set of very informative video clips demonstrating the crack propagation and
bridging process.
It is manifested that increasing overlay thickness can effectively reduce the susceptibility to
shear failure and reflective cracking and minimise the differential displacement at underlying
joints/ cracks.
It is essential to apply the Cohesive Zone Model (CZM) in fracture simulations of SNSBT
to reflect the fibre bridging effect and aggregate interlock at the crack interface. Modelling
the crack as a complete discontinuity led to inappropriately high SIF results. On the contrary,
by incorporating the residual material mechanical properties of the cracks using the CZM,
the simulation not only showed good results but also demonstrated the transformation from
mode II into mode I as the crack proceeded.
Nomographs deduced from finite element analysis showed the influence of the overlay
thickness on the maximum vertical crack movement and shear resistance under traffic loads.
It was predicted that when the crack tip is located in the underlying substrate, the overlay
thickness design can benefit little from employing the new overlay material. A reduction of
thickness from 170 mm to 150 m was achieved using finite element modelling techniques. The
superior crack resistance of the new overlay is significantly more pronounced when the crack
extends into the overlay itself.
Finally, reflective cracking is inevitable in the long term. The multi-cracking feature of
SFRC overlay can provide a safe buffering zone and an effective crack control.
REFERENCES
Delatte Jr., N.J. 2008. Concrete Pavement Design, Construction, and Performance. UK: Taylor &
Francis, 110.
Hughes, B.P. 2006. Optimum Design of Sustainable Concrete Pavements. Proceedings of the Institute of
Civil Engineers: Engineering Sustainability 159 (3), 127–132.
Lin, Y. 2013. Optimum Design for “Green” Concrete Overlays - Flexural Failure. PhD thesis (unpub-
lished). Coventry, UK: Coventry University.
Lin, Y., Karadelis, J.N. and Xu, Y. 2013. A new mix design method for steel fibre-reinforced, roller
compacted and polymer modified bonded concrete overlays. Construction and Building Materials,
48: 333–41.
Nguyen-Minh, L. and Rovnak, M. 2011. New Formula for the Estimation of Shear Resistance of Fibre
Reinforced Beams. Canadian Journal of Civil Engineering 38 (1), 23–35.
Schlangen, E. (1993) ‘Computational Aspects of Fracture Simulations with Lattice Models’. Fracture
Mechanics of Concrete Structures. Proceedings FRAMCOS-2, 913–928.
Tada, H., Paris, P.C., and Irwin, G.R. 2000. The Stress Analysis of Cracks Handbook. third edn. New
York, the USA: The American Society of Mechanical Engineers.
Voo, J.Y.L. and Foster, S.J. 2003. Variable Engagement Model for Fibre Reinforced Concrete in Tension.
in Uniciv Report no. R-420. ed. by AnonSydney, Australia: The University of New South Wales, 88.
Xu, Y. 2014. Optimum Design of Sustainable “Green” Concrete Overlays - Shear Failure at Cracks and
Inadequate Resistance to Reflection Cracking. PhD thesis (unpublished). Coventry, UK: Coventry
University.
1464
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Yi Xu
AECOM Infrastructure and Environment Ltd., Ashford, UK
Yougui Lin
Coventry University, Coventry, West Midlands, UK
Paul Phillips
Aggregate Industries UK Ltd., Ashbourne, Deryshire, UK
ABSTRACT: This study provides the theories and viable modelling techniques for predicting
and simulating intrinsic causes of delamination failure in Bonded Concrete Overlays. Optimum
overlay mixture was formulated using Composite Desirability Analysis (CDA). The experi-
mental treatments considered both structural and elastic stability requirements of the BCOs.
Through serial material and interface fracture testing, the optimum overlay material was charac-
terised and used to predict the restraint capacities of the overlay and the interface to delamina-
tion failure under Mixed-Mode fracture process. The numerical analysis was implemented using
Interface Cohesive Zone Model (ICZM). The FEA results showed that the delamination driving
force increases with increase in BCO structural scale and mismatched elastic parameter.
1 INTRODUCTION
Analytical methods and construction techniques employed for providing durable interfacial
bonding between two cementitious materials cast at different times have been a subject of
interest among engineers and researchers. Viable techniques required for composite action of
Bonded Concrete Overlays (BCOs) are contingent on ensuring appropriate matching of the
bonded materials, and simultaneously providing effective stress transfer across the interface
(Nawy and Ukadike, 1983; Suprenant, 1988). However, in BCO systems, delamination failure
remains a long-standing problem confronting practitioners. The main cause of delamination,
aside from construction deficiency, is intrinsic in nature; and includes all cases involving com-
posite instability between the overlay and the substrate (Olubanwo and Karadelis, 2015).
In this respect, this research provides an exploratory material mixture experiments approach
to minimising delamination problem associated with composite instability, and consequently
aims to quantify complex delamination driven energies at the plane of the interface. The
research employs Composite Desirability Analysis (CDA) within the concept of mixture
optimisation techniques and Interface Cohesive Zone Model (ICZM).
2 THEORETICAL BASIS
face crack traction is derived as an extrapolation of standard Gauss nodal stresses between
two adjacent elements (Olubanwo and Karadelis, 2015). For instance, in ANSYS FEA code
employed for this research, delamination was implemented using the nonlinear traction-sep-
aration (bi-linear) law described in Alfano and Crisfield (2001). Conceptually, as depicted in
Figure 1, the interface is assumed to deform elastically with an initial stiffness (kn). When the
applied stress across the interface reaches the cohesive strength (σ0) at point A, delamination
is initiated, and then grows as a linear function of the de-bonding parameter (dn). As seen,
the value of (dn ) evolves cumulatively from 0 to 1 until the interfacial stress reduces to zero at
its critical crack point C (unc).
From Figure 1b, the delamination parameter (dn) is defined by:
⎧0 for un = un
dn = ⎨ (1)
⎩0 < d ffor un > un
Thus, the corresponding traction (ST) in each mode—tension (Mode I) and shear (Mode
II), is given by:
⎧σ = knun ( − d n ) f Mode I
for
ST = ⎨ (2)
⎩τ = kt ut ( − dt ) ffor Mode II
where, σ and τ are the cohesive stresses in the normal and tangential directions respectively,
and kn and kt are the corresponding contact stiffnesses. un and ut represent the accompany
interfacial separation, while dn and dt are the resulting delamination parameters. ūn is critical
separation for delamination initiation, while un is separation of the interface elements over the
entire loading history. By expanding the conjugate variables of the delamination in the normal
and tangential directions due to Mode-Mixity developing at the crack-tip, the effective trac-
tion vector and its corresponding effective displacement can be determined respectively by:
σ τ
λm σ 2 +τ 2 = =
Cos Sin Ψ (3)
u ut
um = un 2 + ut 2 = n
=
Cos Ψ Sin
i Ψ (4)
Hence, for a local Mixed-Mode delamination, the critical magnitude of the traction vector
now depends on the ratio between the shear and normal tractions, which by definition is the
phase angle given in equation 5.
⎛τ ⎞
Ψ = tan −1 ⎜ ⎟ (5)
⎝σ ⎠
1466
By assuming a constant steady-state phase angle for a small delamination length, the inter-
face toughness is determined as a function of the phase angle such that the interface attains
its critical fracture condition when the Mixed-Mode energy release rate Gic equals the fracture
toughness of the interface Gic ( Ψ) , given by:
Thus, for a reasonable FEA parametric study of a finite BCO system undergoing some
degree of composite instability in form of differential strain and material mismatched prop-
erties between the bonded layers, a two-dimensional plane strain analysis can be implemented
(Mei et al, 2007) to estimate the magnitude of the interface delamination driving energy as
a function of the overlay structural scale (η ) and elastic mismatched parameters of the bi-
material, using the expression given in equation 7:
D f ( ,α , β ) (7)
From above, the resulting Interfacial Cohesive Zone length (llcz ) along the interface is esti-
mated by:
⎛ E *GFm ⎞
lcz Dc (lCh ) Dc ⎜ (8)
⎝ λm2 ⎟⎠
While, the corresponding energy release rate in terms of the overlay structural size is deter-
mined by:
1 λ2h a
GFm λm umc = Dc (η,α β ) m overlay (9)
2 E*
E *Gic ( )
Gic Dc ( ,α , β ) = (10)
λm2 hoverlay
a
elastic modulus of the bi-material, GFm = Mixed-Mode fracture energy. Note, the mis-
m
matched elastic properties are generally controlled by Dundur’s parameters α and β . Where
E′ E′ E ′ ( − )( − ) − E ′ ( − )()( − ) ; E’ = E / (1–v ) is plain strain Young’s Modulus for mate-
α E1 E2 β = 1 2 (1−v1 )( − )(2E1′ − E2′ ) i i i
hoverlaay
rial i, vi is Poisson’s ratio for material i, and η = ( h ) .
total
D [d ( y ) * d ( y ) * * d ( yn )]1/ n (11)
In equation 2.11, n is total number of all individual responses. Note that the desirability
scale for each criterion (response) is generally satisfied within 0 d ( ) 1. In this case, 1 cor-
responds to the ideal response, while 0 indicates that one or more criteria fall outside their
acceptable values. The conditions for acceptance or rejection then depend on the direction of
optimisation—i.e. maximum, minimum or target.
3 EXPERIMENTAL INVESTIGATIONS
Table 1. Material specifications and physical properties (Olubanwo and Karadelis, 2014).
SBR
Cem
(%) ( WCemt )(%)
Range limit SBR (kg) Water (kg) Cem. (kg) Total amount (kg)
Lower 10 18 63.5 114.3 635 812.8
Upper 15 22 95.25 139.7 577.85 812.8
1468
namely: SBR Polymer Emulsion and Steel-Fibre. In terms of its multi-criteria performance,
the following were prescribed: (1) No sinking attribute during vibratory compaction, (2) mix
should provide sufficient mechanical and dimensional compatibility stability with the sub-
strate, and (3) mix should exhibit early durable interfacial bond performance. By these require-
ments, a hypothetical phase classification of the mixture matrix was made. Here, two phases
were identified, namely: the Paste-Phase (PP) comprising Cement, Polymer and Water; and
Solid Inclusion-Phase (SIP) made of Coarse aggregate, Fine aggregate and Steel-Fibre. The
Paste-Phase controls the ease of mix applicability during vibrating compaction, and hence
was considered central for both mix consistency and bond requirements of the overlay mate-
rial. The specifications and properties of the materials used are given in Table 1, while Table 2
represents the variability rage of the mixture components proportion. As seen in Table 2, the
trial range of variabilities for Water-Cement (W/C) and Polymer-Cement (P/C) ratios was con-
strained between 18%−22% and 10%−15% respectively, while the maximum cement content
was limited to 635 kg. The corresponding paste components proportion is shown in Table 3.
By setting the lower and upper bounds such that the sum of each possible Paste combina-
tion is constrained to 1, an Extreme Vertices Design was implemented with 13 possible mix
proportions. The implementation of the mixture model was done on the initial assumption that
a second-degree (quadratic) design will be sufficient. In the design, 4 vertex design points were
created with 4 augmented axial points. In addition, in order to ensure a more robust model,
4 interior and 1 centre points were incorporated. A total of five runs were carried out for
each design point highlighted in Figure 2(a) per specified response. The Analysis of Variance
(ANOVA) and Composite Desirability analysis were subsequently performed with Minitab sta-
tistical software (Version, 16) based on the multi-response desirability limits shown in Table 4.
In the ANOVA, components and models with p-value ≤0.05 were selected as viable.
Figure 2. (a) Extreme vertices design, (b) Representative mixture fully consolidated with a ring of
mortar around the disk, (c) Resulting surface of Mix 3 after consolidation.
1469
Figure 3. (a) Response plots for compressive strength and elastic modulus (b) Composite optimization
response plot.
Figure 4. (a) Direct L-Prism shear-slip test (b) Wedge splitting test (c) Prismatic tensile splitting.
respectively. As seen in the Cox Response plots in Figure 3a, compressive strength decreases
clearly with increase in WATER and SBR proportions, but increases as the proportions of
CEM I increase; while the elastic modulus shows that SBR has a clear reducing effect on the
overall elastic response.
From the analyses, the optimum overlay mixture was selected based on the exposure desir-
ability requirements specified in EC2 (1992). The EC2 specifies a minimum compressive
strength of C30/37*. As such, a minimum target characteristic cylinder strength of 30 MPa
was set within the first 72 hours of placing. In terms of elastic property compatibility require-
ments, the elastic modulus is required to be similar to that of the substrate (Emberson, and
Mays, 1990). Figure 3b depicts the composite desirability response curves where all responses
were optimized based on the input variable settings given in Table 4 and the composite desir-
ability analysis performed based on equation 11 with equal weight of 1 assigned to all the
responses. The overall desirability yields 0.89. As seen, the composite desirability “D” and the
individual desirability depicted by “d” for each predicted property show sufficient closeness
1470
Figure 5. Shear and tensile strengths.
Table 6. Interfacial fracture parameters for Mode I and Mode II (Olubanwo, 2013).
to 1. From the results, the predicted optimum response “y” associated with each property cor-
responds is also given. The optimum mixture proportion is indicated by the red square brack-
ets at the top of each column where SBR = 0.0938, WATER = 0.1523, and CEM 1 = 0.7540.
From here, each optimum component value is multiply by 812.8 kg which gives the needed
paste weight in the total mix. Table 5 gives the resulting complete optimum amount for the
overlay and the substrate OPCC by weight.
Since the overall desirability level shown in Figure 3b is considerably satisfactory, the
interfacial bond capacity of the optimum mixture with the underlying OPCC substrate was
subsequently accessed by employing methods of direct shear-slip and indirect tensile tests.
For the shear-slip test, the L-prism specimen shown in Figure 3a was employed to measure
the shear strength and shear slip. Here, the loading was controlled at 0.02 MPa/s by a clip-
displacement positioned adjacent to the plane of the interface until failure. The substrate
average roughness value determined by sand-patch measurement method (BS 598–3, 1985;
TRRL (1969) for all specimens was approximately 2.3 mm.
For the indirect tensile tests, two distinct test specimens were adopted, namely: Wedge
Splitting Test (WST) (see Figure 3b), and Brazilian splitting test (see Figure 3c). Note, the
Brazilian splitting test loading adhered to BS EN 12390–6:2000 and ASTM C496/ C496M.
The loading rate was controlled at 0.04 MPa/s. In the WST, the set-up requires casting or
sawing an initial groove and a notch through its top edge while a vertical load is applied
through which a horizontal tensile force is induced via the wedge / rollers contact mecha-
nism. During the monotonic displacement loading, the complete applied load vs. crack open-
ing displacement data was captured via the computer data logging device connected to the
load cell and the clip gage placed at the top-mouth of the notch. Here, the loading rate of
0.0016 mm/s was used.
For early age testing of 3 days, the observed mean shear and tensile bond strengths from
seven test specimens in each case were 3.80 MPa and 2.07 MPa respectively. In addition,
as shown in Figure 5, the mean observed tensile strength from four test specimens due to
3 days cylinder splitting tests performed on the overlay material was 5.7 MPa. In all cases,
strength increases with age as depicted in Figure 5. In Table 6, the corresponding interfacial
fracture parameters obtained from L-Prism shear-slip and Wedge splitting tests are shown.
1471
Note that the cohesive tensile strength corresponds to the Prismatic tensile splitting shown
in Figures 4(c) and 5.
The numerical analysis undertaken here investigates the intrinsic causes of delamination in
BCOs involving composite instability due to mismatched elastic properties and differential
strain between the bonded overlay and the substrate. The resulting early-age delamination
driving energy as a function of the BCO structural scale was distinctly studied using equation
10 and interfacial fracture parameters specified for age 3 in Table 6.
Figure 6. (a) Overlay edge deformation (b) Delamination model based on cohesive law.
1472
4.2 Material and interface failure models
The substrate layer was characterised by: Elastic modulus, E = 22.3GPa, Poisson’s ration,
V = 0.2, and Density, ρ = 2400 kg/m3. The elastic foundation was characterised by:
E = 15 MPa, V = 0.2 and ρ = 18.5 kg/m3. The characterisation of the overlay was based
on variable prescribed elastic properties shown in Table 7 to simulate the effects of Dun-
dur’s first parameter (α). Other properties used for the overlay included ρ = 2400 kg/m3,
K = 2.26 W/m.K and αcoef = 13.7 × 10−6 /K.
Figure 7. Typical (a) Von Mises contour stress (b) Tensile Interfacial cohessive stress.
Figure 8. Typical (a) Mixed-mode interfacial responses (b) Variation of delamination coefficient as a
function of structural scale for different mismatched elastic properties.
1473
Table 8. Equivalent delamination coefficient and cohesive zone length.
η DC (where α = 0) lcz
representing the delamination driving energy is presented. As seen in Figure 8b, the interface
response curves generally indicate that the delamination coefficient (D Dc ) increases as struc-
hoverlaay
tural scale ( hTotal
) increases in value. This is due to the fact that the restraint capacity of
the overlay is shown to increase with thickness. Similarly, since Dundur’s first parameter (α )
measures the relative stiffness of the two bonded materials, it is apparent that the driving
energy (force) required to initiate and propagate delamination grows incrementally when the
elastic mismatched parameter of the bi-material tends to nil.
By fitting a linear regression model, the resulting values of the delamination coefficient
Dc ) and the corresponding cohesive lengths (llcz ) when (α = ) are given in Table 8.
(D
5 CONCLUSIONS
The research showed that the use of computational methods is essential in order to predict
and resolve the many manifold problems associated with the intrinsic nature of delamination
in BCOs. From the results, the following distinct conclusions were drawn:
• The optimum mixture satisfied full consolidation at 34.87 seconds consistency-time with
an acceptable apparent maximum density of 97.11% TAFD.
• The optimum mixture achieved about 36.70 MPa and 59.60 MPa compressive strength at
3 and 28 days, with its tensile strength ranging between 12% and 16% of the compressive
strength.
• The optimum mixture induced intrinsic mismatched elastic property problem at its early age.
The resulting high stresses within the vicinity of the interface are generally undesirable.
• The delamination coefficient (D Dc ) varies numerically, and depends largely on: BCO struc-
tural scale and the mismatched elastic parameter of the dissimilar bonded materials. Clearly,
the restraint capacities of the overlay and the interface are shown to increase with overlay
thickness and relative stiffness values. Hence, the FEA showed that a higher driving force is
generally required to delaminate more compliant bonded materials when α → 0.
REFERENCES
ACI 211.3R (2002) Guide for selecting proportions for no-slump concrete. ACI Committee Report,
Farmington Hills, Ml, USA.
ANSYS, Inc. 2016.
ASTM C1170 / C1170M - 08 (2008) Standard Test Method for Determining Consistency and Density
of Roller-Compacted Concrete Using a Vibrating Table. ASTM International, West Conshohocken,
USA.
ASTM C138 (2001) Standard test method for density (unit weight), yield, and air content (Gravimetric).
ASTM International, West Conshohocken, USA.
ASTM C1439–99 (1999) Standard test methods for polymer-modified mortar and concrete, ASTM
International, West Conshohocken, USA.
ASTM C469/C469M (1994) Standard Test Method for Static Modulus of Elasticity and Poisson’s Ratio
of Concrete in Compression. ASTM International, West Conshohocken, USA.
1474
ASTM D792 (2008) Standard Test Methods for Density and Specific Gravity (Relative Density) of
Plastics by Displacement. ASTM International, West Conshohocken, USA.
Alfano, G. and Crisfield, M.A. (2001) “Finite Element Interface Models for the Delamination Anaylsis
of Laminated Composites: Mechanical and Computational Issues,” International Journal for Numeri-
cal Methods in Engineering, 501701–1736.
BS 598–3 (1985) Sampling and examination of bituminous mixtures for roads and other paved areas,
Methods for design and physical testing. British Standard Institute.
Buyukozturk O. and Hearing, B. (1998) “Crack propagation in concrete composites influenced by inter-
face fracture parameters,” International Journal of Solids and Structures, 35(31–32) 4055–4066.
Dundurs, J. (1969) “Edge-Bonded Dissimilar Orthogonal Elastic Wedges Under Normal and Shear
Loading,” ASME Journal of Applied Mechanics, 36, 650–652.
Emberson, N.K. and Mays, G.C. (1990) Significance of Property Mismatch in the Patch Repair of
Structural Concrete – Part 1: Properties of Repair Systems. Magazine of Concrete Research, 42, No.
152, 147–160.
Hillerborg, A., Modeer, M. and Petersson, P.E. (1976) “Analysis of crack formation and crack growth
in concrete by means of fracture mechanics and finite elements,” Cement and Concrete Research, 6,
773–782.
McLean, R.A. and Anderson, V.L. (1966), “Extreme Vertices Design of Mixture Experiments,” Tech-
nometrics, 8(3) 447–454.
Mei, H., Pang, Y. and Huang, R. (2007) “Influence of Interfacial delamination on Channel cracking of
elastic thin-films,” Int. Journal of Fracture, 148, 331–342.
Nawy, E.G. and Ukadike, M.M. (1983) “Shear Transfer in Concrete and Polymer Modified Concrete
Members Subjected to Shearing Loads,” Journal of Testing and Evaluation, JTEVA, 11(2), 89–98
Olubanwo A.O. (2013) Optimum design for Sustainable ‘Green’ Bonded Concrete Overlays: failure due
to shear and delamination. PhD Thesis, Department of Civil Engineering, Architecture, and Build-
ing, Coventry University.
Olubanwo, A. O and Karadelis, N.K. (2014) “Applied mixture optimisation techniques for paste design
of bonded roller-compacted fibre reinforced polymer modified concrete (BRCFRPMC) overlays,”
RILEM Journal of Material.
Olubanwo, A.O. and Karadelis, N.K. (2015) Interfacial Delamination Failure in Bonded Concrete
Overlay Systems - A Review of Theories and Modelling Methods. International Journal of Civil
Engineering & Technology (IJCIET).Volume:6, Issue:5, Pages: 85–99.
Petersson, P.E. (1981) “Crack Growth and Development of Fracture Zones in Plain Concrete and Simi-
lar Materials,” Report TVBM – 1006, Division of Building Materials, Lund Inst. of Technology,
Lund, Sweden.
Suprenant, B. (1988) “Bonding new concrete to old,” Concrete Construction, 33, No. 7, 676–680
1475
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Liangying Li
Lanzhou Jiaotong University, Lanzhou, China
Bo Li
Key Laboratory of Highway Network Monitoring of Gansu Province, Lanzhou, China
Zhengwei Zhang
Chang’an University, Xi’an, China
Xiaolong Yang
Lanzhou Jiaotong University, Lanzhou, China
1 INTRODUCTION
With increasing traffic volume, road traffic noise has become a major source of environmental
noise pollution (Sandberg et al. 1996; Babisch et al. 2008). Among the various kinds of road
traffic noise, the noise generated by conventional cement concrete pavements is the most sig-
nificant one, and the term “pavement noise” has become synonymous with cement concrete
pavements (Ya-min Liu et al. 2012; Ying Liu et al. 2012). Given that continuous improvement
in road conditions is now a necessity from the viewpoint of driver safety, this kind of noise
has become a crucial factor in the development of cement concrete pavements. Reduction of
the noise of cement concrete pavements, especially of the traffic noise in densely populated
areas, on high-speed major routes near cities and in tunnels has become a leading issue that
needs to be addressed by engineers (Van Keulen et al. 2005; Cong et al. 2012).
In order to mitigate the adverse effects of concrete pavement noise on humans and the
environment, approaches such as the construction of soundproof walls and planting of trees
are often adopted locally and globally. These infrastructures, however, are not only expensive
but also difficult to maintain after damage (Hanson et al. 2004), which makes it impossible
to solve the problem of cement concrete pavement noise fundamentally. Because of the noise
generation mechanism between the tire and the road, the stiffness and surface texture of a
1477
concrete pavement have a large influence on its noise level (Li et al. 2008). Tian et al. (2014)
and Jing et al. (2010) reported that the noise level of Portland Cement Concrete Pavements
(PCCPs) with a porosity of 17% can be reduced by 4–8 dB (A) in comparison to that of tined
cement concrete pavements. Through long-term research on Exposed-Aggregate Cement
Concrete Pavements (EACCPs), researchers such as Han et al. (2008) and Jia et al. (2008)
found that the use of EACCPs can reduce the noise level by 4–5 dB (A) in comparison to
that of transverse-tined cement concrete pavements, because the surface of EACCPs has a
texture comprising regular concave and convex patterns, which are sufficient for resisting
sliding and reducing noise effectively. Further, researchers such as Ling et al. (2011) have
found that adding ceramsite to concrete can reduce traffic noise via lowering of the surface
stiffness, while maintaining the mechanical and abrasion resistance properties of the pave-
ment. Therefore, use of new types of concrete pavements can reduce noise significantly. How-
ever, the enormous costs associated with engineering and development can cause difficulties
in the popularization and application of the related technologies. Therefore, at present, to
reduce concrete pavement noise, we need to consider mainly the noise generated between the
hard grooving (the anti-skid structure of concrete pavements currently adopted in China)
and vehicle tires. The Second Strategic Highway Research Program of the US, which is
based on research on the functions and properties of road surfaces, such as safety, reliability,
regeneration, and performance, reports that the noise level of diamond grinding compos-
ite pavements is much lower than those of Hot-Mix Asphalt (HMA) pavements, EACCPs,
and conventional grinding pavements (Akkari et al. 2012). Researchers in Arizona (Ferragut
(Cofield, L. 2003)) tested the noise of cement concrete pavements with different textures and
reported that noise levels could be ranked from low to high in longitudinal diamond-ground
pavements, longitudinal tining pavements (uniformly spaced), transverse tining pavements
(randomly spaced), and transverse tining pavements (uniformly spaced). Generally speaking,
related studies on concrete pavement noise still focus mainly on the validation of the noise
reduction effects of test roads that have adopted new types of concrete pavements and new
texture technologies; however, such studies do not perform systematic and comprehensive
statistical analyses of the noise levels of different concrete pavements and textures. Then, on
the basis of these limited studies and inconsistent methods, it is not possible to define the
zone boundaries of concrete pavement noise and hence the noise limit values of newly con-
structed concrete pavements.
The practice of controlling traffic noise dates back to 1973, when the Federal Highway
Administration (FHWA) laid down requirements on traffic noise control. In the subsequent
two decades, these requirements were put into practice by highway bureaus in many coun-
tries. In 1992, FHWA (1995) proposed the establishment of an industry standard for regulat-
ing control of traffic noise. Currently, in the ISU-FHWA-ACPA Concrete Pavement Surface
Characteristics Project launched jointly by Iowa State University (ISU), FHWA, and the
American Concrete Pavement Association (ACPA), according to the Traffic Noise Model
and road performance, put forth the regional limit, which is the only limit value standard of
concrete pavement noise (Ferragut et al. 2007; Rasmussen et al. 2012; Chai et al. 2010). How-
ever, the highway administration of China has not yet established any regulations or norms
for traffic noise control. Therefore, establishment of a limit value standard suitable for the
design and construction of concrete pavements is presently an urgent requirement.
To this end, a program has been initiated in China for the establishment of a noise con-
trol standard for concrete pavements in China on the basis of evaluation of the pavement
characteristics affecting the tire/pavement noise levels as measured by the On-Board Sound
Intensity (OBSI) method. Specifically, in this study, this method was employed for testing
the noise levels of concrete pavements of many highways. Additionally, the K-means cluster-
ing and threshold limit analysis methods were used for performing statistical analysis of the
noise zoning and referential limit standard. The results of this study are expected to be useful
in determining the most effective management method and in seeking a set of unified noise
control measures, which can provide a basis for improving the textures of concrete pavements
and reducing their noise levels.
1478
2 RESEARCH METHODOLOGY
1479
Figure 1. OBSI test system.
on the concept of the two-microphone sound intensity technique, and it employs the charac-
teristic of directivity; therefore, it can accurately measure the noise of concrete pavements in
tunnels (Lodico and Donavan 2013).
Figure 1 shows the OBSI test system; the car used in the test was a HuaChen car. In the
test, the car passed the test sections in a straight line at a constant speed of 60 km/h. For each
group of valid data, their references were the distances that the car covered in 6 s at a constant
speed of 60 km/h, and these data were then averaged. The frequency for data collection of the
concrete pavement noise was set at 10000 Hz. Dewesoft 7.0 and Coinv Dasp E.T. software
programs were used for the noise data collection.
1480
Table 2. Description of sample statistics for concrete pavement noise test.
A 20 103.6 103.5 104.0 2.649 0.791 0.512 0.149 0.992 101.4 107.3
B 20 103.9 103.5 103.3 3.512 0.308 0.512 −0.413 0.992 100.7 107.7
C 20 103.1 103.2 101.4 3.134 0.028 0.512 −0.357 0.992 99.8 106.6
D 20 100.6 100.5 99.8 1.565 0.533 0.512 0.151 0.992 98.5 103.5
E 20 101.2 101.1 99.8 1.681 0.580 0.512 1.098 0.992 98.7 104.4
F 16 101.9 101.8 102.3 3.449 0.123 0.564 −0.581 1.091 98.9 105.4
G 25 102.9 103.1 102.3 4.693 0.464 0.902 98.6 107.2
1481
−0.112 −0.317
H 22 102.6 102.6 102.5 4.158 −0.123 0.491 −0.196 0.953 98.5 106.7
I 22 102.7 102.7 103.7 5.158 0.159 0.491 −0.480 0.953 98.9 107.4
J 24 102.2 102.5 102.5 4.704 −0.196 0.472 −0.400 0.918 97.8 106.4
K 24 102.3 102.6 100.8 4.806 −0.142 0.472 −0.540 0.918 98.2 106.4
L 22 102.2 102.3 98.7 3.884 −0.010 0.491 −0.629 0.953 98.7 105.7
M 36 99.5 99.0 97.6 4.855 0.524 0.393 −0.524 0.768 95.8 104.5
N 30 98.8 98.7 98.2 2.548 0.292 0.427 0.201 0.833 95.6 102.7
O 20 99.5 99.2 98.2 3.177 0.493 0.512 −0.146 0.992 96.4 103.4
P 20 99.3 99.0 98.9 3.871 0.523 0.512 0.176 0.992 95.8 103.7
Q 34 99.1 98.8 97.3 4.134 0.373 0.403 −0.733 0.788 95.9 103.0
mum, and maximum, of the K-means initial clustering centers of the concrete pavement
noise data are presented in Table 3.
From Table 3, it can be seen that the data samples of the concrete pavement noise bunch
into three clusters. The mean and median value condensation centers of Cluster 1 are 99.2 dB
(A) and 98.9 dB (A), respectively, which are close to the noise levels for HMA and EACCP.
The mean and median value condensation centers of Cluster 3 are 103.5 dB (A) and 103.4
dB (A), respectively, which are close to the noise levels of the cement concrete pavement with
transverse tining. Finally, the mean and median value condensation centers of Cluster 2 are
102.1 dB (A) and 102.2 dB (A), respectively, which are close to the noise levels of transverse
combination tining and longitudinal-transverse combination tining pavements.
The results of K-means initial clustering analysis of the concrete pavement noise data are
presented in Table 4.
From Table 4, it can be seen that the concrete pavement noises coded as A, B, and C cluster
into one type (Cluster 3), which include transverse tining pavements. Further, noises coded
as D, E, F, H, I, J, K, L, and M cluster into a second type (Cluster 2), which include many
longitudinal tining pavements, transverse tining pavements, and pavements with combined
longitudinal- transverse tining. Finally, noises coded with N, O, P, Q, and R cluster into a
third type (Cluster 1), which include EACCPs, SMA pavements, NovaChip pavements, and
AC pavements.
Cluster
Variable 1 2 3
Table 4. Results of K-means initial clustering analysis of concrete pavement noise data.
1 A 3 0.782
2 B 3 0.632
3 C 3 1.132
4 D 2 3.253
5 E 2 2.024
6 F 2 0.723
7 G 2 1.823
8 H 2 1.069
9 I 2 1.763
10 L 2 0.970
11 M 2 0.796
12 N 2 0.337
13 Q 1 1.078
14 O 1 0.959
15 P 1 0.624
16 Q 1 0.273
1482
Table 5. ANOVA table of results of K-means initial clustering.
Clustering Deviation
OBSI noise levels of porous AC and AC pavements are lower than the critical limit value
of this zone (Kowalski et al. 2009; Wang et al. 2012; Way et al. 2010; Zheng et al. 2006).
2. Zone 2, “Mid-controlled” Zone: This is the zone in which the value of the concrete pave-
ment noise in the OBSI test is 99/100–105/106 dB (A). The kinds of surface textures in
this zone mainly include most of the longitudinal tining textures, combined longitudinal-
transverse tining textures, transverse combination tining textures, and some transverse
combination tining textures. It can be seen that the OBSI noise values corresponding to
the construction technology levels of ordinary concrete pavements and textures lie mainly
in Zone 2. During the OBSI test, the surface texture of the concrete pavement has strong
variability, mainly because the depth of the texture lacks unified specifications, which
leads to corresponding changes in the concrete pavement noise. Therefore, in the process
of concrete pavement construction, the construction quality of the texture of concrete
pavements should be controlled stringently so as to reduce the variability of the produced
texture and to achieve texture types having high durability and low noise (Ongel et al.
2011; Donavan et al. 2009).
3. Zone 3, “High-limit” Zone: This is the zone in which the value of the concrete pavement
noise in the OBSI test is higher than 105/106 dB (A). The concrete pavements falling in
Zone 3 are mostly those that are transverse-tined with low smoothness and severe damage
at joints (Herman et al. 2006; Bennert et al. 2005). Therefore, the value of the zoning limit
between Zone 2 and Zone 3 is taken as the limit value of concrete pavements. The results
of this study suggest that in the present design code for concrete pavements, the follow-
ing stipulations should be made for concrete pavement noise. The noise levels of newly
constructed concrete pavements are not permitted to fall in Zone 3, especially for some
noise-sensitive projects. Else, suitable measures must be taken to reduce the noise levels of
the concrete pavements falling in Zone 3.
In addition, the premise of reducing concrete pavement noise should be keeping the sur-
face friction of the pavement. For undertaking noise-sensitive or noise-eliminating projects in
Zone 3 successfully, after suitable measures have been taken, tests on the pavement’s smooth-
ness, friction, and noise should also be performed to ensure that these three factors are bal-
anced and to transfer them to Zone 2, or even Zone 1.
1485
5 CONCLUSIONS
1. By adopting the K-means clustering analysis method, we were able to divide the concrete
pavement noise measured by the OBSI method into three zones: the “innovation” zone,
“mid-controlled” zone, and “high-limit” zone.
2. Zone 1, “Innovation” Zone: This is the zone in which the value of the concrete pavement
noise in the OBSI test is generally lower than 99/100 dB (A). The noise values of concrete
pavements constructed using new types of concrete as their thin paving coats, such as
porous concrete, exposed-aggregate concrete, and polymer concrete, fall in this zone.
3. Zone 2, “Mid-controlled” Zone: This is the zone in which the value of the concrete pave-
ment noise in the OBSI test is 99/100–105/106 dB (A). The OBSI noise values correspond-
ing to ordinary concrete pavements and surface textures mainly fall in Zone 2. This zone
mainly includes pavements with longitudinal tining, combined longitudinal-transverse
tining, transverse combination tining, and transverse combination tining. In the process
of concrete pavement construction, the construction quality of the texture of concrete
pavements should be controlled stringently so as to reduce the variability of the produced
texture and to achieve texture types having high durability and low noise.
4. Zone 3, “High-limit” Zone: This is the zone in which the value of the concrete pavement
noise in the OBSI test is higher than 105/106 dB (A). The concrete pavements that fall in
Zone 3 are mostly those that are transverse-tined with low smoothness and severe damage
at joints.
5. The obtained results suggest that the value of the zoning limit of noise between Zone 2
and Zone 3 should be used as the limit value standard for newly constructed pavements.
In addition, the following stipulations should be made for concrete pavement noise. The
noise levels of newly constructed concrete pavements should not fall in Zone 3, especially
for some noise-sensitive projects. Else, suitable measures such as optimization of the con-
crete texture, repair of pavement joints, and recovery of pavement smoothness must be
taken to reduce the noise levels of concrete pavements.
ACKNOWLEDGEMENT
The research work reported in this paper was sponsored by Gansu DOT Research Project
(2011–08), Changjiang Scholars and Innovative Research Team in University (IRT1139) and
Youth Scientific Foundation of Lanzhou Jiaotong University (2011031). The authors would
like to express thanks to all who have provided valuable help in this study.
REFERENCES
Alexanda Akkari, Bernard Izevbekhai. 2012. Composite Pavements and Exposed Aggregate Texturing
at MnROAD: Cells 70, 71 and 72 Construction Report and Early Performance Evaluation. Minne-
sota Department of Transportation,.
Bennert, Thomas, Hanson, Doug; Maher, Ali; Vitillo, Nicholas. 2005. Influence of pavement surface
type on tire/pavement generated noise.Journal of Testing and Evaluation, 33(2):94–100.
Chai Hua, Zhong Rong-gui, Xie Jun. 2010. Standard Vehicle Type Classification Criteria Based on
Highway Operating speed Design. China Journal of Highway and Transport, 23(Suppl.): 13–18.
Cofield, L. 2003. SR202 PCCP Whisper Grinding Test Sections. Arizona DOT.
Cong Zhuo-hong, Wang Sha-zheng, Nan Xian, etc. 2012. Asphalt mixture design with anti-skidding and
low noise performance in tunnel pavement. China Journal of Highway and Transport, 25(4):36–41.
Donavan, Paul R.; Rymer, Bruce. 2009. Effects of aging on tire-pavement noise generation for concrete
pavements of different textures. Transportation Research Record, (2123): 137–144.
FHWA. 1995. Highway Traffic Noise Analysis and Abatement–Policy and Guidance. Washington, D.C:
Federal Highway Administration.
Han Sen, Dong Yu-ming, Chen Hai-feng, etc. 2008.,Noise reduction performance of exposed-aggregate
cement concrete pavement. Journal of Traffic and Transportation Engineering, 2005, 5(2): 32–34.
1486
Hanson, D.I., R.S. James, etc. 2004.Tire-Pavement Noise Study. AL: National Center for Asphalt
Technology,
Herman, Lloyd; Withers, Jared; Pinckney, Elvin. 2006. Surface retexturing to reduce tire-road noise for
existing concrete pavements. Transportation Research Record, (1983):51–58.
Jia Jin-xiu, Han Sen. 2008.Research on the Skid-resistance and Noise Reduction of Tunnel Cement
Concrete Pavement. Journal of Highway and Transportation Research and Development(Applied Tech-
nology Edition), (4): 42–45.
Jing Lu-bo, Tian Bo, Liu Ying, etc. 2010. Research on Noise Reduction Properties of Porous Cement
Concrete Pavement. Highway, (7): 75–79.
Karol J. Kowalski, Rebecca S. McDaniel, Ayesha Shah. 2009. Long-Term Monitoring of Noise and
Frictional Properties of Three Pavements. Journal of the Transportation Research Board, 2127(2):
12–19.
Li Bo, Han Sen. 2008. Some Methods on Concrete Pavement Noise Reduction. Noise and Vibration
Control, (3): 127–130.
Ling Tian-qing, Wang Ruyan, Zhang Jian, etc. 2011. Influence of Ceramsite Mortar and Ceramic Con-
crete on Reducing Traffic Noise in Tunnel. China Journal of Highway and Transport, 24(1): 21–24.
Liu Ya-min, Han Sen, Tao Zhi-jin, etc. 2012. Noise Evaluation of Concrete Pavement with Different
Texture. Journal of Chang’an University( Natural Science Edition), 36(4):16–20.
Liu Ying, Tian Bo, Niu Kai-min. 2012, Skid-resistance and Denoising Properties of Cement Concrete
Pavement with Different Surface Texture. Journal of Highway and Transportation Research and Devel-
opment, 29 (1): 28–33.
Lodico Dana, Donavan Paul. 2013. Using onboard sound intensity measurements to interpret results
of traffic noise modeling. Transportation Research Record, (2362): 9–15.
Ongel, Aybike; Harvey, John. 2011. Pavement characteristics affecting the frequency content of tire/
pavement noise. Noise Control Engineering Journal, 58(6): 563–571.
Robert Otto Rasmussen, Paul D. Wiegand, Gary J. Fick, etc. 2012. How to Reduce Tire-Pavement
Noise: Better Practices for Constructing and Texturing Concrete Pavement Surfaces. Ames: National
Concrete Pavement Technology Center.
Sandberg,U., J.A.Ejsmont. 1996. Tire/Road Noise Reference Book. Sweden: INFORMEX, 2002.
Shi Li-wen. 2013. SPSS 19.0 Statistical Analysis from Entry to the Maste. Bei jing: Tsinghua University
Press,
Theodore Ferragut, Robert Otto Rasmussen, Paul Wiegand, etc. 2007. ISU-FHWA-ACPA Concrete
Pavement Surface Characteristics Program Part 2: Preliminary Field Data Collection. Ames: National
Concrete Pavement Technology Center.
Tian Bo, Liu Ying, Niu Kai-min, etc. 2014. Reduction of tire-pavement noise by porous concrete pave-
ment. Journal of Materials in Civil Engineering, 26(2):233–239.
Van Keulen W. 2005. Silent Roads for Cost Effective Noise Reduction. I-INCE. 34th International
Congress on Noise Control Engineering 2005. CA: Institute of Noise Control Engineering of the
USA, 3854–3862.
Wang, George; Smith, Gregory, Shores; Richard. 2012. Pavement noise investigation on North Caro-
lina highways: An on-board sound intensity approach. Canadian Journal of Civil Engineering, 39(8):
878–886.
Way, George B; Carlson, Douglas D; Sousa, Jorge B; etc. 2010. Introduction to asphalt-rubber pave-
ment noise reducing characteristics.39th International Congress on Noise Control Engineering 2010.
CA: Sociedade Portuguesa de Acustica, 7408–7417.
Wolfgang Babisch. 2008. Road Traffic Noise and Cardiovascular Risk. Noise&Health, 10(38):27–33.
Zheng Mu-lian, Chen Shuan-fa, Wang Chong-tao. 2006. Strength character of porous concrete. Journal
of Chang’an University( Natural Science Edition), 26(4): 26–29.
1487
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Mykola Dmitriev
National Transport University, Ukraine
1 INTRODUCTION
Damages occurring in pavement during the operation significantly reduce its service life.
Therefore, continuous monitoring of roads and timely repair are critical for ensuring smooth
and safe traffic.
Thermal monitoring is carried out to detect defects of pavements and define their param-
eters to assess the quality and conformity of monitored objects with the requirements of
normative technical documentation.
Non-destructive thermal monitoring of road pavement is based on passive monitoring
method because at roads operation heat absorption (release) occurs due to environmental
conditions.
Recording of temperature distribution on the surface of the pavement was carried out
using a thermal imager. The conducted analysis was aimed at the verification of compliance
of obtained results with the calculated data regarding hidden defects such as voids, scaling,
cracks and crushing of pavement material. It is based on comparing the results of math-
ematical modeling for specific weather conditions under which the study was conducted.
Obtained termograms allow determining the rutting, settlement of individual sections,
potholes, sinks, traces of poor repair, hidden air voids that are caused by pavement deforma-
tion, scouring in the ground basis, transverse and longitudinal cracks and so on.
2 SETTING OBJECTIVES
1489
the definition of basic riding qualities. Obtained experimental characteristics are the basis
for the formation and filling the road data bank. Regularity of data acquisition related to
the roads state and systematic data processing will allow getting updated data for taking
the management decisions on a rational consumption of material and financial resources
for maintenance.
The aim of work is to improve the method for the assessment of service conditions of
roads along with detecting the defects and damages of concrete pavement using thermal
imaging equipment.
3 THEORETICAL ASPECTS
3.1 The concept of residual life of the road structure during operation
Model of deterioration of monolithic material of pavement layer is taken as a basis (Figure 1)
according to which strength margin is decreased during the service life depending on
1
⎛ N (T ) ⎞ br
Kr K o ⎜1 − Σ * ⎟ , (1)
⎝ N ⎠
4>
>?
where Κo- initial strength margin of material in the structure; NΣ(T) - total number of applied
loads during the period of time T; N* - limit number of loads after which application the
material is deteriorated; br - an indicator characterizing fatigue of the material (for asphalt it
is equal to 4–6; for concrete it makes 16–25).
At operation, strength of monolithic materials of pavement layers is reduced due to dam-
age accumulation and aging, and after a period of time T>0 it makes R (T). This residual
strength can be put in line with equivalent residual thickness H (T) with an initial strength
of R (T = 0).
At operation, strength of monolithic materials of pavement layers decreases due to dam-
age and aging effects accumulation. After a period of time T>0 it makes R(T). This residual
strength can be put in line with equivalent residual thickness H(T) with an initial strength
of R (T = 0).
Basing on the results of theoretical calculations it was concluded that at any moment of
time T>0 the thickness of reinforcing layer ΔH(T) required to provide the initial strength of
the structure is determined by the formula 2:
Figure 1. Change in the strength of material of a monolithic pavement layer at loads application.
1490
⎡ 1
⎤ ⎡ 1
⎤
⎢ ⎛ T ⎞ 2b ⎥ ⎢ ⎛ N Σ (T ) ⎞ 2 b ⎥
ΔH (T ) = H o H (T ) H o 1 − ⎜1 − = H 1 − 1 − (2)
⎢ ⎝ TA; ⎟⎠ ⎥ o⎢
⎝ * ⎟ ⎥
>? ⎠
N 4>
⎢⎣ ⎥⎦ ⎢⎣ ⎥⎦
where Ho - initial length which is set depending on the criterion of limit state of the
structure.
The cost of repair Cp of the road section of length L varies linearly depending on the
thickness of ΔH
Cp (a0 + b0 H ) ⋅ L, (3)
where a0, b0 – coefficients that depend on the road category, cost of materials and works
performance technology.
The above formulas (1–3) provide the opportunity to perform the analysis “Durability—
Cost” and cost analysis during operation to assess design solutions and repair strategies.
Changes in the design thickness of the structure’s material during operation can be detected
basing on the results of visual or instrumental inspection considering the reduction coeffi-
cients based on detected defects and deterioration. The research results proved efficiency of
using thermal imaging method.
4 MAIN TEXT
(ξ ξ ) 2
1 − i
f ξ e 2σ 2 (4)
σ 2π
where ξ - the average value (mathematical expectation) of ξ; σ—its standard (rms) deviation
(Table 1).
The design strength Rp and the strength grade of concrete B were determined by the known
formulas
Rp =Rcp (1-1 C )
(5)
B=Rcp (1 - 1,64 • Cv )0.1,
(6)
Where Rcp - average strength value; Cv- coefficient of variation; 0.1 - coefficient of trans-
forming kgs/cm2 in MPa.
1491
Figure 2. Frequency and function of normal distribution of compressive strength of concrete samples
according to test results of 2011 and 2015.
Table 1. Statistical data processing results related to determining compressive strength of samples in
2011 and 2015.
Results
Basing on the results of statistical processing, the average value of compressive strength
of concrete varied from 408.4 kg/cm2 in 2011 to 384.8 kg/cm2 in 2015. Concrete grade corre-
sponds to M 400. Compressive strength variation coefficient Cv ranges from 13.83 to 17.97%,
which is slightly above the design value of variation coefficient at the construction which is
CV = 13,5%. The design strength considering variation coefficient at Student’s law is within
the range of 25.66 ... 26.89 MPa.
The average grade of concrete strength for a given road section at t = 1,64 (normal law) B
30 was received as a result of testing. This strength grade meets the minimum design grade
of concrete for a monolithic single-layer pavement or a top layer of two-layer pavement for
roads of category II, III according to DBN V.2.3–4: 2015 "Highways. Part I. Design. Part II.
Construction".
Thermal monitoring includes the following: the analysis of the design and technological
documentation of road structures; determination of quantified temperature values on the
points of the road surface; identification of additional characteristics of the road surface
and the environment; qualitative and quantitative analysis of temperature fields on the stud-
1492
ied surface of pavement; identification of the areas of abnormal temperature caused by the
defects in the monitored road section; determining the defects parameters; assessment of the
road surface quality.
Thermal imagers that include the devices registered in the state register of measuring
instruments that have been certified and approved for use in Ukraine are the main means of
thermal monitoring.
Thermal measurements are made when the temperature difference on the surface of the
pavement is higher than the sensitivity of the used thermal imager in the absence of precipita-
tion, fog, smoke.
Optimal season for thermal monitoring is spring-summer-autumn period. The optimum
time for monitoring during the day is between 10 am to 20 pm when there is a maximum
thermal contrast.
1493
4.4 Assigning the category of concrete pavement deterioration
Category of deterioration of existing concrete pavement is to be assigned to it in order to cal-
culate rigid pavement reinforcement. In recent years, thermal imaging survey has been used
to determine the extent and causes of deterioration.
Thermal imaging survey of existing pavement was made in accordance with the “Recom-
mendations for improving the method of assessing the state of pavement using thermal imag-
ing equipment” developed by NTU (National Transport University).
TESTO 875-2 brand device was used for the survey.
Main technical characteristics of portable thermal imager used for thermal imaging testing
are as follows:
− detector 160 x 120 pixels, uncooled microbolometer;
− thermal sensitivity is 0.08 °C;
− spectral range is 8...14 ^m;
− display 3.5' LCD, 320 x 240 pixels;
− image view—IR and real (digital camera);
− temperature measurement range (- 20... + 280 °C);
− measurement error - 2 0C.
The survey detected the following defects and deterioration: chipped corners of slabs
(Figure 3); deformation and damages in the area of concrete layer joints (Figure 4); settle-
ment cracks(Figure 5); chips in cement concrete pavement; potholes in cement concrete pave-
ment; settlement cracks in cement-concrete pavement; settlements to 1–3 cm and other.
Random results of thermal imaging survey and detailed description of defects and dete-
rioration are listed below.
1494
Figure 4. Deformation of joints.
1495
Settlement cracks (Figure 5) are caused by insufficient compaction of the foundation in
case of weak soil under the cement concrete slab and poor drainage from the surface. In such
cases it is necessary to replace the weak soil, adhere to the compaction norms, strengthen or
reinforce the soil by geosynthetics. Thermal imaging survey of rigid pavement is shown in
Figure 6.
Relatively uniform temperature distribution over the surface of the pavement is character-
istic for pavements without deformations (Figure 5). Frequency of temperature distribution
over the test area is approximated by the normal distribution. Reflected cracks within 15 m
(three slabs) do not affect carrying capacity and traffic safety.
Figure 7. Change in the strength of concrete samples along the length of the road in 2011 and 2015
(the left and right traffic lanes).
1496
Table 2. Category of deterioration of existing rigid pavements (SNiP (Construction norms) 2.05.08-85
Airfields).
Chipped edges,
Category of diagonal through
deterioration Number of slabs% Through cracks cracks, through
of slabs of existing Scaling depth Chipped edges (longitudinal longitudinal and
rigid pavement over 1 cm in joints location or transverse) transverse cracks
I Less than 10 - - -
II from 10 to 30 Less than 30 Less than 20 -
III more than 30 30 and more from 20 to 30 Less than 20
IV Not rated more than 30 20 and more
Ways of pavements strengthening are determined considering the categories of design load
and depending on the characteristics of the foundation and soil subgrade and the state of
existing pavement.
The category of deterioration of existing rigid pavements is assigned according to
Table 2.
The category is assigned basing on the indication corresponding to the highest category
of deterioration (Fig. 3).
Through cracks are considered if the average distance between them is less than 5 m and
they are not permissible by the design limit state.
At determining the percentage of deteriorated slabs, a number of slabs subjected to the
influence of loading along the rolled route are considered.
At determining the stiffness and limit bending moment of concrete and reinforced con-
crete pavement layers their calculated thickness hpd should be accepted depending on the
category of deterioration which is assigned basing on Table 2 and the thickness hex of existing
pavement of the appropriate category of deterioration:
I hpd = hex;
II hpd = 0.9 hex;
III hpd = 0.8 hex.
For calculation purposes, existing rigid pavement of IV category of deterioration should
be considered as an artificial basis with common elasticity modulus E = 60 MPa (600 kgs/
cm3).
Basing on the survey results, the 2nd category of deterioration of slabs of existing rigid
pavement was assigned by the indicator “through cracks (longitudinal or transverse) less
than 20%, so hpd = 0,9⋅hex for the calculation of reinforcement.
4 CONCLUSIONS
The disadvantage of thermal monitoring method is the following: it is required that the temper-
ature difference should exceed the instrument error. Surveyed sections must not be shaded (by
buildings, trees or parked vehicles, etc.) that may cause error in temperature fields. Pavement
surface must be clean and pollution-free (from oil stains, dirt, foreign matters, etc.). The size of
subsurface irregularities (culvert, void, settlement) must exceed the depth of their occurrence in
two times to enable their detection. If required, additional wetting or heating of the pavement
surface for more accurate defects and damage identification may be expedient.
Experimental and theoretical studies on thermal imaging survey of pavement showed the
following results:
− thermal imaging survey is an effective method of detecting subsurface defects and dam-
ages of road structures that allows quick examination of road sections for technical
diagnostics;
1497
− changing the surface temperature of pavement which is recorded on the thermograms
allows detecting hidden defects in concrete pavement or under it;
− local temperature variation on individual sections of pavement indicates the spots of accel-
erated deterioration under the influence of weather conditions and traffic loads;
− thermal imager using is very efficient for the detection of sections where pavement design
and thickness of layers undergo changes;
− regular monitoring of road sections with defects allows scheduling the repairs on early
stages of deterioration with minimal rehabilitation costs and in compliance with the
requirements of technical regulations.
REFERENCES
Dmitriev, M., Derkachov, O., Papchenko, O. & Doroshenko, Y. 2004. The use of thermal imager for
nondestructive control of airfield pavements//NTU BULLETIN No. 9, P. 67–70.
Dmitriev, M., Papchenko, O., Derkachov, O. & Rutkovska, 1. 2008. Determination of boundary condi-
tions on the surface of airfield pavement with the problem statement related to the modeling of its
thermal field // NAU Bulletin. - K.:1. P. 161−164.
Dmitriev, M., Gamelyak, I., Popelysh, I. & Babskii, V. 2012. Application of thermal imaging equipment
to assess the condition of road asphalt pavement. K.: Proceedings of the 14-th international scientif-
ic—technical conferences “Services market of integrated transport systems and logistics application
issues”, p. 28−34.
Papchenko, O. & Derkachov, O. 2007. Using the capabilities of modern imager to improve the efficiency
of monitoring the state of airfield pavements. Transport issues. Collection of NTU scientific works,
Issue 4, P. 40−44.
Papchenko, O. 2008. Criteria of airfield pavement thermograms analysis. - K.: NTU BULLETIN. - Vol.
17, part 2, P. 47−50.
1498
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
J. Ramaekers
Hasselt University, Diepenbeek, Belgium
L. Vandewalle
Catholic University of Leuven, Heverlee, Belgium
H. Degée
Hasselt University, Diepenbeek, Belgium
1 INTRODUCTION
Rigid concrete pavement design still often relies on the traditional theories presented by
Westergaard and Bradbury, who formulated closed-form solutions for stresses and displace-
ments in concrete pavements under static traffic loading conditions and linear temperature
distribution throughout the pavement depth (Westergaard 1926, 1927, Bradbury 1938).
Since traffic loading is dynamic in nature and temperature distribution throughout the
concrete slab thickness has shown to be highly nonlinear (e.g. Choubane & Tia 1995), these
theories are not representative of actual concrete pavement behavior. As a result, growing
interest in slab curling due to nonlinear temperature gradients and dynamic analysis of
concrete pavements under moving traffic loading has been observed among many research-
ers. However, contradictory results and conclusions exist with regard to the significance of
dynamic effects in concrete pavement design (e.g. Gillespie et al. 1993, Chatti et al. 1994,
Darestani et al. 2007).
In the finite element (FE) method, concrete pavements are generally modeled as a beam,
thin/thick plate or solid element on the underlying layers (Beskou & Theodorakopoulos 2011).
Two types of modeling approaches are used to describe subgrade behavior i.e. mechanical
models, of which the spring model proposed by Winkler (1867) is the most universal, and
the continuum approach, which represents the subgrade as an elastic layer on top of a rigid
layer. Due to the disadvantages of both modeling approaches, Worku (2013) recently pre-
sented hybrid models that synthesize mechanical models with the corresponding generalized
1499
continuum formulation in order to physically relate the mechanical parameters to the elastic
properties of the soil.
Although FE analysis can accurately predict the actual complex behavior of concrete
pavements, the required computation time often does not outweigh the additional accuracy
obtained by full 3D pavement models. As a result, a rational choice between complexity/
accuracy and computation time should be made based on the specific problem at hand.
This study presents a two—and three-dimensional finite element (2D/3DFE) static and
dynamic analysis of a rigid concrete pavement system under moving axle loads, using the
commercial software package DIANA. In a first attempt to model the pavement structure,
both simply supported beam (2D) and plate (3D) models on linear viscoelastic Winkler foun-
dation are adopted. A convergence study is carried out in order to determine the element size
of the FE mesh. Subsequently, the numerical models are validated with available solutions in
the literature. A parametric study of various pavement and vehicle characteristics that pre-
sumably affect the dynamic response of concrete pavements, is performed. Finally, accuracy
of results for the different modeling approaches (static/dynamic, 2D/3D) is compared.
2 MODEL DESCRIPTION
c ξ c = ξ km (1)
cr
where ξ = viscous damping ratio; ccr = critical damping coefficient per unit length (Ns/m/m);
k = (generalized) spring stiffness per unit length (N/m/m); and m = (generalized) mass per
unit length (kg/m). A viscous damping ratio ξ = 5% is considered to be a representative value
of soil damping (Taheri et al. 1990, Patil et al. 2013).
Table 1 shows the material properties of the pavement layers. It should be noted that the pur-
pose of this study is to evaluate the need for a dynamic analysis, independent of material prop-
erties. Therefore, the same reference values are considered for both static and dynamic analysis.
The effect of variation of material properties with analysis type will be part of future research.
