TITLE OF THE THESIS

A Dissertation

Submitted for the Degree of

Doctor of Philosophy

In the Faculty of Engineering

By

Rizwan Khan

Department of Civil Engineering

Indian Institute of Science
Bangalore-560012, India
November 2023
To my beloved son: Sadaan Ahmad Khan
 Acknowledgement

I want to start by thanking the all-powerful God for giving me the courage to successfully submit

my PhD thesis. The PhD program at IISc has provided me with an extraordinary experience,

significantly enhancing my intellectual and personal development. I am profoundly grateful to

those who provided assistance and support during my tenure at IISc.

I am fortunate to have had the chance to work with Prof. Madhavi Latha Gali. She offered me

unwavering encouragement and support, frequently going above and beyond to help me. She

regards her students as members of her own family and is always available to assist in any way. I

admire her perseverance and unwavering dedication to research. I am grateful for the confidence

she placed in me and her constant encouragement to accomplish my goals and earn my Ph.D.

My heartfelt thanks to all the faculty members of the IISc who taught me and extended their

cooperation during my research. Special thanks go to Prof. Jayant Kumar, Dr. P.

Anbazhagan, Dr. P. Raghuveer Rao, Prof. Sudhakar M. Rao, Prof. Chandra Sekhar

Seelamantula, Prof. R. Narasimhan, Prof. Praveen C. Ramamurthy, Prof Satish V. Kailas.

I wish to heartfelt gratitude to my esteemed colleagues and seniors including Saurabh Singh,

Debasis Mohapatra, Obaidur Rahman, Tarun Naskar, Dhanaji Chavan, Bhardwaj Pandit,

Gaurav Tiwari, Nandhi Verman, Rajeev Gupta, Nitin Gupta, whose extensive discussions,

and insights significantly contributed to my research work. I especially thank Silas Abraham for

his invaluable feedback and constructive comments on my research work. I also appreciate

Anjali Pillai, Hemant Jain, Lakkimsetti Balaji, Ilyas Bhat, KV Anusree for their help in my

academic journey. I would also like to acknowledge Faheem Ahmad, Abdullah Talib,

Ashraful Haq, Sk Habibullah, Khaja Moinuddin, Usman Ul Muazzam, whose presence

made my time at IISc memorable and filled with enjoyable moments.

 Acknowledgement

I would like to extend my gratitude to the Civil Engineering Department's staff and chairman for

their prompt assistance whenever it was needed.

Finally, I wish to convey my deep appreciation to my parents, my sisters, namely Reshma and

Samreen, my brother-in-law, Umar Malik, and my cherished son, Sadaan, for their assistance

and unwavering support during my Ph.D. journey.

 Abstract

To understand the process of mobilization of shear strength in sands and sand-geosynthetic

interfaces at a fundamental level, it is essential to precisely characterize the size and shape of the

grains and the shear-induced surface changes in geosynthetics. Due to the difficulty in defining

and measuring the size and shape descriptors for a wide range of complex particle morphologies,

the majority of earlier studies either completely ignored or used visual comparison charts which

are highly subjective. Those who employed digital image-based techniques also limited their

assessment to morphological characteristics within a two-dimensional (2D) context. The 2D

shape parameters are obtained from the images taken through random projection of particles,

which results in inaccurate values. In the existing literature, only a small number of studies have

employed a three-dimensional (3D) particulate morphology to investigate their effects on sand-

geosynthetic interfaces. In addition, most of the research on the effect of particle morphology on

the shear behaviour of granular material and their interactions with geosynthetics has not been

explored at multiple scales of analysis. Also, very limited studies gave importance to the micro-

structural analysis of geosynthetics surfaces after shearing to understand the micro shear

mechanisms responsible for macroscopic shear behaviour. In this context, the current study aims

at accurate multi-scale 3D morphological characterization of sand grains and understanding the

effects of particle morphology on the shear strength of sands and interface shear strength of

sand-geosynthetic interfaces.

In the first part, this thesis presents a systematic and comprehensive 3D quantification of the

morphology including size and shape, and 3D fractal dimension of real sand particles taken from

four different types of sands with various size and shape characteristics. The first step in

obtaining the 3D morphology was to perform high resolution X-ray µCT scanning on the sand

particles. The individual particles were then extracted and separated using a variety of image

processing techniques. The surfaces of raw µCT images are invariably characterized by saw-

i
 Abstract

tooth patterns leading to overestimation of the actual surface area. Therefore, reconstructing the

particle surface from the µCT images for precise quantification of grain's geometrical

characteristics was suggested in literature. Hence, tThe smooth and continuous 3D particle

surfaces with their overall morphology preserved were then recreated using a sophisticated

approach based on spherical harmonic (SH) analysis. To obtain the geometrical parameters for

calculating the size and shape of sand particles, a robust MATLAB algorithm was written and

implemented in this study. For computing the particle size, the three principal dimensions of the

particle were obtained through principal component analysis (PCA). The particle size

distribution (PSD) of the sands obtained from µCT images and mechanical sieving of the sands

were compared. Quantification of tThe 3D shape descriptors was carried outwere quantified

using computational geometry and image analysis methods based on the SH reconstructed

particle surface. Additionally, a fractal dimension for the 3D closed surface of the sand particle

was discussed and quantified using spherical harmonic-based fractal analysis. The statistical

analysis of the shape descriptors revealed that they are not independent. The correlation between

any two shape descriptors relies mainly on the distance between the characteristic scales of these

parameters.

In the second part of the work, the direct shear apparatus (DSA) was modified to overcome

the limitations of conventional set-up and make it suitable also for measuring visualizing and

studying the particle kinematics behaviour of particles. Then aA series of sand alone direct shear

tests were performed using modified DSA on four types of sands with different morphological

characteristics. High quality videos were captured during shearing to examine the development

of a localized shear zone in the tested sands by analyzing particle displacement using digital

image correlation (DIC). Full-field shear strains were measured and plotted in order to determine

the shear band thickness. The experimental data of sand alone tests reveals that sand particles

with higher irregularity exhibit higher peak shear stress and dilation due to their enhanced

particle interlocking compared to similar sized particles with less irregularity. When considering

ii
 Abstract

particle size effects alone, iIt was observed that the variations in particle size also affect the

resistance to rolling and sliding, as well asthereby influencing the shear strength behaviour of

granular materials, with and found that the peak and residual friction angles increase increasing

with an increase in the particle size. The DIC analysis revealed that shear band thickness is

smaller for irregular particles than the particles with regular morphology, which . This result is

explained by in terms of the difficulty in for rotation that the irregular particles encounter,

because of their stronger interlocking.