1500
Table 1. Material properties.
Young’s Damping
Mass density modulus Poisson’s ratio Spring stiffness coefficient
* Value obtained from equation (1) through division by the pavement width.
2.4.3 FE mesh
Linear one-dimensional beam elements with six degrees of freedom per node are used to model
the beam. The equivalent cross-section is defined by application of a predefined I-shape with
its elastic properties equal to that of the concrete slab. An equivalent mass density is calculated
based on the mass of the original structure in order to maintain the mass inertia of the system.
Nodal values of spring stiffness and viscous damping are computed by multiplying the respec-
tive values per unit length by the supporting length of the components. Figure 1 shows the
general setup of the 2D beam model (in case of a mesh with 10 elements).
25 m/s (90 km/h) from the left end to the right end of the beam, and vertical deflections are
examined at the time of load passage through mid-point of the beam.
Based on the eigenfrequencies of the system, a time step of 0.1 ms is adequate for the
transient analysis. Results of the convergence study show that an element size of 0.2 m (50
elements) leads to accurate results.
where y(x,t) = vertical deflection of the beam as function of position x and time t (m);
ρ = mass density (kg/m); A = cross-sectional area (m2); E = modulus of elasticity (Pa);
I = area moment of inertia (m4); c = viscous damping coefficient per unit length (Ns/m/m);
k = spring stiffness per unit length (N/m/m); P = magnitude of point load (N); v = velocity
of point load (m/s); and δ = Dirac-delta function.
Galerkin’s discretization method has proven to be a very powerful tool to solve this kind
of dynamic problem (e.g. Senalp et al. 2010). Based on the method of separation of variables
and the requirement of base functions that satisfy the boundary conditions, the following
form of the vertical beam deflection y(x,t) is assumed:
∞ ⎛ iπ x ⎞ N ⎛ iπ x ⎞
y( x,t ) q (t )sin ⎜ ⎟ ≈ ∑ q (t ) sin
i ⎜ (3)
i ⎝ L ⎠ i ⎝ L ⎟⎠
i =1 i =1
where qi(t) = amplitude coefficient as function of time t; L = beam length (m); and
N = number of base functions (eigenmodes).
After some basic mathematics, the following set of N independent second-order ordinary
differential equations results:
⎡ ⎛ iπ ⎞ ⎤ 4
2P ⎛ iπ vt ⎞
ρ Aq (t ) cq (t
(t ) ⎢ EI
E ⎜ ⎟ + k q (t ) = sin ⎜ , for i = 1, 2, , N (4)
i i ⎢ ⎝ L ⎠ ⎥ i L ⎝ L ⎟⎠
⎣ ⎦
Solving this set of equations for the initial conditions (t = 0) of zero displacement and
velocity at each position x, the amplitude coefficients qi(t) can be substituted back into
equation (3) in order to obtain the vertical deflection of the beam.
The numerical software program MATLAB® is used to implement the aforementioned
procedure. Fourth-order Runge-Kutta numerical integration method is adopted in order to
solve the system of ordinary differential equations (4). A convergence study showed that
accurate results can already be obtained by using 20 eigenmodes and a time step of 1 ms.
Figure 2 shows the time history of the vertical deflection at mid-point of the beam in case of
a point load of 100 kN moving at a velocity of 25 m/s (90 km/h). Excellent agreement can be
observed between the FE model (element size = 0.2 m) and Galerkin’s discretization method.
1502
Figure 2. Time history of vertical deflection at mid-point of the beam.
2.5.2 FE mesh
Linear two-dimensional plate elements with three degrees of freedom per node, i.e. vertical
translation and rotations around the horizontal axes, are used to model the plate. Nodal values
of spring stiffness and viscous damping are computed by multiplying the respective values per
unit area by the supporting area of the components. The general setup of the 3D plate model
is similar to that of the 2D beam model (Figure 1) by extending the model to the y-direction.
∞ ∞ ⎛ iπ x ⎞ ⎛ j y⎞ M N ⎛ iπ x ⎞ ⎛ j y⎞
w( x, y,t ) = ∑ ∑ q (t ) sin sin
i ⎜ ≈ ∑ ∑ q (t ) sin ⎜ siin ⎜ (6)
ij ⎝ L ⎠ ⎝ B ⎟⎠ ij ⎝ L ⎟⎠ ⎝ B ⎟⎠
i =1 j =1 i =1 j =1
where qij(t) = amplitude coefficient as function of time t; L = plate length (m); B = plate width
(m); M, N = number of base functions (eigenmodes) in x- and y-direction, respectively.
After some basic mathematics, the following system of M × N independent second-order
ordinary differential equations results:
⎡ EH ⎧⎪⎛ iπ ⎞ 4 2 2 4⎫ ⎤
⎛ iπ ⎞ ⎛ jπ ⎞ ⎛ jπ ⎞
3
⎪
ρ Hq (t ) cq (t ) + ⎢ ⎨⎜ ⎟ + 2⎜ ⎟ ⎜ ⎟ + ⎜ ⎟ ⎬ + k ⎥ qij (t ) = …
ij ij ⎢12( υ 2 ) ⎪⎝ L ⎠ ⎝ L⎠ ⎝ B ⎠ ⎝ B ⎠ ⎪⎭ ⎥
⎣ ⎩ ⎦ (7)
4P ⎛ iπ v t ⎞ ⎛j y ⎞
…= sin ⎜ P ⎟ sin ⎜ P , for i 1, , N;
⎟ N; j 1 M
LB ⎝ L ⎠ ⎝ B ⎠
Solving this set of equations for the initial conditions (t = 0) of zero displacement and
velocity at each position x and y, the amplitude coefficients qij(t) can be substituted back
into equation (6) in order to obtain the vertical deflection of the plate.
The system of ordinary differential equations (7) is solved again by application of the
numerical software program MATLAB®. A convergence study showed that accurate results
can already be obtained by using 20 eigenmodes in both x—and y-direction and a time step
of 1 ms.
Figure 3 shows the time history of the vertical deflection at mid-point of the plate in case
of a point load of 100 kN moving at a velocity of 25 m/s (90 km/h). Very good agreement
can be observed between the FE model (element size = 502 mm2) and Galerkin’s discretization
method. However, a max. discrepancy of 13% can be observed when the point load passes
through the mid-point of the plate, which results from the concentration of the point load
and the discretization of the system into a FE model.
1504
3 RESULTS AND DISCUSSION
km/h kN /
50-70-90* 80-100*-120 SADT*-TADT-TRDT
*Reference value/configuration.
mm (×10−1) kPa
1505
at mid-point of the beam decrease with increasing value of Young’s modulus and layer thick-
ness, hence bending stiffness. For example, in case of the concrete layer, changing Young’s
modulus from 30 to 35 and 40 GPa leads to a reduced deflection of 1.43% and 2.36%, respec-
tively. Increasing Young’s modulus of the lean concrete base from 15 to 20 and 25 GPa results
in 3.29% and 5.73% lower deflections, respectively. However, the effect of Young’s modulus
of asphalt is almost negligible due to its relatively small thickness when compared to the
concrete layer and lean concrete base. On the other hand, increasing the thickness of the
asphalt layer from 0 to 5 and 6 cm noticeably lowers the deflection by 5.21% and 6.09%,
respectively. Furthermore, spring stiffness has a significant impact on the vertical deflection
of the beam. Consequently, it is very important to correctly estimate the subgrade stiffness,
since small variations in stiffness might significantly alter the pavement response. A higher
damping ratio also results in lower pavement deflections, albeit to a relatively small extent.
The effect of loading velocity on pavement response is negligible, which supports earlier find-
ings by Chatti et al. (1994) who concluded that dynamic analysis is not required for concrete
pavement design, and that quasi-static analysis will generally lead to a conservative design.
Finally, higher loading magnitudes significantly affect pavement deflections, as expected.
With regard to the tensile stresses at mid-bottom of the concrete slab, it is observed that
an increase in Young’s modulus of the concrete layer corresponds to a higher tensile stress. In
contrast, lower tensile stresses are noticed for larger modulus of elasticity in case of the asphalt
intermediate layer and lean concrete base. For example, in case of the asphalt intermediate
layer, changing Young’s modulus from 5 to 10 and 15 GPa leads to 4.14% and 8.09% lower
tensile stresses, respectively. In other words, the asphalt intermediate layer serves as a stress-
reducing layer for the concrete pavement. Increasing the concrete slab thickness significantly
aggravates the tensile stress at the bottom of the concrete slab. However, thicker underlying
layers result in lower tensile stresses. In case of a lean concrete base thickness of 30 cm, even
a compressive stress of −7.42 kPa occurs at mid-bottom of the slab. Furthermore, spring stiff-
ness has a significant impact on the tensile stress in the concrete layer. As mentioned before, it
is of great importance to adequately determine the subgrade stiffness. A higher damping ratio
slightly increases the tensile stress, albeit to a relatively small degree. In contrast to its negligi-
ble influence on vertical deflection, higher vehicle velocity slightly decreases tensile stresses at
the bottom of the concrete slab with 0.29% and 0.64% when altering the velocity from 50 to
70 and 90 km/h, respectively. Lastly, intensifying the loading magnitude substantially increases
the tensile stress at mid-bottom of the concrete pavement, as expected.
Figure 4 demonstrates the effect of axle group configuration (SADT, TADT, TRDT) on
the vertical deflection at mid-point of the beam for the reference design. The axle loads are
assumed to act as independent point loads with a spacing equal to 1250 mm, which corre-
sponds to the axle spacing defined earlier. Figure 4 clearly shows the influence of multiple axle
loading on the vertical deflection of the beam. Changing the axle group configuration from
SADT to TADT and TRDT increases the maximum deflection with 87.97% and 135.79%,
respectively. Furthermore, the point of maximum deflection is shifted to the right, and occurs
when the last axle load within the axle group passes through the mid-point of the beam.
Figure 4. Time history of vertical deflection at mid-point of the beam for different axle groups
(v = 90 km/h).
Figure 5. Comparison of vertical deflection at mid-point of the beam for a quasi-static and dynamic
load (v = 50 km/h).
1507
Figure 6. Transverse vertical deflection profile of the 2D beam and 3D plate model.
Figure 7. Transverse normal stress profile of the 3D plate model compared to the longitudinal normal
stress in the 2D beam model.
4 CONCLUSIONS
In the present study, a two—and three-dimensional finite element static and dynamic analysis
of a rigid concrete pavement system under moving axle loads was performed. Both sim-
ply supported beam (2D) and plate (3D) models on linear viscoelastic Winkler foundation
were adopted in order to represent the pavement structure. Results of the parametric study
showed the insignificance of vehicle velocity (and damping ratio) on pavement responses,
hence dynamic analysis is generally not required for concrete pavement design (under the
given conditions). A comparison of the 2D beam and 3D plate model revealed the impor-
tance of the transverse dimension in the overall pavement response.
5 FUTURE RESEARCH
Future research will examine the effect of different boundary conditions and wave reflection
on the structural response of a concrete pavement system. Furthermore, a full 3D FE model
of a jointed plain concrete pavement (JPCP) will be developed, and the effect of joint behav-
ior, load transfer efficiency and various other parameters will be considered.
1508
REFERENCES
Beskou, N.D. & Theodorakopoulos, D.D. 2011. Dynamic effects of moving loads on road pavements: A
review. Soil Dynamics and Earthquake Engineering 31: 547–567.
Bradbury, R.D. 1938. Reinforced concrete pavements. Washington, D.C.; Wire Reinforcement
Institute.
Chatti, K., Lysmer, J. & Monismith, C.L. 1994. Dynamic finite-element analysis of jointed concrete
pavements. Transportation Research Record 1449: 79–90.
Choubane, B. & Tia, M. 1995. Analysis and verification of thermal-gradient effects on concrete pave-
ment. Journal of Transportation Engineering 121(1): 75–81.
Darestani, M.Y. 2007. Response of concrete pavements under moving vehicular loads and environmental
effects. PhD dissertation.
Darestani, M.Y., Thambiratnam, D.P., Nataatmadja, A & Baweja, D. 2007. Structural response of con-
crete pavements under moving truck loads. Journal of Transportation Engineering 133(12): 670–676.
Gillespie, T.D., Karamihas, S.M., Cebon, D., Sayers, M.W., Nasim, M.A., Hansen, W. & Ehsan, N.
1992. Effect of heavy vehicle characteristics on pavement response and performance. Final report
project 1–25(1) prepared for National Cooperative Highway Research Program, Transportation
Research Board, National Research Council.
Patil, V.A., Sawant, V.A. & Deb, K. 2013. 2-D finite element analysis of rigid pavement considering
dynamic vehicle-pavement interaction effects. Applied Mathematical Modelling 37: 1282–1294.
PCA, 1966. Thickness design for concrete pavements. Portland Cement Association.
Senalp, A.D., Arikoglu, A., Ozkol, I. & Dogan, V.Z. 2010. Dynamic response of finite length euler-
bernoulli beam on linear and nonlinear viscoelastic foundations to a concentrated moving force.
Journal of Mechanical Science and Technology 24(10): 1957–1961.
Taheri, M.R., Zaman, M.M. & Alvappillai, A. 1990. Dynamic response of concrete pavements to mov-
ing aircraft. Applied Mathematical Modelling 14: 562–575.
Van Cauwelaert, F. 2004. Pavement design and evaluation: the required mathematics and applications.
Brussels: Federation of the Belgian Cement Industry.
Westergaard, H.M. 1926. Computation of stresses in concrete roads. Proc., Fifth Annual Meeting of the
Highway Research Board, Washington, D.C., December 3–4, 1925.
Westergaard, H.M. 1927. Theory of concrete pavement design. Proc., Highway Research Board, Part I:
175–181.
Winkler, E. 1867. Die Lehre von der Elastizität und Festigkeit (in German). Prague.
Worku, A. 2013. Calibrated analytical formulas for foundation model parameters. International Journal
of Geomechanics 13(4): 340–347.
Yang, S., Li, S. & Lu, Y. 2010. Investigation on dynamical interaction between a heavy vehicle and road
pavement. Vehicle System Dynamics: International Journal of Vehicle Mechanics and Mobility 48(8):
923–944.
Zokaei-Ashtiani, A., Tirado, C., Carrasco, C. & Nazarian, S. 2015. Impact of different approaches
to modelling rigid pavement base layers on slab curling stresses. International Journal of Pavement
Engineering: 1–9.
1509
Safety
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Emad Kassem
University of Idaho, Idaho, USA
1 BACKGROUND
Several test sections were constructed at the Riverside Campus of Texas A&M University as
a part of a research project to evaluate the effect of several mixture and construction factors
on compaction of asphalt mixtures (Kassem et al. 2012). Each test section was divided into
sub-test sections that were compacted using different compaction methods. In this study,
the researchers measured the surface frictional characteristics on selected sub-test sections.
Figure 1 shows the schematic layout and rolling pattern for the selected sub-test sections.
1514
Three main test sections were selected for surface frictional measurements. Test Section No.
1 was constructed using Type C mix with binder PG 76–22, Test Section No. 2 was con-
structed using Type D mix with binder PG 64–22, and Test Section No. 3 included Type D
mix with PG 64–22 and warm mix additive (foaming). Both Type C and Type D mixtures
are dense graded mixtures with nominal maximum aggregate sizes of 12.5 mm and 9.5 mm,
respectively. Each test section was divided into two sub-test sections as shown in Figure 1.
One sub-test section was compacted using a steel wheel roller (Figure 2) that operated at a
static mode, while the other one is compacted using the same steel wheel roller but operated
at a vibratory mode. Each sub-test section was 2.7 m (9 ft) wide and 36.6 m (120) ft long,
and the lift thickness was about 50 mm (2 inches). Each sub-test section was then divided
into several smaller strips (approximately 9 m [30 ft] long), which were compacted using a
varying number of roller passes as shown in Figure 1. Only 2.1 m (7 ft) from one edge of the
test section was compacted using the 2.1 m (7 ft) steel wheel roller as shown in Figure 2. The
compaction was performed by driving the roller straight back and forth.
The researchers used the CTMeter and DFT to study the surface frictional characteristics of the
test sections. The CTMeter was used to measure the Mean Profile Depth (MPD), while the
DFT was used to measure the coefficient of friction. The CTMeter is a convenient tool for
1515
quantifying pavement macrotexture profiles in the laboratory and in the field. It uses a laser-
displacement sensor that is mounted on an arm that rotates in a circular track of 284 mm diam-
eter. Figure 3a shows the CTMeter during field testing. This device is designed to measure the
MPD of the same circular tack that is measured using the DFT. The measured profile is divided
into eight segments with a total number of 1024 data samples in one full rotation. The data are
processed and analyzed automatically using the CTMeter software, and the MPD is calculated
according to the ASTM E1845standard (ASTM 2009a). Microtexture of pavement surface is
estimated indirectly by measuring the surface friction as a function of sliding speed. The DFT
was used to measure the coefficient of friction at low speeds. The DFT (Figure 4a) consists of
a disk that rotates on a horizontal plane parallel to the pavement surface (ASTM 2009b). Three
rubber sliders are attached to the bottom on the disk. The disk is brought to the desired rota-
tional velocity and then it is lowered to achieve full contact between the rubber sliders and wet
pavement surface. The coefficient friction is recorded as the speed of the rotating disk gradually
decreases due to the friction between the rubber pads and pavement surface until the disk stops
(ASTM 2009b). The pavement microtexture is quantified by the value of the coefficient of fric-
tion at 20 km/hr (DFT20).
The researchers quantified the skid resistance by calculating the International Friction
Index (IFI). The IFI is a universal method for harmonization of different friction measur-
ing equipment to a common calibrated index (ASTM 2007, Henry 2000). The ASTM E1960
covers the calculation of the IFI as a function of pavement macrotexture and wet pavement
friction (ASTM 2007). The IFI is calculated using the MPD measured using the CTMeter
and DFT20 measured using the DFT as given in Equations (1) and (2).
⎛ −40 ⎞
IFI
F 0.081 0.732DFT
T20 exp ⎜ ⎟ (1)
⎝ Sp ⎠
SP 14 2 89 M
MPD (2)
1516
Figure 4. MPD vs. number of passes for three test sections; (a) Type C HMA, (b) Type D HMA, and
(b) Type D WMA.
static and vibratory rollers. However, the change in the MPD for the test sections compacted
using the vibratory roller was steeper compared to the corresponding test sections compacted
using the static roller. The MPD is a measure of the macrotexture of pavement surface.
A smooth surface yields a low MPD, while a rough surface yields a high MPD. The vibratory
roller was found to yield a smoother surface than the static roller. In addition, the compac-
tion process had less impact on the coefficient of friction at 20 km/hr (DFT20) for a given
asphalt mixture, as shown in Table 1. The DFT20 is considered an indication of the pavement
microtexture; therefore, it is concluded that compaction affects macrotexture more that it
does for the microstructure.
Figure 5 presents the IFI of the different sections. The IFI decreased with an increase
in the number of passes for all the test sections; however, the change in the IFI values were
higher for the sub-test sections compacted using the vibratory roller compared to the ones
compacted using the static roller in each test section. The vibratory roller was found to yield
a smoother surface than the static roller with less IFI. These results demonstrate that using
a static roller would be a better choice during compaction if the desired density can be met
since the static roller would produce a pavement surface with better skid resistance.
1517
Table 1. MPD, DFT20 Measurements, and IFI Values.
The WMA test section had higher IFI values compared to the HMA test section, especially
when the static roller was used (Figure 5). The WMA test section had higher coefficient of
friction at 20 km/hr (DFT20) at the corresponding compaction levels compared to HMA.
Hence, one can postulate that the WMA test section had higher microtexture right after the
construction compared to HMA. However, it is expected that both HMA and WMA would
have comparable microtexture once the asphalt film on the surface is removed under traffic,
1518
Figure 6. Coefficient of friction at 60 km/h vs. MPD.
because both WMA and HMA used the same aggregate type. Also, the WMA test section
had higher MPD when the static roller was used in compaction compared to HMA (Table 1),
while both sections (HMA and WMA) had comparable MPD when the vibratory roller was
used in compaction. It should be noted that the WMA test section was compacted about
11°C lower than the compaction temperature of the HMA. Higher MPD would result in
coarser microtexture, and hence higher IFI.
The IFI results in Figure 5 show that the HMA Type C test section had higher IFI com-
pared to the HMA Type D. Type C and Type D mixtures have nominal maximum aggregate
sizes of 12.5 mm and 9.5 mm, respectively. One would expect coarse mixture would produce
coarse macrotexture and hence higher MPD and IFI.
Finally the researchers found a fair relationship between the MPD and the coefficients
of friction at 60 km/h measured using the DFT, as illustrated in Figure 6. Such relationship
suggested that better traction would be expected between tires and coarse pavement surface
compared to finer pavement surface with low MPD.
The researchers recovered several field cores from each strip for density measurements and
X-ray CT scanning. The X-ray CT was used to examine the air void distribution across the
depth. Two field cores were scanned using the X-ray CT. These cores were taken from the
middle of the mat from test strips that were compacted using different numbers of roller
passes (one pass, three passes, and seven passes). The X-ray CT is a nondestructive tech-
nique used to visualize the internal structure of opaque objects. In this test, a test specimen
is placed between the X-ray source and linear detector. The source sends out X-rays that
penetrate the test specimen and reach the detector on the opposite side. An X-ray CT image
is obtained after the test specimen completes a full rotation with respect to its center. The test
specimen moves upward to take images at different locations across the depth. The research-
ers used Image-Pro® Plus software (Image-Pro 1999) to process the images and analyze the
air void distribution. They developed macros to calculate the percent of air voids and the
average radius of air voids across the depth of a specimen.
Figure 7 shows the air void distribution across the depth at different numbers of passes
for the extracted field cores. The results showed clearly that the vibratory roller was more
effective in reducing the air voids with the number of passes compared to the static roller.
More importantly, the percent air voids toward the top surface of the test sections compacted
using the static roller did not change with the number of passes but decreased significantly
when the vibratory roller was used. There was a higher percent air voids at the top in the field
cores compacted using the static roller compared to the counterparts compacted using the
vibratory roller. The higher percent air voids toward the surface would increase the MPD and
1519
Figure 7. Air void distribution across the depth of field cores.
hence yield higher IFI. This explains the reason that test sections compacted using the static
roller had higher IFI that decreased slightly with the number of passes compared to those
compacted using the vibratory roller. Higher percent air voids at the surface would contrib-
ute to higher surface macrotexture. Furthermore, Figure 7 shows that the WMA test sections
had slightly higher percent air voids at the top compared to the counterparts of the HMA
especially in the sections compacted using static roller which is consistent with the MPD
measurements in Table 1. Overall, the findings of the X-ray analysis explained and correlated
well with the MPD and skid measurements.
The researchers studied the uniformity of the air void distribution in the top half of field
cores (top 2.5 mm) by calculating an index called the uniformity index (UI). The UI is calcu-
lated using Equation (3) (Kassem et al. 2008), as follows:
• Plot percent air voids f(x) against the core depth x.
• Fit a fourth-order polynomial for f(x).
• Calculate the derivate f’(x) of the function f(x).
• Calculate the UI using Equation (3).
b
1
UI =
b a ∫a
[f x ]2dx (3)
1520
Figure 8. Uniformity index at the top half of field cores.
Kassem et al. (2008) found that the fourth-order polynomial to fit the percent air voids
functions very well. The UI is equal to zero for a straight-line function representing uniform
distribution, and it increases with an increase in non-uniformity. The integration limits depend
on the thickness over which the analysis is conducted for the top half (a = 0, b = h/2).
Figure 8 shows the UI for all evaluated test field cores. The UI for test sections compacted
using the vibratory roller was less than the ones compacted using the static roller. This means
that the air voids at the top half of the pavement were more uniform in the test section com-
pacted using the vibratory roller compared to the static roller. Also, the UI decreased with the
number of the passes when the vibratory roller was used, while the UI increased with the number
of passes when the static roller was employed, which means that the air void distribution at the
top half of the pavement surface became more non-uniform. In addition, it can be seen that the
UI of WMA Type D mixture is less than the UI of HMA Type D mixture which indicates that
the air void distribution toward the surface is more uniform in the WMA test section.
6 CONCLUSIONS
This study evaluated the effect of compaction method and level of compaction on the
skid resistance of asphalt pavements. The main findings for this study are summarized as
follows:
• The MPD decreased with the number of passes for the test sections compacted using
both static and vibratory rollers due to densification of asphalt mixtures. Test sections
compacted using the vibratory roller had a steeper change in MPD compared to those
compacted using the static roller.
• The IFI decreased with the number of passes for all test sections. However, the change in
IFI with the number of passes was higher for the sub-test sections compacted using the
vibratory roller compared to the ones compacted using the static roller. The vibratory
roller was found to yield a smoother surface than the static roller.
• The results shows that the static roller would be a better choice during compaction if the
required density can be achieved since the static roller would produce a pavement surface
with better skid resistance.
• Coarse mixtures had better skid resistance compared to finer mixtures. In addition, WMA
had higher IFI compared to HMA.
• The X-ray analysis of the void distribution toward the surface explained and correlated
well with the mean profile depth and skid measurements.
• Test sections compacted using the vibratory roller had more uniform air void distribution
toward the surface compared to test sections compacted using the static roller.
1521
• The researchers recommend conducting further field testing to verify the outcomes of this
study to optimize the compaction of asphalt mixtures to provide an acceptable level of
skid resistance.
REFERENCES
ASTM 2007. ASTM Designation E1960: Standard Practice for Calculating International Friction Index
of a Pavement. American Society for Testing and Materials (ASTM), Philadelphia, Pa.
ASTM 2009a. ASTM Designation E2157: Standard Test Method for Measuring Pavement Macrotex-
ture Properties Using the Circular Track Meter. ASTM, Philadelphia, Pa.
ASTM 2009b. ASTM Designation E1911: Standard Test Method for Measuring Paved Surface Fric-
tional Properties Using the Dynamic Friction Tester, American Society for Testing and Materials,
Philadelphia, Pa.
Corps of Engineers 2000. Hot-Mix Asphalt Paving Handbook 2000, AC 150/5370–14 A, US Army
Corps of Engineers, Washington, D.C.
Dahir, S.H. 1979. A Review of Aggregate Selection Criteria for Improved Wear Resistance and Skid
Resistance of Bituminous Surfaces, Journal of Testing and Evaluation, JTEVA, 7(5), 254–253.
Davis, R.M., Flintsch, G.W., Al-Qadi, I.L.K., and McGhee, K. 2002. Effect of Wearing Surface Char-
acteristics on Measured Pavement Skid Resistance and Texture. Presented at 81st Transportation
Research Board Annual Meeting, Washington, D.C.
Henry, J.J. 2000. NCHRP Synthesis of Highway Practice 291: Evaluation of Pavement Friction Charac-
teristics. Transportation Research Board, National Research Council, Washington, D.C.
Image-Pro Plus, Media Cybernetics, L.P. Version 4.1, Georgia, MD. (1999)
Ivey, D.L., Bullard, D.L., Lock, J.R., and Griffin, L.I. 1992. Texas Skid Initiated Accident Reduction
Program: Final Report, Report 910–1F, TTI: 2–18–89/910, TX-92/910–1F, Texas Transportation
Institute, College Station, Texas
Kandhal, P.S., and Parker, F. Jr. 1998. Aggregate Tests Related to Asphalt Concrete Performance in Pave-
ments, NCHRP Report 405, National Cooperative Highway Research Program, National Academy
Press, Washington, D.C.
Kassem, E., Awed, A., Masad, E., and Little, D. 2013. Development of a Predictive Model for Skid Loss
of Asphalt Pavements, Journal of the Transportation Research Board, 2372, 83–96
Kassem, E., Masad, E., Chowdhury, A., and Claros, G. 2008. Influence of Field Compaction Pattern
on Asphalt Pavement Uniformity, Journal of the Association of Asphalt Paving Technologists, AAPT,
77, 257–298.
Kassem, E., Scullion, T., Masad. E., and Chowdhury, A. 2012. Comprehensive Evaluation of Com-
paction of Asphalt Pavements and a Practical Approach for Density Predictions, Transportation
Research Record: Journal of the Transportation Research Board, 2268, 98–107.
Leiva, F., West, R.C. 2008. Relationship between Laboratory Measured Characteristics of HMA and
Field Compactability, Journal of the Association of Asphalt Paving Technologists, 77, 83–219.
Peterson, B., Mahboub, K., Anderson, M., Masad, E., and Tashman, L. 2004. Comparing Superpave
Gyratory Compactor Data to Field Cores, Journal of Materials in Civil Engineering, ASCE, 16(1),
78–83.
Roberts, F.L., Kandhal, P.S., Brown, E.R., Lee, D., and Kennedy, T.W. 1996. Hot Mix Asphalt Materi-
als, Mixture Design and Construction, National Asphalt Pavement Association, Research and Educa-
tion Foundation, Lanham, Maryland.
Tashman, L., Masad, E., Peterson, B., and Saleh, H. 2001. Internal Structure Analysis of Asphalt Mixes
to Improve the Simulation of Superpave Gyratory Compaction to Field Conditions, Journal of the
Association of Asphalt Paving Technologists, 70, 605–645.
1522
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Pavement ruts present a potential road safety hazard that can lead to skid-
ding or hydroplaning accidents in wet weather. Ruts collect water during wet weather and
cause pavement skid resistance to fall. Although the potential danger of driving on flooded
pavement ruts is well recognized, the magnitudes of reduced skid resistance and the cor-
responding driving risks on rutted pavements during rainy days have not been well stud-
ied and documented. This is mainly due to the complexity of analyzing the skid resistance
characteristics of wet pavements. Using a recently developed theoretical skid resistance and
hydroplaning simulation model, this paper proposes a quantitative procedure of evaluating
the wet-weather driving risks of rutted highway pavements for different rut depths under dif-
ferent surface runoff flow conditions and different vehicle speeds. Driving risks are evaluated
by comparing traffic speeds with the hydroplaning speed and the maximum vehicle speed
allowed based on safe braking distance consideration.
1 INTRODUCTION
Wet-weather road accidents are an important component of total road accidents that is
closely related to the surface characteristics of road pavements. Traffic safety studies from
different parts of the world suggest that approximately 20% of all road accidents occur in wet
weather conditions (Ivan et al. 2012, Mayora and Pina 2009, McGovern et al. 2011, Murad
and Abaza 2006). Researchers and practitioners generally believe that low skid resistance
and hydroplaning speed (i.e. vehicle speed at which hydroplaning occurs) are the two most
significant contributing factors of wet-weather road accidents (Alvarez et al. 2006, Ivan et al.
2012, McGovern et al. 2011).
Traditionally, all studies related to wet-pavement skid resistance and hydroplaning con-
sider a straight forward case of plane pavement surface with known longitudinal and cross
slopes, without taking into account the presence of ruts along the two wheel paths of pave-
ments. This is unsatisfactory because all asphalt pavements will rut under repeated traffic
loading. There are ruts in practically all in-service asphalt pavements, with the exception of
newly paved roads and roads that carry very light traffic.
It is interesting to note that although the driving risks associated with flooded pavement
ruts is well recognized, the magnitudes of the reduced skid resistance and the corresponding
driving risks on rutted pavements during rainy days have not been well studied and docu-
mented. This is likely due to the complexity of analyzing the skid resistance characteris-
tics of wet pavements. This study aims to examine the extent to which the presence of ruts
would affect the driving risks on a highway pavement. Using a recently developed theoretical
skid resistance simulation model, this paper proposes an analytical procedure to evaluate the
skid resistance and hydroplaning speed on rutted highway pavements for different rut depths
under different surface runoff flow conditions and different vehicle speeds. The driving risks
on rutted pavements under two different rainy wet-weather conditions are considered: a case
of flooded pavement ruts with no flowing water on the pavement surface, and another case
of flooded rutted pavement with a film of runoff water flowing on the pavement surface.
1523
2 BASIS OF DRIVING RISKS ANALYSIS
The tire-pavement friction available to a vehicle decreases when the pavement becomes wet.
The reduction in tire-pavement friction leads to a loss of skid resistance. The magnitude of
reduction in skid resistance increases with the thickness of water on the pavement surface
and the vehicle speed. Two quantitative measures can be used to assess the driving risks asso-
ciated with reduced skid resistance: the increase in vehicle braking distance, and the decrease
in hydroplaning speed.
The imposition of braking distance control is related to safety requirements of highway
geometric design. A maximum braking distance is imposed to ensure that a vehicle will be
able to stop within a safe distance when required, such as in front of a stop sign or a rail-
road crossing, or in times of emergence, such as when there is an unexpected obstacle on the
roadway or a sudden lane intrusion by other vehicles or objects. To meet this requirement, a
minimum safe skid resistance is usually specified by highway agencies.
Hydroplaning is another safety concern of weather driving. When a vehicle hydroplanes
on a wet pavement, the driver will lose steering and braking control because the tires are sup-
ported on a layer of water and the frictional resistance available to the tires is close to zero.
This situation happens when the thickness of pavement surface water is sufficiently thick,
and the vehicle speed is sufficiently high. Under this situation, the hydrodynamic uplift force
acting on the tires causes the tires to be separated from the pavement surface (Horne and
Dreher 1963, Browne 1975). The vehicle speed at which hydroplaning occurs is known as the
hydroplaning speed. Studies have found that hydroplaning speed decreases (i.e. hydroplaning
potential increases) as the water film thickness on pavement surface increases. That is, vehicle
hydroplaning risk increases with water film thickness.
In the case of a rutted pavement, the ruts will be filled with water during rainy days. The
water depths along the ruts are thicker than the non-rutted portion of the pavement. Since
ruts are formed along the wheel paths of a pavement lane, the water film thickness along the
ruts have a direct effect on vehicle hydroplaning risk. Computation of vehicle hydroplaning
speed on a rutted pavement will lead to under-estimation of the safety risk if the presence of
ruts is ignored in the evaluation of hydroplaning speed. It is therefore necessary to examine
the safety risk implications of pavement ruts.
The following two conditions of safety risk analysis may be considered:
Wet-Weather Condition I: Ruts are still filled with water but the rainfall has stopped and
there is no flowing water on the pavement surface. The braking distance and hydroplaning
speed can be calculated by considering a water depth equal to the rut depth along the wheel
paths.
Wet-Weather Condition II: Ruts are flooded with water during a rainfall, and there is sur-
face runoff flowing on the pavement surface. Under this condition, the braking distance and
hydroplaning speed can be calculated by considering a water depth equal to the sum of the
rut depth and the thickness of flowing water along the wheel path in question.
The driving risks associated with Wet-Weather Conditions I and II are evaluated against the
respective distributions of traffic speeds under the two conditions. There are some differences
in terms of the traffic speeds under the two conditions. Under Wet-Weather Condition II, the
vehicle speeds are expected to be lower than those under Wet-Weather Condition I, because
the visibility and driving conditions would be more unfavorable to the drivers in the former
case. However, the former would also have shorter braking distance and lower hydroplaning
speed due to the thicker water depth. Therefore, it is necessary to examine both conditions to
identify their respective safety risks for the purpose of ensuring wet-weather driving safety.
The procedure involved in the evaluation of the wet-weather driving risks on a rutted pave-
ment is illustrated in the flow diagram of Figure 1. The evaluation requires the determination
of the following five main elements:
1524
Figure 1. Flow diagram of wet-weather driving risk analysis.
1525
where tw is the water film thickness (mm), n the Manning’s roughness coefficient, L the drain-
age path length (m), i the rainfall intensity (mm/h), f the infiltration rate into the pavement
(mm/h), α a model constant, S the slope of drainage path (m/m), and MTD the mean texture
depth (mm). The Manning's roughness coefficient accounts for the hydraulic effect of pave-
ment surface on surface water flow and its value is surface dependent. Anderson et al. (1998)
have identified the Manning's coefficients for various pavement types in their experimental
studies.
1526
The simulation begins with a low vehicle sliding speed which is gradually increased in the
analysis. As the sliding speed is increased, the hydrodynamic uplift force acting on each tire
also increases. This process is continued until the uplift force is equal to the tire load. The
vehicle sliding speed at which this occurs is the hydroplaning speed.
1527
tions varied with rainfall intensity, road geometric design, traffic flow volume, and driving
culture of the motorists (Unrau and Andrey 2006, Perrin et al. 2001, Hranac et al. 2006,
Brilon & Ponzlet 1996). It is noted that explicit tire-pavement friction measurements have not
been made or referred to these driving speed reduction effects of rainfall.
In view of the number of complex factors involved, it is likely that wet-weather driving
speed reductions would vary from one highway to another, even within the same road net-
work, not to mention the differences from one region to another, or one country to another.
As there does not exist any reliable procedure to predict the wet-weather traffic speed distri-
butions of a road section, it is necessary to rely on actual field survey measurements to estab-
lish such information for the purpose of evaluating driving risks associated with pavement
ruts in wet weather. Ideally, wet-weather traffic speed distributions for a road section could
be established for different rainfall conditions, including the case of Wet-Weather Condition
I (see Section 2) where the rain has stopped but pavement ruts are still filled with water.
3.5 Driving risks analysis associated with hydroplaning and braking distance
Having established (i) the hydroplaning speed and (ii) the vehicle speed-braking distance rela-
tionship of a road section for different water film thicknesses, as well as the wet-weather
speed reduction distributions for different rainfall conditions, wet-weather driving risks asso-
ciated with hydroplaning and emergency stopping distance, respectively, can be evaluated.
These two forms of driving risks can be represented by the proportion or percentage of vehi-
cles traveling at a speed higher than the threshold hydroplaning speed, and the proportion
or percentage of vehicles traveling at a speed that will have stopping distance longer than the
safe stopping sight distance, respectively.
1528
e. Set driving speed as the threshold hydroplaning speed, and from the lower plot of Fig-
ure 3 calculate the driving risk as the percentage of vehicle speeds higher than the thresh-
old hydroplaning speed. The driving risk is represented by the shaded portion under the
traffic speed distribution curve.
In the case of rutted pavement in Wet-Weather Condition II, the driving risk analysis is
more complex than that of Wet-Weather Condition I. The main complications arise from the
following two aspects: (i) The surface water depth on pavement ruts equals to the sum of rut
depth and pavement surface runoff flow depth, and is a function of rainfall intensity, pave-
ment cross slope, and pavement surface texture properties; (ii) The traffic speed distribution
is known to vary with rainfall intensity. In general, traffic flow speeds decrease as the rainfall
intensity increases. While the determination of driving risk following essentially the same
general concept depicted in Figure 3, the following more elaborate steps are involved in the
analysis:
a. For the given rainfall intensity, calculate the runoff water film thickness and add this
thickness to rut depth to obtain the total water depth for hydroplaning speed calculation.
b. Perform hydroplaning speed simulation analysis for different rut depths, and establish the
relationship between threshold hydroplaning speed and total water depth.
b. Plot total water depth against threshold hydroplaning speed (i.e. the upper plot of Fig. 3
with total water depth equal to the sum of runoff water film thickness and rut depth).
c. Conduct traffic speed survey for the pavement section at Wet-Weather Condition II, and
plot the traffic speed distribution as shown in the lower plot of Figure 3. It should be
noted that, unlike Wet-Weather Condition I, traffic speed distribution in Wet-Weather
Condition II is known to change with rainfall condition.
d. For a given rutted pavement, identify the rainfall intensity for driving risk analysis, meas-
ure the rut depth to obtain the water depth, then determine the threshold hydroplaning
speed from the upper plot of Figure 3.
e. Set driving speed as the threshold hydroplaning speed, and from the lower plot of Fig-
ure 3 calculate the driving risk as the percentage of vehicle speeds higher than the thresh-
old hydroplaning speed. The driving risk is represented by the shaded portion under the
traffic speed distribution curve.
1529
3.5.2 Determination of driving risk associated with braking distance
For a given pavement section, the driving risk associated with braking distance is determined
with reference to the safe stopping sight distance for the pavement section. Most highway
agencies specify a minimum stopping sight distance which is equal to the sum of reaction
distance and braking distance (AASHTO 2011). Reaction distance is the distance traveled by
the vehicle concerned during the braking reaction time by the driver. The braking reaction
time is commonly assigned a fixed value in highway safety analysis for stopping sight distance.
The braking distance corresponding to the specified stopping sight distance is the maximum
braking distance allowed. In wet weather, braking distance is affected by two main factors:
skid resistance properties of the pavement, and vehicle speed. for a pavement section with
known skid resistance properties, one can identify the maximum vehicle speed beyond which
the maximum braking distance control would be violated. This maximum vehicle speed is
used in calculation of the driving risk associated with the safe braking distance requirement.
The following steps are involved in the analysis:
a. For the pavement section analyzed, identify the safe stopping distance specified and the
corresponding maximum braking distance allowed.
b. Calculate the total water depths for the weather condition to be analyzed (i.e. Wet-Weather
Condition I or II), according to the procedure described in Section 3.5.1 for hydroplaning
driving risk calculation.
c. Perform skid resistance simulation analysis for different rut depths, and establish the rela-
tionship between skid resistance and water depth.
d. For each water depth, apply Equations 2 and 3 to calculate the maximum vehicle speed
that has braking distance equal to the maximum braking distance allowed.
e. Knowing the maximum braking distance for each water depth, and the traffic speed dis-
tribution for the weather condition analyzed, the driving risk associated with braking dis-
tance is calculated as the percentage of vehicle speeds higher than the maximum vehicle
speed allowed.
The final steps of driving risk determination described above are similar to that illustrated
in Figure 3, but with a different upper plot. The upper plot of water depth versus threshold
hydroplaning speed should now be replaced with a plot of water depth versus maximum
vehicle speed allowed based on braking distance consideration.
This paper has presented a quantitative procedure for evaluating the rain-related wet-weather
driving risks of rutted pavements by considering two wet-weather driving conditions. Wet-
Weather Condition I considers the case where pavement ruts are still filled with water after
rainfall has stopped and there is no flowing water on the pavement surface; while in the
case of Wet-Weather Condition II, ruts are flooded with water during a rainfall, and there
is surface runoff flowing on the pavement surface. Driving risks on a pavement section are
evaluated by comparing traffic speeds with the hydroplaning speed and the maximum vehicle
speed allowed based on safe braking distance consideration, respectively.
The driving risk associated with hydroplaning requires the determination of hydroplaning
speed of a pavement as a function of surface water depth, while the driving risk associated
with braking distance requires the development of the relationship between tire-pavement
skid resistance and surface water depth. These two requirements are met by employing a
three-dimensional finite-element skid resistance and hydroplaning simulation model devel-
oped earlier by the authors. The detailed steps involved in the evaluation of the rainy-day
wet-weather driving risks associated with hydroplaning and braking distance, respectively,
are described. The procedure presented offers highway engineers a quantitative method of
evaluating the wet-weather driving risks on rutted pavements, equipping them with a useful
tool to enhance their ability in the management of wet-weather driving safety.
1530
REFERENCES
AASHTO. 2011. A Policy on Geometric Design of Highways and Streets, 6th Edition, American Associa-
tion of State Highway and Transportation Officials, Washington, D.C.
ADINA R&D Inc. 2012. ADINA Theory and Modeling Guide Volume III: ADINA CFD and FSI,
ADINA R&D Inc., Watertown, Massachusetts.
Alvarez, A.E., Martin, A.E., Estakhri, C.K., Button, J.W., Glover, C.J. & Jung, S.H. 2006. Synthesis of
Current Practice on the Design, Construction, and Maintenance of Porous Friction Courses. FHWA
Report 06/0–5262-1, Texas Transportation Institute, Austin, Texas.
Anderson, D.A., Huebner, R.S., Reed, J.R., Warner, J.C. & Henry, J.J. 1998. Improved surface drainage
of pavement. National Cooperative Highway Research Program Web Document 16, Transportation
Research Board, National Research Council.
ASTM (2006). E274 Standard Test Method for Skid Resistance of Paved Surfaces Using a Full-Scale
Tire, ASTM International, West Conshohocken, PA.
ASTM (2008). E524-08 Standard Specification for Smooth Tire for Pavement Skid-Resistance Tests.
ASTM International, West Conshohocken, PA.
Brilon, W., Ponzlet, M. 1996. Variability of Speed-Flow Relationships on German Autobahns, Trans-
portation Research Record 1555: 91–98.
Browne, A.L. 1975. Mathematical analysis for pneumatic tire hydroplaning. In J.G. Rose, J.G. (ed.), Sur-
face Texture verses Skidding: Measurements, Frictional Aspects, and Safety Features of Tire-Pavement
Interaction. American Society for Testing and Materials, West Conshohochen, Philadelphia, 75–94.
Fwa, T.F. & Ong G.P. 2008. Wet-pavement hydroplaning risk and skid resistance: analysis”. Journal of
Transportation Engineering, 134(5): 182–190.
Hinze, J.O. 1975. Turbulence. Second edition, McGraw-Hill, New York.
Horne, W.B. & Dreher, R.C. 1963. Phenomena of pneumatic tire hydroplaning. NASA Technical Note
D-2056, National Aeronautics and Space Administration, Washinton D.C.
Hranac, R., Sterzin, E., Krechmer, D., Rakha, H. & Farzaneh, M. 2006. Empirical Studies on Traffic
Flow in Inclement Weather. Federal Highway Administration, Washington DC.
Ibrahim, A.T. & Hall, F.L. 1994. Effect of Adverse Weather Conditions on Speed-Flow-Occupancy
Relationships. Transportation Research Record 1427: 184–191.
Ivan, J.N., Ravishanker, N., Jackson, E., Aronov, B. & Guo, S. 2012. A statistical analysis of the effect
of wet-pavement friction on highway traffic safety. Journal of Transportation Safety & Security, 4(2):
116–136.
Kyte, M., Khatib, Z., Shannon, P., & Kitchener, F. 2001. The Effect of Weather on Free-Flow Speed,
Transportation Research Board Annual Meeting, Paper No. 01-3280, Washington, D.C.
Mayora, P.J.M. & Pina, J.R. 2009. An assessment of the skid resistance effect on traffic safety under
wet-pavement conditions. Accident Analysis and Prevention, 41(4): 881–886.
McGovern, C., Rusch, P. & Noyce, D.A. 2011. State Practices to Reduce Wet Weather Skidding Crashes,
Publication FHWA-SA-11-21, federal Highway Administration, U.S. Department of Transporta-
tion, Washington D.C.
Murad, M.M. & Abaza, K.A. 2006. Pavement friction in a program to reduce wet weather traffic acci-
dents at the network level. Transportation Research Record, 1949: 126–136.
Ong, G.P. & Fwa, T.F. 2007. Wet-pavement hydroplaning risk and skid resistance: modeling. Journal of
Transportation Engineering, 133(10): 590–598.
Perrin, P. Jr., Martin, P.T & Hansen, B.G. 2001. Modifying signal timing during inclement weather.
Transportation Research Board 80th Annual Meeting, January 7–11, 2001 Washington, D.C.
Unrau, D. & Andrey, J. 2006. Driver Response to Rainfall on Urban Expressways. Transportation
Research Record, 1980: 24–30.
1531
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Improving drivers visibility in night time conditions is vital. Night-time vis-
ibility represents one of the most important features of road safety. Within such context, the
use of photoluminescent road markings could represent an enhancement with regard to road
safety. Consequently, the objective of the investigation here described was confined into the
analysis of photoluminescent paints by referring to dense-graded and open-graded friction
courses. Measurements, based on photometry technique, were carried out in the laboratory.
Cores extracted from the surface layer of known pavements were used. Transitory effects
(charge and discharge) and decay phenomena were investigated and modelled as a function
of treatment and pavement characteristics (paint quantity, hot mix asphalt volumetrics, etc.).
The results highlight that the photoluminescent performance depends on the volumetric
characteristics of bituminous mixtures. Results can benefit both researchers and practition-
ers and can allow optimising painting treatments for different bituminous mixtures.
1 INTRODUCTION
Improving drivers visibility conditions is vital (Bosurgi et al, 2015). Road pavement markings
are crucial when concerning night-time visibility. In fact, they provide guidance and safety for
drivers. Their most important property is the luminance since it allows improving the view of
any device or object during the night.
Luminance is the intensity of brightness and is measured in candela per unit area of a
surface (cd/m2). The basics which allows estimating photoluminescent effects are below
explained (see also Table. 1)
The luminance is defined by the derivative:
d 2ΦV
LV = (1)
dA d Ω cosθ
where LV is the luminance (cd/m2), ΦV is the luminous flux or luminous power (lm), θ is the
angle between the surface normal and the specified direction, A is the area of the surface
(m2), and Ω is the solid angle (sr).
The luminous flux is defined as follows:
ΦV =V
V W (2)
where V is the human eye spectral sensitivity (lm/Watt) and W is the luminous power (Watt).
The illuminance is the total luminous flux incident on a surface, per unit area:
dΦ
E= (3)
dS
1533
When visual performance analyses are carried out, the luminance contrast (C) can be con-
sidered, as in (Bullough et al, 2014). C is defined by the formula:
C (
Lt − Lb / max Lb,Lt ) (4)
where Lt is the luminance (in cd/m2) of the target or object to be seen and Lb (in cd/m2) is the
luminance of the target’s or object’s background.
Overall, the luminance (L) of an object can be estimated by the illuminance (E) on an object
and its reflectance (ρ: 0 is perfectly black; 1: perfectly white) using the following formula:
L Eρ / π (5)
L (cd/m2) Luminance (L), the intensity of brightness and is measured in candela See equation 1
per unit area of a surface and 5
ΦV (lm) Luminous flux or luminous power, measured in lumen, lm See equation 2
E (lx) Illuminance (E), the measure of how much luminous flux is spread over See equations 3
a given area, or luminous power incident on a surface. It is measured in lx. and 5
C Luminance contrast (dimensionless) See equation 4
Symbols. C: luminance contrast; lm: lumen; cd: candela; sr: steradian or square radian (SI unit of
solid angle; a full sphere has a solid angle of 4π steradians); m: meter; lx: lux.
Note. 1 lm = 1 cd⋅sr; 1 lx = 1 lm/m2 = 1 cd⋅sr/m2.
Turnpenny and Crawford, 2014 Road marking made with glow-in-the-dark paint (with
strontium aluminate pigments).
Bacero et al., 2014 Luminescent paint with strontium aluminate on Rural and
Unilluminated Roads.
Bullough et al., 2014 Photoluminescent materials for signage, marking and delineation
applications in work zones. In particular, tested materials were:
yellow photoluminescent paint (8” square), green
photoluminescent tape (1” width), red photoluminescent
tape (2” width), orange photoluminescent tape (2” width),
yellow photoluminescent tape (2” width).
NYC, 2007 Photoluminescent green safety paint system for line markings.
Botterman and Smet, 2015 Use of persistent phosphors (a specific type of luminescent
materials) for glow-in-the-dark road marks and in combination
with solar cells and photo catalytic processes.
Giuliani and Autelitano, 2014 Use of afterglow photoluminescent pigments for road
wearing courses.
Mishra et al., 2009 Strontium aluminate/polymer composite samples were
attached to the surface of asphalt slabs.
1534
Many materials can be used for marking road pavements, such as paint, tape and spray.
Table 2 summarises the international literature on this topic.
In this study the material taken into account is paint. Paints may consist of several typologies
(Jiang, 2008): waterborne, conventional solvent, thermoplastic and epoxy paints. The use of these
materials is essentially based on their characteristics, like durability and cost. Basically, they differ
in the number of components and in the VOC (Volatile Organic Compound, see Paints Directive
2004/42/EC; Jiang, 2008; Bahar et al., 2006; Gates et al., 2003; Migletz et al., 2002; Andrady, 1997).
Their performance is often related to the amount of pigments in the resin (Barletta et al., 2014).
In this study an epoxy paint was used. Epoxy is a two-component material. The first com-
ponent consists of resin, pigment, extenders, and fillers, while the second component acts as
a catalyst to accelerate setting time. Epoxy paint has many advantages such as exceptional
adhesion to both asphalt and concrete pavements when the pavement surface is properly
cleaned before application (Gates et al., 2003). Such property strongly influences the dura-
bility, which is very high. Among the advantages there is the low VOC content, even if the
chemicals used in the production are classified as hazardous materials. Table 2 focuses on the
international literature about photo-luminescent paints applications to road pavements.
2 OBJECTIVES
The main object of this study is to assess the effect of photoluminescent treatments on surface
luminance of asphalt concretes. In more detail: i) a phosphorescent paint was used; ii) both dense-
graded and open-graded friction courses were considered. Additionally, effects in terms of charge
and discharge time were assessed and related to the volumetric properties of the samples.
The study was carried out following the scheme shown in Figure 1. Figure 1 and Tables 3 to
5 summarise laboratory investigations.
The investigation was organized into six main tasks:
Task1. Design of experiments (DOE);
Task2. Material selection and coring (MS);
Task3. Volumetric and surface characteristics before painting: experimental survey on
effective porosity and skid resistance;
Task4. Painting. Paints were applied on the surface according to the procedure explained
below;
Task5. Luminance measurements. Measurements were carried out in sunny conditions and
dark ones;
Task6. Volumetric and surface characteristics after painting. An experimental survey on
effective porosity and skid resistance was carried out.
Task7. Data analysis.
In the pursuit of carrying out the research, two typologies of wearing course were consid-
ered: Open Graded Friction Courses (OGFC) and Dense Graded Friction Courses (DGFC).
Ten cores per type were used. Their gradations are reported in Figure 1.
Cores were investigated in terms of surface properties and volumetric ones. Then they
were subjected to the painting treatment.
The laboratory investigation consisted of two main parts: painting activity and volumetric
properties survey before and after paint treatment.
Volumetric properties are worth to be investigated because of the paint influence on the
global behaviour of the surface (i.e., permeability, acoustic absorption, etc.).
Symbols. DOE: Design of experiments; MS: Material selection and coring; OGFC: Open
Graded Friction Course; DGFC: Dense Graded Friction Course; P%: percentage passing
(aggregate gradation); d: sieve opening [mm].
1535
The painting treatment consisted of two phases: White Paint (WP) application and Phos-
phorescent Paint (PP) application. Table 3 illustrates timing and quantities of this process.
Symbols. WP: White Paint; PP: Phosphorescent Paint; Timing: gap between each laying
phase.
The surveys were carried out by measuring luminance regularly (with a time interval of
60 minutes during the day).
More specifically, each survey was structured in terms of a sequence of charge and dis-
charge time simulating the real behaviour on site (day-to-night cycle).
Charge time started every day at 9 o’clock, lasting 7 hours.
Afterwards, the discharge time, during the decay of luminance, was observed and
measured.
Luminance measurements were carried out for 3 months. Attention was focused on the
relationship existing among luminance, volumetric properties, and paint quantity. Luminance
measurements were carried out by means of a photometer, specifically a HAGNER Universal
Photometer/Radiometer model S4. It is able to measure luminance (0.01–199,900 cd/m2, in 5
ranges) and illuminance (0.01–199,900 lux, in 5 ranges). The lowest detectable value is 0.01 cd/m2
or lux (+/- 1 digit). Accuracy is better than ±3% for all common light sources and in daylight.
Tables 4 and 5 summarize the experimental plan and the properties of the samples investi-
gated, before (task 3) and after (Table 6) the treatment.
1536
Table 3. Timing and quantities of painting process.
Quantity Timing
WP 17 ± 6 60 ± 5
PP 14 ± 6 30 ± 5
Sample Ci,j
Symbols.Gmb: Mix bulk Specific Gravity; neff: Effective porosity; PTV: Pendulum Test Value; P: Painting;
L: Luminance.
Table 6. Model indicators regarding the OGFCs (left) and the DGFCs (right) based on experiments
(I-VIII cycles).
Cycle t1 t2 tL = t2 – t1 LM Cycle t1 t2 tL = t2 – t1 LM
1537
− Quantity of phosphorescent paint (PP);
− Solar radiation (W/m2);
− cycle of charge-discharge: Charge Time, CT; Discharge Time, DT; luminance time, tL.
Solar radiation and temperatures were measured by means of a specific device (Vaisala
HydroMet™ Automatic Weather Station MAWS201). It is a portable AWS (Automatic Weather
Station) for temporary installations. The MAWS201 is field-proven in a wide range of applications,
among them the solar radiation. It consists of 5 basic sensors, a solar panel and a battery.
By referring to the solar radiation it is noted that, based on latitude, day of the year, under
conditions of absence of clouds, the solar radiation should measure about 1 kW/m2 (instead
of 0.9 as per Figure 2). Indeed, solar radiation represents the amount of power that would
be received by a tracking concentrator in the absence of cloud, based on local solar time,
latitude (38°06'37" N), and the day of the year.
Figure 2 illustrates how the solar radiation (measured through the above mentioned
device) varies over time. Note that experimental data fit with the theoretical solar radiation
as per predictive equations (which are based on the equation of the sun’s position in the sky
throughout the year).
Figures 2–5 illustrate the variation over time of:
− Solar radiation SR, (W/m2), predicted and observed;
− Air temperature T, (°C)
− Daytime luminance Ld, (hcd/m2, where hcd = 102 cd);
− Night-time Luminance Ln, (mcd/m2, where mcd = 10–3 cd).
Note that:
i. the solar radiation has a peak of 0.9 kW/m2 at about 12 am;
ii. there is a lag of about two hours between SR and T peaks;
iii. the daytime luminance has a peak of 322 hcd/m2 at about 1 pm;
iv. the night-time luminance yields its peak when the sample is removed from solar radiation
(at about 4 pm);
v. Ln peak is approximately 10–5 times Ld peak.
4.2 Results
Table 6 illustrates the main results in terms of luminance time and average luminance regard-
ing OGFCs (left) and DGFCs (right).
Note that t1 is the time (in seconds) at which luminance measurement starts. In contrast, t2
represents the conventional end of luminance effectiveness (L < 0.01 cd/m2, Rea et al., 2004;
Bullough et al., 2014; Deng et al., 2001).
Table 7 illustrates the average characteristics of samples, in terms of effective porosity
(neff), skid resistance (PTV) and mix bulk specific gravity (Gmb).
Sample Gmb neff AVe (*) PTV neff AVe (*) PTV AV PTV
Symbols. neff: Effective porosity; PTV: Pendulum Test Value; Gmb: Mix bulk Specific Gravity; DGFC:
Dense Graded Friction Course; OGFC: Open Graded Friction Course. AVe: estimated air voids content
(* Praticò and Moro, 2007); (** ANAS, 2010).
1538
Figure 2. Solar radiation: predicted vs observed. Figure 3. Observed solar radiation and
temperature.
Importantly, a decrease of average air voids content (AV) and skid resistance (PTV) can
be observed.
Furthermore, it is noted that volumetrics, surface properties, and drainability may be
affected by chemical properties of PP. Indeed, the latter is made up of two components,
namely photoluminescent paint and catalyser, which is probably responsible for minor phe-
nomena of solution of asphalt binder, which interacts with paint draindown (see Figure 6).
Figures 7 to 10 show how highest luminance (Lmax, cd/m2), luminance time (tL, s), discharge
slope (DS, cd/s⋅m2), and average luminance (LM, cd/m2) vary as a function of cycle and Hot
Mix Asphalt (HMA) type.
Overall, the following observations can be stated:
− Solar radiation (SR) affects photoluminescent effect (Figures 4 and 5). The higher SR, the
higher the effects result.
− Open-graded friction courses, due to their high porosity, have a maximum luminance
which is lower than the one exhibited by dense-graded friction courses (the remaining fac-
tors being constant, see Figure 7). This disequality applies to all cycles.
− OGFCs exhibit also a luminance time (tL) which is lower than the one recorded for DGFCs
(see Figure 8). The higher the cycle number is, the lower this difference results.
− Based on the above facts, it turns out that the average luminance during the period of
effectiveness (L > 0.01 cd/m2) is different when comparing OGFCs and DGFCs.
− Cycles affect photoluminescent effect and discharge slopes (see Figure 9). A quite evident
tendency towards lower values can be observed.
− Cycles do not affect the maximum luminance according to a clear tendency. The first cycle
and successive cycles yield a behaviour which does not call for a decay tendency.
1539
Figure 6. PP draindown and PP-bitumen interaction (bottom of cores). Symbols. PP: Phosphorescent
paint.
Symbols. DGFC: Dense Graded Friction Course; OGFC: Open Graded Friction Course; Lmax: Maxi-
mum luminance [cd/m2]; tL: Luminance time [s].
− Further research is needed for luminance time, average value and performance indicators
whose variations over time do not show an evident pattern. Importantly, a better knowl-
edge of photoluminescent performance over pavement life cycle and its durability (see
Praticò et al., 2010) is vital in terms of life cycle cost analysis and sustainability (Praticò
and Vaiana, 2012; Celauro et al, 2015).
1540
5 CONCLUSIONS
Based on data gathered and analyses performed, the following conclusions can be drawn:
1. Measurements point out that there is a relationship between luminance and air void con-
tent since this latter influences the effect of the paint on the surface. The paint laid on the
surface gradually pours inside the core voids until it reaches the bottom of it (draindown
effect). Consequently, there is a limit of paint that can be placed on the top, after which it
is not useful going ahead.
2. Samples, either open-graded or dense-graded friction courses, show a quite similar dis-
charge slope for all cycles.
3. Overall, DGFCs perform better than OGFCs.
REFERENCES
ANAS S.p.A. 2010. CAPITOLATO SPECIALE DI APPALTO – Parte 2° Norme Tecniche Pavimen-
tazioni stradali/autostradali.
Andrady, A.L. 1997. Pavement Marking Materials: Assessing Environment-Friendly Performance,
NCHRP Report 392, Washington, D.C., Transportation Research Board.
Andre, J., Owens D.A. 2001. The twilight envelope: A user-centered approach to describing roadway
illumination at night. Human Factors, 43(4): 620–630.
Bacero, R., To, D., Arista, J.P., Dela Cruz, M.K., Villaneva, J.P., Uy, F.A. 2015. Evaluation of Stron-
tium Aluminate in Traffic Paint Pavement Markings for Rural and Unilluminated Roads, 11th Inter-
national Conference of Eastern Asia Society fir Transportation Studies, EASTS 2015, September
11–14, 2015, Cebu, Philippines.
Bahar, G., Masliah, M., Erwin, T., Tan, E. and Hauer, E. 2006. Pavement Marking Materials and Mark-
ers: Real-World Relationship Between Retroreflectivity and Safety Over Time, NCHRP Web Only
Document 92, Transportation Research Board.
Barletta, M., Rubino, R., Tagliaferri, V., Vescoa, S. 2014. Design and manufacture of photoluminescent
coatings on stainless steel substrates, Colloids and Surfaces A: Physicochemical and Engineering
Aspects, 455: 147–155.
Bosurgi, G., D’Andrea, A., Pellegrino, O. 2015. Prediction of Drivers’ Visual Strategy Using an Analyti-
cal Model, Journal of Transportation Safety and Security, 7(2).
Botterman, J., Smet, P.F. 2015. Persistent phosphor SrAl2O4:Eu, Dy in outdoor conditions: saved by
the trap distribution, Optics Express, 23(15): 868–881.
Bullough, J.D., Skinner, N.P., Snyder, J.D., Besenecker, U.C. 2014. Nighttime Highway Construction
Illumination, SPR Research Study No. C-08–14, Final Report.
Celauro, C., Corriere, F., Guerrieri, M., Lo Casto, B. 2105. Environmentally appraising different pave-
ment and construction scenarios: A comparative analysis for a typical local road, Transportation
Research Part D: Transport and Environment, 34: 41–51.
De Beer, M., Maina, J.W., Van Rensburg, Y., Greben, J.M. 2012. Toward using tire-road contact stresses
in pavement design and analysis, Tire Science and Technology, 40(4): 246–271.
DIN—German standard. 2009. 67510, Photoluminescent pigments and products – Part 1–4.
Deng, L., Chen, L., Rea, M.S. 2005. An evaluation of the Hunt94 color appearance model under differ-
ent light sources at low photopic to low mesopic light levels, Color Research and Application, 30(2):
107–117.
Gallagher, D., Cyrus, H., and Bassey, R. 2009. Photoluminescent Material Evaluation, Report n. DOT/
FAA/AR-TN09/35, U.S. Department of Transportation Federal Aviation Administration.
Gates, J.T., Hawkins G.H, and Rose, E.R. 2003. Effective Pavement Marking Materials and Applications
for Portland Cement Concrete Roadways2, FHWA/TX-03/4150–2, Texas Transportation Institute.
Giuliani, F., Autelitano, F. 2014. Photoluminescent road surface dressing: A first laboratory experimen-
tal investigation, Materiaux et Techniques, 102(6–7).
Jiang, Y. 2008. Durability And Retro-Reflectivity Of Pavement Markings (Synthesis Study), Joint
Transportation Research Program, FHWA/In/Jtrp-2007/11.
Migletz, J. and Graham, J. 2002. Long-Term Pavement Marking Practices. A Synthesis of Highway
Practice, NCHRP Synthesis 306, Washington, D.C., Transport Research Board of the National
Academies.
1541
Mishra, S.B., Mishra, A.K, Revaprasadu, N., Hillie, K.T., Steyn, W.J. vdM., Coetsee, E., Swart, H.C.,
2009. Strontium Aluminate/Polymer Composites: Morphology, Luminescent Properties, and Dura-
bility, Journal of Applied Polymer Science, 112: 3347–3354.
NYC Department of Building. 2007. MEA 387–07-M, Report of Materials and Equipment Acceptance
Division.
Paints Directive 2004/42/EC.
Praticò, F.G., Moro, A. 2007. Permeability and Volumetrics of Porous Asphalt Concrete – A Theoreti-
cal and Experimental Investigation, International Journal of Road Materials and Pavement Design,
8: 799–817.
Praticò, F.G., Moro, A., and Ammendola, R. 2010. Potential of fire extinguisher powder as a filler in
bituminous mixes”, J. Hazard. Mater., 173(1–3), 605–613.
Praticò, F.G., Vaiana, R. 2102. Improving infrastructure sustainability in suburban and urban areas:
Is porous asphalt the right answer? And how?, WIT Transactions on the Built Environment, 128:
673–684.
Praticò, F.G., Vaiana, R. 2013. A study on volumetric versus surface properties of wearing courses,
Construction and Building Materials, 38: 766–775.
Rea, M.S., Bullough, J.D., Freyssinier, J.P., Bierman, A. 2004. A proposed unified system of photom-
etry. Lighting Research and Technology, 36(2): 85–111.
Specification sheet Epoxy 9000 Paint.
Turnpenny, K., and Crawford, E. 2014. Investigating the potential for reactive ‘glowing’ roads as an
initiative on the Scottish road network, Final Report, Scottish Road Research Board.
UNI EN. 2005. 13036–4, Road and airfield surface characteristics - Test methods: Method for measure-
ment of slip/skid resistance of a surface - The pendulum test.
1542
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Liang He
Chongqing Jiaotong University, Chongqing, China
ABSTRACT: In this paper, nano-TiO2 was loaded onto activated Al2O3 according to the
reasonable proportions so as to prepare a haze-eliminating material, which could be applied on
the pavement surface to efficiently eliminate haze by absorbing PM2.5 particles and purifying
harmful ingredient in exhaust. Moreover, a novel test system which is composed of exhaust
environmental cabinet, solar simulation lamps, surface simulation plate, PM2.5 detector and
automobile exhaust detector was designed, and then a standard test method and some evalu-
ation indexes were proposed to investigate the performances of haze-eliminating materials.
It is found that the haze-eliminating efficiency reached the highest when the proportion of
activated Al2O3 to nano-TiO2 was 2.75:1, and the purifying efficiency for NOx was up to
83%. Finally, the application feasibility of haze-eliminating material on the asphalt pavement
surface was validated, and the surface functions were also tested. Test results showed that the
haze-eliminating coating improved the impermeability of asphalt pavement, while reduced
the anti-skid performance. However, the anti-skid performance of haze-eliminating coating
could be ensured by adding a certain amount of carborundum.
1 INTRODUCTION
In the city, the combined effect of large traffic volume, slow vehicle speed and long passage
time causes a large number of exhaust emissions, which is the main reason for the formation
of PM2.5 particles. Moreover, the dense distribution of high-rise buildings also prevent the
PM2.5 particles from spreading, which is seriously harmful for human health and traffic safety.
According to relevant estimates, by 2020 the air pollution losses accounted for the proportion of
GDP will be as high as 13%, and the exhaust is the main pollution source (Zhang et al., 2004).
Research shows that automobile exhaust contains large amounts of gaseous phase of carbon
monoxide (CO), nitrogen oxides (NOx), hydrocarbons (HC) and partly solid phase (Pb), naph-
thalene, anthracene, phenanthrene, fluoranthene, pyrene, benzoanthracene (Wang et al., 2007).
Taking Beijing as an example, the contribution proportion of all kinds of emissions for PM2.5 is
shown in Figure 1. It is easy to find that the proportion of exhaust reaches as high as 31.1%, thus
how to effectively purify exhaust has great important significance in dealing with the haze.
Currently, the main measures taken by most countries to purify exhaust are to issue exhaust
emissions regulations, limit the traffic volume, and install the exhaust purification equipment
on the car and other measures. However, the limit of the traffic volume brings great incon-
venience to residents travelling, so this is a temporary solution but not a permanent cure.
Automobile industry actively develop clean energy vehicles in order to achieve the internal
combustion engine purification, but it is still in its infancy, its promotion still needs time to
1543
Figure 1. Sources of PM2.5 in Beijing.
further verify that it is difficult to achieve a wide range of applications in short term. Or, con-
figuring ternary catalyst system to realize exhaust purification, in a certain extent, improved
the exhaust pollution to the atmospheric environment, but there are some difficult problems
such as catalytic conversion rate is low, the catalyst failure, cold start, high cost and so on.
Under this background, the domestic and foreign researchers have turned to photocatalytic
nanomaterials to purify exhaust. Among them, the TiO2 has high chemical stability, corrosion
resistance, high REDOX potential, photocatalytic reaction driving force, high photocatalytic
activity, can make some endothermic chemical reaction to realize and accelerate on the sur-
face of the optical radiation with the characteristics of non-toxic and low cost. And TiO2 has
become a photocatalyst with certain application potential and research value (Zhang et al.,
2009). Aiming Sha (Sha, 2011) and Yiqiu Tan (Tan et al., 2010) studied the asphalt pavement
materials of degrading automotive exhaust, researched the mineral load and coupled type of
light catalytic purification materials which are applied on the road surface. Lianfang Zhao
(Zhao et al., 2003) study the law of oxidation reaction rate of the nitrogen oxide in automo-
bile exhaust on the cement road loaded with Nano-TiO2.
The main exhaust purification material is Nano-TiO2, but the catalyst Nano-TiO2 has some
defects. For example, because the Nano-TiO2 is fine, when the Nano-TiO2 powder suspension
system is catalysing, the active ingredient loss will be big and the catalyst stability is bad. In
order to solve this problem, some catalyst carriers, such as activated Al2O3, activated carbon,
silicone and zeolite were introduced (Akira et al., 2000), which can not only improve the cata-
lytic activity and stability of the catalyst, but also make it easy to modify the surface of catalyst
and prepare various shapes of the reactor. Among several types of catalyst carrier, activated
Al2O3 shows better performances, which is a kind of solid materials with porous structure,
high dispersion, large specific surface area, good adsorption performance, excellent mechani-
cal properties, excellent mechanical properties and good thermal shock resistance. Therefore,
activated Al2O3 was used as catalyst carrier in this paper. In addition, there were no unified
purification test methods and evaluation indexes for haze-eliminating coating materials.
Urban roads, buildings and other infrastructures have a wide interface with haze, which
can be coated with a layer of purifying materials as to solve haze problem. Therefore, this
paper is based on automobile exhaust purification as the main control method, preparing a
loaded pavement coat material, in order to rationally utilize the surface of the urban infra-
structure to absorb and purify automobile exhaust, besides the standard test method and
evaluation indexes were proposed in order to evaluate the purification effect, the research
results has an important practical significance.
1545
Figure 3. XRD spectra of nano-TiO2.
the fan to ensure the automobile exhaust is evenly distributed, then put the PM2.5 detector
into it and combine with automobile exhaust detector; (3)When the PM2.5 concentration is
about 700 μg/m3, put into surface simulation plate with the same time opening solar simula-
tion lamps, then start to test; (4) Record the PM2.5 concentrations and exhaust component
concentration.
1546
C1 -C 2
η= × 100 (1)
T
1547
Figure 8. The influence of different proportions of Al2O3 and Nano-TiO2 on adsorption rate.
1548
Figure 10. The purification efficiency of samples with different proportions.
According to the Eq.(1), the degradation efficiency of the harmful components in auto-
mobile exhaust is calculated by different proportions of coating materials, as shown in
Figure 10. It can be seen that there are some differences in the degradation efficiency of
automobile exhaust with different proportions of the coating materials, among the degrada-
tion efficiency of NOx was the most obvious, and CO2 was relatively weak; Under the best
proportion, the purification efficiency for automobile exhaust HC, NOx, CO and CO2 are
67.0%, 83.3%, 64.4% and 53.6%.
It was found that the haze-eliminating material combine high efficiency of nano-TiO2 and
degradation of activated Al2O3 to realize absorbing PM2.5 particles and purifying harmful
ingredient in exhaust, and eventually realize the effect of haze-eliminating. It was safe, envi-
ronment protection and efficient property that make the haze-eliminating material widely
applied on infrastructure of city. Because of the huge contact surface between infrastructure
and haze, the effect of haze-eliminating is good. However, it is different between road and
building in function and using, so that it need to meet the requirement of pavement per-
formance when the haze-eliminating material use in road, such as anti-skid performance and
impervious performance.
V2 -V
V1
Cw = × 60 (2)
T2 -T
T1
Where, V1 is the first time volume, mL; V2 is the second time volume, mL; T1 is the first
time, s; T2 is the second time, s.
The result was shown in Table 1, and the impervious performance of pavement was deter-
mined by CW. According to the requirements in Highway Asphalt Pavement Construction
Specification JTG F40–2004, it is found that the permeation coefficient is small, and the
impermeability of pavement coat is good.
1549
Table 1. The result of water permeability experiment.
Asphalt pavement 67 10 −3 64
70 10 −3 67
Coating pavement without 55 10 −3 52
carborundum 49 10 −3 46
Coating pavement with 67 10 −3 64
carborundum 66 10 −3 63
6 CONCLUSIONS
1. Nano-TiO2 was loaded onto activated Al2O3 with the reasonable proportions so as to pre-
pare a haze-eliminating material for asphalt pavement coat. The haze-eliminating material
1550
used epoxy resin as a carrier material, coated on the road surface, building roofs and other
urban facilities surface, having a wide interface with haze, so as to adsorb and degrade
automobile exhaust, to achieve the control of urban haze.
2. Several devices were self-made, such as exhaust environmental cabinet, solar simulation
lamps and surface simulation plate, and combined with PM2.5 detector and automobile
exhaust detector to integrate a haze-eliminating purification test system, besides proposed
haze adsorption rate and degradation rate the two types of evaluation index. This system
can realize the controlled distribution and accurately test of haze’s concentration.
3. By testing different proportions of activated Al2O3 and nano-TiO2, the purification test
results show that the haze-eliminating efficiency is best when the haze-eliminating mate-
rial made with 4 g nano-TiO2 and 11 g activated Al2O3, and the haze-eliminating effi-
ciency is up to 28.2 μg/m3⋅min.Under the best proportion, the purification efficiency for
automobile exhaust HC, NOx, CO and CO2are 67.0%, 83.3%, 64.4% and 53.6%, and the
purification efficiency is the highest to the most harmful gas such as the nitrogen oxides
and carbon hydroxide.
4. Analyzed the feasibility of haze-eliminating coating material for road pavement, and ver-
ify the function of haze-eliminating coating. The results showed that the haze-eliminat-
ing coating can improve the impermeability of pavement, however reduce the anti-skid
performance. However, the anti-skid performance of haze-eliminating coating could be
ensured by adding a certain amount of carborundum.
ACKNOWLEDGEMENTS
This work was supported by the National Natural Science Foundation of China (Nos. 51378073
and 51408047), the Natural Science Foundation of Shaanxi Province (No. 2014 JQ7278), and
the Special Fund for Basic Scientific Research of Central College of Chang’an University
(Nos. 310821165009, 310821153502 and 310821152003). The authors gratefully acknowledge
their financial support.
REFERENCES
Akira Fujishima,Tata N. Rao,Donald A. Tryk. 2000. Titanium dioxide photocatalysis. Journal of Pho-
tochemistry & Photobiology C Photochemistry Reviews,1(1), 1–21.
Sha Ai-min. 2011. The Construction and Maintenance Technology of Environmentally Friendly Road.
The Special Focus: 54–59.
Shen Weiren, Zhao Wenkuan, Hefei. 1998. TiO2-Based Photocatalysis and Its Applications For Waste
Water Treatment. Progress in Chemistry, 10 (4): 349–361.
Tan Yi-qiu, Li Luo-ke, Wei Peng, Sun Zheng. 2010. Application Performance Evaluation on Material of
Automobile Exhaust. China Journal of Highway and Transport, 23(6):21–27.
Wang Boguang, Lv Wanming. 2007. Emission Characteristic of PAHs Composition in Motor Vehicles
Exhaust of City Tunnel. China Environmental Science, 27 (4): 482–487.
Zhang Zheng, Cao Hui. 2004. Empirical Analysis of Relationship between Auto Industry and Environ-
ment of Beijing City. China Soft Science, 11:57–58.
Zhang D, Qiu R, Song L, et al. 2009. Role of oxygen active species in the photocatalytic degrada-
tion of phenol using polymer sensitized TiO2 under visible light irradiation. Hazard Mater163 (2–3):
843–847.
Zhao Lianfang, Fu Dafang, Qian Chunxiang. 2003. The Study on the Photo-catalysis Efficiency of
Degradation of Nitrogen Oxide by an Iron-doped Nanometer TiO2. Journal of Southeast University
(Natural Science Edition), 33(5):677–680.
1551
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Zewen Yu
Hunan Communications Research Institute, Changsha Hunan, China
ABSTRACT: Connection sections of successive tunnels are mostly located in the can-
yon between two mountains, where the wind speed is higher than the nearby flat ground
because of terrain effects. One common design pattern of successive-tunnel connection
sections is the parallel twin-deck bridges. Previous studies mainly focused on the wind-load
study of bridges instead of considering the crosswind effects on driving safety. In order to
study the influence of crosswind on vehicles on twin-deck bridges, numerical simulation
models were established to simulate the wind field when the vehicle drives on different
lanes of the twin-deck bridges. It was concluded that the vehicle driving on the first lane
of the twin-deck bridges is with the greatest accident risk. And the crosswind speed change
causes severe changes of vehicle aerodynamic force, which may lead to driving fatigue and
driving error.
1 INTRODUCTION
In recent years, with the continuous development of highway network in China, the number
of mountainous highway tunnels increased rapidly, including tunnel group formed by suc-
cessive tunnels. Connection sections of successive tunnels are mostly located in the deep
canyon between two mountains, where the higher wind speed will perhaps cause vehicles
out of control and lead to traffic accidents. The common design pattern of the mountain-
ous successive-tunnel connection sections is the parallel twin-deck bridges, so under such
situation the upstream bridge and the downstream bridge will influence each other through
the change of airflow and aerodynamic force. Furthermore, with the increase of the span
and height of the bridges, the crosswind speed increases accordingly. As a consequence the
influence of crosswind on driving safety of highway parallel twin-deck bridges should be
studied.
At present, researchers paid more attention to wind-load study of bridges and highway
traffic safety research of single tunnel or bridge, and the traffic safety under strong cross-
wind on the parallel twin-deck bridges of the successive tunnels connection sections needs
more studies. Therefore, this paper focuses on the influence of crosswind on driving safety
on the parallel twin-deck bridges of the successive-tunnel connection sections. The numeri-
cal simulation models of the vehicle on different lanes of the parallel twin-deck bridges were
established and FLUENT software was used for Computational Fluid Dynamics (CFD)
numerical simulation. On the basis of the results of the four simulating situations, the cross-
wind field characteristics and the driving dynamic characteristics were analyzed, to provide
the foundation for the further study of the successive-tunnel connection sections traffic
safety.
1553
2 LITERATURE REVIEW
Traffic safety problems caused by crosswind have attracted the attention of researchers
since 1930s. Affected by the sudden crosswind, drivers sometimes cannot react properly and
promptly and the vehicles will perhaps rollover or sideslip. High speed vehicles would be
much more easily affected by crosswind. In addition to the natural crosswind in the open
highway, some wind environment is also caused by man-made structures including tunnels,
bridges and so on. Researches of the crosswind effect on vehicles are usually studies by meth-
ods of wind tunnel test, analytical analysis and numerical simulation. J. Bettl et al. (2003)
studied the aerodynamic force characteristics of vehicles passing through the bridge tower
equipped with windbreak and did simulated analysis of vehicle dynamic response based on
previous research results. Snxbjornsso et al. (2007) fixed the ultrasonic anemometer on mov-
ing vehicles to measure the relative wind speed and used the GPS receivers to monitor real-
time vehicle speed and position to establish the corresponding coordinate for wind speed and
wind direction analysis. Considering with the wind speed, the vehicle speed, adhesion coef-
ficient and the road humps, the probability of occurrence of traffic accidents was evaluated
by using the accident rate as the performance index. S. Charuvisit et al. (2004) found that the
response of vehicles entering crosswind increased with the increase of wind speed and vehicle
speed, and when the wind yaw angles reached 30 degrees, the response reached a maximum
value. After setting the windbreak, the lateral acceleration and the sideslip angular accel-
eration of vehicles suffering crosswind decreased obviously. For parallel twin-deck bridge
research, Gou et al. (2013) used CFD software FLUENT to calculate the wind coefficients of
parallel twin-deck bridge under situations of different wind attack angles, parallel twin-deck
spacing, and compared with those of single-deck bridge to analyze the aerodynamic interfer-
ence effects of the parallel twin-deck bridges. Chen et al. (2006) did a series of wind tunnel
experiments based on the section modeling of the bridge and found that the critical flutter
wind speed of the main section of parallel twin-deck box girder bridge was closely correlated
with the distance between the two bridges. The results showed that the flutter critical wind
speed increased with the increase of distance between two bridges. Huang (2011) researched
aerodynamic interference effect of the middle section of the Shanmen Bridge under differ-
ent bridge distances and different wind attack angles, which is a parallel twin-deck bridge. It
showed that the impact of the aerodynamic interference effect of the downstream bridge was
more obvious; at the same time, the distance between the bridges and the wind attack angle
were two of the factors that would affect the aerodynamic interference effect of the parallel
twin-deck bridges. These studies on parallel twin-deck bridges focused on the aerodynamic
force of bridge decks instead of vehicles on bridge or driving safety.
Although some studies have studied traffic safety under strong crosswind on single deck
bridge (e.g. Chen et al. 2015), few study has taken the traffic safety under strong crosswind
on the parallel twin-deck bridges into consideration to the best knowledge of the authors.
With the rapid construction of the highways in the mountainous area in China, there will be
more and more parallel twin-deck bridges, especially in the bridge-tunnel connecting section,
therefore it is necessary to pay attention to the driving safety under the influence of wind on
the parallel twin-deck bridges.
1554
In this paper, the geometric model of the parallel twin-deck bridges is determined accord-
ing to the Mishui River Bridge of Yanru Highway. The width of the vehicle lane is 3.75 m, the
width and the height of the roadside barrier is 0.5 m and 1.1 m, the height and the width of
the bridge deck is 3.5 m and 11.25 m, and the distance between the two bridges is 15 m.
Because of its high gravity center, the high-side truck is adopted as the research simulation
object. The shape size of the truck body is 6.95 m length, 2.5 m width, 3.7 m height and the
minimum ground clearance is 1.1 m. In the simulation it is assumed that the vehicle drives in
the middle of the lane.
In order to improve the efficiency of the simulation, some parts of the vehicle and bridges
which have little effect on the aerodynamic analysis are simplified. For example the vehicle
orthographic projection is simplify as rectangle. The dimension of the bridges and truck is
shown in Figure 2.
1555
Based on the results of trial calculation, the computational domain size is determined: the
distance from the air entrance to the windward side of the bridge is 5 times of the width of
the bridge (55 m), the distance from the exit to the leeward side of the bridge is 10 times of
the width of the bridge (110 m), the distance from the lower boundary to the bottom of the
bridge is 5 times of the bridge height (16.5 m), and the distance from the upper boundary to
the root of the vehicle is 9 times of the bridge height (31.5 m). The total length of the calcula-
tion domain is 202 m, and the total height is 56.3 m, as shown in Figure 3.
According to the shape and size of the geometric model, the structural configuration
(block) is built and the grid division of the calculation domain is made, which is shown in
Fig. 4. In order to adapt to the change of the flow field variables, it is decided to refine the
grid around the bridge through trial as shown in Fig. 5.
1556
(15 degrees Celsius). The air entrance is set as the speed entrance, the top and bottom surface
as the wall surfaces, the surfaces of the vehicle and bridges also as wall surfaces, and the exit
as effluent. Because of the low turbulence intensity of the vehicle outflow field, the turbu-
lence intensity is set as 0.5%, the turbulent viscosity coefficient as 10, the density of the fluid
(air) as 1.225 Kg/m3, the moving viscosity coefficient as 1.7894*10–5 m2/s, and the surface
boundary as wall surfaces with condition 0.16 and roughness height 0.05 m. In addition, the
model is discretized by the first order upwind difference scheme. The SST k-ω turbulence
model is used to simulate and the SIMPLEC is used to calculate.
1557
Figure 8. Wind pressure of the vehicle on Lane 2 of the bridges.
1558
the pressure of these areas is lower, too. The results show that the vehicle driving in the wind-
ward side is more dangerous than driving on the leeward side.
Besides, the results show that the low pressure area size between the two bridges will
decrease significantly when the vehicle travelling on the bridge in the windward side. Mean-
while, the low pressure area will appear in the windward side of the downstream bridge and
the high pressure area will appear in the leeside. It can be seen that there is obvious mutual
interference between upstream and downstream bridges.
1559
Figure 14. Wind speed of the vehicle on Lane 4 of the bridges.
1560
Table 1. Aerodynamic coefficient table.
Situation Wind speed (m/s) Lateral force coefficient Lift force coefficient Torque coefficient
The speed contour of every situation obtained by Fluent is shown in Figures 15–18.
The aerodynamic force coefficient results in Table 1 illustrate that the vehicle is under
wind force from left to the right, the downward air pressure, and a clockwise torque around
the vehicle mass center. The aerodynamic coefficient increases as the wind speed increases.
The last row in Table.1 shows the aerodynamic force coefficient results if the vehicle is driv-
ing on regular ground road instead of twin-deck bridge and the lateral force coefficient of
vehicle on twin-deck bridge is much bigger than that of vehicle on ground road (7.43 or
6.29 vs 5.43 with 15 m/s wind speed), which illustrates that it’s more dangerous to drive on
twin-deck bridge than on ground road under strong crosswind. Therefore the aerodynamic
force of vehicle will drastically change and have a great impact on the driver’s driving judg-
ment when driving on twin-deck bridges encountering crosswind. The dramatic changing
aerodynamic will force the driver to constantly adjust the speed and direction and may lead
to driving fatigue and driving error which will perhaps result in traffic accidents.
5 CONCLUSION
This paper uses CFD technology to research the wind field and traffic safety of the vehicle on
the parallel twin-deck bridges in highway successive-tunnel connecting section. Four different
situation when vehicle is driving on different lanes of parallel twin bridge under wind load
influence are simulated, and the related wind speed and pressure distribution were analyzed. In
order to study the influence of crosswind on vehicles on twin-deck bridges, simulation models
were established using Fluent software to simulate the wind field of the different four situations.
The results show that the wind caused lateral force of vehicle on parallel twin-deck bridge is
considerable bigger than that of vehicle on ground road, which indicates that under the same
crosswind condition it’s more dangerous to drive on twin-deck bridge than on ground road.
And the vehicle driving in the first lane on the twin-deck bridges is under the biggest aerody-
namic force and this situation is at greatest risk. Based on the simulation results under different
wind speed, it is concluded that crosswind speed change causes severe changes of vehicle aero-
dynamic force, that will force the driver to constantly adjust the speed and direction and may
lead to driving fatigue and driving error which will perhaps result in traffic accidents. These
results showed the influence of crosswind on traffic safety on twin-deck bridges of successive-
tunnel connection sections and provided theory basis for further driving safety research.
ACKNOWLEDGMENTS
This research was jointly sponsored by Project 5150840 supported by the National Natural
Science Foundation of China and the Project-sponsored by SRF for ROCS, SEM.
1561
REFERENCES
Bettle J., Holloway A.G.L., Venart J.E.S. (2003) A computational study of the aerodynamic forces act-
ing on a tractor-trailer vehicle on a bridge in cross-wind [J]. Journal of Wind Engineering and Indus-
trial Aerodynamics. 91(5): 573–592.
Charuvisit S., Kimura K., Fujino Y. (2004) Effects of wind barrier on a vehicle passing in the wake of
a bridge tower in cross wind and its response [J]. Journal of Wind Engineering and Industrial Aero-
dynamics. 92(7–8): 609–639.
Chen, Ning, Yongle Li, Bin Wang, Yang Su, Huoyue Xiang. (2015) Effects of wind barrier on the safety
of vehicles driven on bridges. J. Wind Eng.Ind.Aerodyn.143:113–127.
Gou Guotao, YE Zhengwei, XIANG Yiqiang, CUI Heli. (2013) Numerical Study on Static Aerody-
namic Coefficients Aerodynamic Interference Effects of the Box girder Section of CHEN Continu-
ous Rigid Frame Bridge with Close Twin Deck. Highway Engineering. 38(2): 196–201.
Huang Haibin. (2011) Research of long-span continuous rigid frame bridge with high-rise piers of ulti-
mate bearing capacity and wind resistance. Central South University.
Snæbjörnsson J.T., Baker C.J., Sigbjörnsson R. (2007) Probabilistic assessment of road vehicle safety
in windy environments [J]. Journal of Wind Engineering and Industrial Aerodynamics. 95(9–11):
1445–1462.
Chen Zhengqing, Niu Huawei, Liu Zhiwen. (2006) Experimental Study on Flutter Stability of Parallel
Box-Girder Bridges. Journal of Vibration and Shock, 25(6): 54–58.
1562
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The protection of road infrastructure from the rockfall phenomenon is very
demanding both for design and for construction of rockfall defense works. In particular,
design approach is strongly influenced by the phenomenon’s aleatory nature and the difficult
determination of the geomechanical parameters that characterize the interaction between the
in motion rocky boulders and the affected slope ground. In this context the Authors propose
a probabilistic methodological approach finalized to rockfall design of works through imple-
mentation of a three-dimensional simulation software called ROTOMAP®. This approach
is described through the application to a case study on the road SS18 “Tirrena Inferiore”
(Italy) recently affected by a drought rockfall. The results allowed verifying the effectiveness
of the proposed methodology in the design process of new defensive rockfall works and in
the verification of defensive existing works. The reliability of the design/verification process
is linked to the execution of a large number of simulations.
1 INTRODUCTION
Rockfalls are one of the dreadful events more frequent along the mountain roads or along
road paths located at the foot of unstable rocky walls, characterized by geological and/or
weather conditions particularly risky in terms of slope stability (Singh & Goel 1999, Agosti-
nacchio & Olita 2002, Peila et al. 1998, Dorren 2003).
Contextually, the vulnerability of large areas of the Italian country is very significant due
to a whole series of isolated or concomitant factors as, for example, the high seismicity, the
frequent, short and intense rainfalls or even the high thermal excursions, typical of the Italian
Alpine arc.
Rockfalls show themselves with a sudden fall off of rocky boulders of a few cubic meters
volume that, abandoning its original equilibrium position slide downstream along the slope.
The extreme dangerousness of the phenomenon was mainly due to the rapid starting and
evolution, as well as the difficulty to predict both the position of the possible detach points
and the line of fall evolution.
For such reasons the deleterious consequences produced by this disaster are not so rare, rang-
ing from the traffic discomfort (Agostinacchio et al. 2014), caused by the road interruption, to
the isolation of entire villages and up to the direct loss of human lives (Palma et al. 2012).
In this paper the Authors analyze the peculiar aspects of the rockfalls phenomenon, and
then expose the results relating to a specific case study. Due to the aleatory context the meth-
odological problem approach is probabilistic and conducted with the implementation of the
ROTOMAP® software code which allows to determine the probability zones of boulder stop,
the distribution of kinetic energies and the impact recurrence time on a fixed area (Scioldo
2006). The identification of the boulder stop zones allows to establish the need, or not, to
1563
provide suitable protection works, while fall speed and trajectory calculation is essential to
locate the optimal geometric position of these works, as well as the impact energy that they
must be able to dissipate.
The case study examined in this paper regards the rocky escarpment of Monte Serra (Fig. 1)
overlying the road SS18 “Tirrena Inferiore” (Fig. 2) between the two villages of Ogliastro and
Rasi of Maratea (PZ), where it has developed a rockfall with stone boulders having maxi-
mum size of about 2 cubic meters (Fig. 3), which the existing rockfall barriers have failed to
contain (Figs 4–5). For this reason, the Local Road Authorities has ordered the temporary
closure of this road section, causing discomfort and economic damage to the population of
Maratea.
1564
3.1 Theoretical aspects
In the study, the RBDB (Rockfall Defensive Barrier Design) methodology has been used for
the verification of existing rockfall barriers and the design of additional one. This method-
ology is based on a probabilistic approach both during the collapse simulation and during
the rockfall defensive barrier design (Artini & Coccolo 2000, Coccolo & Artini 2003). The
innovation of this methodology is linked to an aleatory parameter, called security level, the
choice of which is able to influence the entire design process (Tab. 1).
The Required Safety Level (SLreq) depends on type and importance of work to protect by
rockfall, as well as by economic and social considerations, so its determination is entrusted
to the design engineer sensitivity. In this case, considering the high local seismicity and tak-
ing into account that the SS 18 fits into a seaside resort context of considerable merit, it is
considered reasonable to assume SLreq = 98%.
The Actual Safety Level (SLact), which is essential for the preliminary assessment, has been
formed from a probabilistic analysis based on a very large number of collapse simulations on
the studied slope. For this purpose, the three-dimensional kinematic model implemented in
the ROTOMAP® software was used. This software considers the boulders as material points
and provides, for each falling, the results on medium and more probably path between all
those physically acceptable (Scioldo 2006).
In particular:
n0
Lact =
SL (1)
n
where n0 represents the number of simulation boulders that do not exceed a certain physical
limit, for example, an ideal line located at a fixed obstacle distance or a pre-existing defense
barrier, and n indicates the total number of simulations.
1565
Figure 6. Three-dimensional model of the studied area. Figure 7.
a grid size equal to 2 meters. Practically, the analysis of the problem has required the identifi-
cation of the detach niches. This allows the software to simulate the collapse of boulders, and
their free rolling down the valley. Since the detachment occurs from not identifiable points
within sub-vertical areas, the used software requires the identification of detach lines, which
can be positioned below the niches themselves.
The simulation is done by starting the boulders from points arranged along the fixed
detach lines, with speeds ranging from a minimum to a maximum as a function of the niches
geometry. So it is possible to obtain a broad spectrum of initial conditions, extremely use-
ful for a statistically significant distribution of paths. In this case it was decided to use three
detach lines positioned at different altitudes (respectively 400, 500 and 590 m) and initial
speeds ranging from 0.5 to 5 m/s.
1566
Figure 9. Homogeneus areas. Figure 10. Rotomap® homogeneus areas.
the model are the coefficient of normal (Kn) and tangential energetic restitution (Kt), the
coefficient of roto-slipping friction (tgδr) and the limit angle (α). The coefficients of restitu-
tion normal and tangential are defined as the ratio between the boulder speeds (respectively
normal and tangential) before and after impact.
The coefficient of roto-slipping friction expresses, in the phases of rolling and sliding, the
loss of energy caused by the friction between the slope and the boulder, while the limit angle
is the angle over which occurs the passage by the roto-slip conditions to those of free flight.
Some previous studies (Paronuzzi et al. 1996) have proposed limit angles variable, in function
of the ground roughness, between 3 and 12°.
In this case it was not possible to carry out the scaling of some boulders to film the trajec-
tories and determine the actual values of the coefficients of energetic restitution and the coef-
ficient of roto-slipping friction, so it was decided to refer to the available literature data for
each type of homogeneous zone and to adopt, with major safety, medium-to-high values for
the restitution coefficients and medium-to-low values for the roto-slip coefficient (Tab. 2).
The estimation of the used limit angle was conducted through a statistical approach chang-
ing its value between 3 and 12°, with increments equal to 1°. For each limit angle ROTOMAP®
has simulated the falling of 100 boulders (for a total of 1000 simulations), then a significant
sample of 200 simulations was extracted.
For this sample the chart of the cumulative frequencies of the maximum values of mass
energies (V2/2 - provided by the software) was constructed identifying the median, or the
value corresponding to the cumulative frequency of 0.5 (Paronuzzi et al. 1996).
For each limit angle have been then calculated the deviations between each value of the
mass energy of sample falling and the relative median, as well as the standard deviation of
each values series for a given angle.
The deviations analysis has allowed to establish the value of a limit angle equal to 11°. The
choice to validate this value is supported by the consideration that the slope has a few bumps
at the wall foot, and so the boulders path is probably characterized by prevailing rolling.
existing barriers by a value between 15 and 20%. Therefore, instead of 1750 kJ (Fig. 11), it is
assumed a reduced energy dissipation value equal to 1400 kJ.
Barrier Class 4 5 6
Barrier Class 7 8
Figure 15. Class 8 barrier h of 7 m. Figure 16. Idem hyp 2 with 2 Class 8 (hyp 2 -
SLd = 98.1%). barrier h of 6 m.
1569
more valid the design solution described in Figure 15 that is able to meet the minimum design
validation SLd ≥ SLreq.
However, it should be noted that such intervention must not necessarily be considered the
only possible for the slope in object. For the risk mitigation, additional alternative and/or supple-
mentary solutions are represented by interventions aimed to reforestation of the detrital areas.
4 CONCLUSIONS
In this paper are showed the phases of a study aimed to design a passive defense work able to
avoid, or reduce, the rockfall consequences on the road SS18 “Tirrena Inferiore” in locality
Ogliastro of Maratea (Potenza, Italy).
The Authors have proposed a probabilistic methodological approach through imple-
mentation of the simulation software ROTOMAP®. This calculation code is based on a 3D
approach with which it is possible to implement the most various rockfall simulations, so it is
possible to reconstruct in probabilistic key, the potential behavior of the boulder trajectories,
their speed and heights reached.
Therefore, the reliability of the design/verification process is very satisfactory as it is related
to the possibility of performing a large number of simulations that are capable of providing
statistically significant results. In this way it is possible to achieve the predetermined safety
levels, defined through a random parameter whose choice is fundamental because it is able to
influence the entire design cycle. The study shows that hazardous situations related to rock-
fall can be effectively addressed and resolved through appropriate mathematical modeling.
REFERENCES
1570
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Zhigang Li
Key Laboratory of Road Structure and Material of Ministry of Communication, Chang’an University,
Xi’an, China
Lipeng Dong
Zhenzhou Communications Planning Survey and Design Institute, Zhenzhou, China
Peiwen Hao
Key Laboratory of Road Structure and Material of Ministry of Communication, Chang’an University,
Xi’an, China
1 INTRODUCTION
According to statistics, more than 1 million person is killed due to traffic accidents every year
worldwide. However, this number is still rising. The findings indicate that the most impor-
tant factor which resulted in traffic accident was the skid resistance performance of asphalt
pavement, especially on rainy days (Ibrahim M. Asi, 2007). It can dramatically affect braking
distance and so contributes to safety of traffic. Obviously, improving the skid resistance per-
formance of asphalt pavement is the most effective way to reduce the traffic accidents. Skid
resistance depends on a pavement surface’s microtexture and macrotexture and changes over
time due to wearing (Ibrahim M. Asi, 2007).
Skid resistance is a measure of the resistance of the pavement surface to sliding or skid-
ding of the tyre. Goodman (2009) found that polished stone value (PSV) of different aggre-
gates had a significant influence to skid resistance performance of asphalt pavement, and
microtexture as well as PSV of aggregate surface contributed enormously to skid resistance.
Girum (2011) analyzed the skid resistance of different aggregate and mixture, and found the
correlation between the cumulative standard axle load number, the passing rate of 4.75mm,
the asphalt content, abrasion value and PSV of aggregate:
TESAL = 22513320.25 * x11 022 * x20 983 * x30 907 * x40.899 (1)
Where, x1 is the passing rate of 4.75 mm,%; x2 is the asphalt content,%; x3 is PSV; BPN; x4
is Los Angeles abrasion value,%.
1571
There are lots of factors which affect the skid resistance, and a mass of work has been done
by researcher worldwide. Peter Kotek (2014) reported that the direct characteristics affecting
the skid resistance are the age of wearing course, traffic intensity, and climatic conditions of
section. Ibrahim M. Asi (2007) found that addition of steel slag in asphalt concrete mixes and
using SMA mixes can be implemented to improve skid resistance of road surfaces, especially
at highway intersections. Zhou Zhihong (2012) found that microstructure of mixture had
little relationship with mixture type, void volume, texture depth, but it is strongly correlated
to microtexture and PSV of coarse aggregate. Dynamic modulus and damping performance
were studied and the characteristics of viscoelastic damping pavement were revealed by accel-
erated abrasion test (Wang Xiuyun, 2012). Rub test was carried out to simulate the decay law
of skid resistance under vehicle loads (Qiu Zhixiong, 2013).
A review of all of the research conducted so far indicates that different aggregate surface
contributed enormously to the skid resistance due to different microtexture. As shown above,
researchers have done a lot to reveal that different kinds of aggregate had a significant influ-
ence on the skid resistance of pavement. However, no agencies have been reported to monitor
pavement friction performance in site or lab, especially the affecting to the decay law of skid
resistance of pavement with different kind of aggregate. This paper aims to understand the
decay law of skid resistance under different aggregate conditions. The Asymptotic exponen-
tial model may help us to predict the decayed condition of pavement skid resistance perform-
ance during the practical engineering.
The size of sieve 16 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075
mesh/mm
percentage 100 95 54.8 28 28 19.8 14.1 10 7.1 5
Passing/%
1572
Figure 1. schematic and product picture.
specimens were made in accordance with JTJ052-2011 of China. The test device was car-
ried out polishing by the controller without water, and it was stopped while approached the
certain cycles. Take all the specimens out and Los Angeles abrasion test was carried out to
measure the pendulum type friction coefficient.
To study the effect of aggregate type on the skid resistance properties of asphalt pave-
ment, 30cm width and length by 5cm height samples were compacted using wheel grinding
method.
Usually, pendulum type friction coefficient would be attenuation as the vehicle loading
applying. However, the data of pendulum type friction coefficient during 10000 cumulative
polishing cycles were increasing. Friction on pavement surface is a force which obstructs
relative movement between a tyre and a pavement surface, and the texture has a significant
influence on the data of pendulum type friction coefficient. The texture is a basic element
of interaction between tyre and pavement. It is set by microtexture of aggregate surface and
macrotexture of pavement surface (Kotek, 2014). When the cumulative axis applied during
10000 times, the friction between tyre and pavement made the asphalt membrane of aggre-
gate surface fell off. This progress made the microtexture of aggregate surface been the main
factor which affected the Pendulum type friction coefficient. After 10000 times loading, the
macrotexture of pavement surface became decay. Thus, the pendulum type friction coeffi-
cient began damping.
Experimental data were fitted by using the Asymptotic exponential model. The data of
pendulum type friction coefficient were fitted from 10000 times, to reveal the real decay law of
skid resistance. The fitting results were shown as Fig. 2 and the fitting equation were shown
as Table 2.
The data of texture depth indicated that the reduction of early texture depth had little dif-
ference because of the same gradation. However, there is an obvious difference between the
different properties of aggregate. The texture depth had an obvious reduction at 30000 cumu-
lative cycles approx. As the cumulative polishing cycles continue to increasing, the aggre-
gate began wearing. The largest abrasion value limestone was, the fastest attenuation speed
(maximum coefficient B) and the maximum attenuation amplitude (maximum coefficient A)
of texture depth was. However, 100% diorite is just the opposite. Attenuation speed and
amplitude of texture depth were obviously changed by adding diorite to limestone aggregate.
1573
Figure 2. The fitting results of texture depth and pendulum type friction coefficient.
Table 2. The fitting equation of texture depth and pendulum type friction coefficient.
Furthermore, the larger dosage of limestone was, the faster attenuation speed and the larger
attenuation amplitude were. When the cumulative polishing cycles exceeded 200 thousand
times, texture depth of different aggregates tended to a stability except 100% limestone.
As discussed previously, within 10000 cumulative polishing cycles, the pendulum type fric-
tion coefficient increased, for the asphalt membrane of aggregate surface falling off. Gener-
ally speaking, skid resistance of High Friction Surfaces could be improved by reducing the
dosage of asphalt to some extent. It is evident from Fig. 2 that the influence of skid resistance
performance on coarse aggregate increased, after the asphalt membrane falling off. The larg-
est abrasion value limestone was, the fastest attenuation speed (maximum coefficient B) and
the maximum attenuation amplitude (maximum coefficient A) of Pendulum type friction
coefficient was. However, 100% diorite is just the opposite, and it was similar to texture depth
of pavement. Attenuation speed and amplitude of Pendulum type friction coefficient were
also obviously changed by adding diorite to limestone aggregate. Aggregate itself perform-
ance affected skid resistance performance of pavement in the low speed running and also
high speed running. In terms of attenuation amplitude, skid resistance of aggregate itself is
the most important factor that affected the skid resistance of High Friction Surfaces.
1574
4 CONCLUSION
To ensure safety on roads it is necessary to focus on skid resistance of High Friction Surfaces.
The accelerated abrasion experiment is used to figure out the decay law of skid resistance of
High Friction Surfaces with different kind of aggregate.
Asymptotic exponential model was adopted to analyze the decay law of texture depth
and pendulum type friction coefficient of asphalt pavement with different aggregates. The
model could well reflect the relationship between skid resistance and the cumulative polish-
ing cycles. Asymptotic exponential model could be used to predict the decayed condition of
skid resistance for long-term monitoring of service life of the asphalt pavement. It can be
concluded that the texture depth and pendulum type friction coefficient tend to decrease
gradually as the cumulative polishing cycles increasing. However, pendulum type friction
coefficient increased at first 10000 cumulative polishing cycles for the asphalt membrane of
aggregate surface falling off. The performance of aggregate itself affected skid resistance per-
formance of pavement. Adding some aggregate with skid resistance, will become economic
means to improve the skid resistance of High Friction Surfaces in highway construction.
REFERENCES
Girum. Awoke. 2011. Estimating The Friction Performance of Hot Mix Asphalt Pavements Based on
Aggregate Properties and Route Characteristics: Analysis, Modeling and Validation. Maryland: Uni-
versity of Maryland.
Ibrahim M. Asi.2007. Evaluating skid resistance of different asphalt concrete mixes. Building and Envi-
ronment 42 (2007): 325–329.
Peter Kotek, Zuzana Florková. 2014. Comparison of the Skid Resistance at Different Asphalt Pavement
Surfaces over Time. XXIII R-S-P seminar, Theoretical Foundation of Civil Engineering (23RSP)
(TFoCE 2014). Procedia Engineering 91 (2014) 459–463.
Qiu Zhixiong, Li Zhi, Li Donghai, et al. 2013. Anti-slide properties of meso-stucture of dense skeleton
asphalt mixture. Journal of south China University of Technology.41(4): 107–112.
Stephen. N. Goodman. 2009. A thesis submitted to the Faculty of Graduate Studies and Research in
Partial Fulfillment of the requirements for the degree. Ottawa-Carleton: Department of Civil and
Environmental Engineering Carleton University.
Wang Xiuyun. 2012. Study on the viscoelasticity antiskid pavement used for the steep slope sections.
Chongqing, Chongqing Jiaotong University.
Zhou Zhihong. 2012. Affecting Factors and Experiments of the Skid Resistance of the Surface of
Asphalt Layers. Shanxi science and technology. 27(1):86–87.
1575
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Currently in China, pavement of many road sections had been damaged far
earlier than the service life designed due to heavy-load traffic. Thus, a particular research was
implemented in China, on the support of the freeway from Zhungeer to County Xinghe in
inner mongolia autonomous region, to study on the influence of heavy load traffic on cement
concrete pavement and load control. The added thickness of pavement structure has been
calculated. The influences of factors on the maximum load-bearing capacity were analyzed.
The factors were temperature grade, loading location, concrete flexure strength and reliabil-
ity coefficient. According to the actual situation, the maximum axle loads for pavement in
heavy-load direction and light-load direction were calculated. The control value of axle load
and economical axle load were gained based on load stress level and accumulated action
times of axle load, with modified fatigue stress calculating formula.
1 INTRODUCTION
Transportation system in China has developed greatly recent years. Up to early in the year
2015, there has been freeway built for over 0.1 million kilometers. At the same time, the pro-
portion of heavy-duty trucks has increased gradually. Heavy load has become a general issue,
especially in the regions with large-scale coal mine transportation.
Stick to similar situations, researchers around the world had conducted specific researches
and practices, which were significant references. Typical composite pavement structure for heavy-
load traffic was proposed in Belgium, which added the thickness and strength and adopted base
stabilized by hydraulic binder (G, 2012). Lean cement concrete was applied as the base course for
the pavement of heavy-load highway in Brazil, Australia, France and America (Gulyas, 2012.).
In china, Professor Wang (1999) had done researches about the fatigue calculating formula,
axle load conversion and fatigue coefficient. Professor Wang (2004) in Chang’ an university had
studied on the cement concrete pavement design method under heavy-load traffic.
The support construction for this research was the freeway from Zhungeer to county Hex-
ing in china’s inner mongolia autonomous region. Z-H freeway was the first heavy-load free-
way in china, whose pavement structures were designed differently in the heavy-load direction
and the light-load direction. Due to the high rate of trucks and ubiquitous overweight trans-
portation, pavement diseases had arose at the initial stage of using.
In this paper, the influence of heavy load on pavement fatigue life was analyzed. The fac-
tors that had great effects on the maximum load-bearing capacity were studied, including
temperature grade, loading location, concrete flexure strength and reliability coefficient. The
maximum axle load, the traffic control value of axle load and economical axle load for Z-H
freeway were proposed, which could offer valuable reference for heavy-load pavement design
and axle load control of similar projects.
In china, the damage of traffic to pavement structure was mainly due to the repeated fatigue
damage of over-load vehicle load. The current pavement design took 100 kN as standard
1577
axle load, but practical measured data were much bigger, on which conditions the thickness
of pavement structure must be increased to prevent pavement premature failure. The added
thickness of each structure was calculated with the theory that bottom tension stresses of
sub-base course under axle loads with different quantities were equal to the one under stand-
ard axle load (Table 1).
Table 1 showed that each index except for fatigue life increased about 19% when 20 kN was
added to the load each time, the stress(strain) quantity was doubled when the load quantity
increased from 100 kN to 200 kN, the added thickness was 37 cm when the load increased to
200 kN, and the fatigue life decreased obviously with the increasing of axle load.
To analyze the influence of heavy load traffic on pavement structure fatigue life, each typical
cement concrete pavement structure of freeway and first-class highway with particular design
parameters (Table 2) were chosen as the objects of research.
Based on the original traffic amount, considering 60% vehicles of the total amount were
over-load, the fatigue lives with varying ratios of overloading were calculated (Table 3).
Pavement structure
Freeway 30 5 0.2
First-class highway 20 4 0.2
1578
It’s showed that if pavement structures of different highway classes were designed based on
standard accumulative axles under full load, service life would drop with heavy load traffic.
Considering 60% vehicles of the total amount are over-load, if the ratio of overloading was
30%, the service life would be reduced to 14.99 years for freeway and 11.17 years for first-
class highway. Meanwhile, if the ratio of overloading was 100%, the service life would be 3.09
years for freeway and 2.14 years for first-class highway.
The studied pavement structure in the heavy load traffic direction differed from the one in the
light road direction (Table 4).
According to the regulations about ultimate condition in Specifications for Design of High-
way Cement Concrete Pavement (2011), the maximum axle load could be back-calculated
with Equation 1.
1
⎡⎛ f ⎞ ⎛ 1.45 × 10 −3 0 65 −2 ⎞ ⎤ 0.94
Pm ≤ ⎢ r − σ t,t ⎟⎠ ⎜⎝ k k ⋅ ⋅r h ⎥ (1)
⎣⎝ rr
a r c
1 + Db Dc−1 g c ⎟⎠ ⎦
where Pm = maximum axle load, fr = flexural strength, rr = reliability coefficient, σt,max = maxi-
mum temperature stress, kr = stress reduction coefficient, kc = synthesis coefficient, Db = section
flexural rigidity of the bottom slab, Dc = section flexural rigidity of the top slab, rg = relative
rigidity radius, hc = thicknesses of cement concrete slab. In this research, kr = 0.87, kc = 1.15.
Thicknesses/cm
Structural configuration Heavy load traffic direction Light load traffic direction
Temperature grade ºC/m Heavy load traffic direction Light load traffic direction
0 496 407
20 465 368
40 435 329
60 404 291
80 374 253
86 364 242
89 360 236
92 355 230
100 343 215
1579
According to the results, the influence was obvious. Regardless of the temperature grade,
the maximum axle loads of pavement for heavy load traffic direction and light load traffic
direction were 496 kN and 407 kN, which increased 37.8% and 72.5% respectively, compared
to the condition when temperature grade was 89ºC/m.
The maximum temperature grade was rare in practice, and the general grade was much
less than the specified value for most of the time. The grade was less than 1/3 of the specified
value even in summer when grade reached the top, So 1/2 of the specified value should be
adopted in the calculation.
Considering the influence of pavement thickness on temperature grade, the grads was deter-
mined 32ºC/m for heavy load traffic direction and 36ºC/m for light load traffic direction.
Based on the calculation results, the maximum axle loads of pavement for heavy load traf-
fic direction and light load traffic direction were 447 kN and 337 kN, which increased 34.2%
and 42.8% respectively, compared to the condition when temperature grade was 89ºC/m.
Flexure strength/MPa Heavy load traffic direction Light load traffic direction
5 365.27 239.02
5.5 410.00 271.75
6 454.48 304.48
Flexure strength/MPa Heavy load traffic direction Light load traffic direction
5 409.47 266.03
5.5 456.74 300.56
6 504.02 335.23
1580
4.1.4 Reliability coefficient
The impact of reliability coefficient on the maximum axle load was studied when the coef-
ficient changed from 1.09 to 1.33 (Table 9).
It’s seen that the maximum axle load tended to decrease when reliability coefficient
increased. Considering the actual situation for Z-H freeway, the pavement structure was
capable and safe to a great extent, because of the fact that the thickness of concrete slab was
bigger than the general size in china, and the flexure strength of the adopted was much bigger
than designed. So in this research, the reliability coefficient could vary from 1.26 to 1.12, and
the corresponding maximum axle load was 513 kN for heavy load direction and 391 kN for
the light load direction.
Flexure strength/MPa Heavy load traffic direction Light load traffic direction
5 359.84 236.04
5.5 404.64 268.72
6 449.26 301.45
Reliability coefficient Heavy load traffic direction Light load traffic direction
1581
Table 10. Multi-factor influenced maximum axle load.
Factor
Table 11. Relationship among axle load, tire pressure and contact area.
Maximum
Load stress/ temperature Temperature Maximum Action times
Direction MPa stress/MPa fatigue stress/MPa axle load/kN of axle load
When axle load changed from 60 kN to 300 kN, the relationship among axle load, tire
pressures and contact areas were gain (Table 11).
S lg .084 0.0432 lg N
lg1 (3)
1583
5.2 Economical axle load
In this research, the economical axle load was identified by pavement service life, 30 years.
Considering the actual situation and factors such as traffic amount and traffic growth rate,
the accumulated action times of axle load at the end of 30 years was calculated as 1.396 × 107.