In the last part of the work, interface shear tests were conducted between on four different

test sands in contact with a woven geotextile and or a smooth geomembrane. Experimental data

revealed that the dominant shear mechanism in scenarios involving dilative (sand-geotextile)

interfaces, the dominant mechanism is the interlocking between the sand particles and the

surface asperities of the geosynthetic material, whereas . In contrast, the primary mechanisms for

in non-dilative (sand-geomembrane) systemsinterfaces, the mechanisms are sliding and plowing.

Considering theWhen particle size effects are considered, when fine particles interacted

interacting with the geotextile surfaces , it resulted in higher peak friction angles, due to their

ability to interlock better with the asperities of the geotextile surfaces. However, in the case of

medium and coarse sand particles, their size was larger than the concavity of the geotextile's

asperities, which made it difficult for them to interlock properly during shearing, resulting in .

This resulted in lower peak friction angles and reduced dilation. In sand-geomembrane interfaces

also, fine particles led to higher peak friction angles compared to medium and coarse particles.

Micro-topographical analysis of sheared geomembranes revealed that finer particles tend to

make a more significant number of effective contacts per unit area with the geomembrane

surfaces during shearing, which results inwith more grooves with formed at less spacing,

resulting in higher shear strength. Also, higher normal stress levels caused a decrease in the peak

friction angles observed in sand-geotextile interfaces, whereas the opposite trend was observed

in sand-geomembrane interfaces, which is attributed to the transition in the shearing

iii
 Abstract

mechanisms. This variation can be attributed to the particles engaging in plowing behaviour on

the smoother surface of the geomembrane at higher normal stresses, resulting in an increased

interface friction angle. Findings from this study help in the precise characterization of particle

morphology of granular materials and quantify the effects of different morphological descriptors

on the intergranular and sand-geosynthetic interactions at multiple scales.

iv
 Table of Contents

_Toc148461462Abstract………......................................................................................................i

Table of Contents..........................................................................................................................viv

List of Figures...........................................................................................................................ixviii

List of Tables............................................................................................................................xvxiv

Notations and Abbreviations....................................................................................................xvixv

Chapter 1. Introduction................................................................................................................1

1.1 Background.......................................................................................................................1

1.2 Outline of the thesis........................................................................................................76

Chapter 2. Literature Review..................................................................................................108

2.1 Introduction...................................................................................................................108

2.2 Characterization of morphology of granular materials.................................................119

2.2.1 Disadvantages of shape quantification using 2D images....................................1715

2.2.2 Watershed segmentation technique.....................................................................2018

2.2.3 Characterization of particle morphology using 3D image analysis.....................2220

2.2.4 Fractal dimension/surface roughness measurements...........................................2826

2.3 Experimental analysis.................................................................................................2927

2.3.1 Effect of particle morphology on direct shear behaviour of granular material. . .3331

2.3.2 Effect of particle morphology on interface shear tests........................................3533

v
 Table of contents

2.4 Quantification of shear induced surface changes of geosynthetics.............................4543

Chapter 3. Materials..............................................................................................................4846

3.1 Introduction.................................................................................................................4846

3.2 Granular materials.......................................................................................................4846

3.3 Geosynthetics..............................................................................................................5149

3.4 Summary.....................................................................................................................5351

Chapter 4. 3D characterization of particle morphology.....................................................5452

4.1 Introduction.................................................................................................................5452

4.2 X-ray micro-Computed Tomography (µCT)..............................................................5755

4.2.1 Fundamentals of Computed Tomography...........................................................5856

4.2.2 Micro-CT set up...................................................................................................5957

4.2.3 Description of the scanning process....................................................................6159

4.2.4 Scanning Artefacts...............................................................................................6260

4.3 Particle morphology of sand particles using 3D image analysis................................6260

4.3.1 Micro-CT Image acquisition and processing.......................................................6260

4.3.2 Reconstruction of 3D particle surface.................................................................6866

4.3.3 Three-dimensional morphological descriptors of sand particles.........................7270

4.4 Results and Discussions..............................................................................................8886

4.5 Summary...................................................................................................................10098

vi
 Table of contents

Chapter 5. Effects of 3D Morphological Contrasts on Intergranular Interactions and Shear

Strain Fields in Sands............................................................................................................103101

5.1 Introduction.............................................................................................................103101

5.2 Testing programme.................................................................................................106104

5.2.1 Global behaviour.............................................................................................109107

5.2.2 Local responses................................................................................................112110

5.3 Summary.................................................................................................................122120

Chapter 6. Effect of Particle Size of Sand on Sand-Geosynthetic Interfaces....................124122

6.1 Introduction.............................................................................................................124122

6.2 Experiments............................................................................................................127125

6.3 DIC analysis of sand-geosynthetic interfaces.........................................................138136

6.4 Micro-topographical analysis of sheared geomembrane........................................144142

6.5 Summary.................................................................................................................151149

Chapter 7. Effect of Particle Shape on Sand-Geosynthetic Interface Shear Response.....153151

7.1 Introduction.............................................................................................................153151

7.2 Interface shear tests.................................................................................................156154

7.2.1 Results of interface shear tests.........................................................................158156

7.3 Digital Image Correlation (DIC) analysis...............................................................163161

7.3.1 Results of DIC analysis of the sand-geosynthetic interfaces...........................164162

7.4 Quantification of shear induced surface changes to geomembrane........................166164

vii
 Table of contents

7.5 Summary.................................................................................................................172170

Chapter 8. Summary and Conclusions..............................................................................174172

8.1 Introduction.............................................................................................................174172

8.2 Conclusions.............................................................................................................174172

8.3 Few recommendations for practitioners..................................................................178176

8.4 Scope for future work.............................................................................................180178

References……......................................................................................................................181179

List of Publications................................................................................................................206204

viii
 List of Figures

Fig. 1.1. Different types of geosynthetics (Koerner, 1998).............................................................2

Fig. 2.1. Visual chart for angularity measurement provided by (ASTM D2488-09a, 2009)........10

Fig. 2.2. Visual shape chart for qualitatively describing sphericity and roundness of particles

(after Powers, 1953).......................................................................................................................10

Fig. 2.3. Modified version of visual shape chart for sphericity and roundness (after Cho et al.,

2006)..............................................................................................................................................11

Fig. 2.4. Three characteristic scales of particle shape...................................................................23

Fig. 2.5. Schematic diagram of conventional and modified direct shear apparatus......................29

Fig. 2.6. Mechanism of development of couple by shear force in conventional arrangement (a-c)

and (d-f) development of counter couple in improved arrangement (after Jewell, 1989).............29

Fig. 2.7. Schematic view of the modified interface direct shear apparatus currently in use.........33

Fig. 2.8. Different modifications for lower shear box (after Liu et al. 2009)................................35