The fatigue stress calculating formula,Equation 3, was transformed to Equation 4. The eco-
nomical load stress level was easily gained, 0.55, and the corresponding axle load was 282 kN
in heavy load direction and 215 kN in light load direction (Table 14).
6 CONCLUSIONS
1. Fatigue life and service life decreased obviously with the increasing of axle load.
2. The maximum axle loads was proportional to concrete flexure strength.
3. The control value of axle load should be determined by temperature grade and reliability
coefficient, regardless of load location because of its randomness.
4. Based on the features of curve of S(stress level) and N(accumulated action times of axle
load), the control level of load stress for cement concrete pavement should be 0.75, which
offered references for freeway load control.
5. A 30-year service life considered, the economical load stress level for cement concrete
pavement was easily gain with formula S = 10 lg 1.084 − 0.0432 lg N , and the value was 0.55.
REFERENCES
Andras, Gulyas. 2012. Axle load trends in Hungary and their effects on pavement structural design.
Procedia-social and behavioral sciences 48: 888–896.
Binggang, Wang & Mulian Zheng (ed.) 2004. Design and construction of cement concrete pavement.
Beijing: China communication press.
Luis, G. et al. 2012. Evaluation of truck factors for pavement design in developing countries. Procedia-
social and behavioral sciences 53(3):1139–1148.
Ministry of Transport of the People’s Republic of China (ed.) 2011. Specifications for design of highway
cement concrete pavement JTG D 40-2011. Beijing: China communication press.
Sili, Li, Bo, Tian, & Kaimin, Niu. 2012. Research on critical loading position of cement concrete pave-
ment. Journal of highway and transportation research and development 29(7): 1–8.
Xuancang, Wang et al. 1999. Study of the concrete pavement of running heavy vehicle. China journal of
highway and transport. 12(1): 14–20.
Zhaohui, Liu. 1999. Analysis on the effect of overload truck on cement concrete pavement. Journal of
highway and transportation research and development 16(1)1: 5–8.
1584
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Zhang Jin Xi
Beijing Key Laboratory of Traffic Engineering, Beijing University of Technology, Beijing, China
Jing Chao
Beijing Key Laboratory of Traffic Engineering, Beijing University of Technology, Beijing, China
School of Civil Engineering and Mechanics, Yanshan University, Qinhuangdao, Hebei, China
ABSTRACT: In order to know the working conditions of asphalt pavement, sensors were
installed in test road and used to track and analyze the changes of working conditions. About
3 years data related to working conditions have been collected and analyzed till now. The
results showed that the temperature at the bottom of top layer varies from 33°C to −16.8°C
during a year. The average temperature gradient in pavement is 0.45°C/cm in summer and
−0.1°C/cm in winter. The results also showed that the transverse strain on bottom of top
layer of pavement is between 400–800 με in summer, but it is only between 60–100 με in win-
ter. The compressive stress on top of base course is smaller in winter than that in summer.
Some conclusions about the change trend of working conditions is given in the paper.
1 INTRODUCTION
The working conditions for asphalt pavement refers to the temperature, humidity, stress,
strain and other working conditions of asphalt pavement in use. Research and applications
has shown that asphalt pavement performance and service life greatly depend on its working
conditions, and there is a close relationship between the deterioration of asphalt pavement
performance and its working conditions. Asphalt pavement with different working condi-
tions indicate different deterioration trends. Special working conditions may cause unpre-
dictable damage to the pavement. Therefore, accurate understanding and a good knowledge
of pavement working conditions have great significance for pavement structural design and
performance prediction.
Asphalt pavement working conditions are affected by many factors, including the natu-
ral environment, seasons, other natural conditions, the traffic, traffic axle load, other trans-
port conditions, material’s properties, and structural thickness. Therefore, to have a good
knowledge of pavement working conditions, certain pavements are constantly monitored
for reliable data. At present, domestic and foreign researchers have studied the temperature
field of asphalt pavement with theoretical analysis and statistical analysis and proposed the
correlation between the temperature of asphalt pavement and the environment, time and
depth. Many studies examine the internal stress and strain distribution in pavements under
the load, but most of them have numerical simulation analysis, partly through a static test,
and partly through a short-term continuous test. Few studies investigate the internal mechan-
ical states of pavement through long-term monitoring. In order to understand the variations
in pavement structural working conditions, the article studies the test road built by Beijing
Expressway, measures the temperature, stress, strain and other pavement conditions through
continuous monitoring with embedded monitoring equipment.
1585
2 TEST ROAD CONSTRUCTION
The test road is located in Beijing’s outer ring road of the Sixth Ring Road from K84+000
to K104+000. It was repaired in Sep, 2012 because the road suffered from serious surface
damage and rutting failure due to the heavy traffic and the large numbers of heavy vehicles.
Maintenance and Rehabilitation (M&R) operation includes milling the top layer and the
middle layer and paving them (back) with the same layer thickness. The contrast of Materials
and Structures of the test road before and after M&R is shown in Table 1.
The present study uses the monitoring equipment including strain sensors, pressure sen-
sors and temperature sensors, as shown in Figure 1. The technical parameters of sensors are
shown in Table 2.
The study uses DTMCU80G-20 Remote Monitoring Unit for data acquisition and stor-
age. The minimum sampling interval of collecting equipment is 1 second, and all collectors
are placed in a closed metal cabinet to avoid them being stolen or getting wet as shown in
Figure 2. Repeated tests are conducted to ensure the reliability of data and the performance
of storage devices. The results show that a frequency of 30times/min. should be chosen for
data acquisition and storage.
All the monitoring equipment is arranged at the bottom of the wheel-path area, and the
layout scheme for the vertical profile of the monitoring equipment is shown in Figure 3.The
monitoring equipment of transverse strain is arranged at the bottom of the top and bottom
layers to monitor the transverse strain conditions. Longitudinal strain monitoring equipment
is also arranged at the bottom of the top and bottom layers, to monitor longitudinal strain
conditions. Pressure monitoring equipment (pressure cell) is arranged on top of the base
course to monitor the stress there. The temperature monitoring equipment is arranged at the
bottom of the top layer and the bottom layer and 5 cm into the base course for monitoring
the temperature of different layers of pavement.
1586
Figure 2. DTMCU80G-20 data collector.
1587
Figure 4. Temperature variations in different layers on the selected days.
The data show that the highest and the lowest temperature for the bottom of top layer
(5 cm below the road surface) are 33.8 degrees and −16.8 degrees; the inside of base course
(24 cm below the road surface) are 25.2 degrees and −15.1 degrees and the bottom of bot-
tom layer (19 cm below the road surface) are 27.5 degrees and −15.1 degrees all year around.
The highest and lowest temperature occur in August and January. As shown in Figure 5 (d
represents the distance from the road surface, unit: m), the temperature gradient of Beijing
highway pavement can be obtained from the above data. The average temperature gradient
in the pavement is 0.45°C/cm in summer and −0.10°C/cm in winter.
Vt (T i −T i ) (t (i ) − t2 ( i ) ) (9)
where Tmax(i) = the highest temperature on any day; Tmin(i) = the lowest temperature on any day;
t1(i) = the moment of highest temperature on any day; t2(i) = the moment of lowest temperature
on any day.
Cooling rate variations in different parts of asphalt pavements is shown in Figure 6, for the
period Dec 1, 2012 to Nov 30, 2013. As can be seen in Figure 6, cooling rate variations at differ-
ent depths are basically consistent, but the cooling rate inside pavement decreases with depth.
The maximum cooling rate inside pavement occurs in June. June 21st, 2013 is the summer
solstice (Chinese lunar calendar). The daytime is the longest and the radiation is the strongest
on summer solstice in northern hemisphere, so temperature difference in cooling period of
1588
Figure 5. Temperature gradient in pavements.
a day is the largest. The cooling rate of internal pavement on the summer solstice (Chinese
lunar calendar) reaches maximum values, about 0.6∼1.2 °C/h. It is only about 0.2∼0.4 °C/h in
winter as the road is at a low temperature for a long time. Therefore, the probability of fatigue
cracking caused by temperature cooling is relatively low, and we should pay much attention
to the problem of low-temperature cracking of asphalt mixture.
These results also show that the higher temperature on asphalt pavement in summer is one
of the main causes of rutting. Meanwhile, the cooling rate inside asphalt pavement in sum-
mer is about three times as much as that in winter. Further analysis is required to examine
the impact of rapid temperature drops on asphalt pavement and whether it can aggravate the
temperature fatigue cracking occur.
1589
Figure 7. Histogram of temperature distribution.
Time of low, moderate and high temperature period for the bottom of the top layer is 132,
188 and 45 days which account for 36%, 52% and 12% of a year. In addition, there are 3 days
in which temperature is higher than 30 °C in total.
Time of low, moderate and high temperature period for the bottom of the bottom layer is
140, 179 and 46 days which account for 38%, 49% and 13% of a year.
Time of low, moderate and high temperature period for the base course is 147, 181 and 37
days which account for 40%, 50% and 10% of a year.
As can be seen from the results above, high temperature period for middle and bottom
layer of asphalt pavement also have a long duration, so there may be a problem to some
extent if we consider anti-rutting in the top layer only, neglecting anti-rutting in middle and
bottom layers. Moreover, the top layer of asphalt pavement is in the high temperature period
for about 45 days. This period is the key time for the asphalt pavement rutting, and should be
high-lighted in the M&R and traffic management. These results also show that the pavement
is in the low and moderate temperature period for about 85% of the year, and equal atten-
tion should be paid to the occurrence of asphalt pavement damage and creep properties of
asphalt mixture.
1590
Figure 8. Distribution of strain in the bottom of top layer (at September 13, 2013).
Figure 9. Distribution of strain in the bottom of bottom layer (at September 13, 2013).
Figure 10. Distribution of compressive stress on the top of base course (at September 13, 2013).
1591
Table 3. Distribution of strain and compressive stress in pavement.
Proportion (%)
Though a large proportion of strain and compressive stress are in the normal range
throughout a year, there are a proportion of the monitoring data above normal range, espe-
cially for the strain at the bottom of the top layer, which is proved by a greater proportion
monitoring data above 800 με. A high-strain working condition is generated into pavement
because of the influence of environment or heavy load traffic.
4.2 Distribution of maximum value of strain and compressive stress all year round
This article aimed at monitoring data of pavement working conditions collected from Sep
28th, 2012 to Sep 24th, 2013. A statistic of the maximum absolute value of strain or compres-
sive stress has been made respectively, and the variation day by day charts of the maximum
absolute value of transverse and longitudinal strain on the bottom of the top and bottom lay-
ers and maximum value of compressive stress on the top of base course are shown in Figure 11
to Figure 13. (Monitoring data between March 14th, 2013 and May 12th, 2013 is lack).
Seen from Figure 11 to Figure 13, the strain in the bottom of top layer is larger than that
of the bottom layer and the transverse strain is generally larger than that of the longitudinal
strain in each layer of asphalt pavement. The strain within asphalt pavement in anyone of
June, July, August and September is much higher than other months, especially the strain
in the bottom of top layer is very large in summer. Moreover, the compressive stress on the
top of base course in anyone of June, July, August and September is much higher than other
months in summer.
Monitoring data also indicates:
Seen from the maximum value of strain and compressive stress in pavement, the maximum
value is smaller in winter (from Nov 2012 to Mar 2013). The maximum value of strain on the
bottom of the top and bottom layers and the compressive stress on the top of base course are
all below 300 με, 100 με, and 200 Kpa respectively.
The maximum value of strain and compressive stress in pavement is larger in summer
(from Jun to Oct). The maximum value of strains on the bottom of the top layer are mainly
1592
Figure 11. Maximum longitudinal and transverse strain in the bottom of top layer.
Figure 12. Maximum longitudinal and transverse strain in the bottom of bottom layer.
300 με above (account for 81%), and the highest one reaches up to 5438 με; the maximum
value of strains on the bottom of the bottom layer are mainly 100 με above (account for
90%), and the highest one reaches up to 929 με; the maximum value of the compressive stress
on the top of base course are mainly 200 Kpa above (account for 39%), and the highest one
reaches up to 652 Kpa.
5 CONCLUSION
1. Method of sampling and monitoring at random, the paper used for testing the working
conditions in pavement, can indicate the characteristics of working conditions in asphalt
pavement comprehensively and has an extensive application.
2. In the Beijing area, the highest and the lowest temperature for the bottom of top layer
are 33.8 degrees and −16.8 degrees all year around, respectively. The average temperature
gradient in pavement is 0.45 °C / cm in summer and −0.10 °C / cm in winter.
1593
3. The cooling rate of pavement around the lunar calendar summer solstice is the biggest and
is about 0.6∼1.2 °C/h; while it is only about 0.2∼0.4 °C/h in winter as the pavement is in
low temperature condition for a long time.
4. Time is about 45 days for the top layer of asphalt pavement under the high temperature
period in Beijing while it is about 85% of a year for the pavement in low and moderate
temperature period. It should be paid equal attention to the occurrence rule of asphalt
pavement damage and creep properties of asphalt mixture.
5. The data indicates that the values of strain and compressive stress for the pavement within
each layer in summer are higher than those in winter, which indicates asphalt pavement at
an unfavorable working condition in summer.
6. Though a large proportion of strain and compressive stress is in the normal range through-
out a year, there are a proportion of the monitoring data above normal range, especially
for the strain at the bottom of the top layer, which is proved by a greater proportion moni-
toring data above 800 με. A high-strain working condition is generated into pavement
because of the influence of environment or heavy load traffic, and we need pay attention
to the design and M&R.
It is only one test road pavement whose working conditions are monitored and studied in
this article, and there is a certain degree of limitation for the sample size and extent of the rep-
resentative. Now the process of monitoring is still ongoing and there is not obvious damage
for the test road because of the short operating hours. Although the monitoring data obtained
is still insufficient to establish an exact relationship between the temperature and the asphalt
pavement damage, comprehensive work would be get a further development in the future.
ACKNOWLEDGEMENTS
This research was funded by “Maintenance decision theory and method of asphalt pavement
considering multi-objective demand during whole life cycle” project from National Natural
Science Foundation of China (No.: 51478021) and “Study on Mechanism and Prediction
Model of Asphalt Pavement Deterioration Used for Durability Design and Maintenance
Decision” project from Beijing Natural Science Foundation and Science and technology pro-
gram of Beijing Municipal Commission of Education (No.KZ201510005002).
REFERENCES
Brian Scott et al. 2005. The Aurora Consortium: Laboratory and Field Studies of Pavement Tempera-
ture Sensors. Minnesota Department of Transportation Research Services Section Publishing.
Basang Dun-Zhu & HUANG Xiao-Ming. 2006. Analysis for fatigue life of asphalt pavement temperature
crack in Tibet. Journal of SOUTHEAST University: Natural Science Edition 36(6): 1014–1017.
Huang Li-Kui et al. 2005. Statistical Analysis of Asphalt Pavement Temperature Profile. Central South
Highway Engineering 30(3):8–10.
Hu Xiao-Yuan. 2011. Measure and Analysis on Thermal Field of Asphalt Pavement. Xi’ an: School of
Road Engineering Chang’ an University.
Kang Hai-Gui et al. 2007. Regression Analysis of Actual Measurement of Temperature Field Distribu-
tion Rules of Asphalt Pavement. China Journal of Highway and Transport 2007(11): 14–17.
Qin Ying-Hong & Jacob, E.H. 2011. Modeling temperature distribution in rigid pavement slabs. Con-
struction and Building Materials 25(9): 3753–3761
Sun Li-Jun & Qin Jian. 2006. Prediction Model on Temperature Field in Asphalt Pavement. Journal of
TONGJI University (Natural Science) 34(4): 481–483.
Wang Chang-Heng et al. 2007. Prediction Model on Temperature Field in Asphalt Pavement. Central
South Highway Engineering 32(3): 113–115.
Wei Lu. 2011. Prediction Model on Temperature Field in Asphalt Pavement of Germany. Journal of
China & Foreign Highway 31(6): 118–121.
1594
Traffic Engineering
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: According to the World Health Organization, traffic accidents may cost a
country between 1 and 3 percent annually of Gross national product. In low and middle-
income countries (LMICs), vulnerable road users are dominant in urban areas with the least
physically protected group; pedestrian constituting the highest share in traffic fatalities. These
concerning road safety figures require immediate actions to be taken.The majority of availa-
ble methodologies for assessing pedestrian safety conceptualize in developed countries where
traffic characteristics and pedestrian behavior could be different from LMICs. Accordingly,
there is a need to propose, test and apply a new modelling approach for pedestrian safety
applicable in LMICs urban roads.This paper cover the first research activity to address this
objective by conducting a theoretical review of available models to assess road safety and
explore what effecting variables are correlated with pedestrian perceptions of safety in urban
areas and determine applicability in developing countries.
1 INTRODUCTION
The high toll of fatalities and serious injuries due to traffic accidents is considered as one
of the major health problems. Therefore, policies and engineering interventions need to be
design to mitigate this problem at local and global scale.
In less developed countries, vulnerable road users are dominant in urban areas with the
least physically protected group; pedestrian (Downing eat al., 2000). Consequently, pedes-
trian constitutes significantly higher share in road traffic fatalities and injuries in low and
middle-income countries (LMICs) than in high-income countries (World Health Organaza-
tion, 2013; Ogendi et al., 2013).
Road networks and infrastructure in transitional economies carrying traffic volumes and
mixture far beyond its design Due to rapid urbanisation and motorisation leading to traf-
fic delay, congestion and accidents in urban centres. Furthermore, urban areas experiencing
a high rate of conflict between pedestrian and motorised traffic because of heterogeneous
traffic, incompatible land uses and unplanned growth. This is lead to uneven competition
between different road users groups, high risk exposure, significant traffic speed variation
and driving deterioration the road network (Ross et al., 1994).
Several safety models investigate the traditional measures traditional traffic measures such as
traffic speed, volume and geometric design; however some measures could affect the pedestrian
safety of developing areas such as road side activities, traffic speed variation and land uses.
This study present a number of examples of pedestrian safety models together with imple-
mentation issues with the view to underlying the effecting factors on pedestrian safety and
perceptions.
1598
Ped Vol: annual average daily pedestrian volume,
and X3 … Xn represent other site specific feature such as speed limits, number of lanes,
type of median left turn volume percentage and the presence or absence of crosswalks.
In spite of model functional form similarity for these three studies, Zeeger et al. (2005)
looked at intersection and non-intersection sites whereas Lyon and Persaud (2002) and
Leden (2002) focused on intersection sites.
Based on literature review, most of pedestrian safety modelling studies were based on devel-
oped countries context, which show a different traffic characteristics, user behaviour and
enforcement policies, there is a need to develop models taking the developing case into con-
sideration. It been felt that traffic speed variation, land uses and frequent side friction or
side activities are among the important differences. The effect of these factors on pedestrian
safety on urban roads not yet well understood. Inclusion of these factors may improve the
developed models to assess pedestrian safety.
4 iRAP METHODOLOGY
In 2006, the concept of star rating score to assess road safety was further developed and
extended to be used in low and middle income countries under the International road assess-
ment program (iRAP). This methodology provides a comprehensive approach with the
objective of assessing the road safety provided by road infrastructure for each road user
category to develop tools and cost-effective countermeasures to be applied in LMICs. The
iRAP methodology employs the star rating concept to reflect the effect of road design, mul-
tiple countermeasures and treatments, operating traffic speed and external flow influence in
quantitative rating score.
Star rating score is the calculated outcome of the approach and transformed into equiva-
lent star rating to express the safety provided by the road infrastructure in simple way.
A Star Rating Score (SRS) is calculated for each road user category including pedestrian
for each 100 metre road segment, using the following functional form (iRAP, 2014):
Where:
SRS represents the quantified relative risk of fatality and serious injury for a road user;
and
Crash Type Scores = Likelihood × Severity × Operating speed × External flow influence
Where:
• Likelihood is a quantitative result of multiplying road attribute factors contributing to the
possibility of an accident will be initiated.
• Severity represent the contribution of specific road features for the severity of an
accident
• Operating speed represents factors that account for rate of changes in risk with speed
• external flow influence refers to the level to which a road user`s risk of being subject to
accident is dependent of another road user
For pedestrian, two crash types considered; walking along the road and crossing the road.
However, crossing the road is further divided for calculation purposes into two directions; the
inspected road and the side road.
iRAP methodology providing an approach to prepare improvement procedure (Safer
Roads Investment Plans) and presenting an economic appraisal for the needed infrastructure
countermeasures to upgrade road safety. The concept enable users to estimate the risk of
crashes in each 100 meters road segments, and a further estimate of serious casualties and
fatalities can be done in each segment under the prevailing condition. The approach finally
used to prioritise and optimise the potential countermeasures within available budget.
The economic appraisal can be done by comparing the economic benefits of human casu-
alties might be prevented due to implementing one or more of 93 potential countermeasures
with implementation cost.
However, this approach is set to assess road safety in interurban roads and need to be fur-
ther developed to enable assessment of urban road (Lynam, 2012).
5 DISCUSSION
According to literature review, the road traffic safety problem is more severe in developing
countries. Planners and authorities concentrate on providing the infrastructures to avoid the
1602
delay and congestion of motorised traffic on the expense of other road users. The lack of
infrastructure allocated for pedestrian and other non-motorised road users leads to share the
available facilities which designed for high speed vehicles exposing them to crash risks. As
example, the share of pedestrian deaths ranged between 40–60% of total road traffic fatali-
ties between 1990 and 1999 in Delhi (Tiwari, 2001a). Moreover, despite the increasing focus
on traffic safety studies in Beijing city, the accidents data records revealed that 80% of severe
accidents involved non-motorised road users in mixed traffic urban arterial from 2004 to
2007 (Ma et al., 2010).
This paper cover the first research activity to address this objective by conducting a theo-
retical review of available models to assess road safety and explore what effecting variables
are correlated with pedestrian perceptions of safety in urban areas. The second research
activity includes the development of pedestrian safety model, which will investigate the fac-
tors influencing pedestrian safety in developing urban areas.
The study basically used a stated preference survey in which each roadway segment will
rated based on a fixed scale. The methodology going to have respondents on selected sites
and rate these segments with respect to perceived safety they would be walking and crossing a
under the roadway conditions. This is can used to investigate the significant factors effecting
the individuals’ perceived safety and also to compare the risk of walking alongside roads with
risk of crossing same roads. Video clips could be used for comparison.
The model provide traffic planners and others the ability to rate the roadways according to
pedestrians’ satisfaction and may be used in the process of evaluating existing roads, prioritize
the roads in need for safety interventions, designing new roads, or redesigning existing roads.
REFERENCES
Agarwal, N.K. 2011. Estimation of Pedestrian Safety at Intersections Using Simulation and Surrogate
Safety Measures. University of Kentucky.
Al-Ghamdi, A.S. 2002. Pedestrian–vehicle crashes and analytical techniques for stratified con-
tingency tables. Accident Analysis & Prevention, 34(2), 205–214. http://doi.org/10.1016/
S0001–4575(01)00015-X
Al-omari, H, Obaidat, S. 2013. Analysis of Pedestrian Accidents in Irbid City, Jordan, 1–6.
Archer, J. 2005. Indicators for traffic safety assessment and prediction and their application in micro-
simulation modelling: A study of urban and suburban intersections. Doctoral Dissertation, Royal
Institute of Technology, Department of Infrastructure, Division for Transport and Logistics, Centre
for Transport Research. http://doi.org/ISBN 91–7323–119–3
Baltes, M., & Chu, X. 2002. Pedestrian Level of Service for Midblock Street Crossings. Transportation
Research Record, 1818(1), 125–133. http://doi.org/10.3141/1818–19
Bang, L. 1995. Impact of Side Friction on Speed-Flow Relationships for Rural and Urban Highways.
SWEROAD, Indonesia, (July), 27.
Chand, Sai, Chandra, S. 2014. Impact of Bus Stop on Urban Traffic Characteristics –a Review of
Recent Findings. Journal of Society for Transportation and Traffic Studies (JSTS), 5(2), 57–72.
Cheng, Wei, Zhang, Ning, Li, Wei, Xi, J. 2014. Modeling and Application of Pedestrian Safety Conflict
Index at Signalized Intersections. Discrete Dynamics in Nature and Society, 2014, 1–6. http://doi.
org/10.1155/2014/314207
Chiguma, M.L. 2007. Analysis of side friction impacts on urban road links. Case study, Dar-es-salaam.
Doctoral Thesis submitted to Royal Institute of Technology, Stockholm, Sweden.
Derry, J, Ebel, E, Mock, N, Afukaar, F, Donkor, P. 2010. Pedestrians’ injury patterns in Ghana. Acci-
dent; Analysis and Prevention, 42(4), 1080–8. http://doi.org/10.1016/j.aap.2009.12.016
Downing, A., Jacobs, G., Thomas, A., Sharples, J., Silcock, D., Lottum, C., Walker, R., Ross, A. 2000.
Review of road safety in urban areas. Department for International Development, World Bank (
Unpublished), (3142272).
Elvik, R., Hoye, A., Vaa, T., Sorensen, M. 2009. The Handbok of Road Safety Measures. Emerald
Group Publishing Limited, second edition.
Elvik, R., Mysen, B.A. Vaa, T. 1997. Trafikksikkerhetshåndbok (Traffic safety handbook). Institute of
Transport Economics, Oslo,Norway.
Finch, DJ, Kompfner, P, Lockwood, CR, Maycock, G. 1994. Speed, speed limits and accidents. TRL,
London, UK.
1603
Frith, J, Patterson, L. 2001. Speed Variantion, Absolute Speed and their Contribution to Safety, With
Special Reference to the Work of Solomon. IPENZ Transportation Group, 1–8.
Garber, N.J., Gadiraju, R. 1988. Speed Variance and its Influence on Accidents. AAA Foundation for
Traffic Safety, Washington D.C., USA.
Garber, Nicholas J., Ehrhart, A.A. 2000. The Effect of Speed, Flow, and Geometric Characteristics on
Crash Rates for Different Types of Virginia Highways. Virginia Transportation Research Council.
http://doi.org/10.1017/CBO9781107415324.004
Ha, H.-H., & Thill, J.-C. 2011. Analysis of traffic hazard intensity: A spatial epidemiology case study
of urban pedestrians. Computers, Environment and Urban Systems, 35(3), 230–240. http://doi.
org/10.1016/j.compenvurbsys.2010.12.004
Hall, R. 1986. Accidents at four-arm single carriageway urban traffic signals. Transport and Road
Research Laboratory, Crowthorne., (TRRL Contractor Report CR65).
Harwood, D.W., Torbic, D.J., Gilmore, D.K., Bokenkroger, C.D., Dunn, J.M., & Zeeger, C.V., et al.
2008. NCHRP Web-Only Document 129, Phase 3: Pedestrian Safety Prediction Methodology. Wash-
ington, DC.: Transportation Research Board, (March 2008). Retrieved from http://onlinepubs.trb.
org/onlinepubs/nchrp/nchrp_w129p3.pdf.
Hidayati, N., Liu, R., & Montgomery, F. 2012. The Impact of School Safety Zone and Roadside Activi-
ties on Speed Behaviour: the Indonesian Case. Procedia—Social and Behavioral Sciences, 54, 1339–
1349. http://doi.org/10.1016/j.sbspro.2012.09.848
Hunter, W, Stutts, C, Pein, E, Cox, L. 1996. Pedestrian and Bicycle Crash Types of the Early 1990’S,
149–152.
Hunter, W., Charles, L. Carter, V. Zegeer, and J.R.S. 2007. Pedestrian and Bicyclist Intersection Safety
Indices: User Guide, (FHWA-HRT-06–130).
Jensen, S.U. 2008. Pedestrian and Bicyclist Level of Service on Roadway Segments. Transportation
Research Record, 2031(-1), 43–51. http://doi.org/10.3141/2031–06
Jonsson, T. 2005. Predictive models for accidents on urban links—A focus on vulnerable road users.
Lund Institute of Technology, Department of Technology and Society., 226. Retrieved from https://
lup.lub.lu.se/search/publication/24269
Kang, L., Xiong, Y., & Mannering, F.L. 2013. Statistical analysis of pedestrian perceptions of sidewalk
level of service in the presence of bicycles. Transportation Research Part A: Policy and Practice, 53,
10–21. http://doi.org/10.1016/j.tra.2013.05.002
Koushki, A, Ali, A. 2003. An analysis of pedestrian accidents in Kuwait. Kuwait Journal of Sci-
ence and Engineering, 30(1), 139–151. Retrieved from http://www.scopus.com/inward/record.
url?eid = 2-s2.0–2442640284&partnerID = tZOtx3y1
Lave, C.A. 1985. Speeding, coordination, and the 55-mph limit. The American Economic Review, 75,
1159–1164.
Leden, L. 2002. Pedestrian risk decrease with pedestrian flow. A case study based on data from signal-
ized intersections in Hamilton, Ontario. Accident Analysis & Prevention, 34(4), 457–464. http://doi.
org/10.1016/S0001–4575(01)00043–4
Lynam, D. (2012). Development of Risk Models for the Road Assessment Programme. RAP504.12
and TRL Report CPR1293, Published by iRAP and TRL. Retrieved from www.irap.org and www.
trl.co.uk.
Lyon, C., Persaud, B. 2002. Pedestrian Collision Prediction Models for Urban Intersections. Journal of
the Transportation Research Board, (Transportation Research Record 1818.).
Ma, M., Yan, X., Abdel-Aty, M., Huang, H., Wang, X. 2010. Safety Analysis of Urban Arterials Under
Mixed-Traffic Patterns in Beijing. Transportation Research Record: Journal of the Transportation
Research Board, 2193(-1), 105–115. http://doi.org/10.3141/2193–13
Munawar, A. 2011. Speed and Capacity for Urban Roads, Indonesian Experience. Procedia—Social and
Behavioral Sciences, 16, 382–387. http://doi.org/10.1016/j.sbspro.2011.04.459
Ogendi, J., Odero, W., Mitullah, W., Khayesi, M. 2013. Pattern of pedestrian injuries in the city of
Nairobi: implications for urban safety planning. Journal of Urban Health : Bulletin of the New York
Academy of Medicine, 90(5), 849–56. http://doi.org/10.1007/s11524–013–9789–8
Pulugurtha, S, Sambhara, R. 2011. Pedestrian crash estimation models for signalized intersections.
Accident; Analysis and Prevention, 43(1), 439–46. http://doi.org/10.1016/j.aap.2010.09.014
Ross, A., Baguley, C., Hills, B., McDonald, M., Silcock, D. 1994. Towards safer roads in developing
countries: A guide for planners and engineers. Transportation Research Laboratory, Overseas Devel-
opment administration.
Sawalha, Z., Sayed, T. 2001. Evaluating Safety of Urban Arterial Roadways. Journal of Transportation
Engineering, 2(April), 151–158.
1604
Solomon, D. 1964. Accidents on Main Rural Highways Related to Speed, Driver and Vehicle. US
Department of Commerce and Bureau of Public Roads, Washington D.C., USA.
Tiwari, G. 2001. Pedestrian infrastructure in the city transport system: a case study of Delhi. World
Transport Policy & Practice, 7(4), 13–18.
Ukkusuri, S., Miranda-Moreno, L., Ramadurai, G., Isa-Tavarez, J. 2012. The role of built environ-
ment on pedestrian crash frequency. Safety Science, 50(4), 1141–1151. http://doi.org/10.1016/j.
ssci.2011.09.012
Wang, K. and Liao, M. 2012. Modeling Pedestrian Degree of Satisfaction Based on Structure Equation
Model. ICLEM 2012: pp. 442–448.
Wedagama, M, Bird, N, Metcalfe, V. 2006. The influence of urban land-use on non-motorised trans-
port casualties. Accident; Analysis and Prevention, 38(6), 1049–57. http://doi.org/10.1016/j.
aap.2006.01.006
World Health Organazation. 2013. Pedestrian Safety: a road safety manual for decision-makers and
practitioners. Geneva. http://doi.org/10.4016/12660.01
Zeeger, C., Stewart, J., Huang, H., Lagerway, P., Feaganes, J., & Campbell, B. 2005. Safety Effects of
Marked Versus Unmarked Crosswalks at Uncontrolled Locations. Final Report and Recommended
Guidelines, Report No. FHWA-RT-04–100. Washington, DC.: Federal Highway Administration.
1605
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
A.K. Jameel
Civil Engineering School, University of Birmingham, Birmingham, UK
The Iraqi Ministry of Higher education and Scientific Research, Iraq
H. Evdorides
Civil Engineering School, University of Birmingham, Birmingham, UK
ABSTRACT: Road Star Rating (SR) is a system used by iRAP (International Road Assess-
ment Program) to quantify the level of road safety. It is based on an estimated score (SRS)
which is computed by considering road attributes affecting the road safety and subsequently
used to design investment plans for safer roads. However, the outputs of iRAP are four inde-
pendent SRs, each are for an individual road user group: vehicle occupant, motorcyclist,
pedestrian, and bicyclist. Therefore, this paper aims to develop an aggregated star rating to
quantify the overall level of roads’ risk. The reason behind this topic is simplifying the design
of the investment plan which depends on the iRAP’s SR and increasing the applicability of
the iRAP models for strategic decision making. Four statistical aggregation functions are
suggested to aggregate the four SRSs and each of them is then evaluated with the view to
select the most proper function.
1 INTRODUCTION
The iRAP (International Road Assessment Programme) has been used by government and
nongovernment organisations in about 70 countries including developed and developing
areas. The main aim of this programme is to assist road engineers and experts in improving
the level of road safety and to control the increase of road accidents and their severity lev-
els (iRAP 2015a). WHO (2009, 2013) reported that about 1.3 million people are killed and
50 million are injured yearly in road accidents. The majority of these accidents happened in
developing countries where it is estimated that they will increase by more than 80% in the
next few years (WHO 2009, McInerney & Smith 2010, EuroRAP 2011). Researchers and
road experts have shown that one of the main reasons behind that is the lack of traffic crash
data which is the main input of traffic safety studies (McInerney & Smith 2010). iRAP has
been designed to control the issue of crash data unavailability, especially in low and medium-
income countries by considering road attributes which have a significant effect on traffic
safety level (iRAP 2015a).
Road authorities have demonstrated that the crash severity and probability depend mainly
on roadway designs (Ossenbruggen et al. 2001, NRSS 2010, McInerney & Smith 2010). For
example, wide lanes or poor quality roadsides can significantly increase the crash severity
and likelihood (McInerney & Smith 2010). Therefore, improving the geometric design ele-
ments of highways and establishing modern safety control devices have been considered in
iRAP (Zaloshnja & Miller 2009). In addition, iRAP considers the road infrastructures, such
as footpaths and bicycle paths and their effect on the level of road safety (NRSS 2010). This
is because of the significant reduction in the level of road risk which has been achieved where
such infrastructures have been implemented (NRSS 2010, Stigson et al. 2008). Therefore,
Road attributes and infrastructure is the main variables in iRAP models and road safety
investment plan which are the objectives of iRAP to achieve its main aim.
1607
Figure 1. The conceptual framework of iRAP (iRAP 2015a,b).
The iRAP designers have identified the objectives of this programme which are: quantify-
ing the level of risk in terms of star rating scores, developing a safer roads investment plan
by suggesting economical infrastructure improvement based on diagnosing high risk loca-
tions (McInerney & Smith 2010, iRAP 2015b), and conduct economic studies which low and
middle-income countries have lacked. These objectives consider mainly the road attributes
which represent the road geometric and infrastructure characteristics, as well as the traffic
flow and speed. Figure 1 shows the conceptual framework of iRAP.
This paper will focus on the first objective which is the quantifying of safety level in terms
of star ratings.
The assessment of roads in terms of their safety level can help in diagnosing the dangerous
roads, the roads with high rate of traffic accidents, and in turn suggest a countermeasure
which can reduce the traffic accident rate. The iRAP assessment is in terms of star ratings
which are classified into five levels, one star to five stars. 5 stars and 4 stars represent the saf-
est conditions produced from well-designed roads, such as grade separated intersections and
good quality safety barrier, while 1star and 2 stars are the worst safe conditions generated
from poor road elements, such as undivided pavement and frequent curves and intersections
(Rogers & Bin Hashim 2011, iRAP 2015b). SR is identified using the iRAP star rating bands
which are shown in Table 1 (Rogers et al. 2012, iRAP 2015b).
Table 1 shows two points. The first one is that the identification of the star rating of a
road is based on a star rating score (SRS) which is computed by a model whose main vari-
ables represent the road attributes affecting the severity and likelihood of traffic accidents as
well as traffic flow and speed (EuroRAP 2011, Rogers & Bin Hashim 2011, iRAP 2015b).
Figure 2 shows the vehicle occupant SRS model with the factors affecting the likelihood and
severity of vehicle accidents. Similar models have been developed for pedestrian, bicyclists,
and motorcyclists.
1608
Table 1. iRAP Star Rating bands (iRAP 2015a, b).
The second point demonstrated in Table 1 is that the outputs of rap are four star ratings
for each road section, each of them is for an individual road group which are: vehicle-occu-
pant, motorcyclist, bicyclist, and pedestrian (iRAP 2015b). Each of these SR is determined
by assigning its SRS value and each of the four SRS is computed from a separate model.
Therefore, there are four models for computing the four SRSs, one of these models is shown
in Figure 2 and the other three models are shown with details in iRAP (2015a, b). Each of
these models is based on specific variables representing the factors affecting the likelihood
and severity of specific specific types of accident, and each of them are used in designing the
safer road investment plan as shown in Figure 1.
This point leads to suggesting that these may be a need to develop a single all-embracing
iRAP star rating score and star rating as an overall index of road safety.
Aggregating the iRAP star rating scores (SRS and SR) to quantify the overall level of road
safety may simplify the design of the investment plan and generate the most effective coun-
termeasures. This may in turn increase the applicability of the iRAP models for strategic
decision making and planning.
To achieve this aim, the following objectives may be set:
1. Searching for aggregation functions that have already been used in developing the iRAP
star rating systems
2. Hypothesizing SRS aggregation methods such as sum, average, maximum and minimum
functions and test the validation of each method to select the optimal.
3. Hypothesizing SR aggregation methods as above.
This paper will focus on aggregation techniques to determine the aggregated star rating
scores SRS and star rating SR. Then, the chosen methods will be validated and finally some
conclusion points and recommendations for further research will be drawn.
Finding a single value representing a set of values can be achieved by using aggregation tech-
niques (Grabisch et al. 2009). Therefore, to aggregate the star ratings and star rating scores,
the aggregation techniques which can be used to compute the aggregated value have been
examined.
According to Figure 2, which shows the vehicle occupant SRS model, it can be illustrated
that the SUM function is used to aggregate the SRS of the five accident types. These types
which shown on the left side of the figure are: Run off, Head-on (loss of control, Head-on
(Overtaking), Intersection, and Property access type. Each of these five scores is computed
by individual sub-model (iRAP 2015b). In addition, the AVERAGE function is used to
1609
Figure 2. The iRAP model of computing the vehicle occupant SRSv (iRAP 2015b).
aggregate the SRSs which are computed per each 100 m to one value for the whole road.
Since the AVERAGE and SUM functions are types of statistical aggregation functions, these
functions will also be tested to select the function which gives the most appropriated overall
aggregated value (Jayram et al. 2007). These functions are elaborated below.
1610
4.1 The statistical aggregation functions
Four aggregation functions have been used and tested to find a single output of iRAP mod-
els. These functions are: AVERAGE, SUM, MAXIMUM, and MINIMUM. Although there
are other aggregation functions, such as Standard Deviation and Count, they do not give the
concept of aggregation which is aimed in this research. These four functions will be used in
aggregating the SRS and three of them (AVERAGE, MAXIMUM and MINIMUM) will be
used in aggregation of SRs because the SUM function produces SR values higher than the
five star level which is the highest SR level. The methodology of applying these functions will
be explained below.
1611
4.2 The selection of the aggregation function
Figures 3 and 4 show the methodology for selecting the most appropriate method of aggre-
gation to be used in this research. The validity of each method will be used in the selection
of the best method which can produce a single indication of the overall level of road safety.
Figure 3 shows the steps of testing the validity of the four aggregation function mentioned
above in finding a single SRS value representing the four SRSs. Although the aggregated
SRS will not be used to determine the overall SR, it can be used in designing the investment
plan of safer roads. It is needed to demonstrate how this aggregated value can simplify the
designing of this plan.
To test the hypothesis of aggregation, the aggregated SRS of a road section is compared
to real crash rates from previous IRAP studies. This comparison has been tested in two steps.
The first step is testing the logical increase of the SRSs and SRs with the increase of the real
crash rates for a specific section of road (iRAP/EuroRAP 2011, McInerney & Fletcher 2013).
The next step is finding the P value which represents the probability of getting a strong corre-
lation between the observed and theoretical results. If P < 0.05, it means that the hypothesis is
significant (iRAP/EuroRAP 2011). This method of testing the validation has also been used
in the methodology of finding the aggregated SR which is shown in Figure 4. An example of
following these steps and validating the aggregated SR and SRS values will be shown in the
following section.
To test the validity of the hypothesis of this work and select the aggregation technique, real seri-
ous and fatal crash data is required (iRAP/Euro 2011, McInerney & Fletcher 2013). For this, it is
necessary to select a road section with crash data for at least 5 years used for iRAP studies. Road
A453 (between A42 Junction to A52 Junction) is one of the roads which have been included in
iRAP star rating improvement studies in the UK (Vida-iRAP, 2015). Therefore, it has been taken
as an example for this work’s validation. This section of road A453 is about 14.9 km and it has
been divided into 149 sub-sections, each sub-section is about 100 m long. The output of iRAP
is represented by SRSs and SRs for each 100 m. These values has been aggregated according
the methodology shown in Figures 3 and 4 using the aggregation functions mentioned in sec-
tion 4.1. Then, these aggregated values were compared with the real crash rate observed in this
road and tested the validity as explained in section 4.2 and Figures 3 and 4.
The results of the comparison are shown in Figures 5 and 6, and in Table 2. Figure 5 dem-
onstrates that the ASRS resulting from the SUM, AVERAGE, and MAXIMUM functions
increase with an increasing observed crash rate. Figure 6 demonstrates that ASR resulted
from using the AVERAGE, MAXIMUM, and, MINIMUM functions decrease in sections
with higher crash rate. This means that these functions may produce the ASR and SR which
reflect the actual overall level of safety. However, it may be argued that this comparison could
not be enough to test the matches between the theoretical quantification of the overall level
of road safety represented by ASRS and ASR with the real crash rate. Some studies measure
the percent of crash rate change when moving from a certain star level to the next one, for
example from 2 stars to 3 stars (McInerney & Fletcher 2013, iRAP/Euro 2011). These stud-
ies consider these measures may be a better indication of the relationship between the crash
rate and the road characteristics impact which is represented by SR. These changes have not
been measured in this paper because most of the sections were in one and 2 stars level. Other
studies measure the significance of the aggregation hypothesis using the P value (iRAP/Euro
2011). Therefore, the comparison of this paper has tested using P value. The result of the
correlation test is shown in Table 2. It may be shown that the P values of the four aggrega-
tion methods are greater than 0.05. This means that the suggested aggregation methods may
not produce a valid measure of the overall level of safety. This insignificant correlation may
be attributed to reasons related to the size of data and that the accident causation was not
related to the road characteristics but other factors, such as the driver behavior.
1612
Figure 4. The methodology of SRs aggregation.
SUM 0.156 –
AVERAGE 0.438 0.137
MAXIMUM 0.060 0.051
MINIMUM 0.723 0.221
1613
Figure 5. The comparison between the observed crash rate with the aggregated iRAP SRSs computed
by four different aggregation functions.
Figure 6. The comparison between the observed crash rate with the aggregated iRAP SRs computed
by three different aggregation functions.
1614
6 CONCLUSIONS AND RECOMMENDATIONS FOR FURTHER RESEARCH
ACKNOWLEDGMENT
The authors wish to express their gratitude to Dr. Steve Lawson and Mr. James Bradford
from iRAP, for their considerable support to this research. The views expressed in this paper
do not necessarily represent those for iRAP. The authors also wish to express their grati-
tude to the Iraqi Ministry of higher education and scientific research for their financial
support.
REFERENCES
EuroRAP:European Road Assessment Programme. 2011. Integrating safety rating into design: Moldova
star rating from road design plans, M2-R7 Saratenii-Soroca-Drochia Junction and M21Dubasari-
Goieni. June 2011 RAP504.3.[online]. Available from: http://www.irap.net/en/about-irap-3/research-
and-technical-papers.[Accessed 18 April 2015].
Grabisch, M., Marichal, J.L., Mesiar, R. & Pap, E. 2009. Aggregation functions, The UK, Cambridge:
Cambridge University Press.
1615
iRAP. 2015a. International road assessment programme. [online]. Available from: http://www.irap.net.
[Accessed 8 May 2015].
iRAP. 2015b. iRAP-methodology fact sheet. international road assessment programme. [online]. Avail-
able from: http://www.irap.net. [Accessed 8 March 2015].
iRAP/EuroRAP. 2011. rate – Star Rating comparisons: Review of available evidence. working paper
504.2, May. Available from: http://www.irap.net/en/about-irap-3/research-and-technical-papers.
[Accessed 20 April 2015].
Jayram, T., McGregor, A., Muthukrishnan, S. &Vee, E. Estimating statistical aggregates on probabilis-
tic data streams. ACM Trans Database Syst (TODS) 33(4):26.
McInerney, R., & Fletcher, M. 2013. Relationship between Star Ratings and crash cost per kilom-
eter travelled: the Bruce Highway, Australia. Available from: http://www.irap.net/en/about-irap-3/
research-and-technical-papers. [Accessed 20 April 2015].
McInerney, R. & Smith, G. 2010. Saving lives through investment in safer roads: International Road
Assessment Programme (iRAP). [online]. Available from: http://www.irap.net/en/about-irap-3/
research-and-technical-papers.[Accessed 20 April 2015].
NRSS: National Road Safety Strategy. 2010. Safer roads for development: a policy framework for safe
infrastructure on major road transport networks. International Road Assessment Programme iRAP.
2000–10, Australia. [online]. Available from: www.irap.ne. [Accessed 18 March 2015].
Ossenbruggen, P., Pendharkar, J. & Ivan, J. 2001. Roadway safety in rural and small PTV.GROUP
(2015) PTV Visum Safety. [online] Available from: http://www.traffic-inside.com/tag/road-safety-
impact-assessment/. [Accessed 8 May 2015].
Rogers, L. & Bin Hashim, H. 2011. Star rating road improvements: 2009 Malaysia black spot pro-
gramme. iRAP. iRAP504.10. [online]. Available from: http://www.irap.net/en/about-irap-3/research-
and-technical-papers.[Accessed 20 April 2015].
Rogers, L.R., Devika, K., Smith, G., Kumar, H.S., Prakash, S. & Prakash, G. 2012. Star rating road
designs: performance indicator for roads in India. iRAP. Document reference no. 502.15.17. [online].
Available from: http://www.irap.net/en/about-irap-3/research-and-technical-papers.[Accessed
18 April 2015].
Stigson, H., Krafft, M. & Tingvall, C. 2008. Use of fatal real-life crashes to analyze a safe road transport
system model, including the road user, the vehicle, and the road. Traffic Injuries Prevention. Oct;
9(5):463–71.
Vida-iRAP, 2015. Available from: http://www.irap.org/en/resources/vida-online-software. [Accessed
20 April 2015].
WHO: World Health Organisation. 2009. Global status report on road safety: time for action. Geneva.
[online]. Available from: www.who.int/violence_injury_prevention/road_safety_status/2009.
[Accessed 20 April 2015].
WHO: World Health Organisation. 2013.Global status report on road safety: supporting a decade of
Action. [online]. Available from: www.who.int/violence_injury_prevention/road_safety_status/2013
[Accessed 27 April 2015].
Zaloshnja, E., & Miller, T., R. 2009. Cost of crashes related to road conditions, united states 2006. In
53rd AAAM Annual Conference of Annals of Advances in Automotive Medicine. USA, October
5–7, 2009. pp. 141–153 (PubMed ID: 20184840).
1616
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Lina Wu
Heilongjiang Institute of Technology, Harbin, China
ABSTRACT: This study first constructs models of the lane-crossing behaviors of riders
of non-motored vehicles based on a survival analysis. With data from traffic investigations,
methods such as the nonparametric method, maximum likelihood estimation and hypoth-
esis testing are employed to complete the process of modeling and verification. Second,
safety evaluation models are developed under the assumption that the amount of critical
conflicts in the curb-parking section followed a binomial distribution, and 4 specific roads
are evaluated regarding the non-motored riding process. Finally, using the constructed
lane-crossing model, the routing choice of non-motorists is analyzed for the condition
of curb-parking problems. The analysis provides helpful suggestions for resolving curb-
parking problems.
1 INTRODUCTION
Curb parking is one of the most important features of the urban parking system. However,
curb parking significantly invades the non-motored running space, which reduces the effi-
ciency of non-motored vehicles and increases traffic accidents.
In “curb-parking problem”, AASHTO proposed A policy on geometric design of streets
and highway. Bernhoft et al. treated pedestrian and non-motored riders as research objects
and compared different psychological and physiological processes and various characteris-
tics, such as penalty and age, to obtain a summary of different performances in the traf-
fic running process. Guo et al. presented a model to analyze lane-crossing behaviors in the
curb-parking section. Sener et al. and Torrance et al. collected 6,400 samples from Austin,
Houston and San Antonio and developed multiple regression models to forecast the pos-
sible locations of motor and non-motored vehicles and predict the invasion probabilities of
motor vehicles. Zhou employed a fuzzy mathematics method to compare different routing
choices between motored vehicles and non-motored vehicles. Chen analyzed the group and
compression characteristic of bicycle flow and put forward the model of bicycle compression
wave based on airflopresented a set of road resistance functions for mixed traffic flow by
considering the impact of curb parking. conduct a systematic observation of behavior from
video. They discovered that the curb-parking problem has a significant influence on the route
selection process of motor and non-motored vehicles.
1617
Table 1. Basic conditions of the traffic investigation.
2 DATA ACQUISITION
The survey of the behaviors of riders of non-motored vehicles in a curb-parking section included
(1) traffic flow data, (2) conditions of road facilities in the vicinity of curb-parking sections, and
(3) basic data for curb-parking conditions. Based on specific projects, researchers completed the
observation and investigation processes in Harbin, Cangzhou and Zhangjiakou, China.
F0 (q ) = Pr (q qc ) (1)
The survival function for the behavioral analysis of non-motored vehicles comprises the
remaining part, which excludes the probability distribution function in the lane-crossing
process:
1618
S0 (q ) = 1 − F0 (q ) = P (q qc ) (2)
ki di
S0 (q ) = 1 − F0 (q ) = ∏ qi < q i {B } (3)
ki
Where qi is the flow of non-motored vehicles in the i th observation section, ki is the number
of observation sections in which non-motored vehicles do not cross the lane, di is the number
of observation sections in which non-motored vehicles cross the lane, and {B} is the number
of observation samples.
This paper analyzed lane-crossing behavior from a micro-perspective. To collect flow data
of non-motored vehicles, researchers classified bicycle lanes into 3 sections by lane width—
2.5, 3.0, and 3.5 m—and set an interval of 30 s for the interval observation section.
Regarding the product limit estimation method as the main function, the sample data are
input into the model to obtain the estimation pictorial diagram shown in Figure 1.
According to Figure 1, the distribution tendency that considers the curb-parking condition
is similar to a normal distribution. Figure 1 shows the inverse ratio relationship between the
effective probability and flow data. It also indicates the direct ratio relationship between the
effective flow and lane width. The investigation process presented a low probability of lane
crossing for a small flow of non-motored vehicles and a high probability of lane crossing for
a large flow of non-motored vehicles; this trend is similar to the trend depicted in Figure 1.
1619
Figure 1. Analysis of samples by the nonparametric method.
n ⎡ 1 −
(xi − μ ) 2 ⎤ n
⎡ 1 n 2⎤
L(μ σ 2
) ∏ ⎢ e 2σ 2 ⎥ (2πσσ 2 )− 2 exp ⎢ −2 ∑
( xi − μ ) ⎥ (4)
i =1 ⎢ σ 2π
⎣ ⎥⎦ ⎣ 2σ i =1 ⎦
Then,
n 1 n
∑( )
2
lnL ln(2πσ 2 ) − − (5)
2 2σ 2 i =1
which yields:
⎧ 1 n
⎪ μˆ = n ∑ xi x
⎪ i =1
⎨ n (6)
⎪σˆ 2 = 1 (
⎪⎩ ∑
n i =1
i )2
The sample data for the vehicles that did not cross the lane are used in Equation (6). The
parameter estimation of the sample expectation and the variance are presented in Table 3.
Using the outputs of the parameter estimation, the lane-crossing model is examined based
on the goodness of fit. The curb-parking conditions for when the bicycle lane is 2.5 m wide
were used as an example, and the flow distribution of the non-motored vehicles were evalu-
ated as follows:
In this paper, the real axis was divided into 14 separate sections to enable the sample data
from the investigation to be divided into 14 groups.
Based on Table 3, for a lane width of 2.5 m, μˆ 10; σˆ 2 35 65493; σ 5.97 .
35.65493;
According the following function,
⎛ q − μ⎞
S0 ( q ) = Φ ⎜
⎝ σ ⎟⎠
1620
Table 3. Maximum likelihood estimation results.
⎛ q − μ⎞
p̂ = Φ ⎜ i
⎝ σ ⎟⎠
Set
The results are as follows:
pˆ1 00.1788,
1788 ˆ 2 00.0988,
0988 ˆ3 0.1237,
0 1237 ˆ 4 0.1306,
pˆ5 00.1309,
1309 ˆ6 00.1106,1106 ˆ7 00.0887,
0887 ˆ8 00.0601,
0601 ˆ9 00.0386,
0386 ˆ10 0.0209,.
pˆ11 0 0108 ˆ12 00.0046,
0.0108, 0046 ˆ13 0.0019
The number of samples ni for each section is obtained from the statistics table χ 2
is 14.594. Let k , m = 2, α = 0.05, and search the χ 2 distribution table to obtain
χα (k
2
1) χ 0.0055 (11) 19.675 , Because14.594 < 19.594, H 0 is accepted, which indicates
2
1 (q −10)2
S0 ( q ) =
−
e 71.31
5.97 2π (7)
By inputting the sample data for the remaining two lanes, the modeling process for the
lane-crossing in the curb-parking sections can also be undertaken using these procedures.
Name of road Critical conflict/day First day Second day Third day Fourth day
Kuancheng Bridge 1 15 6 1 3 21
Jingyang Street 0 11 12 1 5 24
Jiefang Road 2 13 6 0 1 27
Changjiang Road 0 10 5 2 3 20
Total 3 49 29 4 12 92
be applied when considering a large amount of sample data. Due to the large number of
investigations of conflicts, the rules of totality can also be applied by various methods. Some
experts have proven that traffic conflicts in a specific time period and section follow a Poisson
distribution. The survey indicates that the observation of critical conflicts within a short time
is challenging, and a Poisson distribution can be used to analyze traffic conflicts according
to probability theory. In the following study, a safety evaluation of non-motored vehicles in
a single curb-parking section is performed.
Assume that the number of conflicts in a specific time period and section is X, where X
is a random variable. Divide this parking section into n sections to ensure that no duplicate
conflicts occur within the specific time period, because each length of section is extremely
short when n is sufficiently large. Then,
⎧1 ( p )
Xi = ⎨ ( ≤i n) (8)
⎩0 ( q )
Where p + q = 1 and X i is the condition of the traffic conflicts in the ith part of this curb-
parking section.
X i = 1 indicates a critical conflict in the ith part with the probability p, and X i = 0 indicates
no critical conflict in the ith part with the probability of q. Thus, this rule complies with a
(0–1) distribution.
P {X = }= p ( x
p)
1− x
,x 0 1(0 < p < 1) (9)
X ∑X i
(10)
P {X = x} C nx p x q ( n x)
x = 0,1, 2..., n (11)
1622
Therefore, when the binomial distribution n 10 p ≤ 00. and the Poisson distribution is
regarded as similar to the binomial distribution. When the sample amount n is sufficiently
large and p is small, the distribution is correct.
λ xe−λ
P {X = x} P {x }= (12)
x!
λ me − λ
ξ / λ} = ∑ ξ
∞
P{ >α (13)
m!
Therefore, the safety evaluation process involves the process of calculating the critical
value ξ to satisfy the condition in which the conflict level λ is less than the significant level
α , which can be calculated by Equation (13).
When the expected conflict value λ is sufficiently large, the calculation process is compli-
cated. Thus, a more universal conclusion is needed.
According to the central limit theorem, if a random variant X can represent a sum of any
n independent random variants when n is sufficiently large, the sum follows a normal distri-
bution. The number of critical conflicts X is typically large. Based on DE Moive-Laplace
theory, the random variant follows
X -np X −λ
Z= ≈ (14)
np( − p ) λ ( − p)
1623
Table 6. Results from the calculation of the critical conflict frequency.
X −λ
Z≈ (15)
λ
Based on these assumptions, the critical conflict value can be approximately calculated by
Equation (16):
ξ λ + Zα λ (16)
2
To ensure the reliability of general projects, a confidence level exceeding 90% satisfies the
precision requirement. Therefore, 90% of the standard normal distribution (Z Z0.05 = 1.65 ) is
input into Equation (16). The normal probability of a daily conflict value less than ξ is 90%
if the value of the amount of traffic conflict exceeds 90%, which indicates that the section of
road is dangerous for riding and safety measures should be implemented to improve upon
the current conditions.
5 CONCLUSIONS
According to the survival analysis, this paper defined lane crossing as a terminal signal and
the running process as a persistent signal. The trend of the effective flow of non-motored
vehicles was illustrated based on discrete sample data obtained from the nonparametric
method, and the trend was estimated to follow a normal distribution. This paper also pre-
sented the results of hypothesis testing for the goodness of fit, in which various probability
distribution models employed different widths of bicycle lanes for situations in which bicy-
clists did not cross the lane.
1624
This paper summarized 5 types of risk avoidance behaviors and demonstrated the process
of definition, recognition, classification and severity evaluation of motored vehicle-bicycle
conflicts in the curb-parking sections. Based on the conflict data acquired in the investi-
gation, safety evaluation models are constructed which based on the assumption that the
amount of critical conflicts in the curb-parking section follows a binomial distribution. The
bicycle riding conditions for four roads were evaluated based on the collected data and devel-
oped models.
ACKNOWLEDGEMENTS
This work was financially supported by the Grant from the Scientific Research Fund of
Helongjiang Provincial Education Department of China (No. 12541650).
REFERENCES
AASHTO. 2001. A policy on geometric design of streets and highways, American Association of State
Highway and Transportation Officials, Washington, D.C.
Bernhoft IM, Carstensen G. 2008. Preferences and behavior of pedestrians and cyclists by age and
gender[J]. Transportation Research Part F, 11: 83–95.
Chen J,Wang W,Mei CY.The analysis on the speed of mixed traffic flow with the curb parking lots[J].
Journal of Transportation Systems Engineering and Information Technology.2005,5(5):34–36.
Chen J,Wang Z.Model of bicycle compression wave based on the effect caused by curb parking on bicy-
cel lane. Journal of Harbin Institute of Technology,2013,45(6):114–118.
Guo HW, Gao ZY, Zhao XM, Yang XB. 2011. Traffic behavior analysis of non-motorized vehicle under
influence of curb parking. Transportation System Engineering and Information, 11(1): 79–84.
Sener IN, Eluru N, Bhat CR. 2009. An analysis of bicycle route choice preferences in Texas, US. Trans-
portation, 36: 511–539.
Torrance K, Sener IN, Machemeh R, Bhat CR, Hallett I, Eluru N, Hlavacek I, Karl A. 2009. The effects
of on-street parking on cyclist route choice and the operational behavior of cyclists and motorists.
FHWA/T X Report.
Van Der Horst ARA, De Goede M,De Hair-Buijssen S,et al.traffic conflicts on bicycle paths: a system-
atic observation of behavior from video [J].Accident Analysis and Prevention,201462:358–368.
Zhou B. 2008. Study on the cyclist’s microscopic behaviors at non-signalized intersections based on
fuzzy mathematics. Master’s Dissertation of Beijing Jiaotong Unversity.
1625
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Hao Liu
Beijing Transportation Information Center, Beijing, P.R. China
Heng Shi
Tongji University, Shanghai, P.R. China
ABSTRACT: Traffic states are important traffic information for travelers. The existing
traffic regulations have a detailed and unique criterion for distinguishing the traffic states.
Mostly, the traffic states are defined based on the objectively estimated travel speeds. Yet,
it is reported that travelers’ perceptions on travel speeds and traffic states might be differ-
ent from the objective ones. This paper employs the empirical perception data collected in
the city of Beijing to investigate travelers’ perceived travel speeds and traffic states in rela-
tion to the objective ones. The effects of travelers’ social characteristics on their perceived
speed differences are explored as well. 120 videos, each recording a 5-second traffic condition
data, are prepared representing a variety of traffic conditions on different road sections. The
objectively estimated travel speed of each video and the associated traffic state based on the
traffic regulations were pre-processed. 94 effective participants were asked to judge the traffic
states of all 120 videos, and then to report their perceived travel speeds and traffic states for
each video. The differences between the perceived and objectively estimated travel speeds and
traffic states are analyzed based on the collected data. The results show that users do have
significant perceived differences in travel speeds and traffic states. High heterogeneities are
found among participants in their perceptions, especially among different gender, age and
risk-attitude groups. This study suggests that the existing criterion for distinguishing the traf-
fic states should be adjusted by considering travelers’ perceived differences in travel speeds.
1 INTRODUCTION
Urbanization in China develops rapidly following the initiation of the reform and opening
policy. By the end of 2012, the mainland of the People’s Republic of China had a total urban
population of 712 million, which is 52.6% of the total population, rising from 26% in 1990
(Europe.chinadaily.com.cn., 2014). The population living in China cities and towns is more
than twice of the population of the United States of America. Consequently, traffic jams are
one of the norms in the expanding Chinese economy, since the inhabitants to congregate in
large urban areas lead to sometimes intolerable levels of traffic congestion on urban streets and
thoroughfares. For example, since the Chinese government abolished the highway charging
during long holidays in October 2012, traffic congestion has also been severe, with the number
of passengers on the highways reaching 85.6 million on the first day of the policy’s implementa-
tion, an increase of 13.3 percent from the previous year. Thousands of tourists were stranded in
Jiuzhaigou, a famous tourist site in Sichuan Province in Southwest China, for 11 hours due to
congestion caused by excessive numbers of visitors on October 2, 2013 (Zhang, 2014).
In order to optimize the usage of the transportation networks, a wide range of traffic
information has been disseminated including traffic and road conditions, closure and detour
information, travel restrictions, incident information, emergency alerts and driver advisories.
Traffic condition is one of the common disseminated traffic information. Occupancy, travel
1627
time, travel volume etc. are most frequently reported as performance indicators of traffic
conditions on urban freeways. Compared to other performance indicators, the traffic state
index (TSI) is adopted often as a measure to reflect the average congestion levels on road
network networks. For example, Travel Time Index and Travel Rate Index proposed by Texas
Transportation Institute are utilized to evaluate the reliability of road networks (Schrank
and Lomax, 2007, Eisele et al., 2011). American Customer Satisfaction Index can reveal the
service quality of transportation systems, while it is difficult to guide the analyses of network
traffic states (Van den Berg et al., 2004). Road Congestion Index can reflect the degree of
urban traffic congestion based on the travel distance and the length of mainline (Boarnet
et al., 1998). Traffic Congestion Index System based on data collected from RTMS of Bei-
jing expressway can evaluate traffic congestion in various time intervals and varied space
(Wang et al., 2010). TSI is proposed as well by Shanghai Transportation Information Center
(SHTIC) (Shanghai Municipal Transportation Information Center, 2013), for evaluating the
traffic performance on the urban freeway corridors.
The traffic states index, TSI, is typically determined on a basis of travel speeds. For instance,
TSI proposed by Shanghai Municipality and Beijing Municipality are both calculated based
on the free-flow speed (vf) and the actual travel speed (v), as shown in Eq.(1).
vf − v
TSI = × 100 (1)
vf
The speed v for determining the TSI value is measured (real) travel speed. However, several
studies have reported travelers’ attitudes to congestions are diverse. A number of studies have
explained this observed diversity (i.e. the different preferences on congestion) as heterogene-
ity among travelers (e.g. (Freeman Iii and Haveman, 1977, Boxall et al., 2003, Felipe and
Michael, 2008)). The effects of heterogeneous preferences on congestion have been inves-
tigated by several studies, for instance the work by Freeman and Haveman for recreational
users (Freeman Iii and Haveman, 1977). They concluded that the distribution of user prefer-
ences has significant impacts on traffic managements, for instance in their paper the optimal
allocations of charging locations and toll levels. However, the heterogeneity might not be
the only explanation for the observed preference differences on congestion. It might be that
travelers’ perceptions on congestion are different from what we have defined according to
the real travel speeds. Limited studies can be found on dealing with this subject on the per-
ceived differences in speed and traffic states, as the perception data are challenging to obtain.
Motivated by this, this paper aims to investigate the perceived differences in travel speeds and
congestions by conducting empirical analyses based on perception data collected in Beijing.
The remaining part of the paper is organized as follows. Section 2 presents the method-
ology for analyzing the perceived speed differences. Section 3 conducts the analyses for the
empirical data collected from the survey and discusses the main findings. Finally, some con-
clusions are drawn and future research is discussed in Section 4.
2 METHODOLOGY
1628
The 120 recorded video data provide a sufficient coverage on all traffic conditions, e.g. free-
flow, congested and jam conditions, to allow for a full range analysis over all traffic state
regimes. The traffic states at merging, diverging, or weaving sections are not recorded for this
survey in order to ensure that changes in traffic stream flows are not influenced by the lane
changing behavior or by any incident.
2.2 Participants
We employed people to participate our investigations. Each participant was asked to assess
the traffic conditions after watching the 5-second video and to respond for all the 120 videos.
At the same time, the participants provide his/her perceptions on the travel speed and con-
gestion levels. Their perceived travel speed was recorded in a stepwise manner with a step of
10 km/h, for example, 20–30 km/h. Their perceived traffic states were categorized into three
levels, i.e. free-flow condition, congested condition and jam.
Participants
1629
Figure 1. The probability density function of the differences between perceived and real speeds.
1630
Figure 2. Perceived differences in speeds for male and female.
Z-statistic
Gender
female
male −11.66
Age
21 21–30
21–30 −11.71
>31 −11.10 −4.61
Risk-attitudes
risk-seeking risk-neutral
risk-neutral −9.76
risk-averse −11.67 0.49
Driving License
with licence
without-licence 4.49
1631
Figure 4. Perceived differences in speeds in relation to risk-attitudes.
1632
Figure 6. Perceived differences in traffic states (for free-flow conditions only).
3.5 Discussions
Several remarks can be made on the basis of above results. Firstly, the analyses suggest that
it might be meaningful for the government to reconsider the way of classifying traffic states
and the speed thresholds for different traffic states. Travelers’ perceived differences in travel
speeds and traffic states might be valuable for defining the traffic states, as traffic state infor-
mation is normally distributed to travelers which influences travelers’ travel decisions. Once
the traffic state information is more closely defined to travelers’ perceptions, they would be
more inclined to follow the information from the traffic managers.
Secondly, travelers’ perceptions in travel speeds are significantly different among differ-
ent age, gender and risk-attitude groups. Different cities might require analyses on travelers’
characteristics and then correspondingly may have different thresholds for defining traffic
states.
Thirdly, a large dispersion in perceived travel speeds and traffic states complicates the
interpretation of these results. For instance, the fact that 31.1% of the participants perceived
a congested traffic condition as a free-flow condition leads to a difficulty in defining the
threshold value for distinguishing the traffic states. Although these findings add complexity
to the distinction of user-based traffic states, they can be useful for planning and management
purposes in the sense that the perceptions of different types of travelers can be considered.
4 CONCLUSIONS
This paper investigates the perceived differences among travelers in travel speeds and traffic
states. Empirical data collected from 94 effective participants have been used for the analyses.
Each participant was asked to report his/her perceived travel speeds and traffic states for 120 vid-
eos for various traffic conditions on various road sections. The main findings are as followings:
1. Travelers’ perceptions on travel speeds are significantly different from the real objective
traffic speeds. On average, participants over-estimate the travel speeds, however with very
large variations;
2. The male and the female have significant differences in their perceived speed differences;
3. The risk-seeking participants perceive significantly different from the risk-neutral and
risk-averse participants in travel speeds;
4. Participants of different ages do have significant differences in their perceived speed
differences;
5. Travelers’ perceptions of traffic states for a given traffic condition vary substantially.
1633
Clearly, differences exist between the subjective perception on traffic condition and the
objectively estimated traffic states. It raises a question on how to define the traffic states by
considering travelers’ perception differences and heterogeneity in it. The main challenge of
this study is collecting the empirical data. In this study, 94 effective respondents with 120 vid-
eos each have been collected. To enhance our findings, both the number of participants and
videos data should be enlarged. Furthermore, we are going to focus on exploring the thresh-
old values for distinguishing the traffic states by taking travelers’ perceptions into account.
ACKNOWLEDGEMENT
This project is supported by National Natural Science Foundation of China (NSFC, Grant
No.51578414) and by the Fundamental Research Funds for the Central Universities.
REFERENCES
Boarnet, M., Kim, E. & Parkany, E. 1998. Measuring Traffic Congestion. Transportation Research
Record: Journal of the Transportation Research Board, 1634, 93–99.
Boxall, P., Rollins, K. & Englin, J. 2003. Heterogeneous preferences for congestion during a wilderness
experience. Resource and Energy Economics, 25, 177–195.
Eisele, W.L., Zhang, Y. & Park, E.S. 2011. Developing and Applying Models for Estimating Arterial
Corridor Travel Time Index for Transportation Planning in Small to Medium-sized Communities.
Transportation Research Record: Journal of the Transportation Research Board, 2244, 81–90.
Europe. Chinadaily. Com. CN. 2014. Census: Population hits 1.37 b,2014–02–16.
Felipe, V.L. & Michael, H.W. 2008. Taste indicators and heterogeneous revealed preferences for conges-
tion in recreation demand. eScholarship, University of California [Online].
Freeman III, A.M. & Haveman, R.M. 1977. Congestion, quality deterioration, and heterogeneous
tastes. Journal of Public Economics, 8, 225–232.
Schrank, D. & Lomax, T. 2007. The 2007 Urban Mobility Report. Texas.
Shanghai Municipal Transportation Information Center 2013. Research on Traffic Performance Index
and Its Applications in Shanghai.
Van den Berg, M., Schutter, B.D., Hegyi, A. & Hellendoorn, J. Model Predictive Control for Mixed
Urban and Freeway Networks. www.trb.org Compendium of papers TRB 83rd Annual Meeting,
2013/01/13–2013/01/17 2004 Washington D.C.
Wang, Z., Que, L., Liu, J. & Guan, J. 2010. Study on the Evaluation System and Prediction Approach on
Traffic Congestion Index of RTMS Data. Traffic and Transportation Studies, 903–914.
Zhang, Y. 2014. Add Holidays to Ease Tourist Congestion: Report(2014-09-01) [Online]. Available:
http://www.dongguantoday.com/news/national/201401/t20140109_3313670.shtml 2014–09–01].
1634
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Real-time crash risk prediction is important in the traffic management system
aiming at improving traffic freeway safety. This paper aims to study the relationship between
crash potential and lane-level data collected by single ultrasonic detectors, which were installed
on the S28 freeway in Jiangsu Province, China. A binary logistic regression model is developed
with multi-vehicle crash data and detector data by using the matched case-control method.
The max eigenvalue is introduced to the model to utilize lane-level data sufficiently. Results
show that the model with data of 0–5 minutes prior to the crash has an accuracy of 61.9%,
which is quite satisfactory. The model can be used to predict real-time crash potential through
the application of advanced traffic management system in China gradually.
1 INTRODUCTION
Road safety is a complicated transportation research field due to the random nature of acci-
dent occurrence. Understanding the various factors that cause road accidents is crucial. Inci-
dent management departments can take measures to avoid the accident occurrence or reduce
the risk when accident precursors have been sensed. Hence, many researches have been con-
ducted to construct an assessment and estimation system in a statistic way and investigate the
inner relationship between the crash data and traffic operation data.
Madanat and Liu (Madanat et al. 1995) proposed the concept “Real-time Incident Likeli-
hood Prediction” to analyze the crash similarity by modeling the crash data on freeway with the
detected flow data and weather data. Cheol Oh et al. (Oh et al. 2001) employed a Non-Paramet-
ric Bayesian Classification Model to estimate accident likelihood. This study demonstrated that
speed variation detected 5 minutes before accident could be used to estimate an accident effec-
tively by using loop detector data and by setting different threshold level different percentage of
accidents can be identified. Lee et al. (Lee et al. 2002) proposed the concept “Crash Precursor”
for traffic flow characteristics observed prior to crash occurrence. An aggregate log-linear model
was developed with the variation of speed and traffic density, which were statistically significant
predictors of crash frequency. In the coming year, Lee et al. (Lee et al. 2003) developed a proba-
bilistic real-time crash prediction model with a rational method to select crash precursors and
optimal observation time slice durations. The study found that the difference between the speed
at the upstream detector and the speed at the downstream detector was significantly higher.
Abdel-Aty et al. (Abdel-Aty et al. 2004) developed a crash likelihood prediction model using
the “matched case-control logistic regression” method. The study suggested that the 5-min aver-
age occupancy observed at the upstream station during the 5 to 10 min before the crash, along
with the 5-min coefficient of variation in speed at the downstream station during the same time
affected crash occurrence most significantly. The model could achieve more than 69% crash
identification. Abdel-Aty et al. (Abdel-Aty et al. 2005) suggested utilizing the hazard ratio values
about 15 minutes before the accident occurrence could effectively reduce the impending risk by
analyzing the flow data 90 minutes before and after the accident occurrence, respectively of four
loop detectors upstream and two loop detectors downstream.
1635
These studies showed high availability and reliability for applied the precursor method to
predicting crash potential. More detailed studies (Thomas et al. 2004, Oh et al. 2005, Pande
et al.2006, Thomas et al. 2008, Hossain et al. 2012, Xu et al. 2012) had identified the crash
potential for different type of crashes by classify the traffic condition with different models.
Different models had different prediction accuracies, among which, the model with neural
network-based classifiers (Pande et al. 2006) could potentially identify almost 75% of rear-
end crashes with reasonable false alarms. Eleni I. Vlahogianni et al. (Vlahogianni et al. 2012)
developed neural network models by analyzing different variables associated with second-
ary accident likelihood and suggested that traffic speed/ duration of the primary accident,
hourly volume, rainfall intensity and number of vehicles involved in the primary accident are
the top associated factors. AVI (Automatic Vehicle Identification) data had been utilized in
trial to analyze the crash potential in recent years (Mohamed Ahmed et al. 2012, Mohamed
Ahmed et al. 2013). The studies found that AVI systems could provide a measure of the risk
of a crash in real time, with the accuracy 75.93% for rear-end crashes. Moreover, when fusing
the AVI data and RTMS (Remote Traffic Microwave Sensors) data, the model with the data
fusion framework had a higher estimation accuracy, robustness and reliability.
Generally, these findings point to the potential use of high-quality and intensive detec-
tor data (approximately 0.5 mile apart) and detailed crash data in the field of traffic safety.
However, in developing countries such as China, there is almost no continuous and intensive
traffic information collecting devices and surveillance devices. In the past in most developed
regions in China, the detectors were installed far apart, they would only be installed around
the core road segments with complex traffic environment, the detectors would always break
down and the data the detectors collected was usually missing or stored with poor quality,
which makes it impossible to estimate the potential risk by the existing models with data
from single detector. Moreover, although researches on the transferability and robustness
have proved that it is possible to use the same model for multiple freeways (Shew et al.
2013), most existing studies are based in the same dataset, studies by researchers using more
diverse datasets could lead to a more robust collective inquiry (Roshandel et al. 2015) and
the transferability of the existing models cannot be validated in China because of different
driving behavior and traffic pattern. However, in recent years, with the application of infor-
mation technology in freeway management in China, more and more integrated control
and management systems have been constructed in some developed regions, data with high
quality can be collected in real time, which makes it possible to conduct crash prediction
research.
The objective of this study is to develop a crash potential model with the lane-level data col-
lected by ultrasonic detectors, which can be implemented in real-time integrated freeway control
and management system. To some extent, the results can promote fundamental understanding
of the impact of traffic flow characteristics on the potential of multi-vehicle crashes and help to
develop more targeted strategies on freeway operation and management afterwards.
Data were obtained from a 35-km segment with four lanes (two lanes for each direction)
on the S28 freeway in City of Yangzhou, Jiangsu Province, China. Ultrasonic detectors
are installed approximately 4 km (2.49 mile) apart (Fig. 1) and provide detailed vehicle
type information with corresponding speed and flow for each lane. The detectors are
installed on the flyovers above the freeway surface, which can make the collected data
accurate. The detectors are working with replaceable batteries to ensure the continuous
data collection process. A detector is installed between every two tolls. In this study, the
corresponding traffic data are related to the segment crashes occurred. For instance,
a crash was reported on the segment between Toll2 and Toll3, then the corresponding
detector would be D3.
The raw traffic data were collected with the time interval 5 minutes. Table 1 shows a set of
sample data.
1636
Where DataTime denotes the data collecting time, CDNO denotes the lane number, CXn
denotes the vehicle type (among which CX1 denotes car, CX2 denotes coach, CX3 denotes
van or minivan, CX4 denotes truck), CSn denotes the corresponding speed with CXn.
Table 2 shows the detailed variables used in this study.
Crash data comes from the firsthand freeway operation data log. A total of 93 crashes were
collected from 2014–12 to 2015–05. Since the crash data covers large amount of single-vehi-
cle crash data, and the recorded crash causes of most of the single-vehicle crashes are vehicle
breaking down and hitting the fixation, only multi-vehicle crash data with corresponding
detector data is utilized in this study, with the number of 42 in total at last. In order to specify
the crash precursors excluding external factors such as road geometry and time of day, which
had been proved to be significant on crash occurrence, the matched case-control method is
adopted in this study. As stated in these existing studies, the control-case ratio around 4:1 is
recommended since the statistical power generally does not increase significantly beyond a
4:1 ratio (Zheng. 2010). Hence, a 4:1 control-case ratio is adopted in this study.
Table 1. Sample of part of the detailed raw traffic data collected by ultrasonic detector.
DataTime CDNO CX1 CX2 CX3 CX4 CS1 CS2 CS3 CS4 sbzh
Symbol Variables
1637
3 METHODOLOGY
Ax = λ x (1)
where A denotes a square matrix, x denotes the eigenvector and λ denotes the eigenvalue.
A − λE = 0 (2)
Lane No. Flow (pcu) Proportion of Truck (%) Average Speed (km/h)
1 f1 p1 v1
2 f2 p2 v2
3 f3 p3 v3
Data Time Lane No. Flow (pcu) PT (%) Average Speed (km/h)
1638
Then each crash case had 4 corresponding non-crash cases. However, because of hardware
problems and other data transmission errors, part of the traffic data is missing or abnormal,
therefore, cases with missing data or abnormal data were filtered out. At last, a dataset with
173 cases for 0–5 time slice(0–5 minutes prior to the occurring time), 173 cases for 5–10 time
slice(5–10 minutes prior to the occurring time) and 173 cases for 10–15 time slice(10–15 min-
utes prior to the occurring time) was formed to model the crash data and traffic data.
eβ0 + ∑ βk xki
P ( y i = |x i ) = p i =
1 + eα + ∑ βk xki
(3)
1
P ( y i = |x
| i ) = 1 − pi =
1 + eβ0 + ∑βk xki (4)
The odds ratio g ( x ) is calculated as:
pi
g ( x) = = e β0 + ∑βk xki (5)
1 − pi
where x ki denotes input variables series; β 0 denotes constant in logistic model; β k denotes
coefficients for independent input variables series.
And β k can be estimated by solving the log-likelihood function by:
( )
n
ln L ( β , ) ∑ ⎡⎣ β 0 β1 βk ln + e β ∑βk xki ⎤ (6)
i
i =1
1i ki
⎦
1639
Table 6. 5–10 time slice logistic regression result.
Input Parameters: IsW, IsH, Flow, PT, AvgV, MaxV, MinV, MaxEig, SndEig, TrdEig.
Input Parameters: IsW, IsH, Flow, PT, AvgV, MaxV, MinV, MaxEig, SndEig, TrdEig.
4 PREDICTION MODEL
The binary logistic regression procedure in SPSS 19 was utilized in this study. Parameters
used in this study were not limited to the traffic characteristics. Another two parameter had
been introduced into the model analyzing process, namely, IsW(is or not during weekend, 1
denotes Weekend and 0 denotes Not Weekend) and IsH(is or not during National Holidays, 1
denotes Holiday, 0 denotes Not-Holiday). Backward LR(likelihood ratio) variable selection
was applied to select the optimal parameters for the model. The ultrasonic detectors collected
data at the time interval of 5 minutes, so in the study, traffic data of three types of time series
had been analyzed, namely 0–5 time slice, 5–10 time slice, 10–15 time slice. And the results
are listed below:
1640
Table 8. Model accuracy with different time slice.
5 CONCLUSION
This study developed a binary logistic regression model by using the case-control method
with the lane-level ultrasonic detector data. The results show that the models can reach as
high as 61.9% crash identification, which is quite satisfactory.
The above three models have a parameter in common, IsH (Whether or Not during National
holidays), and the corresponding β value has the strongest impact on the probability of crash.
This indicates that during national holidays, crashes are more likely to occur. For example, dur-
ing the National Day in 2014(from 10-1-2014 to 10-7-2014) on the same segment, 189 crashes
were recorded. However, 244 crashed were recorded during the whole month, 2014/10. In China
there is a certain traffic management policy that all the cars are free of tolling fee in national
holidays, which usually cause a sudden increase in traffic demand leading the high crash risk.
The three models have a similar accuracy of both crash and non-crash. And the low false alarm
rate makes the models practicable in the field of operation and management of freeway.
1641
Figure 2. Result of the neural network analysis.
Models can be changed according to the practical need and data source. In this study, 0–5
time slice logistic regression model is recommended due to the highest predicting accuracy,
61.9%. The results prove that the method is valid to predict crash potential. Once the poten-
tial is certain, measures for traffic routing and speed limit may be implemented to reduce the
risk. It can help the staffs of the corresponding management department to develop more
targeted and proactive strategies.
6 DISCUSSION
In this study, only logistical regression is used to model the relationship between crash poten-
tial and traffic data. However, logistical regression is an linear analyzing method, while the
parameters maybe more suitable for non-linear analyzing method. By using non-linear ana-
lyzing method, it may achieve a higher prediction accuracy. In previous studies, logistical
regression, neural network and support vector machine were widely used. Different methods
have different advantages and disadvantages. In this section, to ensure the reliability by uti-
lizing the parameters from single ultrasonic detector to estimate and predict crash potential,
neural-network method had been jointly used to analyze the relationship. For the 0–5 time
slice data (traffic data 0–5 minutes prior to the crash), a neural network model (Fig. 2) had
been developed by Matlab neural network pattern recognition and classification tool. The
samples were divided as, Training: Validation: Testing = 70%:15%:15%.
The result shows that the prediction accuracy of crash of the model is 59.5% and the false
alarm rate is 6.1%.
Generally, the method by analyzing the relationship between crash potential and single
ultrasonic detector data is practicable. Both the logistic regression model and the neural net-
work model can evaluate crash potential with a satisfactory prediction accuracy. However,
as the uncertainty of the neural network model is clear, the accuracy may vary by differ-
ent training process. Hence, logistical regression is recommended in this study. And other
machine learning techniques, such as support vector machine, may be used in future study.
As the recorded data is accumulated, more cases can be selected to validate and optimize
the model. Meanwhile, with the spreading application of information technology in freeway
management in China, researches on the transferability of the existing models can be con-
ducted. In future study, more relevant factors, such as weather factors and driving behavior,
can be taken into consideration to promote the prediction accuracy of the model. The fact is
that many devices are under construction such as weather detectors, pavement sensors and the
newest technologies for collecting driving behavior data such as naturalistic driving are being
researched at present. Hence, researches can be conducted to study the evolutionary process
of the crash on freeways more systematically and help us better understand the interactions of
behavioral, vehicle, and roadway factors.
ACKNOWLEDGEMENT
This research is supported by the National “Twelfth Five-Year” Plan for Science &
Technology Support Project in China (2014BAG01B04), the National Natural Science Foun-
dation (71301119) and the Shanghai Natural Science Foundation (12ZR1434100).
1642
REFERENCES
Abdel-Aty, M. & Pande, A. & Hsia, L.Y. & Abdalla, F. 2005.The Potential for Real-Time Traffic Crash
Prediction. ITE Journal.75(12): 69.
Abdel-Aty, M. & Uddin, N. & Pande, A. &Abdalla, M.F. & Hsia, L. 2004. Predicting freeway crashes
from loop detector data by matched case-control logistic regression. Transportation Research Record:
Journal of the Transportation Research Board. 1897(1): 88–95.
Ahmed, M & Abdel-Aty, M. 2013.A data fusion framework for real-time risk assessment on freeways.
Transportation Research Part C: Emerging Technologies.26: 203–213.
Ahmed, M.M. & Abdel-Aty, M. & Yu, R. 2012. Bayesian Updating Approach for Real-Time Safety
Evaluation with Automatic Vehicle Identification Data. Transportation Research Record: Journal of
the Transportation Research Board. 2280(1): 60–67.
Golob,T.F. & Recker, W. & Pavlis,Y. 2008.Probabilistic models of freeway safety performance using traf-
fic flow data as predictors. Safety Science. 46:1306–1333.
Golob,T.F. & Recker, W.W. & Alvarez,V.M 2004.Freeway safety as a function of traffic flow. Accident
Analysis and Prevention. 36:933–946
Hossain,M. & Muromachi,Y. 2012. A Bayesian network based framework for real-time crash prediction
on the basic freeway segments of urban expressways. Accident Analysis and Prevention. 45373–381.
Lee, C. & Hellinga, B. & Saccomanno, F. 2003. Real-time crash prediction model for application to
crash prevention in freeway traffic. Transportation Research Record: Journal of the Transportation
Research Board. 1840(1): 67–77.
Lee,C. & Saccomanno,F. & Hellinga,B. 2002.Analysis of crash precursors on instrumented freeways.
Transportation Research Record: Journal of the Transportation Research Board.1784(1): 1–8.
Madanat, S. & Liu, P.C. 1995.A Prototype System for Real-Time Incident Likelihood Prediction. IDEA
Project Final Report (ITS-2). Transportation Research Board of the National Academies Washington
D.C..
Oh, C. & Oh, J. & Ritchie, S. & Chang, M. 2001.Real-time Estimation of Freeway Accident Likelihood.
CD-ROM, Transportation Research Board of the National Academies Washington D.C.
Oh, J.S. & Oh, C. & Ritchie, S.G. & Chang, M. 2005. Real-time estimation of accident likelihood for
safety enhancement. Journal of transportation engineering.131(5): 358–363.
Pande, A. & Abdel-Aty, M. 2006.Comprehensive analysis of the relationship between real-time traffic
surveillance data and rear-end crashes on freeways. Transportation Research Record: Journal of the
Transportation Research Board.1953(1): 31–40.
Roshandel, S. & Zheng Z. &Washington S. 2015. Impact of real-time traffic characteristics on freeway
rash occurrence: Systematic review and meta-analysis. Accident Analysis and Prevention.79: 198–211.
Shew, C & Pande, A & Nuworsoo, C. 2013.Transferability and robustness of real-time freeway crash risk
assessment. Journal of Safety Research.46:83–90.
Vlahogianni, E.I. & Karlaftis, M.G. & Orfanou, F.P. 2012. Modeling the effects of weather and traffic
on the risk of secondary incidents. Journal of Intelligent Transportation Systems.16(3): 109–117.
Xu, C. & Liu, P. & Wang, W & Li,Z. 2012. Evaluation of the impacts of traffic states on crash risks on
freeways. Accident Analysis & Prevention. 47: 162–171.
Zheng Z.2010. Empirical analysis of freeway traffic oscillation: its safety impact and evolution. Arizona
State University.
1643
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Continuous, real-time monitoring of the road network can assist in prioritis-
ing when roads should be maintained or rehabilitated. Timeous maintenance or rehabilita-
tion of roads improves road safety and comfort and can be used to assist with investing the
available funds and resources more intelligently. In this paper, it will be shown that telematics
technology can potentially be used to monitor the condition of the extended road network
and on a continuous, real-time basis. Telematics technology is continuously evolving and
is currently used in the market to monitor and manage vehicle fleets, to retrieve hijacked/
stolen vehicles and for driver behavioral insurance. Most telematics units incorporate Global
Positioning Systems (GPS), acceleration and gyroscope sensors. Leading telematics compa-
nies generally have large databases and complex algorithms to store this type of information.
In this paper, different statistical analysis techniques have been applied to the existing telemat-
ics data and the most promising outcome utilised to show that this technology contains addi-
tional potential in terms of monitoring the condition of the extended road network on a
continuous, real-time basis. Although road roughness is currently measured using dedicated
Class 1 profilometers, when effective calibration techniques are applied to the data harvested
from telematics units, a reasonable accurate representation of the condition of the road can
be produced. This was shown by sampling data from a few local passenger vehicles fitted with
telematics devices and doing the necessary analysis and calibration techniques on the vertical
(up/down) acceleration to produce data, which can provide a cost-effective indication of the
condition of the extended road network in real-time.
1 INTRODUCTION
Continuous monitoring of the road network can be a successful tool in terms of improving
road safety and comfort. In addition, it assists with the development of better prioritisation
schemes and to invest available funds and resources more intelligently. In this paper, the
practicality of adopting the use of telematics technology to monitor the condition of the
extended road network on a continuous, real-time basis is explored.
2 LITERATURE REVIEW
1645
processing the measured profile by using a mathematical transform which filters and cumu-
lates the wavelengths encountered in the profile. The IRI is more commonly characterised
by a processing algorithm which mimics the physical properties and displacement of a vehi-
cle wheel and suspension system (i.e. from the quarter-car model), while the vehicle travels
at 80 km/h. Advantages of the IRI parameter is that it is stable with time, reproducible, and
a parameter which is widely used and adopted throughout the world (Jooste, 2007; Sayers
and Karamihas, 1998).
The IRI index produces a controlled measurement for a number of roads under numer-
ous conditions from a variety of instruments and methods. It is defined as the ratio of the
accumulated suspension motion of a vehicle, divided by the total distance travelled by the
vehicle, and has units of slope (m/km). An IRI interpretation scale exists which generally
indicates the IRI ranges for different paved and unpaved road sections (Figure 1) as well as
recommended safe operating speeds (Sayers et al., 1986).
Most modern profilometers have the capability of measuring the profile of the road
in both wheelpaths simultaneously. A more theoretically accurate assessment of rough-
ness can be attained if the roughness index is calculated in both wheelpaths (i.e. the Half-
car Roughness Index (HRI)). The same processing algorithm used to calculate the IRI
is adopted for the HRI parameter, except that the HRI uses a point-by-point average of
profiles in left and right travelled wheelpaths. The HRI more closely matches the manner
in which most response type devices measure road roughness (Jooste, 2007; Sayers and
Karamihas, 1998).
1646
Table 1. Measurement methods (Sayers et al., 1986).
Class Definition
2.4 Profilometers
Measuring a road profile allows a continuous representation of a road to be evaluated to
identify local defects, or to be processed to yield a roughness numeric. Profilometric methods
include both manual quasi-static and high-speed profilometers (Sayers et al., 1986).
3 METHODOLOGY
Different statistical analysis techniques were investigated to find a method which could be
used to link the data harvested from the telematics devices in the field to a road condition
index which could be used and distributed to road users on a continuous, real-time basis
throughout the road network.
1650
Figure 4. Coefficient of variation of vertical acceleration for 10 m segments.
Figure 6. Average CoV of vertical acceleration versus actual measured HRI for the road section.
1651
average CoV determined for the 2 km road section analysed was employed. This effectively
removes most of the vertical scatter (Figure 5).
3.7 Average CoV of vertical acceleration versus actual measured HRI for the road section
The average CoV of vertical acceleration calculated from the data harvested from the telemat-
ics devices and the actual measured HRI with the Class 1 profilometer were plotted (Figure 6)
to visually illustrate the close relationship which exists.
3.9 Linear regression analysis of Root Mean Square Vertical Acceleration (RMSVA)
Literature (Sayers et al., 1986; Jones and Forslof, 2014) states that the RMS of the vertical
acceleration is used to determine a suitable model through linear regression analysis. Its suit-
ability was validated in terms of the relationship it produces in comparison with the model
generated for the CoV of the vertical acceleration. Comparing Figure 7 and Figure 8, it is
evident that the Normalised CoV produces a stronger linear relationship.
Figure 7. Linear regression of average normalised CoV of vertical acceleration and the actual meas-
ured HRI with Class 1 profilometer.
1652
Figure 8. Linear regression of RMS of vertical acceleration and actual measured HRI with Class 1
profilometer.
Table 2. Correlation of determination (r2) of normalised CoV of vertical acceleration and actual
HRI.
Speed
Vehicle type 40 km/h 60 km/h 80 km/h Varying Combined
Small Hatchback 0.57 0.47 0.58 0.21 0.55
Large Hatchback 0.47 0.59 0.65 0.60 0.73
Large SUV 0.71 0.78 0.83 0.71 0.85
Combined 0.69 0.77 0.81 0.68 0.80
1653
Table 3. Standard error of normalised CoV of vertical acceleration and actual HRI.
Speed
Vehicle type 40 km/h 60 km/h 80 km/h Varying Combined
Small Hatchback 0.18 0.25 0.23 0.23 0.19
Large Hatchback 0.18 0.24 0.23 0.23 0.16
Large SUV 0.14 0.18 0.17 0.23 0.14
Combined 0.13 0.16 0.16 0.18 0.13
Table 4. Standard HRI classification ranges and the determined approximate average normalised CoV
of vertical acceleration ranges.
Figure 9. Actual measured HRI and estimated HRI plotted in Tracker Connect Maps.
4 RESULTS
The ultimate purpose of analysis is to generate reliable data which can be visually portrayed
to all road users such that they are aware of the road condition and can drive accordingly if
they know the condition beforehand. This assists with the awareness factors of road users
and improves road safety.
1654
4.1 Mapping the condition
The continuous, real-time road roughness of the extended road network can be computed and
displayed and distributed to all road users via Portable Navigation Devices (PNDs) or a smart-
phone application. The roughness of the road section will be continuously calculated as identi-
fied consenting customers’ travel on different roads on the network. The deterioration rate of the
road can also be monitored by comparing the estimated road roughness, calculated previously,
with the estimated road roughness after a specific time period has elapsed. This can provide an
indication of rapid road deterioration and the resultant need for more in-depth investigation of
the road structure. This will thus assist with better prioritisation schemes in terms of which roads
requires immediate maintenance/rehabilitation to ensure that the safety and comfort aspects of
all road users are successfully addressed. The actual measured HRI from the Class 1 profilom-
eter and the estimated HRI from the telematics device data were plotted to illustrate the means
of visual portrayal and correlation (Figure 9). Corresponding overlapping areas of high road
roughness (i.e. the red road sections indicated by the red circle in Figure 9) are evident.
In this paper it should be evident that telematics technology can potentially be utilised to
estimate the condition of roads. With appropriate calibration techniques and continuous
progress, the data harvested from a large sample of telematics devices installed in the field,
can produce information. This information can be valuable to all road users, improving road
safety and assisting with implementation of better prioritisation schemes of roads. Telemat-
ics technology holds the potential of various prospective future applications, such as:
• Calibrating the sensors to detect road anomalies such as potholes, speed-bumps and untrue
bridge deck joints on a larger network level through pattern recognition algorithms;
• Deterioration rate monitoring of roads on a network level;
• Proposals for safe operating speeds in relation to road roughness, and
• Vehicle operating cost models can be developed which links the fuel cost to the estimated
road roughness calculated.
6 ETHICAL ISSUES
Only aggregated data are used for monitoring the condition of the extended road network
and the privacy and confidential information of the telematics company clients/customers
will be protected and not distributed.
ACKNOWLEDGEMENTS
The support of Tracker Connect Pty Ltd (provision of tracking devices and workforce time
and development), Specialised Road Technologies (Pty) Ltd (measuring actual profiles of
the road sections), and the University of Pretoria (financial support) are acknowledged. This
work is based on the research supported in part by the National Research Foundation of
South Africa. The Grantholder acknowledges that opinions, findings and conclusions or
recommendations expressed in any publication generated by the NRF supported research are
that of the authors, and that the NRF accepts no liability whatsoever in this regard.
DISCLAIMER
This paper reflects the views of the authors, who are responsible for the facts and the accu-
racy of the data presented herein. The contents do not necessarily reflect the official views or
policies of the funders.
1655
REFERENCES
Eriksson, J., Girod, L., Hull, B. 2008. The Pothole Patrol: Using a Mobile Sensor Network for Road Sur-
face Monitoring. MobiSys’08, Breckenridge, Colorado, United States of America.
Gillespie T.D., Sayers, M.W., and Segel, L. 1980. Calibration and Correlation of Response-Type Road
Roughness Measuring Systems. National Cooperative Research Program, Transportation Research
Board, National Research Council. Highway Safety research Institute, University of Michigan, Ann
Arbor, Michigan.
Higgins, J. 2003–2010. The Radical Statistician: Unleashing the Power of Applied Statistics in the Real
World. Sacramento, California, United States of America.
Jones, H. and Forslof, L. 2014. Roadroid Continuous Road Condition Monitoring with Smart Phones.
33rd Southern African Transport Conference. Council of Scientific and Industrial Research, Preto-
ria, South Africa.
Jooste, F. 2007. Guidelines for Network Level Measurement of Road Roughness Version 1.1. Committee
of Transport Officials (COTO), Cullinan, South Africa.
Koichi, Y. 2013. Extensional Smartphone Probe for Road Bump Detection. Traffic Hazard Technology,
Tokyo, Japan.
Mohan, P., Padmanabhan, V.N., and Ramjee, R. 2008. Nericell: Rich Monitoring of Road and Traf-
fic Conditions using Mobile Smartphones. SenSys’08, Raleigh, North Carolina, United States of
America.
Ross, S.M. 2009. Introduction to Probability and Statistics for Engineers and Scientists. 4th Edition.
Elsevier Academic Press, Burlington, United States of America.
Sayers, M.S., Gillespie, T.D., and Paterson, W.D.O. 1986. Guidelines for Conducting and Calibrating
Road Roughness Measurements. World Bank Technical Paper Number 46. The World Bank, Wash-
ington, D.C., United States of America.
Sayers, M.W., Gillespie, T.D., and Queiroz, C.A.V. 1986. The International Road Roughness Experiment:
Establishing Correlation and a Calibration Standard for Measurements. The World Bank, Washing-
ton, D.C., United States of America.
Sayers, M.W., and Karamihas, S.M. 1998. The Little Book of Profiling: Basic Information Measuring
and Interpreting Road Profiles. The Regent of the University of Michigan, Ann Arbor, Michigan.
Schlotjes, M.R., Visser, A., and Bennett, C. 2014. Evaluation of a Smartphone Roughness Meter. 33rd
Annual Southern African Transport Conference. Council of Scientific and Industrial Research, Pre-
toria, South Africa.
SkyTrax. 2015. Freedom to drive Taking back tomorrow with class-leading products and services.
Tracker Connect Pty Ltd, South Africa.
Steyn, W.J.vdM. 2013. Freight-Truck-Pavement Interaction, Logistics, and Economics: Final Phase 1
Report (Tasks 7–8). Research Report No.: UCPRC-RR-2013–08.
Tracker Connect Maps. Tracker, Tom Tom, Digital Globe, CDNGI, NASA. (Accessed November 27,
2015).
Tracker Connect Pty Ltd. 2015. Tracker Connect Pty Ltd. http://www.tracker.co.za.
Van As, S.C., and Joubert, H.S. 2008. Applied Statistics for Civil Engineers. SARB Chair in Transporta-
tion Engineering, University of Pretoria, South Africa.
Wang, T., Harvey, J., Lea, J.D., and Kim, C. 2013. Impact of Pavement Roughness on Vehicle Free-Flow
Speed. Technical Memorandum: UCPRC-TM-2013–04.
1656
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: Engineers and researchers in the transportation discipline have tried to build
safe roads following appropriate design standards, but traffic crashes are unavoidable. Crash
pattern could be detected if accurate prediction models capable to predict various traffic
crashes are developed. To obtain the maximum possible crash reduction effects with limited
budgetary resources, it is important that measures are based on engineering and scientific
surveys to study cause of crashes. The research presented here aims to evaluate Road Safety
using Bayesian methods on a section of four-lane National Highway (NH)-58 located in the
state of Uttarakhand, India. This paper compares Poisson Regression, Poisson Log-Normal
and Poisson Gamma modelling techniques using a full Bayes hierarchical approach for Indian
highway crash data. Explanatory variables for highway segment analysis were Geometric
Characteristics like Curvature Change Rate, Transverse Slope and Traffic Characteristics as
Average Daily Traffic, Volume to Capacity ratio were analyzed against dependent variable as
Crash Count per two hundred meter per year.
1 INTRODUCTION
Present challenge in front of transportation engineers is to plan and operate the transpor-
tation network in such a way that it meets various consumer objectives such as reducing
travel time and enhancing safety. Over the years, there has been an unprecedented increase
in road transportation and vehicular traffic in India due to growth in the economy and con-
sumption habits, which has led to increasing safety concerns on our highways. Casualties
due to crashes on these roads are increasing year after year. The road fatalities and injuries
are global phenomena but more sever situation in mixed traffic condition as prevailing on
Indian multilane highways. Owing to the likely under-reporting of crashes, the real situation
is likely to be much worse. To deteriorate the road safety further, there is habit of poor main-
tenance of vehicle, bad driving practice, lack of enforcement, casual attitude of road users
and least concerned towards the basic road safety measures by road maintaining agencies.
Road safety has become a growing concern for the public in general and highway profes-
sional in particular as road crashes are a major source of loss of life. Further; the economic
losses caused by property damage or loss of working days resulting from injuries and fatali-
ties have been estimated at billions of rupees annually. Concept of quality management and
sustainable safety have gained ground in the past two decades and may have been among
the factors that led policymakers and stakeholders to realize the need for dedicated safety
oriented studies to reduce the occurrence of crashes.
Road safety is both the health and development issue of concern due to its nega-
tive impacts on the economy, public health and the general welfare of the people. Road
safety is a multi-sectoral, responsibility of the government, civil society stakeholders
and finally of everyone using the facility. The success of road safety programs in any
country depends upon a broad based support and common action from all associated
organizations.
1657
1.1 Road safety concern in India
With increase of multilane highways in India; the casualties due to crashes on the roads
are increasing alarmingly year by year. The greater share of road transport in India is same
as in the rest of the world. The movement of both passengers and goods is expected to
increase further in the future years with the full implementation of the current road expan-
sion programs being undertaken in the country. Most of the road development programs are
expected to be completed by 2017.
Presently in India more than 500,000 crashes occur annually. About 138,000 people lose
their life in these crashes. Other than above mentioned fatalities, nearly 500,000 people are
injured in road crashes which lead to lifelong misery for the victims and family. Road Safety
Study can ensure that various safety deficiencies in road are reviewed so that these can be
taken care at appropriate stage of highway design or operation in a cost effective way.
Considerable amount of research has been done in recent years to establish relationships
between crashes and various traffic flow characteristics, geometric characteristics for highway
segments and intersections (Cafiso et al. 2010; Daniels et al. 2010; El-Basyouny, K., Sayed,
T. 2012; Geedipally, S.R., Lord, D., 2008; Geedipally, S.R., Lord, D., 2010; Malyshkina,
Mannering 2010b). Past research works that examined the relationship between crashes and
traffic characteristics like volume to capacity (v/c) ratio and Level of Service (LOS) are (Min,
Sisopiku 1997) reported that U-shaped models explained the relationship between v/c and
accident rates for weekdays and weekends, multi-vehicle, rear-end, and property-damage-
only accidents. (Persaud, Nguyen 1998) revealed that crash rates and number of crashes both
increase when LOS decreases from A to F. (Lord et al. 2005) studied the relationship between
crashes and traffic volume and found that not only traffic volume but vehicle density and v/c
ratio have a direct influence on the probability and severity of crashes.
Numerous studies have been performed by many researchers on road safety analysis in
India. Landge et al. 2006 reviewed different modelling approaches adopted worldwide. (Dinu,
Veeraragavan 2011) implemented random parameters count model for crash prediction on
two lane undivided highways. There are limited safety studies adopting Bayesian technique
for crash prediction on Indian Highways.
2 RESEARCH OBJECTIVE
The study aims to identify the most critical safety influencing variables and to find a better
technique within PR, PLN and PG statistical models using Bayesian inference to predict fre-
quency of crash counts on a section of four-lane National Highway-58 in India.
The study objective was accomplished by analyzing the model results developed using col-
lected dataset and comparing PR, PLN and PG models based on Pearson Chi-Square, Devi-
ance Information Criterion, Mean Absolute Deviation and Mean Squared Prediction Error.
(yi − μˆ i )
2
Pearson − χ 2 = ∑ i =1
N
(1)
VAR (yi )
DIC
Cj ( )
D θ j + 2 pD
Dj = D pD j (2)
( ) ( )
where D θ j = the deviance D θ j |y at the posterior mean θ j of the parameters for model
j, called Dhat in WinBUGS, D = the expected devaiance D E ( D y ) , given by the mean
D of the sampled deviances D(t) from Markov Chain Monte Carlo simulations, also called
Dbar in WinBUGS, and pDj = effective number of parameters in the model, computed as the
( )
difference between D j and D θ j , that is, pD
Dj = Dj D θ j . ( )
The DIC is the Bayesian generalization of Akaike Information Criterion. It is a combina-
tion of the measure of model fitting the dataset and the number of explanatory variables in
the model.
While comparing between two models, a difference in DIC value greater than 10 will rule
out the model giving higher value of DIC (Spiegelhalter et al. 2003). Where the difference
1659
is less than 10, the models are reasonably similar. Smaller the DIC value indicates a better
model fitting.
1 n
MAD = ∑ μi − yi
n i =1
(3)
The model closer to zero value is considered to be best among all the available models.
1 n
( )
2
MSPE = ∑
n i =1
i − yi (4)
4 DATABASE DEVELOPMENT
1660
4.3 Details of road geometrics
Main Carriageway is 7.00 m wide with 1.50 m paved and 2.00 m earthen shoulder on either
side of the roadway. The median width is 5 m and raised type in general.
CCR =
∑ i
γi
[ gon m] (5)
L
where γi is the deflection angle for a contiguous element (curve or tangent) i within a section
of length L.
1661
Table 1. Spearman’s rho correlation coefficients.
*Correlation is significant at the 0.05 level (2-tailed). **. Correlation is significant at the 0.01 level (2-tailed).
Crash Count (CC), Curvature Change Rate (CCR), Transverse Slope (TS), Average Daily Traffic
(ADT), Volume to Capacity Ratio (V/C) and Median Width (MW).
Crash Count, Average Daily Traffic (ADT), Curvature Change Rate (CCR), Transverse Slope (TS),
Volume to Capacity Ratio (V/C) and Segment Length (Offset, meters).
characteristic of independent variables with crash frequency. There are sub-populations that
have different variabilities from others.
Table 2 gives the statistical summary of the variables selected to build the final models. The
results in Table 2 revealed that for any subset of the independent variables, the crash count
(CC) exhibits over dispersion.
1662
Figure 1. Scatter plot matrix.
Model A:
Model B:
The model form ‘B’ was developed to check the efficiency of the modelling techniques
when only traffic characteristic variables are available from the field data. As in most of the
situation in India it is a challenging job to collect geometric design and other variables for
crash analysis.
Full Bayesian framework was implemented for modeling and inference. The parameter
estimation and inference can be obtained by means of Markov Chain Monte Carlo (MCMC)
and software such as WinBUGS. As the Bayesian formulation requires priors for all unknown
parameters, non-informative normal priors for β’s and gamma/lognormal priors for error
terms were adopted.
For each model, three Markov chains were used in the coefficient estimation process with
150,000 iterations, and 75,000 iterations were used in burn-in process and were discarded.
Convergences of the models were checked by monitoring the MCMC trace plots and Gelman
Rubin (G-R) diagnostics for the model parameters. If all values were within a zone without
strong periodicities or tendencies, the model was considered convergent. Additionally, the
number of iterations was chosen such that the model Monte Carlo error for each of the
model parameter would be less than 10% of the value of the model standard deviation for
parameter under consideration.
Output analysis and diagnostics for Markov Chain Monte Carlo simulations was carried
out on coda files from WinBUGS using coda package in R. The G-R convergence statistic
is generally used to verify that the simulation runs converged properly. For model compari-
son, it was suggested that convergence was achieved when the G-R statistic was less than 1.2
(Mitra, S., Washington, S 2007).
Bayesian hierarchical framework method considers the coefficients for the covariates as
random variables rather than fixed values as in classical statistical inference. Hence the model
output will be a sampled posterior distribution for each of the estimated parameter. The
estimated coefficient values for the model parameters are as shown in Tables 3 and 4.
1663
Table 3. Parameter estimates and goodness of fit for model form A.
Parameters Poisson Regression (PR) Poisson Log-Normal (PLN) Poisson Gamma (PG)
Intercept (B0) −19.81 3.703 0.129 −10.01 3.695 0.128 −18.5 4.656 0.162
LnADT (B2) 2.16 0.408 0.014 1.079 0.408 0.018 2.018 0.514 0.017
CCR (B3) 3.67 0.657 0.006 2.898 1.086 0.010 3.382 1.102 0.010
TS (B4) 2.272 0.716 0.004 2.965 1.184 0.007 2.156 1.181 0.008
V/C (B5) −2.746 0.7676 0.026 −0.703 0.794 0.026 −2.508 0.983 0.033
Tau(PLN) / - - - 1.582 0.118 8.82E-4 0.718 0.049 2.82E-4
α = 1/Φ (PG)
Pearson Chi 4612.350 4690.182 4581.262
Square
MAD 1.535 1.536 1.535
MSPE 5.725 5.736 5.725
DIC 6263.170 5023.770 4982.950
Dependent Variable: Crash Count. Independent Variables: Natural log of Average Daily Traffic
(LnADT), Curvature Change Rate (CCR), Transverse Slope (TS) and Volume to Capacity Ratio
(V/C).
Parameters Poisson Regression (PR) Poisson Log-Normal (PLN) Poisson Gamma (PG)
Intercept (B0) −12.12 2.879 0.100 −3.603 3.544 0.109 −10.87 5.814 0.180
LnADT (B2) 1.322 0.317 0.011 0.340 0.391 0.012 1.184 0.642 0.019
V/C (B5) −1.351 0.607 0.020 0.557 0.769 0.023 −1.073 1.231 0.037
Tau(PLN) / - - - 1.552 0.116 6.253E-4 0.736 0.050 2.404E-4
α = 1/Φ (PG)
Pearson Chi 4690.470 4870.533 4645.962
Square
MAD 1.553 1.555 1.552
MSPE 5.805 6.060 5.804
DIC 6308.560 5031.210 4985.360
Dependent Variable: Crash Count. Independent Variables: Natural log of Average Daily Traffic
(LnADT) and Volume to Capacity Ratio (V/C).
Figure 2. CURE Plot with ADT Variable and ± 2Std. Dev. for model form A.
1665
Table 5. Sensitivity of PG model parameters in model form ‘A’.
From sensitivity analysis 10% increase in traffic, increases crash frequency by 21%. Cur-
vature Change Rate reported 0.02 gon/m change of CCR value increases crash frequency by
7%. Sensitivity analysis reported 0.01 change in transverse slope has 2% hike in crash count.
Volume to capacity ratio is the highest sensitive variable from graphs next to traffic volume.
Change of 0.1 value of V/C decreases crash frequency by 22%.
ACKNOWLEDGMENT
Corresponding author wishes to express his gratitude to NHAI and WUP Toll Ltd. for pro-
viding the data used for analysis and my friends Mr. Rahul and Mr. Tushar for carrying out
the video graphic data collection and extraction.
REFERENCES
Bolstad, W.M. Introduction to Bayesian Statistics. John Wiley & Sons, Ltd, 2007.
Cafiso, S., Di Graziano, A., Di Silvestro, G., La Cava, G., Persaud, B. Development of comprehensive
cident models for two-lane rural highways using exposure, geometry, consistency and context vari-
ables. Accident Analysis and Prevention, Vol. 42 pp. 1072–1079, 2010.
Congdon, P. Bayesian Statistical Modeling. John Wiley & Sons, Ltd, 2006.
Daniels, S., Brijs, T., Nuyts, E., Wets, G., 2010. Explaining variation in safety performance of rounda-
bouts. Accident Analysis and Prevention, 42 (2), 393–402.
Development Core Team, R., 2006. R: A Language and Environment for Statistical Computing. R Foun-
dation for Statistical Computing, Vienna, Austria. http://www.R-project.org
Dinu, R.R., Veeraragavan, A., 2011. Random parameter models for accident prediction on two lane
undivided highways in India. Journal of Safety Research, 42 (1), 39–42.
El-Basyouny, K. and Sayed, T. Statistical Analysis of Crash Data: Alternative Modeling Methods. LAM-
BERT Academic Publishing, 2012.
El-Basyouny, K. and Sayed, T., 2009. Collision Prediction models using Poisson-lognormal regression.
Accident Analysis and Prevention, Vol. 41 pp. 820–828.
Geedipally, S.R., Lord, D., 2008. Effects of the varying dispersion parameter of Poisson gamma models
on the estimation of confidence intervals of crash prediction models. Transportation Research Record
2061, pp.46–54.
1666
Geedipally, S.R., Lord, D., 2010. Investigating the effect of modeling single-vehicle and multi-vehicle
crashes separately on confidence intervals of Poisson-gamma models. Accident Analysis and Preven-
tion, Vol. 42 pp. 1273–1282.
Hauer, E. Observational Before-After Studies in Road Safety: Estimating the Effect of Highway and
Traffic Engineering Measures on Road Safety. Elsevier Science Ltd, Oxford, 1997.
Landge, V.S., Jain S.S. and Parida M. 2005, “A Critical Appraisal of Road Traffic Safety Analysis and
Modeling for Nonurban Highways”, Journal of Indian Roads Congress, 66, pp. 105–129.
Landge, V.S. (2006). “Development of Methodology for Road Safety for a National Highway.” Ph.D. Thesis.
Lord, D., Manar, A., & Vizioli, A. (2005). Modeling crash-flow-density and crash-flow-V/C ratio rela-
tionships for rural and urban freeway segments. Accident Analysis and Prevention, 37(1), 185–199.
Lord, D. Modeling Motor Vehicle Crashes Using Poisson-Gamma Models: Examining the Effects of
Low Sample Mean Values and Small Sample Size on the Estimation of the Fixed Dispersion Parameter.
Accident Analysis & Prevention, Vol. 38, No. 4, 2006, pp. 751–766.
Lord, D., S.P. Washington, and J.N. Ivan. Poisson, Poisson-Gamma and Zero Inflated Regression Models
of Motor Vehicle Crashes: Balancing Statistical Fit and Theory. Accident Analysis & Prevention, Vol.
37, No. 1, 2005, pp. 35–46.
Lord, D., S.P. Washington and J.N Ivan. Poisson, Poisson-Gamma and Zero-inflated Regression Mod-
els of Motor Vehicle Crashes: Balancing Statistical Fit and Theory, Accident Analysis & Prevention
Vol.37, No.1, 2005, pp.35–46.