Fig. 2.9. Definition of normalized roughness (after Tovar-Valencia et al. 2017).........................39

Fig. 3.1. Microscopic images of sands used in this study: (a) FS (b) MS (c) CS (d) AS..............47

Fig. 3.2. Scanning samples and 3D visualization of sand particles obtained using µCT..............48

Fig. 3.3. Particle size distribution curves of tested sands..............................................................49

Fig. 3.4. SEM images of the geosynthetics: (a) woven geotextile; (b) HDPE geomembrane......50

Fig. 4.1. Photograph of the Micro-CT (SkyScan 1272) used in this study...................................57

Fig. 4.2. Schematic diagram of the µCT (after S. C. Lee et al., 2003)..........................................58

Fig. 4.3. Flowchart illustrating the steps of image processing in X-ray µCT...............................61

ix
 List of figures

Fig. 4.4. Illustration of image processing steps: (a) typical slice of a 3D tomographic image (b)

image after applying 3D median filter (c) image after thresholding segmentation (d) image after

watershed segmentation.................................................................................................................63

Fig. 4.5. Comparison of watershed segmentation technique: (a) input binary image; (b)

conventional watershed segmentation (MATLAB 2019); (c) modified watershed segmentation

technique........................................................................................................................................65

Fig. 4.6. 3D visualization of sand samples after watershed segmentation....................................66

Fig. 4.7. Three-dimensional images of typical sand particles of FS sample.................................67

Fig. 4.8. 3D particle surface reconstruction of a typical AS particle (a) particle morphology

represented by voxel assembly (b) particle morphology represented by reconstructed surface

generated from SH analysis (c-d) close-up views.........................................................................69

Fig. 4.9. 3D surface reconstruction at different spherical harmonic degrees................................70

Fig. 4.10. Three characteristic scales of particle shape.................................................................71

Fig. 4.11. Flowchart of 3D shape characterization of sand particles............................................72

Fig. 4.12. Rotation of particle to make its principal axes parallel to the cartesian axes for

calculating the principal dimensions.............................................................................................73

Fig. 4.13. The shape classification systems for characterizing particle form: (a) proposed by

Zingg (1935) (b) proposed by Angelidakis et al. (2022)...............................................................76

Fig. 4.14. Comparing the two classification systems for characterizing the particle form...........77

Fig. 4.15. Calculation of geometric parameters of the particle: (a) typical sand particle with its

SH reconstructed surface mesh (b) diagram used for computations.............................................79

Fig. 4.16. Illustration of finding the convex hull: (a) 2D convex hull; (b) 3D convex hull..........80

x
 List of figures

Fig. 4.17. Illustration of finding the maximum inscribed circle of a given particle: (a) typical 2D

image of a particle; (b) particle boundary; (c) Euclidean distance transform; (d) maximum

inscribed circle...............................................................................................................................81

Fig. 4.18. Corners identified by different curvatures on an AS particle surface: (a) maximum

principal curvature (kmax) (b) minimum principal curvature (kmin) (c) mean curvature (km). unit:

mm-1...............................................................................................................................................83

Fig. 4.19. Computational geometry techniques for determining particle roundness: (a) maximum

inscribed sphere; (b) mean curvature contours for corners identification.....................................84

Fig. 4.20. Spherical harmonic descriptor (Dn) as a function of the spherical harmonic degree n in

log-log scales for MS particles......................................................................................................86

Fig. 4.21. Comparison of particle size distributions obtained using µCT images and sieve

analysis: (a) FS (b) MS (c) CS (d) AS...........................................................................................88

Fig. 4.22. Particle form characterization using Zingg plot: (a) different particle sizes (b) different

particle shapes................................................................................................................................90

Fig. 4.23. Distributions of shape parameters of tested sand particles: (a) Elongation index (b)

Flatness index (c) Sphericity (d) Convexity (e) Roundness (f) Regularity...................................92

Fig. 4.24. Correlation between shape parameters..........................................................................92

Fig. 4.25. Histograms of fractal dimension of tested sands with their fitted normal distributions.

.......................................................................................................................................................94

Fig. 4.26. Correlation between the 3D shape descriptors: (a-b) sphericity against roundness (c-d)

roundness against fractal dimension (e-f) sphericity against fractal dimension............................96

Fig. 5.1. Diagram illustrating the modified configuration of the direct shear test apparatus......106

xi
 List of figures

Fig. 5.2. Sand specimen during shearing process: (a) shear begin (u = 0 mm); (b) shear end (u =

15 mm).........................................................................................................................................106

Fig. 5.3. Repeatability of direct shear test results of MS specimens: (a) stress-displacement

response; (b) dilation response....................................................................................................107

Fig. 5.4. Particle size effect on direct shear test results: (a, c, e) stress-displacement response; (b,

d, f) dilation response..................................................................................................................108

Fig. 5.5. Particle shape effect on direct shear results: (a) stress-displacement response; (b)

dilation response..........................................................................................................................109

Fig. 5.6. Experimental set-up used in this study: (a) showing the shear box and the arrangement

for video capturing; (b) providing a closer view of the transparent shear box............................112

Fig. 5.7. Localized shear zone thickness in AS specimen: (a) variation of shear strain field; (b)

typical DIC data at 15 mm displacement.....................................................................................113

Fig. 5.8. Distribution of shear strain field in AS specimen at different time intervals (σn = 40

kPa)..............................................................................................................................................114

Fig. 5.9. (a) Distribution of horizontal displacements of AS specimen (σn = 40 kPa); (b)

distribution of vertical displacements of AS specimen (σn = 40 kPa). (unit: mm)......................115

Fig. 5.10. Variation of shear strains as a function of vertical distance from the bottom of ROI at a

horizontal displacement of 15 mm: (a) FS; (b) MS; (c) CS; (d) AS............................................116

Fig. 5.11. Correlation between particle morphology and the direct shear test results at multi-scale

level..............................................................................................................................................120

Fig. 6.1. Schematic diagram of the modified interface test apparatus for DIC analysis.............126

Fig. 6.2. Particle size effect on sand-GTX interfaces: (a, c, e) stress-displacement response; (b, d,

f) dilation response......................................................................................................................129

xii
 List of figures

Fig. 6.3. Interlocking mechanism between sand particles with different sizes and the surface

texture of geosynthetic materials: (a-c) interfaces between sand and geotextile (GTX); (d-f)

interfaces between sand and geomembrane (GMB)....................................................................131

Fig. 6.4. Particle size effect on sand-GMB interfaces: (a, c, e) stress-displacement response; (b,

d, f) dilation response..................................................................................................................131

Fig. 6.5. Surface profile of geosynthetics: (a) Woven geotextile (GTX); (b) Smooth

geomembrane (GMB)..................................................................................................................132