Lord, D. and Park, B.J. Appendix D-Negative Binomial Regression Models and Estimation Methods.
In Crime Stat Version 3.3 Update Notes: Part 2: Regression Modeling, The National Institute of
Justice, Washington, D.C. 2010.
Lord, D., Miranda-Moreno, L.F., 2008. Effects of low sample mean values and small sample size on the
estimation of the fixed dispersion parameter of Poisson-gamma models for modeling motor vehicle
crashes: a Bayesian perspective. Safety Science 46 (5), 751–770.
Maher, M and L. Mountain. The Sensitivity of Estimates of Regression to the Mean. Accident Analysis
& Prevention, Vol. 41, No.4, 2009, pp.861–868.
Malyshkina, N., Mannering, F., 2010b. Empirical assessment of the impact of highway design exceptions
on the frequency and severity of vehicle accidents. Accident Analysis and Prevention 42 (1), 131–139.
Miaou, S.-P., Lum, H., 1993. Modeling vehicle accidents and highway geometric design relationships.
Accident Analysis and Prevention 25 (6), 689–709.
Miaou, S.-P., 1994. The relationship between truck accidents and geometric design of road sections:
Poisson versus negative binomial regressions. Accident Analysis and Prevention 26 (4), 471–482.
Miaou, S.P., and Lord, D. Modeling Traffic-Flow Relationships at Signalized Intersection: Dispersion
Parameter, Functional Form, and Bayes versus Empirical Bayes. Transportation Research Record,
No.1840, 2003, pp. 31–40.
Miaou, S.-P., J.J. Song, and B.K. Mallick. Roadway Traffic Crash Mapping: A Space-Time Modeling
Approach. Journal of Transportation and Statistics,Vol. 6, No. 1, 2003, pp. 33–57.
Min, Z., and V.P. Sisopiku. (1997). Relationship between V/C Ratios and Accident Rates, Transporta-
tion Research Record 1581, TRB, National Research Council, Washington, D.C., pp. 47–52.
Mitra, S. and Washington, S. On The Nature of Over-Dispersion in Motor Vehicle Crash Prediction Mod-
els. Accident Analysis and Prevention 39(3), 2007, pp.459–468.
Oh, J., C. Lyon, S. Washington, B. Persaud and J. Bared. Validation of FHWA Crash Models for Rural
Intersections: Lessons Learned. Transportation Research Record: 1840, 2003, pp. 41–49.
Spiegelhalter, D., Thomas, A., Best, N., Lunn, D. WinBUGS Version 1.4 User Manual. MRC Biostatis-
tics Unit, Cambridge, http://www.mrc-cam.ac.uk/bugs. 2003.
Jain, S.S. and Parida, M. Final Report on Development of a Scientific Approach to Road Safety for
Indian Environment. Submitted to AICTE, New Delhi, 2001. Parida, M., Jain, S.S., and Naveen
Kumar, C., 2012. Road Traffic Crash Prediction on National Highways. Indian Highways Journal,
Indian Road Congress, Vol. 40 (6), pp. 93–103.
Persuad, B., and Nguyen T. (1998). Safety consideration in capacity analysis. Paper presented at the Proceed-
ings of the third international Symposius on Highway Capacity, Copenhagen, Denmark, pp.817–831.
Raghavachari, M., A. Srinivasan and P. Sullo. Poisson Mixture Yield Models for Integrated Circuits: A
Critical Review. Microelectronics Reliability, Vol.37, No.4, 1997, pp.565–580.
Washington, S.P., Karlaftis, M.G., Mannering, F.L. Statistical and Econometric Methods for Transpor-
tation Data Analysis, Second Edition. CRC Press, 2011.
Winkelmann, R. Econometric Analysis of Count Data. Second Edition. Springer-Verlag, Berlin, 2008.
Zamani, H., Ismail, N., 2010. Negative binomial-Lindley distribution and its application. Journal of
Mathematics and Statistics 6 (1), 4–9.
1667
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Yusheng Ci
Harbin Institute of Technology, Harbin, HeiLongjiang, China
Lina Wu
Heilongjiang Institute of Technology, Harbin, China
Guohui Zhang
The University of New Mexico, Albuquerque, USA
ABSTRACT: This study aims to investigate significant contributing factors for crash fre-
quencies and severities on rural two-lane highways and develop corresponding injury-preven-
tion and mitigation strategies for safety improvements. With data from the Department of
Statistics and the Department of Public Security of Heilongjiang, China, from 2006 to 2012,
Principal component analysis (PCA) techniques are used to extract the crash patterns behind
the data and explore various contributing attributes of crash occurrence. Based on the anal-
ysis of model estimation results, five strategic prevention and mitigation countermeasures
are proposed: (1) education and training, (2) management, enforcement, and punishment,
(3) engineering improvements, (4) emergency, (5) accident audit and safety assessment. This
research contributes to enhance traffic safety performance on rural two-lane highways.
1 INTRODUCTION
Statistical data show that the fatal crash rate on rural two-lane highways is more than twice
as high as that for freeways (1). The data in table 1 illustrate that the number of crashes and
crash rates of injury and fatality on rural two-lane highway are 5.9, 5.9 and 6.1 times higher
than that on freeways in Heilongjiang, China during the period from 2006 to 2012. These
statistical data clearly indicate that traffic safety conditions on two-lane highways need sig-
nificant improvements in China. Few efforts were undertaken to analyze significant causal
factors and injury-mitigation strategy development for rural two-lane highway crashes in
China. Therefore, a better understanding of the contributing factors associated with crashes
will facilitate the identification of suitable countermeasures to reduce crash occurrence fre-
quencies and severities rural two-lane highways in China.
The most commonly used models to analyze crash occurrence and impact factors are
Binomial modeling techniques related approaches. Vogt et al. (2) developed Negative Bino-
mial (NB) and Extended Negative Binomial (ENB) methods to build accident models for
segments and three-legged and four-legged intersections stop-controlled on the minor legs.
Schneider et al. (3) presented a negative binomial model and applied the full Bayes method
to improve model performance and to assess the impacts of horizontal curvature and other
geometric features on the frequency of single-vehicle motorcycle crashes along segments of
rural two-lane highways in Ohio. Peng et al. (4) developed a negative binomial model of crash
frequency and a Multinomial Logit (ML) model of crash severity to investigate the relation-
ship between single-vehicle roadway departure crashes and roadside features.
1669
Ansari et al. (5) found that over 65% of the accidents occurred because of vehicles travel-
ling at excess speed and/or drivers disobeying traffic signals in Saudi Arabia. Karlaftis et
al. (6) used Hierarchical Tree-based Regression (HTR) method to build up the relationship
model among rural road geometric characteristics, accident rates and their prediction. Ander-
son et al. (7) performed Linear Regression (LR) method to analyze the statistical relation-
ship between mean accident rate and mean speed reduction and between degree-of-curvature
intervals and mean speed reduction. Farah et al. (8) developed Tobit Regression (TR) models
to explain the minimum time to collision based on the analysis of drivers’ passing decisions
on rural two-lane highways using an interactive driving simulator. Mujalli et al. (9) intro-
duced a Bayesian Networks (BNs) method for simplifying the analysis of traffic accidents
injury severity on two-lane highways. Xi et al. (10) put forward an improved association rule
algorithm based on Particle Swarm Optimization (PSO) to analyze the correlation between
accident attributes and causes.
Further, several relatively single-factor or multi-factor accident causes and simulation anal-
ysis methods were introduced for two-lane highways accidents. Garber et al. (11) presented
Deterministic models developed by the Multivariate Ratio of Polynomials (MRP) that relate
the crash rate with the mean speed, standard deviation of speed, flow per lane, lane width
and shoulder width on two-lane highways in Virginia. Cafiso et al. (12) presented a methodo-
logical approach for the safety evaluation of rural two-lane highway segments that used both
analytical procedures referring to alignment design consistency models and safety inspection
processes. Also a Safety Index (SI) was calculated that quantitatively measures the relative
safety performance of a highway segment from the procedure. Pardillo-Mayora et al. (13)
used roadside slope, non-traversable obstacles distance from the roadway edge, safety barrier
installation and alignment to characterize the main roadside features that have an influence
on the consequences of roadway departures. Cafiso et al. (14) defined and calibrated acci-
dent models by the Generalized Linear Modeling (GLM) for two-lane road sections based
on a unique combination of exposure, geometry, consistency and context variables directly
related to the safety performance in Italy. Mohaymany et al. (15) applied a Traffic Conflict
Technique (TCT) to 12 rural two-lane roads in order to investigate the potential situations
for accidents to occur and thus to identify the geometric and traffic factors affecting traffic
conflicts. Pearson’s chi-square test method was used to evaluate their relationship to traffic
conflicts. Farah (16) collected the data on the overtaking behavior of 100 drivers with an
interactive driving simulator for two-lane highways, which were used to analyze the overtak-
ing behavior of different drivers classified by age and gender.
2 DATA DESCRIPTIONS
The crash data were collected from the Heilongjiang Statistical Yearbook (2006–2012), the
Annual Report for Traffic Accidents at the Traffic Police Corps of Heilongjiang Department
of Public Security (2006–2012) and the Annual Report for Traffic Volume data at the Bureau
of Highway of Heilongjiang Department of Public Security (2006–2012). Further, other rel-
National State
Highway Category Freeway Highway Highway Total
Lanes 4 2–4 2 —
Mileages (km) 4084 6984 9156 20224
Accidents 969 2420 3268 6657
Deaths 545 1511 1700 3756
Injuries 1157 3057 4013 8227
Fatality rate (%) 32.02 33.08 29.76 31.62
The occupancy rate of two-lane national highways is more than 95% in the national highway system.
1670
evant data were collected from technique standards for highways and references. The data in
table 2 show the descriptive statistics of crash records.
3 METHODOLOGY
⎡ x11 x12 x1 p ⎤
⎢x x22 x2 p ⎥⎥
X =⎢
21
(1)
⎢ ⎥
⎢ ⎥
⎢⎣ xn xn 2 xnp ⎥⎦
Then, the processed data matrix, which is dealt with the method of Z-score normalization,
is expressed with X as well.
xij xj
xij* = (i , 2, , n;
n; j 1 2 p) (2)
var( x j )
1 n 1 n
Where, x j = ∑
n i =1
xij xj = ∑ ( xiijj x j )2
n − 1 i =1
Step 2: Calculating the covariance matrix.
The covariance matrix can be got with following equations:
⎡ r11 r12 r1 p ⎤
⎢r r r2 p ⎥⎥ X T X
R=⎢
21 22
= (3)
⎢ ⎥ n −1
⎢ ⎥
⎢⎣rp rp 2 rpp ⎥⎦
1 n
rij = ∑ xti xtj (i, j = 1, 2,, p)
n − 1 t =1 (4)
1671
Step 3: Calculating the eigenvalues and eigenvectors of the covariance matrix.
From solving eigenequations, λ I R = 0 , we can get the eigenvalues ( λ ) and list them as
λ1 λ2 ≥ ≥ λ p ≥ 0
The eigenvectors can be calculated ( ei (ei1, ei , eiipp ),ii 1 2 p ) correspondingly.
p
ei 1 ∑e
j =1
2
ij =1 (5)
∑λ ∑λ
k =1
k
k =1
k (i 1, 2 p) (6)
⎢ ⎥
⎢ ⎥
⎢⎣l p l p 2 l pm ⎥⎦
Where, Fnm is the score of nth sample with mth principal component.
According the final total scores, we can conduct a comparing analysis comprehensively.
1672
⎡1.0000 0.7398 0.4218 ⎤
R = ⎢0.7398 1.0000 0.9157 ⎥
⎢0.4218 0.9157 1.0000 ⎥⎦
⎣
λ = ( .4039, 0.5906, 0.0055)
Speeding, improper yield, illegal passing and crossing the centerline are the main causes
of accidents according to the PCA calculating results on rural two-lane highways. They con-
tribute to 64.98% of the traffic accidents in total. All of these are related to drivers’ driving
behaviors. Furthermore, the percent of accident causes keeps pace with the result on rural
two-lane highways basically.
Also, illegal meeting, drunk driving and fatigue driving are the second major causes. The
proportion of these causes is around 13.20% and most of them involved drivers’ driving
behaviors as well.
Finally, tailgating, vehicle mechanical failure, pedestrian violations and bad road condi-
tions are the minor causes relatively. Lower of the causes are related to vehicles, highways
themselves and other driving behaviors.
In particular, there are only 0.64% and 1.77% of the traffic accidents involved in highways
and vehicles directly. Also, there are nearly 4.30% and 27.48% of the traffic accidents com-
bined with drivers-vehicles and drivers-highways. However, there are more than 78.19% of the
traffic accidents related directly to human errors. Here are the main reasons we concluded.
Firstly, rural two-lane highways have a wide range of service area and a variety of users in
Heilongjiang, China. Rural two-lane highways are also the arterial roads of the towns along
the highways, which existing much more conflicts at both sides. There may be half of the popu-
lation will use the rural two-lane highways; so the jagged levels of drivers’ behaviors and skills
and vehicles’ performance may become latent dangers. As a result, a lot of accidents occurred,
such as the accidents of crossing the centerline, illegal turning, improper vehicle controlling,
illegal backing-up, stopping and U-turning, vehicle mechanical failure and pedestrian viola-
tions. Meanwhile, there are so many “T” or “Y” intersections because of the structure of the
network. Actually, it’s even worse than it looks because of the lack of signal control devices,
stop signs, yield signs, markings, channelization and effective maintenance. Therefore, 11.63%
of the accidents occurred in this area, especially the death and serious injury accidents.
Next, most of rural two-lane highways are national, provincial or county roads. So the align-
ment conditions, pavement quality and roadside interference are better than other lower classes
of rural highways. Table 5 lists the main design indexes for rural two-lane highways in China.
1 0.7912 −0.6111
2 0.9955 0.0760
3 0.8870 0.4597
Source: These data are from Technical Standard of Highway Engineering, China (JTG B01 2003).
1673
Even so, more than 80% of the deaths and injuries occurred along straight segments of rural
two-lane highways. But we also realize that nearly 20% of the accident deaths and injuries
occurred at curve and slope segments. So it reveals that the geometric conditions have signifi-
cant effect on the accidents, especially at the segments of heavy gradient for death accidents
and sharp curve and combination of slope and curve for injury accidents. Table 6 indicates the
accident characteristics under the conditions of different highway geometric features.
But the fact is that there are still some other traditional problems. Rural two-lane highways
don’t have separate passing lanes in Heilongjiang, China. What’s more, most of them don’t set
passing zones or not passing zones and relevant signs; they just have dashed yellow lines or no
any line between two lanes. There are some accident-prone locations as well. But possibly the
poor management and enforcement are the main problems. Nevertheless, all of these are objective
conditions. They will take effect only under the actions of subjective conditions, which are traffic
participants’ behaviors and most of them are drivers’ behaviors. Hence, accidents were caused by
speeding, illegal passing, illegal meeting, overloading, tailgating and bad road conditions.
According to the statistics, the average speeds were 52.7 km/h to 58.1 km/h for class 2
highways and 40.8 km/h to 43.1 km/h for class 3 highways in Heilongjiang, China. So,
speeding will be more serious if you consider that nearly half the year we face ice and snow
conditions.
In figure 2, it shows the month distribution of traffic accidents that occurred on rural two-
lane highways in Heilongjiang, China. We find that the accidents and severity in winter were
more serious than others. Especially, the lethality rate in winter and spring was 31.85% higher
than that 30.52% in summer and autumn. The data in table 7 show that the accident severity
in snow and ice condition was the most serious.
In Figure 3, it shows time distribution of traffic accidents on rural two-lane highways in
Heilongjiang, China. Obviously, the frequency and severities of accidents were much more
serious during the period between later afternoon to midnight (15:00–22:00). Among these
7 hours, 40.18%, 44.59% and 44.35% of the accident property loses, deaths and injuries were
focused. To start with, there is no lighting except in the town areas for rural two-lane highways,
Table 6. Accidents data versus geometric features on rural two-lane highway in Heilongjiang, China.
1674
Figure 2. Monthly distribution of traffic accidents on rural two-lane highways in Heilongjiang, China.
Table 7. Average deaths and injuries rates in different weather conditions on rural two-lane highways
in Heilongjiang, China.
Figure 3. Time distribution of traffic accidents on rural two-lane highways in Heilongjiang, China.
but it’s also the peak travel time. There are even no headlights, tail lights or emergency warning
lamps for some farmer vehicles, so it will be very dangerous when they stop at the segments of
poor geometric conditions. In addition, glare and weather conditions are also of great impor-
tance. Sun glare and vehicle glare have significant influence on opposite drivers. Especially, the
latter has still worse threats because of no anti-glare facilities for rural two-lane highways and
the bad habit of drivers using high beams. Finally, in the winter, it gets dark and freezes earlier.
So, the accidents that occurred are more serious because of these conditions.
Finally, the most serious and unchangeable factor is behaviors of participants. Most of
them have poor attitudes about drunk driving, fatigue driving, overloading, not using seat-
belts and helmets and so on. Among them, drivers’ driving behaviors are so important; it
may cause a death accident in several seconds with illegal or improper actions or behaviors.
Although bad driving behaviors doesn’t mean an accident, just like drunk driving, overload-
ing, pedestrian violations and illegal turning. But once an accident happens, it may be fatal.
Therefore, a lot of lives were lost when they illegally crossed the intersections.
1675
4 CRASH PREVENTION STRATEGIES
To aim at the main causes of rural two-lane highway accidents, we hereby put forth some
corresponding prevention and mitigation strategies. Parts of them have been implemented
effectively, especially for the engineering improvements.
1. Education, training and propaganda
Just as Ci et al. (17) discussed before, traffic safety education, training and propaganda
are key measures for accident prevention and illegal driving punishment. We conclude it’s
also the most fundamental strategy and can improve current traffic safety level by developing
energetically on rural two-lane highways in Heilongjiang, China.
As a society we would agree that this is a social and systematic problem. But we can also
present some targeted measures. It is not only drivers need to be educated or trained, but
other traffic participants as well. They needed to be educated as to which actions are dan-
gerous or illegal and which actions they should do when they meet certain circumstances.
This content should be emphasized in school education, social education and professional
education from traffic management departments for different groups as a knowledge test, a
television quiz or face to face interview.
Meanwhile, we need some powerful and multidimensional propaganda produced by the
government as well. For instance, we can introduce some statistical accident data or a specific
example. More importantly, we need to explain what happened, and why and how to prevent
this in the future. Further, we need to set up several cautionary traffic signs or models along
the highways, especially near the accident-prone locations.
2. Management and enforcement
However, education is much less effective sometimes. So, management, enforcement and
punishment may be essential also. But punishment cannot be used to instead of education
because it has shown to be less effective in the past.
Most important of all, we need more opportunities to use different kinds of Intelligent
Transportation Systems (ITS) techniques to improve the level and efficiency of management
and enforcement. For example, Intelligent Transportation Monitoring System (ITMS), Elec-
tronic Scales System (ESS), Vehicle Detection System (VDS) and so on. Moreover, we need
more polices on highway enforcement as a deterrent force for illegal driving behaviors at the
present stage. So the laws about seatbelts for all passengers and helmets for cyclists need to be
enforced. Finally, the following are urgent and effective in strengthening source management
as well, annual inspection and registration of motor vehicles, management for applying for
driver’s license and punishment for the illegal driving behaviors.
3. Engineering improvements
Certainly, we need to provide much safer highways for drivers even when they make some
mistakes. There are 10.42% of the accidents, 13.09% of the deaths and 12.89% of the injuries
were made by single vehicle Run-off-Road (ROR) crashes.
First, a safe roadside environment is of great importance. It’s hard for rural two-lane high-
ways to supply roadside Clear Recovery Zone (CRZ) in China (18), but we can implement it
in certain places. Second, improvements should be done for “T” or “Y” intersections in rural
two-lane highways. Widening the road, channelization and setting up signs for proper loca-
tions are effective ways. Third, identification and rectification of the accident-prone locations
are essential. Adding paving markings, signs, deceleration strips and roadside barriers are
partly effectively; adjustments of slope, curvature geometrics and sight distance are also com-
monly used. Finally, perfecting the traffic safety facilities are effective and economic, such as
vibration marking, signs for speed limit, signs for passing or no passing and so on.
4. Emergency
Well functional Emergency Medical Service System (EMSS) is very helpful for the injuries,
because we all know the “Golden Hour”. Emergency Medical Service System (EMSS) is a
special system which can provide road users with towing, on-site rescue, vehicle maintenance
and other services. But actually it’s difficult to build up the EMSS in the broad rural areas
of Heilongjiang, China. We can still increase our efforts to make improvements in this area.
1676
Relevant emergency rescue knowledge should be encouraged through educating, training
and propaganda to the drivers and rural residents. Those who have preliminary emergency
rescue knowledge are encouraged to rescue the injuries at the first time. Furthermore, stand-
ardizing on-site management and control of traffic accidents can reduce secondary-accident
positively. Lastly, some major towns along the rural two-lane highways should be allocated
the emergency equipment and the service ranges should be defined.
5. Accident audit and safety assessment
Thousands of crash records, traffic flow data, and other relevant data are collected and
saved by governments year by year, They are scattered in different departments, and the
lack of communication between these departments resulted in fragmentation of traffic data,
the value of traffic data has not been fully utilized. They should be opened to and shared
with research institutes or consulting firms. Besides, some money and efforts should be spent
on accident audits and safety assessments one by one in rural two-lane highways in Hei-
longjiang, China. Only in this way will we find the specific problems and put forward appro-
priate prevention and improvement strategies and solutions for the different segments of the
rural two-lane highways.
5 CONCLUSIONS
Rural two-lane highways are the major components of the rural highway network system
in China. In order to better understand various attributes of crash occurrence and develop
prevention strategies to considerably reduce crash frequencies or to decrease crash severities
on rural two-lane highways, a study from Heilongjiang, China, was conducted to identify sig-
nificant contributing factors using the crash data collected in the period from 2006 to 2012.
After, some relevant countermeasures were brought forward according to the sequence of the
accident causes calculated by the PCA method. Following are the primary findings:
1. Rural two-lane highways had greater frequencies, deaths and injuries of the traffic acci-
dents than any other rural highways and relatively lower lethality rates;
2. The main causes of the accidents on rural two-lane highways were bad or illegal behaviors
of participants, especially from the drivers. More than 78.19% of the traffic accidents were
related to participants’ behaviors directly. The main causes were speeding, improper yield,
illegal passing and crossing the centerline, which contributing to 64.98% of the accidents
totally;
3. Highways and vehicles factors were relatively minor causes. There were only 0.64% and
1.77% of the accidents involved in highways and vehicles themselves separately. However,
there were nearly 4.3% and 27.48% of the accidents combined with drivers-vehicles and
drivers-highways. Furthermore, environment factors were distinct. Lethality rate in win-
ter and spring was 31.85% higher than that 30.52% in summer and autumn; the accident
severity in snow and ice condition was the most serious; and the frequency and severity of
accidents were much more serious between 15:00 and 22:00, which accounted for 40.18%,
44.59% and 44.35% of the property loses, deaths and injuries respectively; and
4. Several strategies were presented for preventing accidents on rural two-lane highways.
Education, training and propaganda were fundamental ways. But management, enforce-
ment and punishment were essential and necessary also. In addition, engineering improve-
ment was an economic, effective, quick and commonly used way. Finally, emergency and
accident audit and safety assessment need to be enhanced.
ACKNOWLEDGEMENTS
This work was financially supported by the Grant from the Scientific Research Fund of
Heilongjiang Provincial Education Department of China (No. 12541650) and the Project of
Science and Technology from the Department of Transport of Heilongjiang Province, China
1677
(MJ20110034). Special thanks to the Department of Statistics and the Department of Public
Security of Heilongjiang, China, for providing the highway networks data and accidents
records. Here, the authors also would like to thank Mr. Jim Duncan for reviewing.
REFERENCES
Zhao JB, Deng W. Traffic accidents on expressways: new threat to China. Traffic Injury Prevention.
2012, 13(3): 230–238.
Vogt A, Bared J. Accident models for two-lane rural segments and intersections. Transportation Research
Record. 1998, 1635: 18–29.
Schneider WH, Savolainen PT, Moore DN. Effects of horizontal curvature on single-vehicle motorcycle
crashes along rural two-lane highways. Transportation Research Record. 2010, 2194: 91–98.
Peng YC, Geedipally SR, Lord D. Single-vehicle roadway departure crashes on rural two-lane roads.
Transportation Research Record. 2012, 2309: 21–29.
Ansari S, Akhdar F, et al. Causes and effects of road traffic accidents in Saudi Arabia. Public Health.
2000, 114(1): 37–39.
Karlaftis MG, Golias I. Effects of road geometry and traffic volumes on rural roadway accident rates.
Accident Analysis & Prevention. 2002, 34(3): 357–365.
Anderson IB, Krammes RA. Speed reduction as a surrogate for accident experience at horizontal curves
on rural two-lane highways. Transportation Research Record. 2000, 1701: 86–94.
Farah H, Bekhor S, Pous A. Risk evaluation by modeling of passing behavior on rural two-lane high-
ways. Accident Analysis & Prevention. 2009, 41(4): 887–894.
Mujalli RO, Ona JD. A method for simplifying the analysis of traffic accidents injury severity on two-
lane highways using Bayesian networks. Journal of Safety Research. 2011, 42(5): 317–326.
Xi JF, Gao ZH, et al. A hybrid algorithm of traffic accident data mining on cause analysis. Mathemati-
cal Problems in Engineering. 2013: 1–8.
Garber NJ, Ehrhart AA. Effect of speed, flow, and geometric characteristics on crash frequency for
two-lane highways. Transportation Research Record. 2000, 1717: 76–83.
Cafiso S, Cava LG, et al. Safety index for evaluation of two-lane rural highways. Transportation Research
Record. 2007, 2019: 136–145.
Pardillo-Mayora JM, Dominguez-Lira CA, et al. Empirical calibration of a roadside hazardousness
index for Spanish two-lane rural roads. Accident Analysis & Prevention. 2010, 42(6): 2018–2023.
Cafiso S, Graziano AD, et al. Development of comprehensive accident models for two-lane rural high-
ways using exposure, geometry, consistency and context variables. Accident Analysis & Prevention.
2010, 42(4): 1072–1079.
Mohaymany AS, Kashani AT, et al. Identifying significant predictors of head-on conflicts on two-lane
rural roads using inductive loop detectors data. Traffic Injury Prevention. 2011, 12(6): 636–641.
Farah H. Age and gender differences in overtaking maneuvers on two-lane rural highways. Transporta-
tion Research Record. 2011, 2248: 30–36.
Ci YS, Wu LN. Effects and implementation styles on traffic safety education in accident prevention and
punishment. Value Engineering. 2012, 31(3): 11–12. (in Chinese)
Wang YG, Chen KM, et al. Safety performance audit for roadside and median barriers using freeway
crash records: Case study in Jiangxi, China. Scientia Iranica. 2011, 18(6): 1222–1230.
1678
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Zhen Yang
College of Transportation Engineering, Tongji University, Shanghai, China
Haifeng Han
Guangzhou Expressway Co. Ltd., Guangzhou, China
1 INTRODUCTION
1679
accompanied by the increasing of the project construction cost, the difficult problems in
technology and even an unsafely alignment. Therefore, the issue on consistency of the align-
ment is necessary to study from safety and flexibility aspects.
2 LITERATURE REVIEWS
Amundsen F.H. & Ranes, G (2003) found that the accident rate at the tunnel entrance and exit
is 2–3 times higher than other places in the tunnel. Scholars have conducted research on the
index of horizontal alignment of the tunnel entrance and exit. Mashimo, H. (2002) confirmed
that many elements of tunnel design such as cross-section, alignments, drainage, road surface
and shape of portal had important influence on tunnel safety. Yang, Z. & Guo, Z.Y. (2006)
found that geometric design has significant influence on operating speed at the entrance and
exit of tunnel. Based on the relationship between operating speed and road safety, Lin, X.C.
(2007) gave some useful proposal on the geometric design for tunnel and regarded sight distance
as one of the important factors on the alignment design. The consistency of alignment design
is often measured by the consistency of operating speed and the harmony between operating
speed and design speed. The difference of operating speed was selected as the medium index to
evaluate the safety of alignment design at tunnel entrance and exit, including curvature radius
and slope (Wang, Y. et al. 2008). Easa, SM. & Mehmood, A. (2007) developed an optimization
model that established highway horizontal alignment to achieve maximum design consistency
based on the operating-speed (OS) profile. Guideline for Highway Safety Audit of (JTG/T
B05-2004) supplies the concrete method to forecast the operating speed and the criterion of
the critical value of the operating speed to evaluate the highway safety. The indexes of the
driver’s physiology and psychology were also widely used to analyze the driving safety of tun-
nel, including eye movement, heart rate and blood pressure, etc. A new eye movement index
based on plane distribution of view point to evaluate the driver’s visual information level, it
was also applied to safety evaluation of horizontal alignment at tunnel entrance, and the result
indicated that the rational radius of horizontal curve should be no less than 450 m (Du, Z.G.
et al. 2008b). Hong, I. et al. (2006) developed a model to evaluated the risk of curved sections
in relation to the driver’s eye movement and driving behavior.
In order to meet the demands of the vertical alignment design, stopping sight distance and
the crossing sight distance, Ministry of Communications of China (2004b) demanded that
there need consistency alignment before and after 3 s traveling distance at the tunnel entrance
and exit, or 5 s if condition permit. But the guide has not given the concrete meaning of the
consistency of the alignment design.
3 METHODOLOGY
The consistency of traveling for 3 s was proposed based on the consideration of the negative
impact on the vehicle sight distance by black-hole effect and white-hole effect at the tunnel
entrance and exit. If the tunnel entrance or exit sit on a spiral curve with the large spiral
transition curve parameter A, for instance, A = 2000, the spiral curve in a relatively short
distance (3 s traveling distance) can be regarded as a circle, namely the deviation of the track
in circle is very close to the spiral curve. In this condition, it is reasonable to judge the align-
ment satisfy the consistency regarded by the Design Specification for Highway Alignment
(JTGD20–2006). The primary issue is to come to the problem that how much deviation can
be accept according to safety and practicality.
Design Specification for Highway Alignment (JTGD20-2006) permits the straight line
being connected with a circle directly on the premise that the deviation of track from the
sweep movement of the circle extended to tangent direction angle is no more than 0.2 m
(p in Fig.1). So if we fix the steering wheel at the tunnel entrance or exit, and the deviation
from the sweep movement of the later 3 s traveling to design track is controlled no more than
0.2 m, it can be acceptable that alignment satisfy consistency requirement.
1680
Figure 1. Standard horizontal alignment.
For instance, if the tunnel portal and the terminal of the 3 s traveling distance are both on
the spiral transition curve (Fig. 2). Point A is the tunnel portal. Point B is the terminal of the
3 s traveling distance along with design track from Point A. Point C is the terminal of the 3 s
traveling distance along with the fixed curvature of point A from point A. If we calculate the
distance from B to C (ΔD), and compare it with 0.2 m, we can evaluate the consistency of
the horizontal alignment. If ΔD< = 0.2 m, the 3 s traveling distance according to design speed
around the tunnel portal keep consistence, vice versa.
Quantitative calculation was applied to investigate relationship between the spiral curve
parameter A and ΔD.
Combinations of the positions of point A and point B can be concluded into 3 cases. Case
A is the situation that point A and point B both locate on the spiral curve, case B is the situa-
tion that point A locates on the tangent line and point B locates on the spiral curve, and case
C is the situation that point A locates on the spiral curve and point B locates on the circle.
1681
Moreover, each case was divided into several categories with different design speed (80 km/h,
100 km/h and 120 km/h).
Design Specification for Highway Alignment (JTGD20-2006) required the minimum hori-
zontal curve radius should not less than 250 m for 80 km/h design speed, and also required
the minimum length of spiral curve to be 70 m. So the minimum spiral curve parameter A is
132.28 (A2 = R × L); when the design speed is 100 km/h, the minimum radius of horizontal
curve is 400 m, the minimum length of Spiral curve is 85 m, so the minimum spiral curve
parameter is 184.39; when the design speed is 120 Km/h, the minimum radius of horizontal
curve is 650 m, the minimum length of spiral curve is 100 m, so the minimum spiral curve
parameter is 254.95. Moreover, the distance from point A to B was defined as 67 m, 85 m and
100 m according to the design speed of 80 km/h, 100 km/h and 120 km/h.
3.1 Case A
In this case, point A and point B both locate on the spiral transition curve (Fig. 2). MAT-
LAB programming was applied to develop the model to calculate ΔD, the value of the spiral
curve parameter A and l (the distance from the point A to point ZH where straight line and
spiral transition curve are connected) were continuously changed to investigate relationship
between the two influence factors. The formulas are listed as follows.
l2
β=
2 A2
l5 l9
x0 l− +
40 A4 3456 A8
l3 l7
y0 = −
6 A2 336 A6
A2
R=
l
xc x0 − R × sin(β)
yc = y0 + R × cos(β)
l′ = l + S
l ′5 l ′9
x l′ − +
40 A4 3456 A8
l ′3 l ′7
y= −
6 A 336 A6
2
D (( x − xc )2 + ( y − yc )2 )0 5 R (1)
where, β = intersection angle between x-axis and tangent line on point A; S = distance between
point A and B (or C), which is defined as 67 m, 85 m or 100 m according to the design speed
of 80 km/h, 100 km/h and 120 km/h; x0, y0 = the coordinate of the point A; R = curvature
radius of point A; xc, yc = coordinate of the circle center according to point A; x, y = coordi-
nate of the point B.
It can be found that the value of ΔD remained unchanged when fixing the spiral curve
parameter A, even change the value of l. It can be conclude that the value of ΔD is only influ-
enced by the value of the parameter A, not the value of l.
The calculation results for 80 km/h are listed in Table 1. Other design speeds cases are also
calculated and will be presented later.
1682
Table 1. The calculation results of ΔD in case A (the design speed is
80 km/h).
A ΔD(m) S(m)
3.2 Case B
In this case, point A locates on the straight line and point B locates on the spiral transition
curve (Fig. 3).
The formulas are listed as follows.
l S − l0
l2
β=
2 A2
l5 l9
x0 l− +
40 A 3456 A8
4
l3 l7
ΔD y0 = − (2)
6 A2 336 A6
where, l = distance from point ZH (the connection point of straight line and spiral transition
curve) to point B; l0 = distance from point ZH to point A; x0, y0 = coordinate of the point A.
In order to get accurate value of l0 to satisfy the ΔD to be equal 0.2, iterative arithmetic is
developed in MATLAB, and the results for 80 km/h design speed is listed in Table 2. Other
design speeds cases are also calculated and will be presented later.
3.3 Case C
In this case, point A locates on the spiral curve and point B locates on the circle (Fig. 4). The
formulas are listed as follows.
1683
Figure 3. Graphical calculation (Case B).
l2
β0 =
2 A2
l5 l9
x0 l− +
40 A4 3456 A8
l3 l7
y0 = −
6 A2 336 A6
1684
Figure 4. Graphical calculation (Case C).
A2
R0 =
l
xc x0 − R0 × sin(β0 )
yc = y0 + R0 × cos(β0 )
LS 5 LS 9
x LS − +
40 A 3456 A8
4
LS 3 LS 7
y= −
6 A2 336 A6
LS 2
β=
2 A2
A2
R=
LS
thDa = (S l − Ls ) / R
xx = R sin(thDa )
yy = R ( thDa )
xxx = x + xx × β) − yy
y × sin(β )
yyy = y + xx × β ) + yy
y × cos(β )
1685
Table 3. Calculation results of ΔD With fixed Ls – l in case C (80 km/h design speed).
where Ls = the length of the Spiral transition curve; l = distance from point ZH to point A;
β 0 = the intersection angle between x-axis and tangent line on point A; x0, y0 = coordinate
of the point A; R0 as the curvature radius of point A; xc, yc = coordinate of the circle center
according to point A; x, y = coordinate of point HY(connected point of spiral curve and cir-
cle curve); β = intersection angle between x-axis and tangent line on point ZH; R = curvature
radius of point HY; thDa = central angle from point HY to point B which is no more than
180; xx, yy = relative coordinates of point B based on point HY; xxx, yyy = coordinate of
point B.
Data were calculated with the parameter A unchanged while Ls and l changed (Table 3).
Apparently, ΔD was only relative to Ls – l, not the single Ls or l.
Iterative arithmetic is also developed in MATLAB, in order to find the accurate value of
Ls – l to satisfy the ΔD to be equal 0.2. The results for 80 km/h design speed is listed in Table 4.
Other design speeds cases are also calculated and will be presented later.
4.1 Case A
The regression analysis of calculation results of three different design speed (80 km/h,
100 km/h, and 120 km/h) demonstrated a very good relevance. Obviously, ΔD is inversely
related to the square of the parameter A (Fig. 5).
1686
Figure 5. Regression curve (ΔD vs A).
1687
The analysis indicate that alignment consistency will be satisfied if the spiral transition
curve parameter A is larger than a critical value for a certain design speed in case A. The
critical values are 500, 715, and 913 respectively for design speed of 80 km/h, 100 km/h, and
120 km/h.
4.2 Case B
Based on the calculation results for three different design speed situation, it can be find that
ΔD has very good linear correlation to the value l0 (the distance from entrance or exit point to
the connected point of straight line and spiral transition curve), the smaller l0 will get larger
ΔD (Fig. 6).
4.3 Case C
Double exponent fitting give a good description to express the relationship between the
parameter A and Ls – l (Fig. 7).
The following characters we can deduced.
1. The influence of Ls – l on ΔD seemed less sensitive when the parameter gets bigger.
2. There has a critical value to A (spiral transition curve parameter) for a certain design
speed. If the A is bigger than the critical value, ΔD is always less than 0.2 m no matter
what value of Ls – l. The critical values are 500, 715, and 913 respectively for design speed
of 80 km/h, 100 km/h, and 120 km/h, the same as the value in Case B.
5 CONCLUSIONS
Deviation distance from design trajectory to actual track, which cause by 3 seconds freez-
ing steering wheel driving, is an acceptable index to evaluate the consistency of horizontal
alignment at tunnel entrance and exit. 0.2 m is chosen as the critical value which deuce from
Design Specification for Highway Alignment (JTGD20-2006). All alignment design condition
can be included into three cases, and the following conclusions are list to guarantee alignment
consistency based on quantitative calculation and analysis.
1. There are critical values of spiral curve parameter A for each design speed. If the A is big-
ger than the critical value, the alignment will always satisfy the consistency. These value are
500, 715, and 913 respectively for design speed of 80 km/h, 100 km/h, and 120 km/h.
2. If the parameter A < 500 and A> = 132 for 80 km/h design speed, or A < 715and A> = 184
for 100 km/h, or A < 913 and A> = 255 for 120 km/h, the following additional conditions
should be satisfied:
a. If point A locates on the straight line and point B locates on the spiral transition curve, the
maximum value of l0 (the distance from point ZH to the entrance or exit) can be estimate by
the following formulas: l0 = −0.1062 * A + 52.11 for 80 km/h design speed, 0.09582 * A + 66.59
for 100 km/h and –0.08686 * A + 77.47 for 120 km/h.
b. If point A locates on the spiral curve and point B on the circle curve, Ls – l should be great
than the value decided by the following formulas: 0.8254*exp (0.007828 * A)+4.915e-012* exp
(0.05751* A) for 80 km/h design speed, 1.371 * exp (0.004708 * A) +4.69e-015 * exp (0.05129
* A) for 100, and1.69 * exp (0.003853 * A)− 3.732e-015 * exp (0.04026 * A) for 120 km/h
3. The above rules can be deducted to other situations: point A locates on the spiral curve
while point B locates on the straight line; point A locates on the circle while point B
locates on the spiral curves.
1688
ACKNOWLEDGMENT
This paper is supported by the National Natural Science Foundation of China (Grant NO.
51478352) and Science and technology project of Guangdong Provincial Department of
transportation (Grant NO. 2013-02-058).
REFERENCES
Amundsen, F.H., Ranes, G. 2003. Studies on Traffic Accident in Norwegian Road Tunnels. Tunnelling
and Underground Space Technology, 15: 3–11.
Du, Z.G., Pan, X.D. & Yang, Z. 2008a. Relationship between eye movement index and horizontal
alignment at entrance of highway tunnel. Journal of Tongji University: Natural Science, 36(12):
1661–1664.
Du, Z.G., Pan, X.D. & Guo, B.X. 2008b. Evaluation Index’s Application Studies on Safety at Highway
Tunnel’s Entrance and Exit. Journal of Tongji University: Natural Science, 36(3): 325–329
Easa, SM., Mehmood, A. 2007. Establishing Highway Horizontal Alignment to Maximize Design Con-
sistency. Canadian Journal of Civil, 34(9): 1159–1168.
FHWA 2009. Technical Manual for Design and Construction of Road Tunnels FHWA-NHI-10–034
Hong, I., Iwasaki, M., Furuichi, T. & Kadoma, T. 2006. Eye Movement and Driving Behaviour in
Curved Section Passages of an Urban Motorway. Proceedings of The Institution of Mechanical Engi-
neers, Part D: Journal of Automobile Engineering, 220(10): 1319–1331.
Mashimo, H. 2002. State of the Road Tunnel Safety Technology in Japan. Tunnelling and Under Ground
Space Technology. 17: 145–152.
Ministry of Communications of the People’s of Republic of China. 2004a. Guideline for Highway Safety
Audit of (JTG/T B05-2004). China Communications Press, Beijing,
Ministry of Communications of the People’s of Republic of China. 2004b Code for Design of Road
Runnel (JTG D70-2004). China Communications Press, Beijing.
Ministry of Communications of the People’s of Republic of China 2006. Design Specification for High-
way Alignment (JTGD20–2006). China Communications Press, Beijing, 2006.
Lin, X.C. 2007 Discussion on the Horizontal Alignment of the Highway Tunnel’s Entrance and Exit.
Highway, No.3, 2007.
Wang, Y., Kong, L.Q., Guo, Z.Y. & HAN, C.L. 2008. Alignment Design at Tunnel Entrance and Exit
Zone Based on Operating Safety. Journal of Highway and Transportation Research and Development.
25(3): 134–138.
Yang, Z, Guo, Z.Y. 2006. Research on Speed Alteration at Entrance and Exit of Tunnel. Shanghai
Highway, No.1, 2006: 48–51.
1689
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Zhaohui Sun
The Traffic Engineering School of Shenyang, Jianzhu University, China
Tiebin Wang
Liaoning Provincial Traffic Engineering Quality and Safety Supervision Bureau, China
Zefeng Wu
Liaoning Provincial Highway Administration Bureau, China
Guangqiang Zhu
The Traffic Engineering School of Shenyang, Jianzhu University, China
ABSTRACT: The space distribution of aggregate, asphalt binder and pore in the structure of asphalt
mixture is affected by raw materials, construction technology and other factors, it has great variability,
asphalt mixing material differences in macroeconomic performance is asphalt mixture phase differences
in the microscopic distribution in the final analysis. This paper attempts to apply the fractal theory to
test the gradation, in order to reduce the amount of experiment in the course of gradation test. The
correlation models of fractal volume parameters and fractal dimension of GAC-20 modified asphalt
mixture are established. The volume parameters of GAC-20 modified asphalt mixture can be predicted
by the known fractal dimension, which is advantageous to test the grading and determine the structure
type of asphalt mixture. Thereby the workload can be reduced and the design.
1 INTRODUCTION
The fractal characteristics of aggregate particle size distribution, the fractal characteristics of
aggregate mass-grading, the fractal characteristics of aggregate volume distribution, and the
fractal characteristics of porosity of asphalt mixture were analyzed based on fractal theory,
through comparative study between the gradation inspection method based on fractal theory
and the VCADRF method, The research indicates that the gradation inspection based on frac-
tal theory is coincident with the gradation inspection based on experiment. Thus, applying
fractal theory to examine the aggregate is an effective method, which can reduce experimental
quantity, and improve test efficiency.(SUN Zhaohui et al., 2010)
There is a correlation between fractal volume parameters and fractal dimension of
GAC-20 modified asphalt mixture. We would try to establish the correlation model between
them.
The coarse aggregate fractal volume Vc, coarse aggregate fractal voidage Vco and the frac-
tal volume Vf of fine aggregate in the coarse aggregate are analyzed according to the early
research results of the gradation test method of asphalt mixture based on fractal theory. The
fractal volume index is shown in formula (1), (2) and (3).(SUN Zhaohui et al., 2010)
The fractal volume of coarse aggregate in the whole VC
1691
MO ( Dc ) x 3−DDc 4.753 Dc
VC = −
ρ ∫4 75 xmax 3− DC
ddx = max
ρ xmax 3− DC
MO (1)
Type: xmax is maximum particle sieve size; DC is coarse aggregate fractal dimension; Mo is
the overall quality percentage; R is coarse aggregate tamping compaction density;The coarse
aggregate fractal voidage Vco:
VC MC / ρc
VCO = (2)
Vc
Liaohe petroleum asphalt grade A No.90, which is widely used in the northeast of China and
the basic performance test results are shown in Table 1, the basic performance test results of
SBS modified asphalt are shown in Table 2.
The coarse and fine aggregate of GAC-20 mixture use limestone gravel produced by
Liaoyang Xiaotun victory quarry. The basic performance test results are shown in table 3.
Grade A No.90 road petroleum asphalt, SBS modified additives and limestone were tested
in accordance with the requirements of the road usage.
1692
Table 3. Technical index of limestone coarse aggregate.
No. Coarse Fine Filler Gross bulk Gross bulk Bulk The fractal The fractal The fractal
aggregate aggregate quality density density density of dimension dimension dimension
mass mass (%) of coarse of fine asphalt of of coarse of fine
fraction fraction aggregate aggregate mixture gradationD aggregate aggregate
(%) (%) (g/cm3) (g/cm3) (g/cm3) Dc Df
Vc, Vco and Vf are calculated according to formula (1),(2) and (3).The coarse aggregate
can form an effective framework (SUN Zhaohui et al., 2010) when the fractal voidage Vco
of coarse aggregate is larger than the fractal volume Vf of the fine aggregate in the coarse
aggregate, the skeleton of coarse aggregate can be formed.The Vco and Vf value are obtained
from the 25 groups of GAC-20 mixture, and the gradation data of the skeleton formed can
be shown in table 5.
First, the three elements linear regression model is established between the coarse aggre-
gate fractal void rate Vco and the fractal dimension D,Dc and Df by using MATLAB pro-
gramming. The residual plot of Vco and the fractal dimension was drawn in Figure 1.
1693
Table 5. The fractal volume and fractal dimension.
It can be seen from Figure 1, excluding the abnormal data points 1, the three element linear
regression model of Vco and the fractal dimension is obtained in the formula (4).
1694
Table 6. Correlation between fractal voidage of coarse aggregate and fractal dimension.
D Dc Df Vco
It can be seen from table 6, the correlation among Vco and D, Dc, Df ordered from small to
big is Dc < D < Df, the correlation between Vco and Df is relatively high, and the correlation
model of Vco and Df can be established and shown in the formula (5)
1695
Table 7. The correlation among Vf and D, Dc, Df.
D Dc Df Vf
Table 8. The comparison and selection for prediction model of Vco and Vf.
Figure 2. The correlation models of fractal volume parameters and fractal dimension.
5 CONCLUSIONS
The correlation models of fractal volume parameters and fractal dimension of GAC-20 gra-
dation is established in the research.
The prediction model of coarse aggregate fractal voidage of GAC-20 gradation is estab-
lished, that is VCO 1 5783 0 4599DC + 1 2698D f . The prediction model of the fractal
volume of fine aggregate in coarse aggregate for GAC-20 gradation is established, that is
Vf 1 0892 0 6795DC − 0.1456 D f ,which is advantageous to test the grading and deter-
mine the structure type of asphalt mixture. Thereby the workload can be reduced and the
design efficiency can be improved.
1696
ACKNOWLEDGEMENTS
This work was financially supported by the Natural Science Foundation of China
(51178278)
REFERENCES
Occupation Standard of the People’s Republic of China., JTG E20-2011 “Standard Test Methods of
Bitumen and Bituminous Mixtures for Highway Engineering”
Highway Science Research Institute, Ministry of Communications JTG F40-2004 “Technical Specifica-
tions for Construction of Highway Asphalt Pavements”
Occupation Standard of the People’s Republic of China, JTG E42-2005 “Test Methods of Aggregate
for Highway Engineering”
SUN Zhaohui, Yang Guofeng,Yu Baoyang Wang Tiebin. 2010 Study on the Gradation Inspection
Methold Based on Fractal Theory[J] Journal of Shenyang Jianzhu University (Natural Science)
26(5):875–880
Lan 1i, Paul Chan.1993 Quantitative Analysis of Aggregate Shape Basedon Fractal ACI Materials Jour-
nal, July-August.
Mohan Yeggoni, Joe W. Button, Dan G. Zollinger.1996 Fractals of Aggregates Correlated with Creep
in Asphalt Concrete, Journal of Transportation Engineering 122(1):22–28
Feodor M. Borodich.1999 Fractals and fractal scaling in fracture mechanics International Journal of
Fracture,(95):239–259
Faleoner,1990 Fractal Geometry: Mathematical Foundation and Applications Wiely,NewYork.
Lee. H., Carr, Barr J. & Haas J, 1990 The Fractal Dimension as a Measure of the Roughness of Rock
Discontinuity profiles, International Journal of Rock.
Ribble, C., Szecsy, R. and Zollinger, D.G. 1992 Aggregate macro shape and micro texture in concrete
mix design. ASCE Spring Meeting, New York.
Joseph K. Anochie-Boateng, Julius Komba, and Erol Tutumluer 2012 Aggregate Surface Areas Quanti-
fied through Laser Measurements for South African Asphalt Mixtures Journal of Transportation
Engineering 138(8):1006–1015
W.H. Campen,1959 Relationship Between Voids, Surface Area, Film Thickness, and stability in Bitumi-
nous Paving Mixtures, Proc. AAPT, 28
Adhara Castelblanco Torres. 2004 Probabilistic Analysis of Air Void Structure and Its Relationship to
Permeability and Moisture Damage of Hot Mix Asphalt Texas A&M University.
Xinjun Li, R.Christopher Williams, Mihai O. Marasteanu, Timothy R. Clyne, Eddie Johnson. Investi-
gation of In-Place Asphalt Film Thickness and Performance of Hot-Mix Asphalt Mixtures Journal
of Materials in Civil Engineering (21):262–270
Leonardi, G. 2010 Fractal dimension for the characterization of the porosity of asphalt concretes.
Archives of Civil Engineering, 56(4):321–333.
1697
Other Topics
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Xianhua Chen
Southeast University, Nanjing, China
ABSTRACT: Understanding the stresses conditions of flexible pavement under wheel loading
is critical for the undergraduate and graduate students to learn the principle of pavement design.
The paper presented a simulation tool for the students to evaluate the real 3D stresses conditions
of flexible pavements. The tool is developed by the implementation of Fourier-aided series Finite
Element Methods in Matlab® with GUI programming method. The layered flexible pavements are
meshed with 2-D eight-node quadratic elements, and the wheel loads has been approximated with
an eighty items of Fourier Series. The parallel computing technology has been applied to speed up
the program. It is concluded that the tool is fast and easy-to-use with high accuracy, which could
be used in pavement engineering courses for the students to evaluate the load responses of flexible
pavements, as well as in routine pavement design and in scientific research works.
1 INTRODUCTION
Understanding the stresses conditions of flexible pavement under wheel loading is critical for
the undergraduate students to learn the principle of asphalt pavement design. In traditional
lecturing the pavement engineering, a number of tools are available for the students to per-
form such purposes with high efficiency. Most of them were based on layered elastic theory
and were developed at least 20 years ago. The layered elastic theory, originated from Boussin-
esq’s problem on elastic half-space, uses the critical assumptions of linear elastic homogenous
materials of each layer, and axial symmetry on the wheel loading and pavement structure
(Huang 2003, Das 2015, Deng et al.2007), as illustrated in FIG 1. With these critical assump-
tions the three-dimensional problem can be reduced into a two dimensional case.
The pavement structure is apparently not infinite in width, and the real wheel load stamps
are more like a ellipsoidal block or a rectangle than a circular and the vertical stress in con-
tact are not in an uniform distribution (Beer et al. 1996, Siddharthan et al. 2002, HU et al.
Figure 1. Critical assumptions of layered elastic theory (illustrated with single circular load).
1701
2011), as shown in FIG 2. Instead of isotropic materials, many researchers (Ewan et al.
2007, Zhang et al. 2011, Dong et al. 2012) suggested that to use a transversely isotropic
materials whose physical properties are symmetric about the vertical direction that is normal
to the horizontal plane of isotropy. Those cases are far beyond the capabilities of classical
layered elastic theory.
To overcoming the shortages of layered elastic theory, it is very popular in the road engi-
neering society adopting commercial finite element analysis (FEA) tools such as Ansys or
Abaqus to investigate of pavement responses under varies conditions. These tools are cer-
tainly capable in simulating the responses of pavement structures. However, for the daily
pavement engineering lecturing, the finite element method (FEM) has made a comparatively
small impact. The reasons are not only because of the expensive cost to get valid licenses and
time consuming works in three dimensional analyses, but also due to the complexity of the
tools, which requires students have a sound theoretical background and spend much times
to learn the tools before they can make a reasonable simulation. In addition, there is also a
significant increase in computational requirements and analysis time of three dimensional
FEM compared to layered elastic models. It thus appears that there at present few simulation
tools that are relatively inexpensive and complex for the simulation in pavement engineering
courses, and problems that require a fast and efficient three-dimensional visualization of the
stress or strain field within the pavement structures.
The solution seems promising through the Fourier’s series Aided FEM or so-called Semi-
analytical FEM (SAFEM) (Jooste 2002, Zienkiewicz and Taylor 2005). Conventionally,
the SAFEM has been applied to problems with an axi-symmetric geometry (but non axi-
symmetric loading and/or variation of material properties) and it is the application of the
method to such problems that considered. However, the method can also be implemented for
problems expressed in terms of Cartesian geometries. The SAFEM is an effective method
for modeling the load response of structure in which the material properties and problem
geometry do not change in one coordinate direction. By making use of orthogonal functions
and Fourier series, problems of this class permit a two-dimensional plane strain analysis
Figure 2. Typical tyre/pavement contact stamps and the vertical contact stress distribution.
1702
to be substituted for a full three-dimensional treatment of the problem. The method offers
considerable savings in computational requirements compared to a full three-dimensional
finite element analysis, yet allows the modeling of complex pavement design situations such
as those involving cracks or non-homogeneous layers.
The paper presented a tool for the undergraduate students to evaluate the real 3D stresses
conditions of flexible pavements by the implementation of SAFEM in MATLAB program-
ming with GUI method. The layered flexible pavements are meshed with 2-D eight-node
quadratic elements, and the wheel loads has been approximated with an eighty items of Fou-
rier series. The parallel computing technology has been applied to speed up the program. It
is concluded that the tool is fast and easy-to-use with high accuracy, which could be used in
pavement engineering courses for the students to evaluate the load responses of flexible pave-
ments, as well as in routine pavement design and in scientific research works.
2 THEORETICAL BACKGROUND
As pointed out by Zienkiewicz et al.(2005), the SAFEM substitutes the full three dimensional
problem with a series of simplified two-dimensional solutions by the aid of Fourier series.
And the SAFEM is suitable for the analysis of problems in which the problem geometry
and material properties do not vary significantly along one co-ordinate direction, but with
a significant variation in load terms in that direction. That is the scenario of the pavements
structure under wheel load.
In the typical two-dimensional finite element formulation, displacements at a point within
an element can be expressed as:
n
u ∑ N (xx y)u .
i =1
i i
(1)
By selecting alternative shape functions that can be expressed in terms of a Fourier series,
and by expressing the load in a similar way, the two dimensional problem be transformed to
an equivalent three dimensional problem, as shown in FIG 4.
For the situation shown in FIG 4, the problem needs to be fully described in terms of
(x,y,z) coordinates in which z is the coordinate along which the geometry and material prop-
erties do not change, and which is limited to lie between zero and a. The shape functions
defining the variation of displacements are written as a Fourier series:
n ~ L
⎡ l z lπ z ⎤
u ∑ N ( x, y, z ) u ∑ ⎣N (x
i =1
i i ( x, y )).coss
l =1 a
+ N " ( x, y ).sin
i
a ⎥⎦
. (2)
1703
In this type of representation completeness is preserved in view of the capability of Fourier
series to represent any continuous function within a given region (naturally assuming that the
shape functions N ′ and N ′′ in the domain x, y satisfy the same requirements).
The loading terms will similarly be given a form,
L
⎡ _ l z = lπ z ⎤
f ∑ ⎣ p (xx y) cos
l =1
l
a
pl ( x, y ).sin
i
a ⎥⎦
. (3)
Applying the standard process for the determination of the element contribution to the equa-
tion minimizing the potential energy, a typical sub-matrix of the element stiffness matrix is,
( l
) ∫(
Ω
l T
) DB m d dz
d . (4)
( l
) ∫(
Ω
) pl d Ωdz.
l T (5)
By virtue of the formulation of the element displacements given by equation (2), the
matrix provided by equation (4) will contain the following integrals as products of various
submatrices,
lπ z a mπ z
I1 =∫ sin cos dz
a 0 a
a lπ z mπ z
I 2 =∫ sin sin
i dz. (6)
0 a a
a lπ πz
I 3 =∫ cos cos dz
0 a a
These integrals arise from products of the derivatives contained in the definition of strain-
displacement matrix and, owing to the well-known orthogonal property, give,
⎧1
⎪ a, l = m
I2 I3 = ⎨2 . (7)
⎪⎩ 0, l ≠ m
The integral I1 vanishes in most applications and leads the element stiffness matrix to
become diagonal due to the orthogonal property, and the final global assembled equations
for the problem have consequently the form:
⎡ 11 ⎤⎧ ~ ⎫ ⎧ ⎫
⎢K ⎥ ⎪ u1 ⎪ ⎪ F1 ⎪
⎢ ⎥⎪ ~ ⎪ ⎪ ⎪
⎢ K 22 ⎥ ⎪ u2 ⎪ ⎪ F2 ⎪
⎢ . ⎥⎪ . ⎪ ⎪ . ⎪ (8)
⎢ ⎥ ⎨ ⎬ + ⎨ ⎬ = 0.
⎢ . ⎥⎪ . ⎪ ⎪ . ⎪
⎢ . ⎥⎪ . ⎪ ⎪ . ⎪
⎢ ⎥⎪ ⎪ ⎪ ⎪
⎢ ⎥⎪ ~ ⎪ ⎪ ⎪
⎣ K ⎦ ⎩uL ⎭ ⎩FL ⎭
LL
Equation (8) shows that the large system of equations splits into L separate problems,
where L is the number of Fourier terms. Due to the orthogonality of the integrals in equa-
tions (4) to (5), and in combination with the formulation of the displacements and loads
1704
given by equations (2) and (3), the system has consequently a property that the force term of
the lth series term affects only the lth system of equations in equation. This property is of
extreme importance, because, if the expansion of the loading factors involves only one term
for a particular harmonic, then only one set of simultaneous equations need to be solved.
The solution of this will tend to the exact one with increasing subdivision in the x-y domain
only. Thus, what was originally a three-dimensional problem now has been reduced to a two-
dimensional one with consequent reduction of computational effort.
Equations (2) and (3) represent the generalized form of the displacement and load terms.
For a particular application, the functions describing the displacements and the load terms
are chosen such that the boundary conditions at the problem ends are satisfied. For example,
if the displacements in the x-y plane are zero at the problem ends (in the z-direction), then
the shape functions will be chosen such that the lth displacement in the x and y directions
are given by:
⎧ ~l ⎫
u
⎧u ⎫
l
⎡sin γ l z 0 0 ⎤ ⎪ b⎪
⎪ ⎪ ⎪~⎪
u l = ⎨ v l ⎬ = ∑ Nb ⎢⎢ 0 0 ⎥⎥ . ⎨ vbl ⎬.
(9)
si γ l z
sin
⎪w l ⎪ b ⎢⎣ 0 0 cos γ l z ⎥⎦ ⎪ ~l ⎪
⎩ ⎭ ⎪wb ⎪
⎩ ⎭
Where γ l lπ / a . In this, Nb are simply the (scalar) shape functions appropriate to the ele-
ments used in the x-y plane. The displacement expansion ensures zero u and v displacements
at the ends and the zero traction condition can be imposed in a standard manner.
The load terms are selected to ensure that the applied load shape is correctly simulated in
the z domain. For pavement applications, the load terms are represented as below:
L
F ( z ) =∑ bn si γ l z. (10)
l =1
with:
4
2 Pt ⎡ l Ft1t1 l Ft 2 ⎤
bn =∑ ⎢⎣cos a cos . (11)
t =1 l π a ⎥⎦
where Pt is the tth tire load pressure; Ft1 is the z coordinates where the tth tire load starts; Ft2
is the z coordinates where the tth tire load ends.
3 PROGRAM DESCRIPTION
According to Xu 2006 and Ferreira 2008, MATLAB is chosen to develop the tool as its pow-
erful matrix computational capabilities. The whole tool includes those basic models writing
into Matlab functions with a total size less than 1 MB:
1. Structure initiation model: input the information such as, pavement structure thickness
and basic elastic material parameters, as shown in FIG 5. It should be mentioned that
transversely isotropic materials (shown as in FIG 6) can be considered in the simulation
tool;
2. Element and load configuration model: 8 nodes iso-parameter element with linear shape
function (Smith et al.2013) is used. Dual –tyre of a single axle load and rectangle wheel
stamp areas are adopted in the tool, as shown in FIG 7 and FIG 8;
3. Mesh model: a fine mesh in loading area is adopted considering multi-layer and overload-
ing conditions, shown as in FIG 7. Output of this model is convex matrix of elements and
nodes;
1705
Figure 5. Basic input interface of the simulation tool for isotropic materials.
1706
4. Element stiffness matrix model: this model is coded according to equation (4);
5. Assemble global stiffness matrix model: The global stiffness matrix is assembled by direct
stiffness method (Smith et al.2013).
6. Displacement Constrain model: In the system, the bottom line has zero d.o.f. in all three
directions, and the boundary lines perpendicular to x-axis are constrained to deform in z
and y directions;
7. Load vector model: A 100-item of Fourier series is used to construct the vectors of dual
wheel stamp of a single axis load, as demonstrated in FIG 9; It should be noted that the
simulation tool takes only 80 items of Fourier series for the purpose of computational
efficiency.
8. Solution model: The method of Gauss elimination is used;
9. Post-process model: output the displacement, stress and strain at given section, and gener-
ate report in excel sheet for further investigations.
4 PROGRAM VALIDATION
According to Equation (8), the global stiffness matrix should be a symmetric sparse matrix
and only diagonal elements with a certain band have non-zero value. Therefore, we check the
properties of global stiffness matrix after the assembling model. As demonstrated in FIG 11,
the assembled global stiffness matrix indeed reveals the periodic symmetric sparse properties,
which means that the kernel parts of the simulation tool are correctly coded.
Figure 9. An illustration of meshed pavement structure with nodes and elements number.
Figure 10. The load sketch and 100 items Fourier series expansion effect.
1707
Figure 11. Global stiffness matrix of the system and a zoon-in view with non-zero elements.
Figure 12. Surface displacement along the center of the wheel load.
It is further confirmed by the symmetric deflection curve along the center of the wheel
load, shown as in FIG 12. The effects of Fourier series’ item on the results are also investi-
gated, as shown in FIG 12. When the number of items increased from 50 to 70, the maximum
deflection decreases from −0.058 mm to −0.05459 mm, the curve becomes smoother and the
visible oscillation in non-loading area almost vanishes, for the items increases from 80 to 100,
the deflection value changes from −0.05458 mm to −0.05454 mm. The computational times
changes from 14 seconds to 22 seconds with the Fourier items number increasing from 50 to
100. To balance between computational efficiency and precision, the tool finally adopts an 80
items Fourier series for simulation.
To speed up the computational efficiency, parallel computing technique and vector pro-
gramming method are adopted. For the system containing 252 elements and 899 nodes, and
80 terms of Fourier series, which is the equivalent three dimensional system of 20160 elements,
it takes less than 20 seconds to solve the problem on a laptop of i7 CPU and 8 Gb flash mem-
ory. In equivalent case of 3-D analysis in ANSYS, the solution time is over than 300 seconds.
1708
5 CONCLUSION
Up to now, the framework of the software is ready. The computational results of the soft-
ware are verified by symmetric characteristics. The Fourier series items is optimized, parallel
computing technique and vector programming method are adopted to improve the computa-
tional efficiency. It is concluded that the tool is fast and easy-to-use with high accuracy, which
could be used in pavement engineering courses for the students to evaluate the load responses
of flexible pavements, as well as in routine pavement design and in scientific research works.
Further works will focus on optimization of the meshing scheme, and take the non-linearity
materials and discontinuity such as interfacial slippage and reflective cracking into the tools
model.
ACKNOWLEDGEMENT
This paper is based on the works that the author accomplished during his visiting of RWTH-
AACHEN. The corresponding author would like to thank CSC/DAAD for the nice spon-
sorship of his visiting ISAC, RWTH-Aachen. The author also would like to thank NSFC
(51478114) for the financial support on hardware.
REFERENCES
Animesh Das, 2015, Analysis of Pavement Structures, CRC Press, Taylor & Francis Group, Boca
Raton.
de Beer, M., C. Fisher, F.J. Jooste, 1996, Determination of pneumatic tyre/pavement interface contact
stresses under moving loads and some effects on pavements with thin asphalt surfacing layers, Pro-
ceedings of the Eighth International Conference on Asphalt Pavements. Seattle, WA, pp. 179–227.
Ewan, Y.G., E.P. Chen, 2007, Responses of a Transversely Isotropic Layered Half-Space to Multiple
Horizontal Loads, Proceedings of Symposium on Pavement Mechanics and Materials at the 18th
ASCE Engineering Mechanics Division Conference, June 3–6, Blacksburg, Virginia, pp. 63–77.
Ferreira, A.J.M. 2008, MATLAB Codes for Finite Element Analysis- Solids and Structures, Springer.
Fritz J. Jooste, 2002, Flexible Pavement Response Evaluation using the Semi-analytical Finite Element
Method, Road Materials and Pavement Design, 3(2): 211–225.
Huang, Y.H. 2003, Pavement Analysis and Design (2nd Edition), Pearson Prentice Hall, New Jersey.
Rongqiao X.U, 2006, Finite Element Method for structural analysis and MATLAB programming, China
Communications Press, Beijing. (in Chinese).
Siddharthan, R., N. Krishnamenon, M. El-Mously, P. Sebaaly, 2002, Investigation of Tire Contact
Stress Distributions on Pavement Response, J. Transp. Eng., 128(2): 136–144.
Smith, I.M., D.V. Griffiths, L. Margetts, 2013, Programming the Finite Element Method (5th Edition),
John Wiley & Sons, Inc.
Xiao-di HU, L.F. Walubita, 2011, Modeling mechanistic responses in asphalt pavements under three-
dimensional tire-pavement contact pressure, J. Cent. South Univ. Technol. 18(1): 250−258.
Xuejun Deng, Xiaoming Huang, 2007, Principles and design methods of pavement (2nd Edition), China
Communications Press, Beijing.(in Chinese).
Yuqing Zhang, Rong Luo, Robert L. Lytton, 2012, Anisotropic viscoelastic properties of undamaged
asphalt mixtures, Journal of Transportation Engineering, 138(1): 75–89.
Ze-jiao Dong, Mei-li Liu, Hao Zheng, Xiang-bing Gong, 2012, Dynamic Mechanical Analysis of
Asphalt Pavement Based on Cross-isotropic Properties, China Journal of Highway and Transport,
25(5): 18–23. (in Chinese).
Zienkiewicz, O.C., R.L. Taylor, 2005, The Finite Element Method for Solid and Structural Mechanics
(6th Edition), Elsevier Butterworth-Heinemann, Oxford.
1709
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Yixiang Cao
Guangdong Provincial Academy of Building Research Group Co. Ltd., China
Peng Cao
Department of Hydraulic Engineering, Tsinghua University, China
Ruxin Jing
Delft University of Technology, Delft, The Netherlands
ABSTRACT: Due to cement hydration, the change rule of bearing capacity of bored pile in the
refreezing process is a main theoretical foundation for the consequent superstructure construc-
tion. In order to accurately calculate the bearing capacity of pile in refreezing anaphase and telo-
phase, the basic differential equation of load transfer function is adopted. The hyperbolic model
and bilinear model were employed in a computation program to calculate the soil movements
surrounding and at the end of the pile, respectively. Then the friction force, resistance of pile axis
force, and the single bearing capacity of the end of pile in the refreezing anaphase and telophase
were obtained. The results were validated by test data in Kunlun permafrost regions. It is also
concluded that the bearing capability of the bored piles in the Kunlun area can provide enough
supports to superstructures construction during the refreezing metaphase and telophase.
1 INTRODUCTION
The total area of cold regions in China is 417.4 km2, and most locate in Qinghai-Tibet Plateau,
Xing’an Mountains and the western mountain area (Zhang et al. 2010). Due to the western
development program and One Belt and One Road of Chinese government, infrastructures in
the southwest of China are developing in these years. It is important to build infrastructures
with foundations with enough bearing capacity. Bored pile foundation usually has high bear-
ing capacity, strong frost resistance. And it is the most commonly used in seasonal frozen and
permafrost areas. During the construction process of bored pile foundation, it bring extra heat
into underground frozen soil, including the heat released from cement hydration. The thermal
disturbance would influence the stability of frozen soil and decrease the bearing capacity of
bored pile foundation. When the thermal disturbance is balanced, the bearing capacity of
bored pile foundation will be recoved, as shown in Figure 1.Where, qs is the skin friction, pb is
the tip resistance of bored pile, ζ and η present the increases of skin friction and tip resistance
of bored pile, respectively. The fluctuations of ground temperature in permafrost region leads
to the variation of bearing capacity of cast-in-place bored pile frame.
The development of bearing capacity of bored pile foundation in the permafrost area is
a hot research topic in these years. Zhang et al. (2010) studied the cooling characteristics
of crushed-rock interlayer embankment with vertilated duct along a highway in permafrost
regions through a 3 dimensional nonlinear numerical model. They claimed that the crushed-
rock interlayer embankment with bentilated duct is an effective way to decrease the under-
lying ground temperature and keep the highway structure stable in permafrost regions. Xu
1711
Figure 1. The variation of bearing capacity of pile frame.
et al. (2007) used the nonlinear hyperbola soil constitutive model to simulate the load-settle-
ment relationship of frozen soil. Wang et al. (2005) divided the refreezing process into three
stages (primary, middle, and final stages). Jia et al. (2011) studied the influences of cement
hydration heat and cast temperature of fresh concrete in the refreezing process of bored piles.
Vyalov et al. (2000) studied the bearing capacities of single pile and pile groups in laboratory
and field. They proposed some formulas for single pile and pile group settlements based on
theoretical investigations. The strongly expressed rheological properties of frozen soils were
considered in the proposed formulas by using the strain-hardening creep theory.
The thermal disturbance in the pile construction influences the stability of frozen soil and
makes the soil refreezing a long process. Although special construction technology and mate-
rial have been adopted, the refreezing time of the soil around pile lasts 30∼60 days (Hideki
1990). It is not realistic to construct superstructure until the thermal disturbance totally dis-
appears, which will influence the construction schedule, budget and progress. Wu et al. (Cao
2009) proved that as long as the bearing capacity of bored foundation pile reached the design
load requirement, the subsequent construction process could be carried out. Recently, the
bearing capacity test of pile foundation is the most direct and accurate method during con-
struction. However, it is not only expensive, but also takes a long time. Therefore, it is neces-
sary to develop an efficient and effective computational method to capture the development
of bearing capacity of bored pile foundation in permafrost area during construction.
This study aimed to propose an efficient and effective method to evaluate the bearing capacity
of single bored pile in permafrost regions. Based on the load transfer functions of skin friction
and tip resistance of bored pile, a Microsoft Visual Basic6.0 program was developed. The pro-
posed method was validated with the bearing capacity experiment in field in Kunlun Mountain.
2.2 The hyperbolic model for the relative displacement between pile and surrounding soil
Numerous experimental results indicate that the hyperbolic function model can be used to
describe the nonlinear relationship between skin friction of bored pile and soil around it
(Randolph & Wroth 1978), as shown in Eq. (1)
1712
S(z)
qs ( z ) = (1)
a bS ( z )
Where
qs(z) = the skin friction of bored pile;
S(z) = the relative displacement between pile and soil;
a = the reciprocal of the tangent of the curve at the origin, that is 1/a = tan α;
b = the reciprocal of ultimate skin friction of pile.
The calculation method of a and b can be found in previous research (Shen et al. 2003;
Jiang et al. 2010; Zhang & Zhang 2006; Wang et al. 2004).
When bored pile is under vertical load, the axial force of pile can be expressed as
⎡ S (z) a b ⎤
Pe ( z ) = 2uuP AE
EP ⎢ e − 2 ln(
l ( Se ( z )) ⎥ (2)
⎣ b b a ⎦
Assume α = 2u p AE p , then
α ⎛ b ⎞
Pe ( z ) = ( z ) a ln ⎜1 + Se ( z )⎟
bSe (z (3)
b ⎝ a ⎠
Where
α = constant related to material and the cross section size of bored pile;
up = circumference of bored pile;
A = the pile cross section area;
Ep = the elastic modulus of pile.
⎧k S (S Sbbu )
σb = ⎨ 1 b b (4)
⎩k1Sbbu + k2 (Sb − Sbu ) (Sb ≥ Sbbu )
1713
Figure 2. Bilinear model of soil body under the pile tip.
Where
σb = the bored pile tip resistance stress (kPa);
Sb = the displacement of pile tip (mm);
Sbu = the limit of the pile tip displacement (mm);
k1, k2 = the compressive stiffness coefficients of soil under pile tip before and after the pile
displacement reaches Sbu (kPa/mm), respectively.
The parameters Sbu, K1 and K2 can be obtained through field tests. In this study, they were
selected according to references (CCCC First Highway Consultants Co., LTD. 2004).
S Se ( z ) Sb (5)
Assume A is the area of pile tip, Pb is the pile tip resistance force, then,
Pb Aσ b (6)
The inner force (P(z)) on an arbitrary cross section of bored pile consists of two parts,
the axial load causing the vertical displacement Sb, named Pb, and the axial load causing the
elastic compression Se(z), named Pe(z). Then,
P Pb + Pe ( z ) (7)
Substituting Eq. (3) and Eq. (6) into the Eq. (7), the following equation can be obtained:
α b
P ( z ) = Aσ b + bSe ( z ) − a ln(
l ( Se ( z )) (8)
b a
On the top of the pile, z = 0, Se(0) = Se0, where Se0 is the total elastic compression for the
whole pile. Therefore, the relationship between load and settlement, i.e., P-S relationship can
be expressed as:
α b
P0 A b + bSe 0 − a l (1+
1 ⋅ Se 0 ) (9)
b a
1714
Figure 3. The fundamental parameter calculation procedure.
1715
For a group of given Sb, Se0, a group of S0 and P0 can be obtained, then plots P-S curve of
pile by these data, finally the pile ultimate load can be obtained. When the bored pile is sur-
rounded by different soil layers, the relationship between load and settlement is:
n
1 b
P0 = Aσ b + α ∑ bsi Sei − asi l (1+
1 sii ⋅ Sei ) (10)
b
i =1 si asi
where, asi,bsi are the parameters in load transfer functions for the i-layer soil related to charac-
teristics of the soil; n is the number of layers of different soils the pile goes through.
5 CASE STUDY
In order to verify the results of load transfer method, the bearing capacity of bored piles in
the permafrost region of Qinghai-Tibetan Plateau was calculated and compared with static
load test data on bored piles (CCCC First Highway Consultants Co., LTD. 2004).
Table 1. The skin friction of bored pile with depth of 7.3–8.9 m in refreezing metaphase.
1716
Table 2. The skin friction of bored pile with depth of 7.3–8.9 m in refreezing metaphase.
Table 3. Calculated and measured pile tip resistance values (in refreezing metaphase).
Displacement (mm) measured values numerical solution The relative error (%)
0 0 0 0
0.49 1235 1245.77 0.8
1.77 2121 2120.36 0.4
2.65 2553 2567.82 0.5
4.87 3595 3696.64 2.7
1717
1. Skin friction of bored pile in refreezing metaphase
From Table 4 and Table 5, it can be observed that the maximum skin friction locates at the
depth among 13.5 m to 15 m. When the relative displacement of pile to soil was 4.02 mm, the
maximum measured skin friction was 243.10 kPa while the maximum calculated skin friction
was 219.90 kPa. The calculated and measured skin friction of the bored pile was observed
close to each other with no more than 10% error. Also the surrounding soil in deep location
took a larger part of skin friction than the soil in shallow location. The hyperbolic model
is appropriate to describe the pile skin friction and displacement relationship in refreezing
telophase.
2. Pile tip resistance in refreezing telophase
The pile tip resistance gradually increases with the increasing of load. The calculated and
measured tip resistance of pile were 3900 kN and 4012 kN, respectively, when the displace-
ment of pile tip was 4.21 mm. It can be observed that the calculated and measured pile tip
resistance values are similar in Table 6.
Table 4. The displacement and skin friction of bored pile in refreezing telophase (soil layer depth
7.3 m ∼8.9 m).
Skin friction(kPa)
The relative
Depth (m) Displacement (mm) The measured values numerical solution error (%)
7.30 0 0 0 0
8.11 1.53 63.02 62.38 1.0
8.45 2.55 79.30 81.41 2.5
8.68 3.15 92.28 89.18 3.3
8.90 3.60 98.47 93.95 4.5
Table 5. The displacement and skin friction of bored pile in refreezing telophase (soil layer depth
13.5 m ∼15 m).
Skin friction(kPa)
The relative
Depth (m) Displacement (mm) The measured values numerical solution error (%)
13.50 0 0 0 0
14.25 1.37 86.42 85.86 0.6
14.46 2.13 111.56 112.36 0.7
14.90 4.02 243.1 219.90 9.5
15.00 4.31 170.70 156.36 8.4
Table 6. Comparison on measured and numerical bearing capacities of the pile bottom in different
depths (in refreezing telophase).
0.00 0 0 0
0.37 1084 1091.64 0.6
1.66 2446 2395.70 2.0
2.89 3139 3121.47 0.6
3.31 3413 3369.32 1.2
4.21 4012 3900.38 2.7
1718
Figure 6. P-s curve of cast-in-place bored pile in the refreezing telophase.
6 CONCLUSIONS
Based on Microsoft Visual Basic 6.0, this paper developed a program to calculate the skin
friction, tip resistance, and bearing capacity of bored pile. The calculation results were com-
pared with static test data in field. Conclusions are shown below:
1. The hyperbolic model and the bilinear model can describe the change of bearing capacity
of bored pile in refreezing process appropriately.
2. The skin friction and tip resistance of bored pile increases with the decreasing of the tem-
perature of surrounding soil. The bearing capacity of bored pile increases in the refreezing
process.
3. The calculation program gives the calculated bearing capacity of single pile close to meas-
ured values with no more than 10% error.
4. In the environment of the Qinghai Tibet Plateau permafrost area, the bearing capacity of
bored tip in the refreezing metaphase is enough to support subsequent construction.
REFERENCES
Cao Wei. 2009. An improved load transfer hyperbolic model for pile-soil interface and its application.
Chinese Journal of Rock Mechanics and Engineering 01:144–151.
Delpak R., Omer J.R. & Robinson R.B. 2000. Load/settlement prediction for large-diameter bored piles
in mercia mudstone. Proceedings of the ICE-Geotechnical Engineering 143(4):201–224.
Hideki H. 1990. Load-Settlement Analysis for Bored Piles Using Hyperbolic Transfer Functions.Solis
and foundations. 30(1):55–64.
Jia Y., Tian H. & Guo H. 2011. Refrozen process of cast-in-place piles considering the influence of
molding temperature and hydration heat. Engineering Mechanics. 28(S1):44–47.
1719
Jiang J., Zhang Y. & Gao G. 2010. Study on Bearing Behaviors of Pile Tip Resistance of Super-Length
Piles. Engineering Mechanics 23(2):149–160.
Lai Y.M., Xu X.T. & Dong Y.H. 2013. Present situation and prospect of mechanical research on frozen
soils in China.Cold Regions Science and Technology 87(1): 6–18.
Randolph M.F. & Wroth C.P. 1978. Analysis of deformation of vertically loaded piles. Journal of the
Geotechnical Engineering Division 104(12): 1465–1488.
Shen W., Chi Y. & Song E. 2003. Settlement calculation of composite ground with rigid piles including
pile-soil-cushion interaction. Engineering Mechanics 23(2): 36–42.
The research team on pile foundation of bridge and culvert in permafrost areas in CCCC First Highway
Consultants Co., LTD. 2004. Experiment report of bridge test piles in permafrost areas. Xian: 1–70.
Vyalov S.S., Slepak M.E. & Lunev M.V. 2000. Single Pile and Pile Group in Permafrost. Journal of Cold
Region Engineering34 (1): 44–53.
Wang X., Jiang D. & Zhao X. 2004. An experimental study on refreezing characteristics of large-diam-
eter bored pile in different permafrost areas of the qinghai—tibet plateau. Chinese Journal of Rock
Mechanics and Engineering 23(24): 4206–4211.
Wang X., Jiang D. & Zhao X. 2005. Experimental study on bearing features of bored pile under non-re-
freezing condition in permafrost region. Chinese Journal of Geotechnical Engineering 27(1): 81–84.
Xu C., Xu X. & Qiu M. 2007. Numerical analysis of adfreezing force of engineering pile in permafrost.
Journal of Harbin Institute of Technology 39(4): 542–545.
Yang G. 2006. Introduction to elasticity. Beijing: Tsinghua University Press.
Zhang G., Zhang J. 2006. Elastoplastic damage model of soil-structure interface in single pile-soil inter-
action analysis. Engineering Mechanics 23(2): 72–77.
Zhang M., Lai Y. & Dong Y. 2010. Three-Dimensional Nonlinear Analysis for the Cooling Charac-
teristics of Crushed-Rock Interlayer Embankment with Ventilated Duct along the Qinghai-Tibet
Expressway in Permafrost Regions. Journal of Cold Regions Engineering 34(4): 126–141.
1720
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Xingyi Zhu, Lihan Li, Daquan Sun, Ying Yuan & Huanran Wang
The Key Laboratory of Road and Traffic Engineering, Ministry of Education, Tongji University,
Shanghai, P.R. China
Department of Mechanics and Engineering Science, Ningbo University, Ningbo, Zhejiang,
P.R. China
ABSTRACT: Cracking due to tensile stress is a typical damage pattern, which plays a cru-
cial role in the pavement durability. Cracking propagation is always accompanied by the
evolution of microstructure of asphalt mixtures, most of time they are in a coupled system.
Here, the indirect tensile (IDT) test, numerical simulation, and together with image process-
ing technique were conducted to record the variation trajectory of microstructure and stress
distribution. Further, a relationship between microstructure evolution and crack propagation
was established. Based on the variation of defects area and CT values with loading time, it
is shown that the mixture experiences a compaction stage before cracking occurred. Besides,
most vertical direction cracks initiate at the interface due to the rotation of aggregate. Some
aggregates are even fractured because of the squeezing force when rotation was restrained.
Therefore, interface strength and aggregate strength have a great influence on cracking prop-
agation behavior of asphalt mixture.
1 INTRODUCTION
Tensile damage is one of the most common damage in asphalt mixture, and is greatly affected
by its constitutions. Understanding tensile damage mechanism of asphalt mixtures is crucial
for developing a mechanistic-based design method for asphalt pavement. Asphalt mixture is
a kind of composite material which consists of asphalt binder, aggregates, filler and defects
(including voids and cracks). Each component plays an important role in the mechanical
performance of asphalt mixture. Besides, different microstructures will be formed based on
the different constitutions and blending ratios, resulting in the different capabilities to resist
the external loadings. It is well documented that crack initiation and propagation are strongly
affected by several characteristics of the microstructure, such as interlocking skeleton struc-
ture, aggregate orientation, etc. It is thus necessary to investigate the evolution of microstruc-
ture during the damage process so as to correlate the relevant microstructural parameters
with the resistance to crack initiation under tensile force.
Many test methods have been applied to investigate the tensile damage behavior in asphalt
mixture. Indirect tensile test (IDT) of asphalt mixture is a common one with the advantage
of load control and reflecting the real condition. Due to the limitation imposed by giving the
distribution of mechanical response in IDT, various researchers have proposed numerical
approaches, particularly Finite Element Method (FEM), to simulate the damage process of
asphalt mixture in IDT. Due to the viscoelastic properties of asphalt binder, different damage
mechanism i.e. cracking and plastic deformation may exist at intermediate temperature. The
Cohesive Zone Models (de Souza FV et al., 2004, Kim and Aragão, 2013, Souza et al., 2012)
and Extended Finite Element Models (XFEM) (Hainian Wang et al., 2013, Wang et al., 2014,
Ng and Dai, 2011) appear to provide a valuable tool to simulate cracking or facture, yet the
convergence difficulty is usually encountered when the crack path is not pre-defined or the
1721
shape function is used. As for the plastic deformation, limited researches have been con-
ducted to simulate it due to the difficulty of obtaining plastic constitutive model of asphalt
mixture. Therefore, in this case, the viscoelastic model obtained by the creep test was applied
to simulate the micro behaviors of asphalt mixture in IDT, and the simulation results were
used to analyze the crack propagation.
In addition, digital image processing (DIP) techniques have also been widely developed
and applied in this issue. This method is mainly for characterizing microstructure and
providing geometry information to numerical simulations. Through different advanced
DIP analysis, the gradation, distribution, shape and angular orientation of the aggregates
skeleton were estimated (Yue et al., 1995); the distribution and interaction of aggregates,
asphalt mastics and air voids were investigated (Partl et al., 2007, E. Masad et al., 2002,
Sefidmazgi et al., 2013); the fraction of air void phase from a RGB (Red-Grey-Blue)
scale asphalt mixture was calculated (Kose et al., 2000); the voids fraction was measured,
which consists of the air voids fraction, the effective volume fraction and the volume
fraction of aggregates smaller than 75 μm (Moon and Falchetto, 2015). Also, DIP tech-
niques provide more realistic and accurate image of asphalt mixture’s microstructure to
numerical simulations in a follow-up study. However, note that the previous studies have
paid little attention on rotation angle of aggregates. Therefore, in this study, the DIP
technique was applied in calculating the rotation angle of aggregates and generating a
numerical model.
As a rule of thumb, damage in asphalt mixture is a result of changes in the microstructure
of the asphalt mixture due to the application of external load, and these changes in micro-
structure can be generally described as the variation in the shape, size and spatial distribution
of the voids, the angularity, texture and form of the aggregate particles (Onifade et al., 2015,
Chandan et al., 2004). Most of them are of great interest when the tensile damage behavior
of asphalt mixture is studied. In view of the importance of defects acting as stress raisers,
rotations as a part of displacement besides translation, the advantage of numerical simula-
tions, with the aim of investigating the change of defect areas and rotation angles of aggre-
gates as well as crack propagation under tensile force, computed tomography (CT) coupled
with DIP techniques were firstly applied to measure the defect areas and the rotation angles
of aggregates, then the finite element method (FEM) was utilized to obtain stress, strain and
displacement, and finally the relationship between rotation angles of aggregate and stress
distribution was discussed.
The 70#virgin asphalt binder produced by China National Offshore Oil Corp was used
in this study. The basic material parameters, i.e. penetration, softening point and ductil-
ity, of the asphalt binder are listed in Table 1. The limestone aggregate was used and its
physical properties are illustrated in Table 2. Continuously graded asphalt mixture was
employed in this test, see Figure 1. The optimum design asphalt binder content is 4.08%.
All the mechanical properties of asphalt mixture can meet the specification of the pave-
ment performance.
1722
Table 1. Specifications of the 70#asphalt binder.
Particle size/mm
Item
0–2.36 2.36–4.75 4.75–9.5 9.5–19
1723
Figure 2. Indirect tensile test.
Δθt = i = θt = i − θt = 0 (1)
Where Δθt = i is the variation of rotation angle at a specific time compared to the one at initial
time. θt i is the azimuth angle at a specific time i and i equals to 20 s, 40 s, 60 s, 70 s and 80 s,
respectively. θt =0 is the azimuth angle at initial time. It should be noted that the negative value
represents counter-clockwise rotation, while positive value represents clockwise rotation. The
maximum rotation angle of each aggregate was calculated though Eq. (2),
Δθ max
= θm − θn (2)
1724
Figure 4. The processing for an aggregate.
1725
Figure 6. The fitted function of creep curve.
Therefore, a two voigt with one spring model (see Figure 7) was chosen as a constitutive
model of the mastic. Based on the viscoelastic theory, the differential constitutive equation
and the time-dependent relaxation modulus in complex number field of this model can be
described as Eq. (3) and Eq. (4) respectively. The mechanical parameters including E0, E1,
E2, η1, η2 can be obtained by the fitted function. Consequently, the parameters including p1,
p2, q0, q1, q2 and Y (S ) can be obtained. Here the Laplace transform method was applied to
simplify calculation with the advantage of transforming differential equations to algebraic
ones.
σ dσ / dt,σ d 2σ / dt 2
ε d ε / dt, ε d 2ε / dt 2
E η + E2η2 + E0η1 + E0η2 η1η2 E0 E1E2
p1 = 1 1 ,p2 = ,q0 =
E1E2 + E0 E1 + E0 E2 E0 E1 + E0 E2 + E1E2 E0 E1 + E0 E2 E1E2
E0 E2η2 + E0 E2η1 E0η1η2
q1 = ,q2 =
E0 E1 + E0 E2 + E1E2 E1E2 + E0 E1 + E0 E2
Q( s ) q + q1s + q 2 s 2
Y (s) = = 0 2
(4)
sP ( s ) s( 1s 2s )
1726
Where s is the Laplace transform parameter. Y is relaxation modulus in Laplace space. Q
and P are differential operators in Laplace space.
Then, the Laplace adverse transform method is applied to calculate relaxation modulus in
real number field which is required for numerical simulations. The calculation is shown in Eq.
(5) and the relaxation modulus is given by Eq. (6).
Y (t ) = L 1[Y ( s )]
⎡ ⎤ ⎡ ⎤
⎢ ⎥ ⎢ ⎥
⎛q ⎞ ⎢ M ⎥ −1 ⎢
N ⎥
= L−1 ⎜ 0 ⎟ + L−1 ⎢ ⎥ + L (5)
⎝ s ⎠ ⎛ ⎞ ⎢ − − p12 − 4p2 ⎥
⎢ ⎜ s − −p1 + p1 − 4p2 ⎟ ⎥ p
2
⎢s − 1 ⎥
⎢⎜ 2p 2 ⎟ ⎥ ⎢⎣ 2p 2 ⎥⎦
⎣⎝ ⎠⎦
⎛ −p + 12 − 4p2 ⎞ ⎛ −p − p12 − 4p2 ⎞
= q 0 + M exp ⎜ 1 t ⎟ + N exp ⎜ 1 ⎟
⎜ 2p 2 ⎟ ⎜ 2p 2 ⎟
⎝ ⎠ ⎝ ⎠
q pq 2p q p p q − p1q 2 q p q 2p q p p q − p1q 2
M= + 2 0+ 2 1 1 2 0 ,N = 2 − 2 0 − 2 1 1 2 0
2 2 2
2 p1 4p2 2 2 2 p12 4p2
Mastic
Aggregates
Elastic Elastic
Parameters modulus(MPa) y1 τ1 y2 τ2 modulus(GPa)
1727
Figure 8. Finite element model of asphalt mixture.
1728
Figure 9. Surface topography at different time.
and No. 13 exceed 8°, while No. 5, No. 11 and No. 16 were in the range of 7° to 8°. Generally,
the internal stress is the main reason for rotation. When an aggregate is subjected to a large
stress, it is more likely to move or rotate. Meanwhile, the cracking is expected when the tensile
stress or deformation is large enough. For example, the aggregates, No.12 and No.13, showed
large rotation due to the large tensile stress which was almost equal to 0.11 MPa according to
1729
Table 4. The defect information at different time.
Rotation angle
Aggregate Maximum
number 0s 20 s 40 s 60 s 70 s 80 s rotation angle
1730
Figure 12. Comparison between the test and numerical result.
Figure 13. The minor principal stress and horizontal displacement of the asphalt mixture.
the numerical simulation results. Besides, smaller rotation was also observed at No. 5, No.11
and No. 16, because the the tensile stress was relatively smaller. It is interesting to note that, the
cracking occurred at the inside of the aggregates, i.e. No. 7, No. 9 and No. 14, even the rotation
angle was smaller than 4°. This phenomenon indicates that the strength of aggregate is another
important potential factor to the damage of asphalt mixture when there is no more space for
aggregate to rotate. The cracking at the insides of the aggregate can be expected due to the
squeezing force between each aggregate, when the rotation deformation is restrained.
5 CONCLUSIONS
Image processing analysis and the numerical simulation were employed to investigate the
damage evolution and the cracking propagation of asphalt mixture during the indirect tensile
test. The simulation result is consistent with the damage observed during the test. The follow-
ing are the main conclusions obtained from the research:
1. The variation of defect area and CT values indicated that the asphalt mixture showed a
compaction stage at the beginning of the indirect tensile test before the cracking occurred.
2. The internal stress distribution and the strength of aggregate are two significant factors
to the damage of asphalt mixture. The aggregate subjected to larger internal stress is more
likely to move or rotate, resulting in the cracking at the interface of aggregate and asphalt.
If the rotation deformation is restrained, the cracking at the insides of the aggregate can
be expected due to the squeezing force between each aggregate.
3. The cracks generally occurred at the middle of asphalt mixture, and the direction of the
cracks is about parallel to the vertical axis.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the financial support provided by the Innovation Pro-
gram of Shanghai Municipal Education Commission (No. 15ZZ017), the National Natural
Science Foundation of China (Nos. 11102104 and51378393), and a grant from the Mechan-
ics Initiative, a COE Program of Zhejiang Provincial Government in the Department of
Mechanics and Engineering Science, Ningbo University (Grant No. xklx1511).
REFERENCES
Chandan, C., Sivakumar, K., Masad, E. & Fletcher, T. 2004. Application of Imaging Techniques to
Geometry Analysis of Aggregate Particles. J. Comput. Civ. Eng., 18, 75–82.
De Souza FV, Soares JB, Allen DH & Evangelista F 2004. Model for predicting damage evolution in
heterogeneous viscoelastic asphaltic mixtures. Transport Res Rec, 1891, 131–9.
1732
E. Masad, V.K. Jandhyala, N. Dasgupta, N. Somadevan & Shashidhar, N. 2002. Characterization of air
void distribution in asphalt mixes using X-ray computed tomography. Journal of Materials in Civil
Engineering Fracture Mechanics, 14, 122–129.
Hainian Wang, Chen Zhang, Lian Yang & You, Z. 2013. Study on the rubber-modified asphalt
mixtures’cracking propagation using the extended finite element method. Construction and Building
Materials, 47, 223–230.
Kim, Y.R. & Arag O, F.T.S. 2013. Microstructure modeling of rate-dependent fracture behavior in bitu-
minous paving mixtures. Finite Elements in Analysis and Design, 63, 23–32.
Kose, S., Guler, M., Bahia, H. & Masad, E. 2000. Distribution of strains within binders in HMA using
imaging and finite element techniques. Transp. Res. Rec., 1728, 21–27.
Moon, K.H. & Falchetto, A.C. 2015. Microstructural investigation of Hot Mix Asphalt (HMA) mix-
tures using Digital Image Processing (DIP). KSCE Journal of Civil Engineering, 19, 1727–1737.
Ng, K. & Dai, Q. 2011. Investigation of Fracture Behavior of Heterogeneous Infrastructure Materials
with Extended-Finite-Element Method and Image Analysis. Journal of Materials in Civil Engineer-
ing, 23, 1662–1671.
Onifade, I., Birgisson, B. & Balieu, R. 2015. Energy-based damage and fracture framework for viscoe-
lastic asphalt concrete. Engineering Fracture Mechanics, 145, 67–85.
Partl, M.N., Flisch, A. & Jonsson, M. 2007. Comparison of laboratory compaction methods using x-ray
computer tomography. Road Materials and Pavement Design, 8, 139–164.
Sefidmazgi, N.R., Teymourpour, P. & Bahia, H.U. 2013. Effect of particle mobility on aggregate forma-
tion in asphalt mixtures. Road Materials and Pavement Design, 14, 16–34.
Souza, L.T., Yong-Rak Kim, A.M., Souza, F.V. & Castro, L.S. 2012. Experimental Testing and Finite-
Element Modeling to Evaluate the Effects of Aggregate Angularity on Bituminous Mixture Perform-
ance. Journal of Material in Civil Engineering, 24, 249–258.
Wang, H., Wang, J. & Chen, J. 2014. Micromechanical analysis of asphalt mixture fracture with adhe-
sive and cohesive failure. Engineering Fracture Mechanics, 132, 104–119.
Yue, Z., Bekking, W. & Morin, I. 1995. Application of digital image processing to quantitative study of
asphalt concrete microstructure. Transportation Research Record, 1492, 53–60.
1733
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Joao Castro-Gomes
University of Beira Interior, Portugal
Marco Gabriel
Beira Serra, Associação de Desenvolvimento Local, Portugal
ABSTRACT: Based on the initial results obtained in the research program REMINE
(H2020 RISE-Marie Curie Action) in progress, authors discuss the potential use of Pan-
asqueira mine waste aggregates and fillers in the production of construction materials for
transportation infrastructures. An estimate of the economic and social impact that this form
of recycling could have on the local communities and on the perspective of other mining
activities in Portugal and Europe is given. The main goals of the project are to explore mate-
rials design methods of alkali-activated composites from mining/quarrying wastes based on
the demanding requirement of rheology to fit for various processing techniques and appli-
cations. The development of artificial aggregates through alkali-activation of mining waste
presents itself as a viable technical solution to compete with other commonly adopted mate-
rials and may lead to the manufacturing of less porous and harder aggregates for the produc-
tion of most of the road paving materials.
1735
pile and two mud dams now exist at this site. One of the dams is still being fed with material
without tungsten ore (mud, and coarse sterile material) obtained from the ore dressing opera-
tions (see Figure 1) (Ávila et al., 2008).
Accordingly to the initial records, during the first 50 years of mining over 27 million tons
of rock were extracted, from which approximately a total of 92,800 tons of tungsten con-
centrate, 4,800 tons of tin concentrate, and 28,600 tons of copper concentrates were pro-
duced (Smith, 2006). Currently the production quantity is variable in the range of 85,000
to 95,000 tons of tungsten concentrate per year. The plant produces a high quality tungsten
concentrate with 74% to 75% WO3, which is one of the highest grades available on the market
(Franco et al., 2014).
Figure 1. Left: Rio plant and old tailings piles. Right: Barroca Grande plant.
1736
clear that oxidation of sulphides tailings and flow from open impoundments are the main
source of pollution in the surrounding area and are responsible for the mobilization and
migration of metals from the mine wastes into the environment (Ávila et al., 2008); (Godinho
et al., 2010). Also, about 30% of arsenopyrite (the main present sulphide) is rejected with the
tailings (Ávila et al., 2008).
Thus, the accumulation of waste from Panasqueira mining activities, over the years, has
lead to the formation of massive deposits. These deposits present today a potential risk of
environmental pollution and cause serious landscape impacts, thus also affecting the quality
of life of local populations.
1737
Table 2. Geometrical, physical and mechanical properties of coarse waste.
Table 2 presents physical and mechanical properties of waste-rock materials. Particle shapes
were characterized on the basis of EN933–3 and EN933–4 standards for concrete aggregates.
Coarse wastes are constituted by a major percentage of greywacke schist and about 10% of
white quartz. Particles maximum and minimum dimensions were determined according to
EN933–1 and EN12620. Mechanical and physical properties were determined by EN1097–2,
EN1097–3, EN1097–6 and BS812–110.
From the test results, it is evident that waste coarse particles have good and consistent
properties, equivalent to those found in good quality granite aggregates (Pereira et al., 2009).
Waste-rock tailings in Barroca Grande plant have the size like coarse aggregates (on average,
diameter size of 5 mm and 25 mm) and mud containing very fine particles (diameter size of
less than 2 mm) (Antunes, 2009) (Castro-Gomes et al., 2012).
1739
Figure 2. Potential use of tailings in road construction (left). Example of coarse tailings reuse (right).
Table 3. Estimate of tailings consumption for the production of 3 layered bituminous pavement.
cost-effective use of materials: crushing, sieving, mixing with other aggregates sources should
lead to the development of industrial set-up and procedures where impacts and wastes are
minimized.
Environmental risks and contaminations should be evaluated according to the regulations
in force. CE marking of aggregates and aggregates mixtures should be done in accordance
with the European standards for unbound and bituminous bound products.
1741
Table 4. Estimate of material volumes for the production of single pre-cast segments.
Table 4 shows a rough estimate of the volume quantities of alkali-activated and coarse
aggregates admixtures needed for the production of some different pre-cast segments. Their
production in Panasqueira site could make use of the traditional pre-casting processes of con-
crete elements, taking into account that alkali-activated production and curing are different.
Product properties should undergo standard testing procedures for certification and CE
marking, this is particularly relevant for road safety rigid barriers that could be moulded
(after commercial agreements) according to existing and certified products. Constituent
material changes, in fact, will only require computer aided virtual crash testing of the barrier,
avoiding major costs of real scale crash testing.
It is evident that larger pre-cast segments such as barriers and box culverts will use larger
volumes of alkali-activated tailings from Panasqueira piles and lagoons.
5 CONCLUSIONS
The mine of Panasqueira in the Beira Interior region has been active for over 125 years.
Tungsten is the main extracted ore and the mining activity future depends on its world com-
mercial value. The mine also produces huge amounts of waste tailings every day that are piled
in dams and pumped in mud lagoons. These waste materials could be seen as the source of
employment for the local people in case the mining activity is stopped or reduced. Conver-
sion to the production of construction materials is a viable alternative also for the recycling
of the million of tons of waste rocks.
The use of coarse tailings for road construction has been proven feasible provided that
the installation of a stable asphalt concrete plant and aggregates crushing and sieving is
implemented in the mine areas. Cold, warm and hot bituminous mixtures could be designed,
1742
produced and transported within a profitable market range in the districts of Covilha,
Fundao, Belmonte, Castelo Branco and Coimbra.
A rough calculation estimates that more than 3,000 tons of tailings could be used for the
construction of just 1 km of two lane rural road, only in the asphalt concrete layers. This
could rise to more than double if tailings are used also for foundation layers.
Another form of cost-effective recycling makes use of mud tailings for the production of
alkali-activated binders and mortars. Tailings have been proven to be excellent precursors in
the alkali-activation processes that could be a viable alternative to the use of OPC that, on the
contrary, is not fully compatible with the Panasqueira schists and shales.
The fabrication of artificial aggregates by alkali-activation was attempted and gave posi-
tive results suggesting that part of the muds can be recycled into engineered particles having
special features such as water treatment capabilities, enhanced strength and shape as well as
highly textured surfaces for improved skid-resistance bituminous pavements. Larger artificial
aggregates could be seen as paving blocks for low traffic pavements and footpath surfaces.
Finally, large volumes of waste rock tailings could be recycled in the industrial production
of pre-cast segments for civil engineering construction purposes, in particular for transporta-
tion infrastructures. Alkali-activation of fines from mud and other precursors could form
the binding paste for concrete-like elements such as box-culverts, gutters, channels, curbs
and especially safety barriers, containing also coarse tailing aggregates. The correct design of
these admixtures could lead to the production of segments with enhanced characteristics if
compared to the cement concrete ones.
All these aspects are being studied within the on-going REMINE research program
(H2020 RISE-Marie Curie Action) that aims to the reuse of mining waste into innovative
geopolymeric-based structural panels, precast, ready mixes and in-situ applications.
REFERENCES
Antunes, C.S.R., 2009. Development of polymeric-based revetment tiles reusing Panasqueira waste-
rock coarse particles, Master dissertation, University of Beira Interior, Covilhã, July (only available
in Portuguese).
Ávila, P.F., Silva, E.F., Salgueiro, A.R. & Farinha, J.A. 2008. Geochemistry and mineralogy of mill tail-
ings impoundments from the Panasqueira mine (Portugal): implications for the surrounding environ-
ment. Mine Water and the Environment 27 (4), 210–224.
Bakharev, T., Sanjayan J.G., Cheng Y.B. 1999. Effect of elevated temperature curing on properties of
alkali-activated slag concrete. Cem Concr Res. 29: 1619–25.
Castro-Gomes, J.P., Silva, A.P., Cano, R.P., Durán Suarez, J. & Albuquerque, A. 2012. Potential for
reuse of tungsten mining waste-rock in technical-artistic value added products. Journal of Cleaner
Production, 25, 34–41.
Cavey, G. & Gunning, D., 2006. Updated Technical Report on the Panasqueira mine, Distrito de Cas-
telo Branco, Portugal, ORE-QUEST, p. 67.
Dinis de Almeida, M., Castro Gomes, J.P. & Pereira de Oliveira, L.A.. 2006. Valorização de resíduos de
minas em pavimentos de baixo custo, Revista Valorização de Resíduos, Ano 3, Número 9, Centro de
Valorização de Resíduos.
Dinis-Almeida, M., Castro-Gomes, J., Sangiorgi, C., Zoorob, S.E., Afonso, M.L. 2016. “Performance
of Warm Mix Recycled Asphalt containing up to 100% RAP”. Construction and Building Materials,
112:1–6.
Dondi G., Sangiorgi C., Lantieri C., Simone A., Vignali V. & Lamperti R. 2014a. Performance eval-
uation of Construction and Demolition and other waste materials. 3rd International Conference
on Transportation Infrastructures—ICTI 2014, Pisa, April 22–25. Tay&Fran Group, ISBN 978–1-
138–00147–3.
Dondi, G., Mazzotta, F., Sangiorgi, C., Pettinari, M., Simone, A., Vignali, V. & Tataranni, P. 2014b.
Influence of cement and limestone filler on the rheological properties of mastic in cold bituminous
recycled mixtures. Sustainability, Eco-efficiency and Conservation in Transportation Infrastructure
Asset Management—Losa & Papag. (Eds), pp.61–67, Tay&Fran Group. ISBN 978–1-138–00147–3.
Dondi, G., Tataranni, P., Pettinari, M., Sangiorgi, C., Simone, A. and Vignali, V. 2014c. Crumb Rubber
in Cold Recycled Bituminous Mixes: comparison between Traditional Crumb Rubber and Cryogenic
Crumb Rubber. Construction and Building Materials, Volume 68, 15 October 2014, Pages 370–375.
1743
Feret, R. 1939. Revue des Materiaux de Construction et Travaux Publics; 352:1.
Franco, A., Vieira, R. & Bunting, R. 2014. The Panasqueira Mine at a Glance. Tungsten, 3(June),
1–12.
Garcia, JIE., Fluentes, AF., Gorokhovsky, A., Fraire-Luna, PE. & Mendoza-Suarez, G. 2003. Hydra-
tion products and reactivity of blast–furnace slag activated by various alkalis. J Am Ceram Soc. 86
(12):2148–53.
García-Lodeiro, I., Palomo, A., Fernández-Jimenez, A. & Macphee, DE. 2001. Compatibility studies
between N–A–S–H and C–A–S–H gels. Study in the ternary diagram Na2O–CaO–Al2O3–SiO2–
H2O. Cem Concr Res. 41:923–31.
Godinho, B., Abreu, M.M. & Magalhaes, M.C., 2010. Avaliação biogeoquímica dos solos e do medron-
heiro na área mineira da Panasqueira. Revista de Ciências Agrárias 33 (1), 226–235. ISSN 0871–018X
(only available in Portuguese).
Juenger, MCG, Winnefeld, F, Provis, JL, Ideker, JH. 2011. Advances in alternative cementitious binders.
Cem Concr Res. 41:1232–43.
Krivenko, P. 2002. Alkaline cements: from research to application. In: Geopolymers 2002 – turn potential
into profit, Melbourne. Proceedings. Hotel Sofitel, Melbourne, Australia; p. 28–29.
Lamperti, R., Grenfell, J., Sangiorgi, C., Lantieri, C. & Airey G. 2015a. Influence of waxes on adhesion
properties of bituminous binders. Construction and Building Materials. Elsevier. Volume 76, 1 Febru-
ary 2015, Pages 404–412.
Lamperti, R., Lantieri, C., Sangiorgi, C., Bitelli, G. & Simone, A. 2015b. Semi-automatic evaluation of
the degree of bitumen coverage on bitumen-coated aggregates. RILEM Bookseries Volume 11, 26
September 2015, Pages 15–24.
Malolepszy, J. 1993. Proceedings of 3rd international symposium on cement and concrete. Beijing,
China.
Mazzotta, F., Sangiorgi, C., Vignali, V., Lantieri, C., Dondi, G. 2015. Rheological characterization of
bituminous mastics containing waste bleaching clays. RILEM Bookseries, Volume 11, 26 September
2016, Pages 595–606. ISSN: 22110844.
Pacheco-Torgal, F., Castro-Gomes, J.P. & Jalali, S. 2009. Utilization of mining wastes to produce
geopolymer binders, in: J.L. Provis, J.S.J. van Deventer (Eds.), Geopolymers—Struct. Process. Prop.
Ind. Appl., Woodhead Publishing Limited, 2009: pp. 267–293.
Palomo, A., Grutzeck, M.W. & Blanco, M.T. 1999. Alkali-acivated Fly Ashes. A Cement for Future.
Cement and Concrete Research. n. 29, pp. 1323–1329.
Pereira, C.N.G., Oliveira, L.P. & Castro-Gomes, J.P., 2009. Influence of natural coarse aggregates size,
mineralogy and water content on the permeability of structural concrete. Construction and Building
Materials 23 (2), 602–608.
Purdon, AO. 1940. The action of alkalis on blast–furnace slag. J Soc Chem Indust. 59:191–202.
Sangiorgi, C., Tataranni, P., Simone, A., Vignali, V., Lantieri, C. & Dondi, G. 2014. Waste bleaching
clays as fillers in hot bituminous mixture. Construction and Building Materials, 73, 320–325.
Sangiorgi, C., Tataranni, P., Simone, A., Vignali, V., Lantieri, C. & Dondi, G. 2016. Assessment of waste
bleaching clay as alternative filler for the production of porous asphalts. Construction and Building
Materials, 109, 1–7.
Silva, I., Castro-Gomes, J.P., & Albuquerque, A., 2012. Mineral Waste Geopolymeric Artificial Aggre-
gates as Alternative Materials for Wastewater-Treatment Processes: Study of Structural Stability and
pH Variation in Water, J. Mater. Civ. Eng. 1–6.
Simone, A., Vignali, V., Lantieri, C. 2012. A new “frugal” approach to road maintenance: 100% Recy-
cling of a deteriorated flexible pavement. 7th International Conference on Maintenance and Rehabil-
itation of Pavements and Technological Control, MAIREPAV 2012. New Zealand; 28 August 2012
through 30 August 2012; Code 110271.
Smith, M. 2006. Panasqueira the tungsten giant at 100+. Operation focus. International Mining, 10–14.
Van Jaarsveld, JGS, Van Deventer, JSJ & Lorenzen L. 1997. The potential use of geopolymeric materi-
als to immobilize toxic metals: part I. Theory and applications. Minerals Eng. 10(7):659–69.
Vargas, A.S., Dal Molin, D.C.C., Masuero, Â.B., Vilela, A.C.F., Castro-Gomes, J., & de Gutierrez, R.M.
2014. Strength development of alkali-activated fly ash produced with combined NaOH and Ca(OH)2
activators. Cement and Concrete Composites. 53, 341–349.
Vignali, V., Mazzotta, F., Sangiorgi, C., Simone, A., Lantieri, C., Dondi, G. 2014. Rheological and 3D
DEM Characterization of Potential Rutting of Cold Bituminous Mastics. Construction and Build-
ing Materials. Volume 73, 2014, Pages 339–349.
1744
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
F.G. Praticò
University Mediterranea of Reggio Calabria, Italy
D. Catalfamo
Provincia di Reggio Calabria, Italy
C. Lanciano
Professional Engineer, Reggio Calabria, Italy
ABSTRACT: The EU directive 2008/96/EC (in Italy implemented through laws n.35/2011
and n.137/ 2012) requires the establishment and implementation of procedures relating to
road safety impact assessments. Inspections should be carried out in the sections of the
road and questionnaires should be filled where 200 m-long sections are considered. Such
an approach provides insights to road departments in terms of main issues. Unfortunately,
the above procedure doesn’t bring to a quantitative approach and ranking among different
section/roads. Consequently, the objectives and scope of this study were confined into: a)
modelling the reliability of a road based on on-site investigations; b) applying the model to
several case-studies. A model was set up and applied to two rural roads in the Province of
Reggio Calabria (Italy). Results demonstrate that the method can quantify the safety level of
a road and can benefit both researchers and practitioners.
1 INTRODUCTION
Many issues and factors affect road safety, among which vehicle, driver, road, and context
can be listed.
When aiming at a society free from serious road traffic injuries, many countries and organ-
izations set up time-limited and quantified targets for the reduction of fatalities and injuries
(e.g., Organization for Economic Co-operation and Development, 2008, Wittemberg et al,
2013).
EU adopted a road safety program to cut road death in half by 2020 (European Commis-
sion, 2010; Strandroth, 2015).
In setting these targets, EU, single countries and organizations recognize the importance
of monitoring and predicting the development toward the target as well as the efficiency of
road safety policies and interventions (Strandroth, 2015).
Analysing the current status and predicting the future status of the road transport system
is crucial not only with respect to target monitoring but also in the process of operational
planning, in the prioritization of future actions, and in the choice of the most effective safety
measures.
In Italy, the following main decrees and laws apply: i) European directive 2008/96/EC
(European Parliament, 2008); ii) Italian Decree 35/2011 (Ministry of Infrastructure and
Transport, 2011); iii) Italian Decree 137/2012 (Ministry of Infrastructure and Transport,
2012), guidelines for safety management according to the article 8 of the Decree 35/2011.
The latter recalls that by the end of 2020, Italian Regions are due to issue their standards for
the safety management of road infrastructures (section 1.5.1.2). Note that the standpoint of
a road agency mainly addresses road-related safety issues.
1745
Figure 1. From 2008/96/EC to decision making.
Directive 2008/96/EC interacts with guidelines as per DM 137/2012. The latter sets up
detailed forms for carrying out road safety inspections (preliminary and overall).
Safety inspections focus on road main components (e.g., guardrails, pavement, markings,
signs, etc.).
The results of road safety inspections is the basis for the selection of safety measures (e.g.,
pavement overlay), priorities (sections to rehabilitate/maintain over time), analyses (e.g., cost-
to-benefit analysis), and decision making (see Figure 1). Under these conditions, evaluations
are critical to further underpin road safety management decisions (Wegman et al, 2015).
– Transferability of research results in road safety (external validity) deserves further
research
– High-quality road safety data and statistics are essential for road safety management
– We found clear indications that decision makers are willing to accept results from scientific
research
2 OBJECTIVES
Based on the above facts, the objectives of this study were confined into:
a. modelling the reliability of a road based on-site investigations;
b. applying the model to several case-studies.
3 MODELING
Figure 2 illustrates the main components and items (causes) that impact the indicator I
(effect), herein termed “safety level”.
This Shikawa diagram (also called fishbone diagram, or herringbone diagram) shows that
there are eight main classes (major categories) of causes which impact I, namely: i) roadway
(Lamm et al.,1999); ii) road signs (Costa et al., 2014; Lamm et al.,1999); iii) side accesses; iv)
1746
lighting; v) pavement (Praticò and Vaiana, 2012 and 2013; Praticò and Moro, 2011); vi) other
aspects (Praticò and Giunta,2011); vii) singular points; viii) critical points.