Fig. 6.6. Definition of Rmax and Rt (redrawn from Tovar-Valencia et al., 2018).........................133

Fig. 6.7. Experimental setup: a) video capturing arrangement; (b) modified shear box with

transparent wall............................................................................................................................138

Fig. 6.8. Measurement of shear zone thickness in FS-GTX interface: (a) shear strain field

contours; (b) typical DIC data for measuring shear zone thickness............................................139

Fig. 6.9. Shear strains with vertical distance from the geosynthetic plane (u = 15 mm): (a) FS-

GTX; (b) MS-GTX; (c) CS-GTX; (d) FS-GMB; (e) MS-GMB; (f) CS-GMB...........................140

Fig. 6.10. Effect of particle size on shear band thickness: (a) variation of shear band thickness

with d50; (b) variation of ts/d50 ratios...........................................................................................141

Fig. 6.11. Microscopic images of sheared geomembranes: (a) sheared by FS; (b) sheared by MS;

(c) sheared by CS.........................................................................................................................145

Fig. 6.12. 3D surface topography of sheared geomembranes at different normal stresses.........146

Fig. 6.13. Zoomed view of the sheared geomembranes (a) Enlarged view of GMB; (b) groove

geometry......................................................................................................................................146

Fig. 6.14. Effect of particle size on the surface roughness of sheared geomembrane (a) tested

with FS; (b) tested with MS; (c) tested with CS..........................................................................148

xiii
 List of figures

Fig. 7.1. Shear behaviour of sand-geosynthetic interfaces (a) Stress-displacement response of

sand-GTX interfaces (b) Dilation response of sand-GTX interfaces (c) Stress-displacement

response of sand-GMB interfaces (d) Dilation response of sand-GMB interfaces.....................157

Fig. 7.2. Variation of interface friction coefficient with normal stress: (a) AS-GMB interface; (b)

MS-GMB interface......................................................................................................................161

Fig. 7.3. Shear strains as a function of vertical distance from the geosynthetic outer surface at a

horizontal displacement of 15 mm for interfaces (a) AS-GTX; (b) MS-GTX; (c) AS-GMB; and

(d) MS-GMB...............................................................................................................................163

Fig. 7.4. Micro-topographical study of typical geomembrane samples (a) virgin geomembrane;

(b) after shearing with MS particles at 80 kPa; and (c) after shearing with AS particles at 80 kPa.

.....................................................................................................................................................166

Fig. 7.5. 3D surface topography of sheared geomembranes at different normal stresses...........167

Fig. 7.6. Correlation between the 3D shape descriptors and the tested sand-geosynthetic

interfaces at multi-scale level......................................................................................................168

xiv
 List of Tables

Table 1.1 Types of geosynthetics based on their functions (Koerner, R.M. 2012)........................2

Table 2.1 Three-dimensional sphericity formulae available in literature.....................................24

Table 2.2 Three-dimensional roundness formulae available in literature.....................................25

Table 3.1 Physical properties of tested sands...............................................................................50

Table 3.2 Properties of geosynthetics...........................................................................................51

Table 4.1 Available formulae for computing particle form..........................................................75

Table 4.2 Particle size distribution parameters obtained through µCT image analysis and sieving.

.......................................................................................................................................................88

Table 4.3 Statistics of 3D shape descriptors of sand particles......................................................94

Table 5.1 Peak and residual friction angles in direct shear tests.................................................110

Table 5.2 Shear zone thickness for the tested sands...................................................................117

Table 6.1 Peak and residual friction angles................................................................................135

Table 6.2 Shear zone thickness for the tested sand-geosynthetic interfaces...............................141

Table 6.3 Surface roughness parameters of sheared geomembranes..........................................148

Table 7.1 Peak and residual friction angles in direct and interface shear tests...........................159

Table 7.2 Shear zone thickness for the selected sand-geosynthetic interfaces...........................163

Table 7.3 Surface roughness parameters of virgin and sheared geomembrane specimens........167

xv
 Notations and Abbreviations

Notations
(Basic SI units are given in parentheses.)
a, b, c Major, intermediate and minor dimension of the particle (mm)

m
an Spherical harmonic coefficients

µ Coefficient of friction

C 3D Convexity

Cc Coefficient of curvature

Cu Coefficient of uniformity

d10 Effective particle size (mm)

d30 Particle size at 30% finer (mm)

d50 Mean particle size (mm)

Dr Relative density (%)

Dn Spherical harmonic descriptor

Ef Frictional efficiency parameter

emax Maximum void ratio

emin Minimum void ratio

e0 Initial void ratio

f (θ, 𝜙) Polar radius of the particle (mm)

xvi
 Notations and abbreviations

H Hurst coefficient

kmax, kmin, km Maximum, minimum, and mean principal curvature (mm-1)

l Sampling length (mm)

ln Spherical harmonic frequency

m Spherical harmonic degree

n Spherical harmonic order

rins Radius of maximum inscribed sphere (mm)

m
Pn Associated Legendre functions

R 3D Roundness

Ra Average profile roughness (µm)

Rmax Maximum peak to valley height (µm)

Rn Normalized roughness parameters

S 3D sphericity

Sq Root mean square surface roughness (µm)

Sz Maximum peak to valley height (µm)

V Volume of the particle (mm3)

Vhull Volume of convex hull (mm3)

m
Y n (θ , ϕ ) Spherical harmonic function

xvii
 Notations and abbreviations

θ Latitudinal coordinates (degrees)

ϕ Longitudinal coordinates (degrees)

δ Interface friction angle (degrees)

σn Normal stress (kPa)

τ Shear stress (kPa)

ϕ Friction angle (degrees)

ϕp Peak friction angle (degrees)

ϕr Ultimate/Residual friction angle (degrees)

γ Shear strain

xviii
 Notations and abbreviations

Abbreviations

2D Two Dimensional

3D Three Dimensional

AFM Atomic Force Microscopy

AS Angular Sand

CS Coarse Sand

CT Computed Tomography

DEM Discrete Element Method

DIC Digital Image Correlation

DIA Digital Image Analysis

DSA Direct Shear Apparatus

DST Direct Shear Test

IDST Interface Direct Shear Test

FD Fractal Dimension

FS Fine Sand

GMB Geomembrane

GTX Geotextile

SLDST Symmetrical loading Direct Shear Test setup

xix
 Notations and abbreviations

MATLAB Matrix Laboratory

MS Medium Sand

PCA Principal Component Analysis

PIV Particle Image Velocimetry

PSD Power Spectral Density

SA Surface Area

SEM Scanning Electron Microscope

SH Spherical Harmonics

TPM Triangular Prism Method

xx
Chapter 1. Introduction

1
 Introduction

1.1 Background

Natural materials like sands possess complicated multi-scale physical structure, with each

particle being unique in terms of its size and shape characteristics. Sands can manifest aThe

diverse array of variations in the physical features of sands, characterized by the distinct shapes

and textures of particles, which are influenced bythe outcome of their geological origin and the

processes by which they are deposited. The process of sand formation commences with the

mechanical and chemical weathering of rocks, leading to the gradual fragmentation of larger

particles into smaller constituents. These aforementioned fragments are subsequently transported

through a variety of geological processes, which exert a substantial influence on the shaping and

texturing of the sand particles. Historically, the size of sand particles is determined through sieve

analysis using limited range of sieve sizes, thus getting single-scale one-dimensional

approximate information about the geometric dimensions of the particles. While most of the

studies and designs in geotechnical engineering completely ignore the shape of the sand

particles, some of them limit themselves to qualitative terms like rounded, subrounded and

angular to describe the overall shape of the particles.

2
 Introduction

Particle morphology, which collectively describes the particle size and shape,

significantly influences the interparticle and intraparticle interactions in geotechnical systems,

thereby affecting their overall response. Analysis of particle morphology at different length

scales provides important clues to the mechanical behaviour of granular materials, as suggested

by earlier researchers. The grain-level particle kinematics like rolling, sliding, and interlocking

are primarily controlled by the particle size and shape. In this sense, it is crucial to accurately

describe and measure the size and shape of the grains using suitable definitions of morphological

descriptors. Most of the research up until this point has analyzed particle shape in a two-

dimensional (2D) framework and size in a one-dimensional (2D) framework. With the advent of

image-based measurements and robust computational tools, many researchers of current

generation are developing accurate three-dimensional (3D) quantification techniques for particle

morphology and analyzing its effects on different mechanical processes that rely heavily on

particle kinematics. In this context, this thesis presents a new framework based on integrated

image analysis for the comprehensive multi-scale 3D quantification of sand particle morphology

and understands the intergranular interactions of sand particles and interactions of sand-

geosynthetic interfaces during shear, in the light of the multi-scale particle morphology.

3
 Introduction

Geosynthetics, which are a class of polymer-based materials, are extensively utilized in

various geotechnical engineering contextssystems, including but not limited to drainage systems,

reinforcement of earth structures, slope stabilization, and the lining of reservoirs and landfills.

These materials are durable, affordable, and easy to use. Geosynthetics are frequently utilized in

combination with soil or granular particles to create composite structures that contribute to the

collective mechanical characteristics, offering enhanced strength, stability, and performance to

diverse construction projects. Sand is extensively utilized most commonly used in various

construction activities, such as geosynthetic structures, due to its commendable frictional

characteristics and superior drainage attributes. Geosynthetics exhibit a diverse array of shapes

and forms, as depicted in Fig. 1.1, which correspond to their designated applications and

functions. Table 1.1 provides an overview of the various geosynthetic types, describing their

distinct functions, applications, and objectives.

Fig. 1.1. Different types of geosynthetics (Koerner, 1998).

4
 Introduction

Table 1.1 Types of geosynthetics based on their functions (Koerner, 1998)

Geosynthetics
Separation Reinforcement Drainage Filtration Containment
type

Geotextile
   
Geogrid

Geonet

Geomembrane

Geosynthetic

clay liner

Geofoam

Geocells
 
Geocomposite
    

Sand is extensively utilized in various construction activities, such as geosynthetic

structures, due to its commendable frictional characteristics and superior drainage attributes.

Understanding the interaction between sand and geosynthetic interfaces at various scales is

necessary for the efficient designing design of geosynthetic reinforced soil structures. The

achievement of this objective can be facilitatedThis can be achieved only through the precise

characterization of sand an examination of particle morphology and quantification of its effects

on the interfacial shear responsean evaluation of the geosynthetic surface's response to shear

forces. Sands can manifest a diverse array of variations, characterized by distinct shapes and

textures, which are influenced by their geological origin and the processes by which they are

5
 Introduction

deposited. The process of sand formation commences with the mechanical and chemical

weathering of rocks, leading to the gradual fragmentation of larger particles into smaller

constituents. The aforementioned fragments are subsequently transported through a variety of

geological processes, which exert a substantial influence on the shaping and texturing of the sand

particles. The challenge of defining and quantifying the size and shape descriptors for complex

particle morphologies has led forced many previous studies researchers to either completely

ignore these descriptors or only assess them these descriptors only within a two-dimensional

(2D) framework. Historically, the assessment of particle morphology has relied on the

utilization of 2D projected images acquired through scanning electron microscopy (SEM),

optical microscopy, or laser beam systems. The reliability of shape parameters derived from 2D

images is compromised due to their dependence on images obtained from particle projections in

random orientations. Therefore, the characterization of particle shape in three dimensions holds

significant importance in comprehending the true interactions of particles within sands or with

between sands and other construction materials. Particle scanning through 3D microcomputed

tomography (3D CT) and quantification of morphological descriptors using image-based

computational algorithms hold great promise in this pursuit.

The stress-strain response of sand-geosynthetic interfaces is significantly influenced by

various factors, including the shape and gradation of soil particles, density, effective stress, and

surface properties of the contact surface. While woven geotextiles represent the class of planar

geosynthetics that have continuous and perforated surface with structured in-bult texture,

untextured geomembranes represent the smooth and impermeable geosynthetics. Understanding

the shear beaviour of sand interfacing with these two materials will be valuable to cover the

dilative and non-dilative interfaces and related particle kinematics. In order to comprehend the

6
 Introduction

sShear strength behaviour of the sand-geosynthetic interfaces can be studied through , it is

possible to employ a the direct shear apparatus with appropriate improvements. Several prior

studies have attempted to alter the direct shear apparatus to investigate the interaction

characteristics of sand-geosynthetic interfaces. The implementation of this modification was

deemed necessary in order to address the constraints inherent in conventional direct shear

apparatuses during the examination of geosynthetic-soil particle interfaces. The stress-strain

response of sand-geosynthetic interfaces is significantly influenced by various factors, including

the shape and gradation of soil particles, density, effective stress, and surface properties of the

contact surface. Geosynthetic roughness significantly affects the sand-geosynthetic interaction

mechanisms, with rougher geosynthetics enhancing the . Iinterlocking and frictional resistance

enhance shear strength on rough geosynthetic surfaces. Optical methods have become important

for precisely measuring the surface roughness of the planar geosynthetics material surface

roughness. Changes to surface roughness due to particle shearing can be These measurements

can correlated to the macroscopic shear behaviour of sand-geosynthetic interfaces macroscopic

shear behaviour. These micro-to-macro correlations are emerging as important tools for the deep

understanding of granular physics.