In turn, each class out of the abovementioned eight includes single factors (categories).
The method considers the following main components (and weights):
– Roadway (15)
– Signs (20). Sign-related issues include: road markings, road signs, lit road signs or road
markings. Road signs issues include speed limits and Lamm criterion I (Praticò and
Giunta, 2011).
– Un-signalised intersections (20)
– Pavement (20). This class of issues includes items which refer to the friction course and to
road expansion joints. Note that friction course-related issues include friction in terms of
the third Lamm criterion.
– Lighting (5)
– Other issues (5). This class includes safety-related issues (such as fire-fighting equipment,
water supply, fire hydrants, portable fire extinguishers, fire hose, etc.), utility locations,
overhead power lines, and Lamm criterion II.
– Black spots (5)
– Singular points (10). They include intersections, tunnels, overbridges, workzones.
– In the case of an ideal road, it will result:
– I = 15+20+20+5+20+5+5+10 = 100. In contrast in the case of an ideally unsatisfactory road,
it will result I = Σ0 = 0.
Figure 2. Main parameters which impact the indicator I (estimate of the safety level of the road).
1747
The overall algorithm is below reported (equation 1):
R ( I ) ⋅V
V E (1)
where R is the reliability of the road, V is its vulnerability, while E stands for exposure, and I
(safety level) is the score assessed through the method above described.
Note that, based on the above, 100-I represents the probability of having an accident (in
percentage). Therefore, I represents the probability of not having an accident. Consequently,
I is an indicator which refers to the “safety level” of the road: the higher the score (I) is, the
higher the safety level is, the lower the accident probability is.
Indicator I is based on the abovementioned laws and regulations. Authors are aware of the
complexity of the topic and suggest a recursive method of implementation based on continu-
ous upgrading a weight calibration (Chen et al., 2016; Aarts & Van Schagen, 2006).
I is assessed through a scoring process based on multi-criteria method and entails the
derivation of indicators for 20 sections, each one 200 m-long. Each section corresponds to a
given value of s, while each sub-indicator corresponds to a given i.
The first step refers to the derivation of the sub-indicator Ii for each one out of the differ-
ent (20) sections of the road. Two main types of procedures are used: judgement-based and
measurement-based.
The following algorithm is used for both cases:
Ii ∑w ⋅w
s
i s cs = w i ∑w s
s ⋅ cs w i ⋅ ∑ I is
s
(2)
Where ws has a default value of 20–1, wi is the weight of the given i-th indicator, and cs
represents the score associated to the s-th section for the i-th indicator.
For example, if the sub-indicator Ii refers to the condition of the guard-rails and if it has
an overall weight of 3 (wi = 3), then having a perfect state of the guard-rail will correspond to
a running total (summation of ws cs, for the different s) of 1 and then to a corresponding con-
tribution of 3. Therefore, if the qualitative judgment of the condition is “low severity”, then
equation (3) applies, while in the case of “medium severity”, equation (4) applies. Finally, in
the case of “high severity”, equation (5) applies.
cs = (20 ⋅ ws )−1
(3)
cs = 0.5 ( ws )−1
(4)
cs = 0
(5)
In the case of measurement-based indicators, e.g., design consistency according to Lamm
criterion I, a measurement is carried out. Consequently, a given quantity is derived (in this
case Vd-V85), and, based on that, another quantity (cs) is associated to the given speed differ-
ence: the higher the speed difference is, the lower the associated quantity cs is, for the given
section.
In this second case cs is derived based on a simple curve which associates high values of cs
to low values of speed difference and vice versa:
For example, when Lamm criteria I and II, are involved, the following structure applies:
km
cs =(( ws ) 1 if 10
h (6)
km
cs =0 if 20
h (7)
km
cs =(( ws ) 1
(2-0.1⋅
2 ) if 10 < Δ < 20
h (8)
1748
Based on the given cs (which corresponds to the given section s), the following running
total is derived:
Ii ∑w ⋅w
s
i s cs (9)
For example, if ws = 20–1 and wi = 0.5, if the speed difference is always zero, than cs will be
always 1, and Ii will be 0.5.
Once the Ii are known, the overall “safety level” of the road can be derived through the
following running total:
N
I ∑I i
i (10)
cs =(( ws ) 1
if f ≥ 0.01 (12)
cs =0 if f ≤ -0.04 (13)
cs =(( ws ) 1
0 + 20 ⋅ f ) if -0.04 < f < 0.01
(0.8 (14)
When road black spots and singular points (intersection, bridge, tunnel), are involved, the
following structure applies. If just one a critical point (e.g., black spot) has been assessed in
the given section s, than xs = 1, while if the given section is not associated to a black spot, than
it is xs = 0. The coefficient cs is derived according to:
⎛ ⎞
Ii wi ⋅ ⎜1 − ws
⎝
∑ x ⎟⎠
s
s (15)
N
I ∑I i
i
(16)
1 (17)
ws =
20
Note that both R and I (equation 1) refer to a given road as per the given direction.
In the pursuit of validating the model above set up, an experimental investigation was
designed and carried out.
Two rural roads were considered, both located in Southern Italy in the province of Reggio
Calabria.
The main characteristics of the considered roads are summarised in Table 1 and Figure 3.
Figure 4 illustrates the scores (y-axis) obtained for each class of issues (e.g., Guard-rail,
x-axis), for both the roads (SP5-black bars and SP93-white bars).
It must be pointed out that each class of issues is here represented in the range 0–100%, in
the aim of better comparing the roads under investigation. For example, guard rail (or New
1749
Table 1. Main characteristics of the roads.
Figure 3. Roads selected for method validation (Province of Reggio Calabria, Italy).
Jersey) have a weight of 0.56 15 ≈ 8.4 but in Figure 4 this indicator doesn’t range from 0 to
8.4 (see Figure 2) but it is normalized and ranges from 0 to 100.
SP5 performs very well in terms of shoulders, overtaking lane, drainage, first Lamm
criterion, and markings, while SP93 performs better in terms of fence, and second Lamm
criterion.
Figure 5 represents a radar chart of the percentage scores. Each spoke represents one of
the abovementioned classes, while the data length of a spoke (e.g., vertical signs) is propor-
tional to the magnitude of the variable for the given road.
1750
The star plot illustrates a condition in which SP5 (solid line) usually performs better than
SP93 (dotted line).
Figure 6 shows the singles scores obtained for both the roads under investigation. Note
that in the classes which have the highest impact (e.g., side access, singularities, deforma-
tion, guard-rails, etc.), SP93 usually yields higher values. Importantly, SP5 guard-rails have
a condition that is better than in the case of SP93 (6 versus 8, out of a maximum of 8.4).
Furthermore, in the case of SP93, the condition of the signs is absolutely unsatisfactory and
this implies scores close to zero (see dotted boxes in Figure 6). Consequently, this entails that
the overall score of SP5 overcomes the one exhibited by SP93.
Finally, in Table 2, single scores are ranked in decreasing order for both roads.
Additionally, differences are reported (last column).
Figure 5. Single scores expressed as a percentage of the concerned weight, in decreasing order.
1751
Table 2. Ranking per single topic.
SP 5 SP 93 SP5-SP 93
It can be highlighted that, even if the first classes report a negative difference (SP93 is bet-
ter than SP5), the fourth and the seventh class present a clear inversion in terms of ranking
(in favour of SP5), with differences which are much higher than in the remaining classes.
Table above illustrates how the ranking per topic can help safety manager plan rehabilita-
tion and maintenance based on differences among the roads.
In more detail, SP 93 yields a lower score in terms of horizontal and vertical signs. This
occurrence has an outstanding impact on overall safety performance as the international
literature points out (Costa et al., 2015).
Indeed, this issue affects the concerned items and impacts the overall score of about 8
points out of 100.
Furthermore, agency engineer and safety managers are provided with a straightforward
tool to plan their activities based on method application.
5 CONCLUSIONS
The establishment and implementation of procedures relating to road safety impact assess-
ments requires inspections, high level analyses and quantitative methods, based on the most
influential variables (surface, markings, signs, etc.). The method herein set up provides a
straightforward method for assessing the reliability of a road section based on easy-to-use
algorithms and concepts.
Results demonstrate that: i) the method considers many variables and seems to have the
potential for carrying out a reliable ranking of road sections based on their characteristics; ii)
even if a road section has on average better characteristics with respect to another road sec-
tion, when high criticalities in vital characteristics such as road marking are involved, method
comes up with a ranking which highlights the complexity of road safety. Indeed, method is
a function of many variables and it doesn’t neglect that criticalities in several areas can be
1752
of outstanding importance; iii) the method can quantify the safety level of a road and can
benefit both researchers and practitioners.
REFERENCES
Aarts, L. & Van Schagen, I. (2006). Driving speed and the risk of road crashes: A review. Accident
Analysis and Prevention 38, 215–224.
Chen, F., Wu, J., Chen, X., Wang, J. & Wang, D. 2016. Benchmarking road safety performance: Identi-
fying a meaningful reference (best-in-class). Accident Analysis and Prevention 86 (2016) 76–89.
Costa, M., Simone, A., Vignali, V., Lantieri, C., Bucchi, A. & Dondi, G. 2014, Looking behavior for
vertical road signs. Transportation Research Part F: Traffic Psychology and Behaviour, Volume 23,
March 2014, Pages 147–155.
European Commission, 2010. Road Safety Programme 2011–2020: detailed measures. MEMO/10/343.
Strandroth, J. 2015. Validation of a method to evaluate future impact of road safety interventions, a
comparison between fatal passenger car crashes in Sweden 2000 and 2010. Accident Analysis & Pre-
vention, Volume 76, Pages 133–140.
European Parliament, 2008. On road infrastructure safety management. Directive 2008/96/EC.
Lamm, R., Psarianos, B., Mailaender, T., Choueiri, E.M., Heger, R. & Steyer, R. 1999. Highway Design
and Traffic Safety Engineering Handbook. McGraw-Hill, New York.
Ministry of Infrastructure and Transport (Italy), 2011. Road infrastructure Safety Management. Leg-
islative Decree n.35/2011.
Ministry of Infrastructure and Transport (Italy), 2012. Guidelines for the management of Road infra-
structure Safety Management. Ministerial Decree n.137/2012.
Praticò, F.G. & Giunta, M. 2011. Speed distribution in low volume roads: from inferences to rehabilita-
tion design criteria. Transportation Research Record Journal of Transportation Research Board, No
2203 vol. 1 pp 79–84.
Praticò, F.G. & Moro, A. 2011. In-lab and on-site measurements of hot mix asphalt density: Conver-
gence and divergence hypotheses. Constr. Build. Mater., 25(2), 1065–1071.
Praticò, F.G. & Vaiana, R. 2012. Improving infrastructure sustainability in suburban and urban areas:
is Porous asphalt the right answer? And how?. WIT Transactions on Ecology and the Environment,
pp. 673–684.
Praticò, F.G. & Vaiana, R. 2013. A study on volumetric versus surface properties of wearing courses.
Construction and Building Materials, 38, p.766–775.
Praticò, F., Giunta, M. 2011. Speed distribution on low-volume roads in Italy. Transportation Research
Record, Issue 2203, 1 December 2011, Pages 79–84.
Wittenberg, P., Sever, K., Knoth, S., Sahin, N. & Bondarenko, J. 2013. Determining quantitative road
safety targets by applying statistical prediction techniques and a multi-stage adjustment procedure,
Accident Analysis & Prevention, Volume 50, January 2013, Pages 566–577.
Wegman, F., Berg, H.-Y., Cameron, I., Thompson, C., Siegrist, S. & Weijermars, W., 2015. Evidence-
based and data-driven road safety management, IATSS Research, Volume 39, Issue 1, July 2015,
Pages 19–25.
1753
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Xiaochun Qin
Beijing Key Laboratory of Track Engineering, Beijing Jiaotong University, Beijing, China
ABSTRACT: The proper aesthetic disposal and landscape design of road sound barri-
ers would well compensate the negative visual influences on the environment and improve
road landscape environment effectively. Based on the analysis of the relationship between
visual disturbance and landscape compensation, the spatial characteristics of road sound
barriers were identified. The landscape design requirements of road sound barriers with
different spatial form and different wall parts were all studied respectively in detail in con-
sideration of the acoustic and non-acoustic functions of sound barriers. And the dynamic
visual aesthetic features of road sound barrier were also determined from the perspectives
of the relationship between speed and vision and the relationship between texture and
material. The landscape compensation design methods of road sound barrier on the basis
of visual aesthetic factors were proposed respectively from the five aspects of sequence
and rhythm, spatial enclosure, visual orientation force, environmental compatibility and
regional culture.
1 INTRODUCTION
1.1 The visual impacts and landscape compensation of road sound barrier
Road sound barriers have the strong visual influence and disturbance on the surrounding
environment and the passengers due to their large size and unnecessary interruption, and
their landscape quality also has strong impact on the aesthetic perception of the passengers.
From the viewpoint of the visual dimension the huge sound barrier will form the unnecessary
blocks and obstacles, which will have serious visual interferences in the surrounding environ-
ment. Therefore, the proper aesthetic disposal and landscape design of road sound barriers
considering the regional natural and social characteristics as well as environmental transport
noise conditions could well compensate the negative visual effect (Blum, R.F. 1976; Benz,
K. & Colin, E. 2009; Luigi, M. et al, 2013).
1.2 The acoustic function and the non-acoustic function of road sound barrier
Sound barrier is the blocking structure with enough surface density which is located
between the sound source and sound-accepted point. By blocking the transmission of
sound to form a “sound shadow”, the sound barrier can make the sound level get a sig-
nificant additional attenuation in the sound shadow than in a corresponding free area,
so as to weaken the noise in a certain area and improve the acoustic environmental qual-
ity of the surrounding environment (John A. Volpe, 1999). Road sound barrier should
not only provide the noise reduction function and meet the requirements of the noise
reduction target, but also fully take into account the non-acoustic features which need
meet the needs of the local natural landscape in the layout, form, texture, structure and
1755
exterior decoration so as to be coordinated with the surrounding natural environment and
strengthen its visual quality to become a scenery line along the road (Joo, Y.H. et al, 2014;
Like, J et al, 2016).
Therefore, the road sound barrier should be studied and designed from two aspects
of the acoustic performance and the non-acoustic performance considering the actual
condition of traffic noise and the specific condition of nature, traffic and society. The
acoustical design of road sound barrier is based on the noise reduction target to design
the materials, location, size, structure, form of sound barrier and to take noise predic-
tion for various solutions (American National Standard, 1998). The non-acoustical design
mainly from the perspective of landscape, combining the reaction by pedestrians’ vision
and perception to the surrounding environment, consider the relationship between their
own landscape and the road landscape environment. Sound barrier landscape in a wide
sense of the word refers to an integrated landscape system constituted by the sound bar-
rier and its surrounding landscape, which is a general spatial concept consist of road
sound barrier road itself, ancillary facilities, the surrounding natural environment and
human activities.
2 METHODS
1756
Table 1. Landscape requirements based on different space forms of sound barrier.
1757
Table 2. Landscape requirements based on the different forms of sound barrier wall.
Form of
sound
barrier
wall Function Characteristics Landscape requirements
Top The top of screen is usually Screen cap is extremely In the environment with
the key-part considered important for the good landscape effect,
in the acoustic function expression of visual we generally make the
design of sound barrier. landscape on sound top of the sound barrier
T, Y type, antlers type and barrier. The height of the “into” the sky or green
other folded plate types has sound barrier is generally processing, Or using
the function of noise 2.5 to 3.0 meters, while transparent screen in
reduction. the height of sight of order to minimize the
pedestrians in car is visual impact on the
usually l. 2 to 1.5 m. top of the screen.
The sound barrier is In the poor environment
upward. Thus, the vision effect or a cluttered
field of screen top itself environment, we had
and the background better use the deep
is very outstanding. appearance, clear outline
of screen, while trying to
ensure the top of the
sound barrier soft
instead of such big one
that is easy to give a
person top-heavy sense.
Screen Screen is often the most According to different The screen landscape
concentrated part of the environment, the inside design has large degrees
pedestrian and residents. screen and the lateral of freedom so that we
It is not only the main screen could use could select different
visual body of the sound different screen form based on meeting
barrier but also The main landscape design to the requirement of noise
noise reduction. be fully integrated reduction. For area with
into the surrounding good natural or cultural
environment. landscape, ensure the
visibility of screen
surface is the best way to
build landscape effect of
sound barrier. Large area
transparent screen can
really show the landscape
behind the sound barrier
and largely reduce
pedestrian visual barrier.
Meanwhile, transparent
design could ensure the
residences’ daylighting
behind the sound barrier
in reserve is not affected.
(Continued)
rier surrounding close narrow vision field doesn’t change, driver’s sense of the speed becomes
gradually slow for a long distance, which can lead to an inappropriate relaxation and fatigue.
The relationship between the driving speed, vision field and visibility distance is shown in
Table 3 and Figure 1.
1758
Table 2. (Continued).
Form of
sound
barrier
wall Function Characteristics Landscape requirements
Bottom The foundation and pillar Confined to the different General roadsound barrier
of the sound barrier play functional requirements. can make foundation
a role in supporting and screen of the sound happens to be in the
sound insulation. barrier and the bottom shadow of the wave
of the screen will beam guardrail in the
common have obvious pedestrians vision
landscape differences on through reasonable
in materials and design, design, This is a relatively
and the transition is passive approach. The
relatively blunt. especially commonly used method
concrete foundation is to plant shrubs, small
exposed. trees in front of the screen
or to introduce vines at
the bottom to soften the
hard feeling at the bottom
of the sound barrier.
Connector The connector of the sound The connector of the On the premise of not
barrier is the most impact sound barrier reducing acoustic
part on visual of achieved the change function, there are some
pedestrians. from an open space handling methods on the
to a finite space landscape aesthetic design
on the vision. about sound barrier
connector.
(l) by step gradient or
slope gradient gradually
up to design height from
the ground.
(2) By covering by the lop
or clim of green plants.
(3) Comply with the
nature terrain directly.
Table 3. The relationship between the driving speed, vision field and visibility distance.
Speed (km/h) 40 60 80 10 12
1760
Table 5. Texture of the plant material.
3 RESULTS
Figure 3. Weaken the hard edge of the sound barrier and forming sequence rhythm by softening of
planting cut down the large scale sense of barriers in the vision by forming sequence rhythm of plants
color.
1762
Figure 5. Visual traction of vertical orientation force.
Figure 7. Perspective” and “borrow scene” design technique is used to display the road landscape.
Figure 8. Noise reduction cutting and subgrade are combined with the surrounding environment
perfectly.
1763
Figure 9. Red and yellow are used in sound barriers as the symbol of Tibetans customs.
Figure 10. Rich texture of local materials and clever collocation of plants.
4 CONCLUSIONS
The subject of this research consisted in Sound barriers Landscape Design and Visually
Compensation Method based on Dynamic Aesthetics Factors. The following items may be
emphasized.
1. The landscape design of road sound barriers should perfectly complement each other with
the space form. The landscape design requirements with different characteristics should
be put forward for the different space forms of sound barriers.
2. The landscape requirement based on the different forms of sound barrier walls should be
put forward according to the characteristics of the acoustic function and visual consisting
of three parts of the top of screen, screen, and the bottom of the screen.
3. According to the different material characteristics of the sound barrier, the landscape
design of the sound barrier based on the different textures considering the dynamic visual
aesthetic characteristics of the pedestrians under different speed.
4. The landscape compensation design methods of road sound barrier on the basis of visual
aesthetic factors were proposed respectively from the five aspects of sequence and rhythm,
spatial enclosure, visual orientation force, environmental compatibility and regional culture.
1764
ACKNOWLEDGEMENT
The research work was supported by the Fundamental Research Funds for the Central Uni-
versities No.2015 JBM067 and No.2015RC023. The authors are very grateful for the helpful
comments and criticisms of the anonymous reviewers.
REFERENCES
American National Standard. 1998. Methods for Determination of Insertion Loss of Outdoor Noise
Barriers. American National Standards Institute. New York.
Blum, R.F. 1976. A. Guide to Visual Quality in Noise Barrier Design, U. S. Department of Transporta-
tion, Federal Road Administration: 77–12.
Benz, K. & Colin, E. 2009. Environmental Noise Barriers: A Guide To Their Acoustic and Visual
Design, Taylor & Francis Ltd.
David, J.O. & Christopher, A.E. 2015. A parametric investigation of the performance of multiple edge
road noise barriers and proposals for design guidance. Applied Acoustics 96: 139–152.
David, J. & Oldham, & Christopher, A. & Egan. 2011. A parametric investigation of the performance
of T-profiled road noise barriers and the identification of a potential predictive approach. Applied
Acoustics 72(11): 803–813.
Hamilton, J.R. and Thurstone, L.L. 1937. Safe driving: Human limitations in automobile driving. Gar-
den City, N.Y., Doubleday, Doran & company, Inc.
Herni, H. & Ramdzani, A. 2015. Effectiveness of Existing Noise Barriers: Comparison between Vegeta-
tion, Concrete Hollow Block, and Panel Concrete. Procedia Environmental Sciences 30: 217–221.
Julie, F. & Edward, B. 1990. Noise Barrier Design Guidelines, University of Wisconsin. Milwaukee.
Joo, Y. H. & Jin, Y.J. 2014. The effects of audio–visual factors on perceptions of environmental noise
barrier performance. Landscape and Urban Planning 125: 28–37.
John, A. Volpe. 1999. Acoustics and Your Environment: The Basics of Sound and Road Traffic Noise.
National Transportation Systems Center, Cambridge, Acoustics Facility.
Kumar, K. 2014. Optimized height of noise barrier for non-urban road using artificial neural network.
International Journal of Environmental Science and Technology 11(3): 719–730.
Luigi, M. & Massimiliano, M. & Francesco, A. & Maria, D.G. 2013. The influence of visual character-
istics of barriers on railway noise perception. Science of The Total Environment, 445(15): 41–47.
Like, J. & Jian, K. 2016. Combined acoustical and visual performance of noise barriers in mitigating the
environmental impact of motorways. Science of The Total Environment, 543, Part A: 52–60.
U.S. Department of Transportation. 2002. FHWA Road Noise Barrier Design Hand book. Federal
Road Administration.
1765
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
A. Cannone Falchetto
Technische Universität Braunschweig—ISBS, Braunschweig, Germany
K.H. Moon
Korea Expressway Corporation, Gyeonggi-do, South Korea
M.P. Wistuba
Technische Universität Braunschweig—ISBS, Braunschweig, Germany
ABSTRACT: In this paper, a simplified approach based on size effect theory is used to
extrapolate the strength values on small asphalt mixture beams obtained with a modified
Bending Beam Rheometer (BBR) at low temperature. For this purpose, eleven mixtures hav-
ing different mix design are first prepared. Then, the reliability of the measurements and
the effect of temperature, conditioning time and loading rate on the material strength are
addressed through Weibull statistics and factorial design on a representative set of mixtures.
Finally, Indirect Tension (IDT), Direct Tension (DT) and BBR tests are performed at three
different temperatures on the prepared asphalt mixtures and the strength results are initially
found to be statistically different. Based on size effect theory and on the weakest link model,
the strength values are transformed and further compared showing that BBR strength meas-
urements are statistically equivalent to the values obtained with IDT and DT test methods.
1 INTRODUCTION
Low temperature cracking is a serious distress for pavement built in regions experiencing severe
cold climate. Large tensile stresses develop as temperature drops; this is associated to a significant
material embrittlement ultimately leading to premature failure. The Bending Beam Rheometer
(BBR) (AASHTO T313, 2012) and the Indirect Tensile Test (IDT) (AASHTO T322, 2007) are
currently used to characterize and select asphalt binder and mixture for pavement applications
at low temperatures. A simpler test method for estimating the creep stiffness of asphalt mixture
at low temperature was proposed in the recent past based on the BBR procedure (Marasteanu
et al., 2009). This testing method makes use of small asphalt mixture beams with the same size
of the BBR binder specimens. In spite of the limited size of the samples it was demonstrated,
through microstructural analysis, that BBR mixture specimens are representative of the entire
asphalt mixture matrix for the materials analyzed up to a Nominal Maximum Aggregate Size
(NMAS) of 12.5 mm (Marasteanu et al., 2009). If this method can be extended to the evalua-
tion of asphalt mixture strength both creep and strength properties of asphalt mixture could be
obtained using the same device (BBR) and, potentially, the same specimen.
This paper presents a simplified method to determine the strength properties of asphalt
mixtures based on a modified BBR capable of applying higher loads. First an extensive exper-
imental work is performed. IDT (AASHTO T322, 2007), Direct Tension (DT) (Boltzman &
Huber, 1993; Christensen & Bonaquist, 2004) and the proposed BBR strength tests are con-
ducted on eleven asphalt mixtures. Fundamental factors, such as statistical distribution, tem-
perature effect and loading rate, are evaluated. Then, size and geometry effects, together with
testing configuration differences among IDT, DT and BBR tests, are investigated through
Weibull statistics (Weibull, 1939) and the weakest link model (Bažant & Planas, 1998).
1767
2 MATERIALS AND TESTING METHODS
2.1 Materials
A set of eleven asphalt mixtures, listed in Table 1, was prepared for the core of this research
effort. Mixtures were produced using different asphalt binders having two low performance
grades (PG) (AASHTO M320, 2010): −28°C and −34°C. Different types of aggregates and
various amounts of Reclaimed Asphalt Pavement (RAP) and roofing shingles (both tear-
off—TOSS—and manufactured waste scrap shingles—MWSS) were used in the mix design.
All mixtures were Superpave gyratory compacted to 7% air voids. The NMAS was between
4.75 mm for mixture C and 12.5 mm for the other mixtures.
Two additional asphalt mixtures, X and Y, were prepared with granite aggregate having
NMAS = 12.5 mm and two different binders having PG 58–28 and PG 64–34, respectively.
Together with mixture C, mixtures X and Y were used to evaluate the reliability of the BBR
strength measurements, depending on the combination of binder low PG and NMAS,
through the estimation of the statistical failure distribution of the material response (i.e.
strength). The material selected for this research was provided from the Minnesota Depart-
ment of Transportation (MnDOT). Additional information can be found elsewhere (Johnson
et al., 2008).
2P
σ IDT = (1)
π ⋅ bD
where P is the failure load at which the difference between vertical and horizontal defor-
mation is maximum. However, to avoid damage to the displacement measurement devices
attached to the specimen, P is usually assumed as the peak load; for this reason a correction
formula was proposed in a different research (Christensen & Bonaquist, 2004) to evaluate the
tensile nominal strength obtained with this method:
Binder
RAP Shingles
ID Description PG Content Content/Type Content/Type
1768
It must be noted that in the same research effort (Christensen & Bonaquist, 2004), the
results obtained from Equation 2 were compared with the strength results from DT tests
performed on larger specimens than the IDT one, and measuring 150 mm in length, 100 in
diameter and having a volume VDT = 1178 cm3. Therefore, VDT has to be considered as the
reference specimen volume for further analysis.
IDT strength tests were conducted on all eleven mixtures at three different temperatures
depending on the binder grade used to prepare the mixtures: low PG + 22°C, low PG + 10°C,
and low PG − 2°C. Specimens were conditioned for 2 hours at the testing temperature and
three replicates were tested at each temperature.
P
σ DT = (3)
A
where P is the failure load and A is the average cross section area. Due to limited availability
of material only four replicates of mixtures C, D and I were tested at a temperature corre-
sponding to low PG − 2°C after 2 hours conditioning in the climatic chamber.
1769
In this study a modified BBR device, shown in Figure 1, was used to perform strength
tests on asphalt mixture beams having the same size of the asphalt binder beam specimens
(l = 125 mm, b = 12.5 mm, h = 6.25 mm and volume of 9.7 cm3). Specimens were prepared
according to procedure proposed by Marasteanu et al., (2009) and further refined in other
research (Cannone Falchetto & Moon, 2015).
The loading frame of the upgraded BBR can apply and accurately measure loads as high
as 44 N. The new BBR is also equipped with a proportional valve air bearing system capable
of a much more complex control of the pressure, such as applying loads at different loading
rates, which are required when performing strength tests.
The flexural nominal strength σ 3 PB obtained with the BBR can be calculated as:
3PL
σ 3 PB = (4)
2bh2
where P is the peak load, L = 101.6 mm is the beam span, b is the width of the beam and h
is the thickness.
Three sets of tests were performed with the BBR. First, the statistical failure distribu-
tion, corresponding to the cumulative density function (cdf) of the response, was investigated
through histogram strength testing on asphalt mixture. The effect of temperature, condition-
ing time, and loading rate on the measured strength of three of the eleven mixtures was then
evaluated. Finally, BBR strength tests were conducted on the entire set of mixtures at the
same three temperatures used for IDT tests.
A preliminary evaluation of the BBR measurements was performed through strength histo-
gram testing. Histogram testing is a mean strength test performed on a large number of rep-
licates (20 or more). The strength results can then be analyzed on the Weibull scale (Weibull,
1939) to identify the type of failure distribution of the material (brittle, quasibrittle, ductile)
and, hence, used to predict the strength size effect (Bažant & Planas, 1998).
Since the mixtures shown in Table 1 were prepared with binder having two different low
PGs (−28°C and −34°C) and two NMAS (4.75 mm and 12.5 mm), the following three repre-
sentative mixtures were selected for statistical evaluation: C, X and Y (see Section 2.1). Test-
ing temperature was imposed to be equal to low PG+10°C, while the number of replicates
was different for each of the three mixtures depending on material availability. Loading rate
was imposed to be 3 N/min, in order to avoid very short tests which could present a very high
variability. Conditioning time was set to 1 h because of the reduced dimension of the BBR
specimen (Marasteanu et al., 2009). Table 2 presents a summary of materials and testing
condition for histogram testing.
Strength results did follow neither a normal nor a lognormal distribution as previously
demonstrated for other composite materials such as cement concrete (Bažant 2005). Weibull
statistical distribution was then selected and strength data were rescaled in the Weibull plane,
according to the mid-point position method (Rinne 2009) where the failure probability Pf is
plotted against the natural logarithm of the nominal flexural strength and Pf is calculated as:
i − 0.5
Pf ( N )= (5)
n
where i = 1 to n is the rank of the single strength measurement and n is the total number of
tested specimens. From a statistical point of view Pf represents the cdf of strength. Figure 2
provides the three different histograms for mixtures C, X and Y.
The experimental results form an almost straight line and appear to follow a Weibull sta-
tistical distribution for mixture C (binder PG 64–34 and aggregate NMAS = 4.75 mm). A
similar trend is observed for mixture X (binder PG 58–28 and aggregate NMAS = 12.5 mm)
1770
Table 2. Asphalt mixtures for histogram testing.
C 64–34 4.75 mm 24
X 58–28 12.5 mm 26
Y 64–34 12.5 mm 20
Figure 2. BBR strength histogram for mixtures C (left), X (middle) and Y (right).
with a moderate deviation in the central part of the plot. Both histograms of mixtures C and
X, suggest a strong brittle failure behavior, which is most likely associated to the size of the
aggregate particles for C and to the higher low PG of mixture X.
A different trend is shown by the histogram of mixture Y (binder PG 64–34 and aggregate
NMAS = 12.5 mm), which presents a deviation (upper part of the curve) from the Weibull dis-
tribution (the straight lower part of the plot) for higher stress level. This is typically observed
for quasibrittle materials where the size of the homogeneity (NMAS) affects the failure dis-
tribution limiting the material brittleness. However, in this specific case, it is hypothesized
that the lower low PG limit (−34°C) of the binder used for mixture Y, in combination with
the actual NMAS = 12.5 mm, induces such a material behavior which cannot be observed for
mixture X, where a binder with low PG limit-28°C is used.
From the three histograms, the Weibull modulus m, representing the slope of the line fit-
ting the experimental data was calculated and the following three values of 12.68, 11.59 and
13.89 were obtained for mixtures C, X and Y, respectively. Based on hierarchical statistical-
mathematical model proposed by Bažant and his research team (Bažant 2005), m has to be
an integer number, and, therefore, the fitted values were rounded up to 13, 12 and 14, respec-
tively. These values are similar with previous results obtained by other authors for cement
concrete (Zech & Wittmann, 1977). It should be noted that more recent research in concrete
found higher values of m (Bažant 2005).
The Weibull modulus represents a significant parameter in the size effect theory, since it con-
trols the tail of the failure distribution. This part of the cdf has a crucial role in the size effect law
especially as the size of the structure increases, which corresponds to a more brittle structural
behavior. Further detail on size effect theory can be found in section 5 of this manuscript.
The effect of temperature, conditioning time and loading rate, on the BBR strength, was
investigated for a limited set of mixtures: C, D and I (Table 1). A statistical analysis with a 23
factorial design was performed to evaluate the single and combined effect of these factors on
the material response. Three replicates per factor-level combination were used.
1771
Testing temperatures were selected based on binder PG grade: PG + 10°C and PG − 2°C
(Marasteanu et al., 2009). A higher loading rate of 13 N/min was added to the 3 N/min rate
used for histogram testing. Table 3 presents the factorial design with statistical factors and
levels, while the BBR mean nominal strength for mixture C, D and I are shown in Table 4 for
all factor level combinations.
Table 5 summarizes the p-values for the factorial statistical analysis; when the specific
p-value is smaller than the selected significance level (α = 0.05) the influence of the individual
factor on the response is significant. Since differences due to mix design were expected, the
analysis was run separately for each mixture.
The analysis indicates that only the main factors are statistically significant. Temper-
ature affects the strength of mixtures C and D with lower strength associated to lower
temperatures. Loading rate appears to affect the strength of mixture I, where higher rate
corresponds to higher strength measurements. Conditioning time does not affect the mean
BBR strength.
Mixture C
1772
5 COMPARISON BETWEEN BBR, IDT, AND DT MIXTURE STRENGTH
In this section a comparison between two of the current tests method, IDT (AASHTO T322,
2007) and DT (AASHTO TP10, 1993), commonly used for estimating the strength of asphalt
mixture, and the BBR 3PB strength is performed, based on size effect theory (Bažant 2005).
⎡ ⎛ σ ⎞m⎤
Pf = 1 − exp ⎢ − ⎜ ⎟ ⎥ (6)
⎢⎣ ⎝ σ 0 ⎠ ⎥⎦
where σ0 is the scale parameter of the Weibull distribution and m is the Weibull modulus
(shape parameter).
Since the number of possible material defects and, hence, of the randomness of the struc-
tural strength, depends on the volume of specimen/structure, such a parameter must be
included in the size effect analysis. Therefore, the strength of a large structure is statistically
lower than what is measured on a smaller sample of the same material. Using Weibull sta-
tistics, a relationship between strength obtained from the same material on specimens with
different volumes can be obtained:
1/ m
⎛V i n ⎞
σ structure σ specimen ⋅ ⎜ sspecime (7)
⎝ Vsstructure ⎟⎠
Together with volume, specimen geometry and stress field must be addressed. For such a
purpose Equation 6 can be used and the failure distribution can be rewritten as function of
the adimensionalized stress s(x) imposed to the structure, where x represents the vector of
the stress field:
⎧⎪ ⎧ ⎫⎛ σ ⎞
m
⎫⎪
1 e p ⎨ ⎨∫ [ s ] m
Pf dV ( x ) ⎬ ⎜ N ⎟ ⎬ (8)
⎩⎪ ⎩V ⎭⎝ σ 0 ⎠ ⎭⎪
where σN is the nominal strength and V is the effective volume of the specimen under the
stress filed of interest.
In case of a beam under 3PB test the nominal stress in Equation 8 can be replaced by
Equation 4 and the integral can be solved over the volume of the beam subject to tensile
1773
stress (effective volume). For a uniform tension case, such as direct tension tests Equation 8
presents a very simple formulation:
⎡ V ⎛ σU ⎞ m ⎤
Pf = 1 − exp ⎢ − U ⎜ N ⎟ ⎥ (9)
⎢⎣ V0 ⎝ σ 0 ⎠ ⎥⎦
where σ UN is the nominal strength for uniform stress field conditions, V0 is a reference volume
and VU is the volume of the tested specimen.
Equations 7, 8 and 9 can be manipulated to relate the mean nominal strength obtained
from a 3PB configuration, with a given volume V3PB, to the mean nominal tensile strength
measured from direct tension test, with a different volume VU, as:
1/ m
σ N3 PB ⎡ 2 (1/ m ) ⎛ VU ⎞
= 2 (1+ m ) ⎤ ⎜⎝ V ⎟⎠ (10)
σ UN ⎣ ⎦
3 PB
1774
Figure 4. Comparison IDT, BBR and DT strength values after size effect correction.
ing a dimensions 150 mm in length and 100 in diameter with a volume VDT = 1178 cm3. There-
fore, all the IDT, DT, and BBR strength values were converted to equivalent uniaxial tensile
strength obtained on an equivalent specimen of volume VDT. In such a way a consistent
comparison of the experimental results could be performed. For such a purpose Equations 2,
7 9 and 10 were applied to the measured strength data and the specific Weibull modulus, esti-
mated in section 3, was used depending on aggregate NMAS and binder low PG. Figures 3
and 4 show the mean strength values before and after the size effect conversion.
Statistical analysis of experimental measurements indicates that BBR strength is much
higher than IDT and DT values (see Fig. 3). However, after size effect was applied to the
strength results, no significant difference were found between BBR, IDT and DT strength
for a 0.05 significance level. It has to be noted that the use of the reference volume, VDT, is
fundamental for a correct comparison of the strength data across the different test types:
BBR, IDT and DT.
In the present paper a simplified method to extrapolate the strength values obtained at low
temperature on asphalt mixture beams with a modified bending beam rheometer was pro-
posed. Statistical failure distribution and the effect of temperature, conditioning time and
loading rate were first evaluated. Then, the strength values obtained with the bending beam
rheometer were compared with the results measured with two current strength tests meth-
ods: indirect tension and direct tension tests. Finally, size effect theory was used to take into
account the effect of geometry and stress field and the experimental measurements were
further compared.
The following conclusions can be drawn:
• Bending beam rheometer strength results present a Weibull distribution governing the tail
of the cumulative distribution function.
• Temperature and, to a minor extent, loading rate appear to affect the strength
measurements.
• Bending beam rheometer experimental strength values are much higher than those obtained
with indirect tension and direct tension tests.
1775
• Size effect theory indicates that the strength measurements obtained from the three types
of tests are statistically equivalent.
An ongoing research on size effect is currently in progress with the objective of refining the
size effect modeling solution and the extrapolation procedures in the Pavement Engineering
Centre (ISBS) at the Technische Universität Braunschweig, Germany.
REFERENCES
AASHTO M320–10-UL 2010. Standard method of test for performance graded asphalt binder. Ameri-
can Association of State Highway and Transportation Officials.
AASHTO T313–12-UL 2012. Standard method of test for determining the flexural creep stiffness of
asphalt binder using the bending beam rheometer (BBR). American Association of State Highway
and Transportation Officials.
AASHTO T321–14-UL 2014. Standard method of test for determining the fatigue life of compacted
asphalt mixtures subjected to repeated flexural bending. American Association of State Highway and
Transportation Officials.
AASHTO T322–07-UL 2007. Standard method of test for determining the creep compliance and
strength of hot mix asphalt (HMA) using the indirect tensile test device. American Association of
State Highway and Transportation Officials.
AASHTO TP10–93 1993. Standard method of test for thermal stress restrained specimen tensile
strength. American Association of State Highway and Transportation Officials.
Bažant, Z.P. & Planas, J. 1998. Fracture and Size Effect in Concrete. London: CRC Press.
Bažant, Z.P., (1st ed.) 2005. Scaling of structural strength. London: Elsevier.
Boltzman, P. & Huber, G. 1993. Direct tension test experiments. Strategic highway research program
SHRP Report A-641, Washington DC.
Cannone Falchetto, A. & Moon K.H. 2015. Micromechanical–analogical modelling of asphalt binder
and asphalt mixture creep stiffness properties at low temperature. Road materials and pavement
design, 16(1): 111–137.
Christensen, D.W. & Bonaquist, R.F. 2004. Evaluation of indirect tensile test (IDT) procedures for
low-temperature performance of hot mix asphalt. National cooperative highway research program
NCHRP Report 530, Washington DC.
Fisher, R.A. & Tippet, L H.C. 1928. Limiting forms of the frequency distribution of the largest and
smallest number of a sample. Proceeding of the Cambridge philosophical society, 24: 180–190.
Fréchet, M. 1927. Sur la loi de probabilité de l’écart maximum. Annales societatis mathematicae polo-
nae, 6: 93–116.
Galilei, G. 1638. Discorsi e dimostrazioni matematiche intorno a due nuove scienze. English transaltion by
T. Weston, London: Elsevirii Leiden.
Johnson, A., Clyne, T.R. & Worel, B.J. 2008. 2008 MnROAD phase II construction report. Minnesota
department of transportation report, St. Paul, MN.
Marasteanu, M.O., Velasquez R., Cannone Falchetto A. & Zofka, A. 2009. Development of a sim-
ple performance test to determine the low temperature creep compliance of asphalt mixture. IDEA
Program Final Report NCHRP-133 – Transportation Research Board of the National Academies,
Washington DC.
McCurdy, E. (1st ed.) 2002. The notebboks of Leonardo da Vinci. London: Konecky & Konecky.
Oehlert, G.W. (1st ed.) 2000. A first course in design and analysis of experiments. Minneapolis: W.H.
Freeman.
Rinne, H. (1st ed.) 2009. The Weibull distribution. A handbook. London: CRC Press.
Weibull, W. 1939. The phenomenon of rupture in solids. Proceedings of royal swedish institute of engi-
neering research, 151: 1–45, Stockholm.
Zech, B & Wittmann F.H. 1977. A complex study on reliability assessment of the containment of a
PWR, Part II. Probabilistic approach to describe the behavior of materials. In 4th International Con-
ference on Structural Mechanics in reactor Technology, H.T.A. Jaeger and B.A. Boely, eds. European
community, Brussels, 1–14.
1776
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
ABSTRACT: The design of road curb height and distance are important techniques to
effect the urban flooding. This paper reviews a number of researches on pavement runoff.
A 2D hydrodynamic model is developed based on shallow water equations, which allows an
integral simulation of pavement surface runoff under different curb height, distance and
rainfall intensity. The equations are discretized using the finite volume method on unstruc-
tured grids. Numerical results show that, accurate curb distance and height is more effective
in decreasing runoff volume and the pressure of rainwater drainage network during heavy
rainfall event, particularly the water depth and flooded area near the curb. The performance
of curbs inlets on pavement runoff is affected by rainfall intensity, curb distance, curb height
and the location of the road. Additional, the model significantly appropriate to simulate the
water depth, the inundation area and describe the flow field on the pavement. It can be dem-
onstrated that the road with curb opening fully reflects the ideas of low impact development,
which is essential for transport management.
1 INTRODUCTION
The rapid rate of urbanization leads to a huge change of the underlying geology and the
dominant soil type in urban area, which are in determining the rainfall-runoff volume and
proportion of the surface. In addition, climate change has a correlation with extreme weather
phenomena, in particular, the storm events occurring more frequently in recent years (Kapil,
2007). It is apparent that runoff coefficient and flow discharge are increasing and exceeding
the drainage capacity during excessive rainfall events, leading to urban flooding. Ponded water
on pavement surfaces is not only relevant in terms of durability. It also respects to traffic safety
in the presence of high water depths, splash and drastically decrease visibility (Jha et al, 2012).
Hence, effective drainage system of roadway pavement is important to the traffic safety and
the storm-water runoff and the water depth on road is an urgent and essential issue.
In high vehicle speeds, hydroplaning additional constitutes a serious safety risk as the water
on pavement decrease the road resistance. Therefore, with an accurate material and structure
of pavement, the water can be drained fast (Ping et al, 2013; Cerezo et al, 2014). In addition,
some experiments conducted to research the impacts of curbs inlet on road pavement runoff
(Comport et al 2014; Uyumaz, 2014). Hydrodynamic models describe the flow behavior by
equations derived from the Navier Stokes Equations referred to as Saint Venant or Shallow
Water Equations. The kinematic wave equation was applied to pavement surface runoff (Cris-
tina et al, 2003; Malal et al, 2012). Wolff (2013) simulated the real pavement surface runoff,
including features like the simulation on real pavement surface with an irregular topography,
the possibility to treat different surface textures and the treatment of drainage facilities.
It can state that water on pavement significantly affects traffic safety and the amounts of
water on pavement will even increase. New drainage system should be used with the trend
of climate change and high development of urbanization. Low impact development (LID)
techniques are regarded as a more sustainable solution for urban road storm-water runoff
and flooding management than traditional drainage system. The basic principle of this
1777
technique is to maintain post-development hydrology of a site close to its natural condition
present before development occurs (Ahiablame et al, 2012). For highly urbanized areas, such
as urban road, permeable pavements, rainwater harvesting, can be used to reduce runoff vol-
umes, decrease peak flows and collect rainwater effectively (Damodaram et al, 2010; Wang
and Xu et al, 2014).
These previous studies mostly focused on impacts of road pavement material structure
to rainfall runoff. This paper focuses on the effects of curb distance and height techniques
on urban road flooding. Using a case study, a three lane road section was selected for test-
ing, and a pavement surface runoff model based on two-dimensional (2-D) depth-averaged
shallow water equation (SWE) model was developed to simulate the rainfall runoff charac-
teristics. Then, the water depth, inundation area and flow field were analyzed and compared
between new curb designs and conventional drainage system.
2 MODELLING METHODOLOGY
∂U ∂F ( ) + ∂G ( ) = S
+ ( ) (1)
∂t ∂x ∂y
where x and y are the Cartesian coordinates; t is the time; U is the state vector; F and G are
the x- and y- directional flux vector respectively; S is the source term vector. These vectors
are defined as follows:
U =( )T = (h,qqx )
T
qy
T
⎛ 1 ⎞
F = hu, hu
h 2 + gh2 , huv⎟
⎝ 2 ⎠
T
⎛ 1 ⎞
G = hv, huv
h , hv 2 + gh2 ⎟
⎝ 2 ⎠
( ( ) ( ))
T
S= − −
where h is the water depth; u and v are x- and y- directional velocity components; qx = hu and
gy= hv are the unit discharge in the x- and y- directions respectively; q is the discharge per unit
area; S0x is the bottom slope and Sfx is the energy loss due to bottom friction along the x axis
and similarly, S0y and Sfy for the y axis. The friction slopes are evaluated using the Manning
formula.
1778
⎛ ∂U ⎞
∫ ⎜⎝
Aι
∂t
+∇ ⎟⎠ = ∫ Sds
Aι
(2)
where E ( F H ) , Αι is the area of the cell Ωι , and s is the integration variable defined on Ωι .
Using the Green formula, we obtain
∂U i
Ai + ∫ E ndΓ = Si (3)
∂t Γi
where Γ i is the boundary of the cell Ωι ; Γ is the integration variable along the boundary;
n is the outward unit normal ( )
vector of Γ i , n = nx , ny , with nx and ny being the x and y com-
ponents of n, respectively; Si is the integral of the source term and U i is the mean value of U
in the cell Ωι and is stored at the center of the cell Ωι :
1
Ai A∫ι
Ui U ds (4)
Ei
∫ E n ∫ (F x H y )ds = ∫ E ds
ds ∑ E n
j =1
l
nij ij (5)
Γi Γi Γi
where lij is the length of the jth side of the cell Ωι ; E n E i n = F x Hny ; and E niji is the flux
through the jth side of the cell Ωι , which is discretized with a Roe-type approximate Riemann
solver, which is shock capturing, and describes the hydraulic discontinuities accurately.
The source terms are divided to the bed slope terms S 0 and friction slope term. S f For
improve the model’s stability the friction slope term is divided by implicated method; the
bed slop term is decomposed in the characteristic direction so that the numerical scheme can
exactly satisfy the conservative property.
⎛ ∂S f ⎞ ⎡ Δt ⎛ i
E
⎞⎤
U item = ⎜ I −
⎝
⎢U i
∂U ⎟⎠ ⎢⎣
∑ E niji (U )lij
Ai ⎝ j =1
S0 (U n )⎟ ⎥
⎠ ⎥⎦
1⎡ n ⎛ ∂S f ⎞ ⎡ Δt ⎛ i ⎞ ⎤⎤
E
Ui 1
⎢U i + ⎜ I −
2 ⎢⎣ ⎝
⎢U i
∂U ⎟⎠ ⎢⎣
∑ E niji (U
Ai ⎝ j =1
)lij S0 (U tetm )⎟ ⎥ ⎥
⎠ ⎥⎦ ⎥⎦
(6)
1779
Figure 1. Sketch map of the example road.
Figure 2 shows the shape file of traffic lane segment of the section. Taking this segment
of the road for the model application, with a refined grid of quadrilateral for representing
the site. There are 22800 grids in total, and the length of each element is about 0.5 m. Based
on the topographic data, the mesh is interpolated. As is shown in Figure 2, different colors
indicate different bottom elevation. Red zone means lower elevation, and green zone means
higher elevation. The bottom elevations of the example road section ranging from +12.3 m
to +13.29 m. It can be demonstrated that there exists a correlation in urban road between the
terrain elevation and accumulated rain water.
2993[ + − ]
q= (7)
(t + ) 0.79
Where q is rainfall intensity (mm/s); P is return period (year); and t is rainfall duration
(hour).
The Chicago storm profile was used in this study which is usually adapted for urban drain-
age design (Keifer et al, 1957). The Chicago design storm profile are derived from Eq. 7, and
peak time ratio r equals 0.4. The value r (where 0 < r < 1) describes the location of peak time
of rainfall intensity in duration. Both of the return period rainfall events have the same rain-
fall duration (2 h) and location of peak rainfall intensity. The Chicago storm profile is divided
into two formulas before and after peak time, which are described as Eq. 8a and Eq.8b:
1780
11.93 ( + P) ⎡ 0.79ta ⎤
ia = ⎢1 − ⎥
a + 5.38 ⎦
0.79
⎛ ta ⎞ ⎣ t (8a)
+ 13.45
⎝ 0.4 ⎠
11.93 ( + P) ⎡ 0.79tb ⎤
ib = ⎢1 − ⎥
⎛ tb ⎞
0.79
⎣ tb + 8.07 ⎦ (8b)
+ 13.45
⎝ 0.6 ⎠
where ia, ib are rainfall intensity before and after peak intensity (mm/s); ta , tb are time before
and after peak intensity respectively(min).
Rainfall intensity values adapted over a range of time intervals. The time series of two
rainfall intensity curve (2- and 5-year) are as shown in Figure 3.
Bottom resistance reflected by Manning’s roughness coefficient, and asphalt surface tex-
ture is a key parameter to affect the resistance. Manning’s roughness coefficient reflects the
influence between pavement and flow water, and an average Manning’s roughness coefficient
value of 0.015∼0.016 was determined in this paper. (Arcement et al, 1989).
1781
To evaluate the effectiveness of LID techniques, three measuring points on the site denoted
A, B and C as shown in Figure 2 were chosen which represents typical hydraulic conditions.
Point A located at the right and downhill of the road, but close to curb, the elevation is higher
relatively represents an area. Point B approaching the curb, represents an area where the ele-
vation is low and rainfall-runoff accumulated in simulation time. Point C locates at the left of
the example road section, be at the middle of the traffic street, and the elevation is higher.
Factors Features
Figure 5. Water depth under different rainfall intensity. (a) Point A; (b) Point B; (c) Point C.
1782
Figure 6 shows the water depth of the three points in rainfall duration. Point A locates at
the front of the road section and the bottom elevation is relatively high. Point B located at
the flood-prone area, with lowest elevation of the example road section. Point C locates at
the middle of the road and close to median at the center, with highest elevation. As is shown
in Figure6 (a), under 5-year rainfall intensity and curbs opening, maximum water depth of
point B is 5 mm, but 4 mm of A and C respectively. For 5-year return period but with con-
ventional curbs, as is shown in Figure6 (b), the maximum water depth of point B is up to
70 mm, water depth of point A and B is 20 and 5 mm. It is obvious that, taking use of the
conventional curbs, the effects of the location to water depth is greater than which with curbs
opening. Location of flooded prone area is easy for water accumulating, and water depth is
largest all the time.
Figure 6. Water depth with different location. (a) Curbs opening; (b) Conventional curbs.
Figure 7. Effects of curb distance. (a) Point A; (b) Point B; (c) Point C.
1783
Figure 8. Effects of curb height. (a) Point A; (b) Point B; (c) Point C.
Figure 9. Inundation area and water depth distribution on the road pavement. (a) Curb opening; (b)
Conventional curbs.
On one hand, with the curbs opening, as is shown in Figure8 (a), the water depth h of
15 cm curbs is larger than h of 12 cm curbs. However, in the flood-prone area of point B, as
is shown in Figure8 (b), the h of two curb is close and 15 cm is slightly higher than which of
12 cm. At the point C, two curve of water depth is appearing to overlap. It infers that within
the certain rainfall event, the curb height affect the road pavement is not obvious, as storm
flow into swales from the opening inlets quickly. On the other hand, when the curb used con-
ventional design without opening, the maximum water depth at three points are same, the
values are 20, 65 and 2.6 mm. This is because the maximum h is less than curbs height, which
is 120 mm and 150 mm. In consequence, there is on influence of curb height without opening
on road pavement runoff.
1784
Figure 9 (a) is calculated with curbs opening, Figure 9 (b) is calculated with conventional
curbs. Firstly, comparing the upper and lower plots, it is obvious that the inundation area of
the upper plot is much smaller, and water depth distribution is narrow. It can be stated that
practice of curbs opening in reducing inundation area is effective and markedly. Secondly, as
is shown in Figure9 (a), at the incipient stage, the inundation area is small and water depth
distribution shallow. As the rainfall event going, the inundation area and water depth gradu-
ally increasing and reach to peak. The third stage is flooded recession; the spatial distribution
of water depth is going to narrow. Figure 9 (b) shows again the three stages of accumulation
and recession on road pavement.
4 CONCLUSIONS
In this paper, the effects of urban road curb distance and height on road pavement runoff are
analyzed. The main results obtained are summarized as follows:
1. An appropriate curb inlet is more effective in decreasing runoff volume and the pressure
of rainwater drainage network during heavy rainfall events, particularly the water depth
and inundation area on roadside. The techniques of curb inlets fully reflect the ideas of
low impact development.
2. The performance of curbs inlets on pavement runoff is affected by rainfall intensity, curb
distance, curb height and the location of the road. The results of water depth and inunda-
tion could be extremely useful to analyze the changing of pavement flow during heavy rain
event and determine the curb height and inlet spacing.
3. The 2D hydrodynamic model give an integral simulation of pavement surface runoff. It is
significantly appropriate to simulate the water depth, the inundation area and describe the
flow field on the pavement.
REFERENCES
Ahiablame L.M., Engel B.A., Chaubey I. 2012. Effectiveness of Low Impact Development Prac-
tices: Literature Review and Suggestions for Future Research. Water, Air & Soil Pollution, 223(7):
4253–4273.
Arcement G.J.J., Schneider V.R. 1989. Guide for selecting Manning’s roughness coefficients for natural
channels and floodplains. United States Geological Survey Water-supply Paper 2339.
Cerezo V., Do M.T., Prevost D., et al. 2014. Friction/water depth relationship – In-situ observations and
its integration in tire/road friction models. Journal of Engineering Tribology, 228(11): 1285–1297.
Cristina C.M., Sansalone J.J. 2003. Kinematic Wave Model of Urban Pavement Rainfall-Runoff Subject
to Traffic Loadings. Journal of Environmental Engineering, 129(7): 629–636.
Comport B.C., Thornton C.I. 2014. Hydraulic Efficiency of Grate and Curb Inlets for Urban Storm
Drainage. Journal of Hydraulic Engineering, 138(10): 878–884.
Damodaram C., Giacomoni M.H., Khedun C.P., et al. 2010. Simulation of Combined Best Manage-
ment Practices and Low Impact Development for Sustainable Stormwater Management. Journal of
the American Water Resources Association, 46(5): 907–918.
Geng Y.F., Wang Z.L., Lu Y.J. 2009. Discretization of Two-dimensional Advection-Diffusion Equation
with Unstructured Cell Center Finite Volume Method. Chinese Journal of Computational Physics,
01:17–26.
Jha A.K., Bloch R., Lamond J. 2012. Cities and Flooding: A Guide to Integrated Urban Flood Risk
Management for the 21st Century, the World Bank, pp.885–887.
Kapil G. 2007. Urban flood resilience planning and management and lessons for the future: a case study
of Mumbai, India. Urban Water Journal, 4(3): 183–194.
Keifer C.J., Chu H.H. 1957. Synthetic Storm Pattern for Drainage Design. Journal of the Hydraulics
Division, 83:1–25.
Malal K., Minh T.D. 2012. Kinematic wave approach to model water depth on road surfaces during and
after rainfall events. International Journal of Pavement Engineering, 13(5): 385–394.
Ping C.H., Chen Y. 2013. Surface water flow over a pervious pavement. International Journal for Numer-
ical & Analytical Methods in Geomechanics, 37(9): 1095–1105.
1785
Uyumaz A. 2014. Urban Drainage with Curb-Opening Inlets. Global Solutions for Urban Drainage.
ASCE, pp.1–9.
Wang Z.L., Geng Y.F. 2013. Two-dimensional shallow water equations with porosity and their numeri-
cal scheme on unstructured grids. Water Science & Engineering, 6(1): 91–105.
Wang X.X., Xu S.F., S Z, et al. 2014. Study of the Road Rainwater Control and Utilization Technology
Based on Low-Impact Development(LID). Journal of Beijing University of Civil Engineering and
Architecture, 30(4): 33–47.
Wolff A. 2013. Simulation of pavement surface runoff using the depth-averaged shallow water equa-
tions. Highway Institute, University of Stuttgart.
1786
Functional Pavement Design – Erkens et al. (Eds)
© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02924-8
Author index
1790