The onset of progressive failure at in soil-structure interfaces is accompanied by a crucial

stage of deformation occurring at the mesoscale level. This stage is distinguished by the

formation of localized shear zones that have a high concentration close to the material that

constitutes the interface. The process of defining localized shear zones offers useful insights into

the causes of failure and the overall stability of a variety of soil-reinforcement interfaces, such as

piles, nails, and geosynthetic reinforcement. Analyzing the results of the operation can help with

the acquisition of this information. A deeper comprehension of the behaviour and mechanics of

7
 Introduction

interfaces is made possible by the process of characterizing the localized shear zones. Particle

image velocimetry (PIV) and digital image correlation (DIC) are the two modern techniques

available for tracking particles that have proven to be effective inand studying localized

deformations in granular materials and at interfaces with planar continuum materials. Particle

image velocimetry (PIV) is an optical measurement technique, which is employed commonly for

the purpose of ascertaining the velocity field of a fluid flow through the analysis ofby following

the movement of tracer particles that are suspended within the fluid. In contrast, dDigital image

correlation (DIC) is a methodology reliant on imagestechnique that uses progressive imaging and

analysis , which serves to measure the displacement and strain fields of an object. This is

achieved by the comparison of digital images obtained prior to and following deformation. These

technologies enable the collection and examination of displacement, strain, and velocity fields,

thereby improving researchers' comprehension of the complex dynamics exhibited by granular

materials and their interactions with other materials. DIC can be effectively used to visualize the

particle-particle and particle-geosynthetic interactions in direct shear and interface shear tests

and gain insights on the shear zone formations, which can be related to the particle kinematics at

a micro scale and the overall strength of the assembly at a macro scale.

The assessment of shear-induced changes on geosynthetic surfaces allows provides for a

more in-depth understanding of the microscopic shear mechanisms that occur at the interfaces

between sand and geosynthetics. Researchers can gain essential insights into the underlying

mechanisms and interactions taking place at the interface by observing the surface modifications

of geosynthetic material that take place when the material is subjected to shear. The

identification of interlocking mechanisms, the measurement of frictional resistance, and the

evaluation of wear and damage patterns in the constituent materials of the interface can all be

8
 Introduction

facilitated as a result of this. Recently, non-contact optical profilometry has seen significant

developments, and it has emerged as a valuable approach tool for properly precisely measuring

the surface modifications on in geosynthetic materials. as a result of these advancements. It is

now possible, through the usage of this technology, to assess changes that are taking place on

geosynthetic surfaces with increased degrees of accuracy and precision. A more thorough

understanding of the underlying mechanisms and interactions can be achieved through a

comprehensive analysis, quantification, and correlation of the micro-scale changes on

geosynthetic materials with the overall shear behavior of sand-geosynthetic interfaces. This

objective can be accomplished by conducting a comprehensive examination of the correlation

between both of the factors.

The studies carried out in this thesis converge to provide quantitative estimations of the

effects of particle morphology on the intergranular and interfacial shear strength of sands by

developing a computational framework that integrates the micro-CT scanning, direct shear

testing, DIC analysis and optical profilometry. The insights gained through these studies are

valuable for the selection of right granular materials for geotechnical systems and to precisely

estimate the internal and interfacial shear strength parameters for the design by considering the

morphological effects. Further, the developed framework can be extended to develop advanced

constitutive models that account for particle morphology and generate precise particle

geometries for 3D printing and Discrete Element Modelling (DEM). The information supplied

has the capacity to make significant contributions to the advancement of constitutive models that

demonstrate increased precision, enhanced design requirements, and improved performance

evaluation for geotechnical constructions, specifically those involving sand-geosynthetic

interfaces.

9
 Introduction

The objective of this study is to examine and quantify the influence effects of three-

dimensional (3D) morphological characteristics of sand particles on their internal and interfacial

interactions shear behaviourbetween sand and geosynthetic materials. The scope of the work

includes the following.

1. 3D characterization of morphological descriptors of four different sands using micro-CT

scanning and Spherical Harminic (SH) analysis.

2. Analyzing the interdependency of morphological descriptors through statistical analyses.

3. Understanding and quantifying the effects of particle morphology on intergranular

interactions in sand through direct shear tests.

4. Understanding and quantifying the effects of particle morphology on sand-geosynthetic

interactions through interface direct shear tests, DIC and optical profilometry.

The objective will be accomplished through the utilization of a modified direct shear

apparatus, which enables analysis across multiple scales. Furthermore, the implementation

of profilometry techniques will be employed to assess and measure the changes to the surface of

the geosynthetic material caused by shear forces.

[1.2] Outline of the thesis

This thesis is divided into eight chapters.

The first chapter of the thesis provides a comprehensive introduction to the topicthe background

for the conceptualization of the study and outlines the goals and objectives of the research.

Chapter 2 of the thesis covers a comprehensive examination of the existing body of review of

literature, with a particular specific focus on research studies pertaining to the experimental and

computational analysis of particle morphology, direct shear experiments conducted on sandstests

10
 Introduction

on sands and sand-geosynthetic, interfaces between sand and geosynthetics, and the

measurement and imaging techniques for the and assessment of damage caused by shear forces

ontexture and surface changes contact surfaces. This Chapter also brings out the importance of

particle morphology for understanding the mechanical response of granular materials, in the light

of published studies. Taking into account prior research approaches, the chapter explores

experimental evidence that highlights the significance of particle size and shape in the

mechanical properties of granular materials and sand-geosynthetic interfaces. In addition, the

research thesis examines the underlying motivation for conducting the study and the identified

gaps in the existing literature are summarized in this Chapter, providing the motivation and basis

for formulating the research goals.

In Chapter 3 provides the description of physical and mechanical properties of the sands and

geosynthetic materials used in this study. , an extensive analysis is presented regarding the

characteristics of the granular materials and geosynthetics utilized in the context of this study.

In Chapter 4, a brief introduction is presented on X-ray micro-computed tomography (µCT) and

its scanning methodology for the acquisition of three-dimensional (3D) images of sand

specimens. In addition, the chapter introduces a computational methodology for quantifying the

dimensions and morphology of particles across various spatial length scales in a three-

dimensional contextspace. Moreover, aThe Chapter concludes with the statistical analysis is

performed on a set of 3D morphological descriptors, which to demonstrated that these

descriptors exhibit interdependence rather than independence.

Chapter 5 is dedicated topresents the investigations of the impact of 3D particle morphology on

intergranular interactions and shear strain fields in sands. The chapter covers the carrying out of

11
 Introduction

direct shear tests on four distinct types of sands, each exhibiting unique morphological

characteristics and . Tthe quantification of the localized shear zone is accomplished through the

utilization of theusing digital image correlation (DIC)DIC technique. An attempt is undertaken to

establish a A correlation is established between the morphology of particles and the global and

local responses exhibited by sands when subjected to shear forces.

Chapter 6 demonstrates the influence of particle size and surface asperities of geosynthetic

materials on the interactions between sand and geosynthetic materials across various scales. The

present study includes the execution of aA sequence of interface shear tests and their results are

presented along with the, examination ofanalysis of shear bands, and micro-topographical

analysis of the geosynthetic surfaces.

In Chapter 7 presents the results of interface direct shear tests aimed to examine , the

examination is conducted on the impact of particle shape on the shear characteristics of sand-

geosynthetic interfaces, as well as the occurrence of localized shear zones at said these

interfaces. The chapter primarily focuses on the examination of tTwo sand samples exhibiting

comparable particle size distributions yet differing in terms of their shapes are used in these tests,

with an objective of eliminating particle size effects on the results. The main aim of this study is

to enhance our comprehension of the specific impact of particle shape on the shear behaviour of

the interface and the resulting surface alterations observed at the sand-geosynthetic interface.

Chapter 8 offers presents a comprehensive summary of the research undertaken in this thesis and

presents the drawnlists the major conclusions drawn based on the obtained findings. Moreover,

Further, this chapter provides valuable insights, innovative ideas, and pertinent recommendations

for potential avenues of future research within this field.

12
 Chapter 2. Literature Review

2.1 Introduction

This chapter provides a thorough review of previous research that is relevant to the present study.

The main aim of this study is to investigate the multiscale behaviour of sands and sand-

geosynthetic interfaces, focusing specifically on the influence of particle morphology. To

achieve the stated purpose, this work employs X-ray micro-computed tomography and conducts

spherical harmonic analysis to characterize the morphology of particles. Furthermore,

experiments are conducted to evaluate the behaviour of sands and sand-geosynthetic interfaces

using a symmetrical loading direct shear apparatus. In addition, the study used profilometry to

quantify the small changes taking place on geosynthetic surfaces after interface tests. As a

result, an extensive review was carried out on the currently available academic literature to

determine its applicability to the present study and to identify any gaps or limitations within the

current research landscape.

The shape and surface topography of granular particles are of utmost importance in

understanding the mechanical behaviour of these materials and their interactions with planar

continuous materials, such as steel or geosynthetics. Therefore, it is crucial to accurately define

the morphology of particles to obtain a thorough understanding of the various shearing

mechanisms that govern the overall macroscopic behaviour. Many studies have employed

traditional visual charts or digital imaging methods that rely on two-dimensional (2D) depictions

of sand particles. The investigation of the three-dimensional morphology and surface properties

of granular particles has, however, only been undertaken in a few studies.

13
 Literature review

The direct shear test is the preferred method for studying the engineering characteristics of

interfaces between sand and geosynthetics. Several researchers have acknowledged the

limitations and flaws of conventional direct shear apparatus when looking at sand-geosynthetic

interfaces. As a result, they have made appropriate modifications to overcome these

shortcomings and improve the apparatus for investigating such interfaces. The primary emphasis

of this study is to develop a fundamental understanding of the interactions between sand particles

and between sand and geosynthetic materials at multiple scales. To achieve this, various types of

sands with distinct morphologies and geosynthetics with different surface characteristics are

utilized and investigated.

This chapter critically examines the recent studies on assessing surface changes in geosynthetic

materials caused by shear forces and their correlation with the overall stress-strain behaviour of

sand-geosynthetic interfaces. Traditionally, contact-based profilometry techniques like stylus

profilometers have been used to evaluate the damage and wear on contact surfaces. However,

recent research has shifted towards non-contact-based profilometers, which offer greater

accuracy and efficiency in assessing such damages. The literature review in this chapter focuses

on the characterization of particle morphology and its influence on the macroscopic, mesoscopic,

and microscopic behaviour of interactions between sand and geosynthetic materials.

2.2 Characterization of morphology of granular materials

In many different types of engineering construction, including slopes, building foundations,

embankments, tunnels, etc., sand is probably the most prevalent naturally occurring granular

material. For a very long time, geologists and sedimentologists have focused on the analysis of

particle morphology and its effects on the mechanical behaviour of granular materials. It should

be noted that the terms shape and morphology are sometimes used synonymously in literature, in

14
 Literature review

this communication thesis, we define morphology is defined as the combination of particle size

and shape.

Particle shape has been assessed using different approaches. In the past, qualitative methods

relying on visual charts were employed (Krumbein and Sloss, 1963; Powers, 1953). However, in

more recent times, the measurement of particle morphology has shifted towards quantitative

analysis using digital images. Several advanced instruments techniques are now utilized to

capture these images, including scanning electron microscopes microscopy (SEM), optical

microscopy, high definition (HD) camerasvideography, atomic force microscopy (AFM), and,

more recently, X-ray computed tomography are being utilized to capture these images,. The

accuracy of the measurement of shape descriptors depends on the quality of the images obtained.

Fig. 2.2. Visual chart for angularity measurement provided by (ASTM D2488-09a, 2009).

15
 Literature review

Fig. 2.3. Visual shape chart for qualitatively describing sphericity and roundness of particles
(after Powers, 1953).

In the past, the shape of grains was typically assessed through visual examination. A

standard, ASTM D2488-09a, even included images of bulky grains to help estimate particle

angularity based on visual observations as shown in Fig. 2.2. Several studies have presented

visual comparison charts that allow forhelp in the assessment of roundness and sphericity values

of particles geometry. Qualitative descriptors obtained through various shape charts provide a

general idea of particle shape, including the terms such as elongation, flatness, and irregularity

(Blott & Pye, 2008; Sneed & Folk, 1958; Zingg, 1935). A variety of shape diagrams have been

proposed by researchers to qualitatively assess the morphological attributes of particles through

the collective plotting of their dimensional ratios. Powers (1953) created a graphical

representation known as a shape chart to facilitate the visual appraisal and comparison of the

roundness and sphericity characteristics of sand grains (see Fig. 2.3).

16
 Literature review

Fig. 2.4. Modified version of visual shape chart for sphericity and roundness (after Cho et al.,
2006).

Krumbein & Sloss (1963) developed a shape chart that established a framework for

assigning predetermined particle form values to images. This system enabled the objective

evaluation of roundness and sphericity by quantitative analysis. The assessment of circularity

was performed utilizing Wadell's theoretical framework, while the determination of sphericity

was achieved by computing the ratio of the particle's length to its width. In their study, Cho et al.

(2006) proposed a modified definition of sphericity as shown in Fig. 2.4. This revised definition

quantifies sphericity as the ratio of the radius of the largest inscribed circle to the radius of the

smallest inscribed circle within the particle. Zhou et al. (2018) used the concept of "regularity"

(ρ) in their research, which combines the quantification of both roundness and sphericity. The

earlier methods for determining the roundness of sedimentary particles can be attributed to the

research conducted by Wentworth (1919). According to Wentworth, the concept of roundness

involves the computation of the ratio between the radius of curvature of the most convex portion

17
 Literature review

of a particle's surface and half of the longest diameter that intersects that particular point.

However, Wadell (1932) presented a critique of this specific concept by emphasizing that the

corners of a cube, for instance, exhibit a radius of curvature of zero, despite being considered the

most convex areas. Blott & Pye (2008) argue that the definition suggested by Wentworth may

not comprehensively capture the spherical aspects of a particle. The idea of 2D roundness refers

to the average ratio between the curvature radius of individual corners and the radius of the

greatest inscribed circle within a given particle, as discussed in the studies conducted by Wadell

(1932) and Blott & Pye (2008). Many researchers have adopted this alternate definition in order

to quantitatively determine the roundness of particles by utilizing two-dimensional images of

sands obtained from in single or multiple orientations.

The characterization of particle shapes has seen a significant transformation with the

development and use of advanced image analysis tools. This change has resulted in a shift away

from manual measurements and a transition towards more quantitative approaches. The current

procedure involves the acquisition of digital images and the application of diverse image

processing and analysis techniques, such as binarization, segmentation, and feature extraction, to

determine the geometric properties of particles. The utilization of digital image analysis

techniques allows for the automatic measuring of shapes without the need for personal

intervention, a requirement that was present in classical shape descriptors (Vangla et al., 2018; J.

Zheng & Hryciw, 2015). The utilization of image-based shape characterization offers advantages

in terms of efficiency and dependability. However, the analysis of numerous particles inside a

soil sample poses a notable serious challenge. The challenges associated with this task include

the extraction of geometry of individual particles from images of particle assemblies and the

determination of an optimal minimum resolution. Ghalib & Hryciw (1999) were early

18
 Literature review

researchers who addressed this issue by employing the watershed segmentation technique to

effectively separate and recover interconnected particles. Nevertheless, the watershed

segmentation technique frequently encounters the issue of over segmentation, resulting in the

erroneous classification of a single particles as several distinct entities. In order to address this

drawback, researchers have undertaken various attempts to improve the watershed segmentation

methodology. In their study, Zheng & Hryciw (2016) put out a change that incorporated a

threshold value to regulate the extent of overlap observed among adjacent particles. This

modification facilitates the segmentation of particle interactions, leading to enhanced precision in

the overall analysis. Another approach introduced for shape characterization of a large number of

particles is Dynamic Image Analysis (DIA). An example of such a system is the QICPIC

imaging system (Sympatec, 2008), which is based on DIA principles. In this system, high frame

rate cameras capture images of particles as they fall through a fall shaft (Altuhafi et al., 2013).

The advantage of the DIA system is that it enables imaging of particles in random orientations,

unlike static imaging systems where particle projections are captured when they are at in their

maximum stabilitystable orientations. The QICPIC system quicky provided provides the shape

parameters of thousands of particles like sphericity, convexity and particle size for thousands of

particles. However, this device is unable to giveincapable of computing the Wadell’s roundness,

of the particles which is an important shape parameter of the particles. In their work, Zheng &

Hryciw (2015) presented a comprehensive algorithm specifically designed for calculating the

roundness of sand particles in a two-dimensional (2D) framework. Their algorithm incorporates

a local weighted regression to eliminate surface roughness, identifies particle corners, and fits

circles to these corners by minimizing the distance between the corner points and the circles.

Vangla et al. (2018) also employed a similar algorithm to obtain 2D roundness values for sand

19
 Literature review

grains. Furthermore, machine learning and deep learning techniques have been increasingly

utilized to derive shape descriptors from input images of individual particles or particle

assemblies. Kim et al. (2022) and Zheng et al. (2022) have explored the application of these

techniques. They compared their generated descriptors with mathematically computed shape

descriptors and found good agreement, except for roundness, as roundness values may not

always be deterministic. In a recent development, Zheng et al. (2022) introduced a smartphone

application based on machine learning tools called "Laboratory on a smartphone" based on

machine learning tools. This application can classify granular soils into six categories (very

angular, angular, subangular, subrounded, rounded, and well-rounded) according to Powers'

classification (1953). The classification is performed using images captured by a smartphone

camera, demonstrating the potential for utilizing machine learning in soil analysis.

2.2.1 Disadvantages of shape quantification using 2D images

Although vVisual comparison charts , although providingprovide a qualitative understanding of

particle shape, they lack the ability to provide accurate and quantitative measurements. These

charts heavily rely on individual observations and are subject to variation among different

observers. In contrast, computer-based image techniques offer a more efficient and relatively

accurate method for morphological quantification. By utilizing image analysis and computational

methods, 2D shape descriptors can be measured using 2D projections of particles (Alshibli &

Alsaleh, 2004; Altuhafi et al., 2013; Mora & Kwan, 2000; Pillai & Latha, 2022; Sukumaran &

Ashmawy, 2001; Vangla et al., 2018). These 2D shape parameters are derived from images

obtained through random particle projections, but this approach can yield inaccurate results. The

2D image of an irregular particle depends on the angle at which the image is taken. Shape

descriptors obtained through 2D projections will not give erroneous results if the particle is truly

20
 Literature review

spherical which is rare for soil particles. Even today, shape characterization is mainly carried out

using 2D projections of images since 3D imaging systems are expensive, hardly accessible and

complex compared to 2D methods. Most of the 2D shape characterization studies in literature

used images taken while the particle is resting on its plane of maximum stability or the plane of

maximum dimension. Such measurements miss on all other orientations for particle shape

characterization. Even when multiple orientations are used for image capturing, the mean shape

parameters obtained through the analysis of these images cannot accurately represent the overall

shape of the particle. Jia & Garboczi (2016) highlighted the constraints associated with shape

analysis that relies on two-dimensional projections of particles. The shape of particles can vary

significantly depending on the viewing angle, as demonstrated by using a cylinder as an

example. Different angles of view can result in the object being perceived as a sphere, disc, cube,

rectangular plate, or spherocylinder. To obtain reliable quantitative measurements of particle

shape, more advanced and precise methods are required.

Researchers have attempted to overcome this problem by designing camera systems with the

ability to capture particle images from many different angles. As an example, Arasan et al.

(2011) devised an experimental arrangement using a rotational camera system, enabling the

capture of both top and front perspectives of particles. In their study, Kuo et al. (1996) employed

transparent trays that were utilized to adhere to particles, facilitating the recording of images

from two orthogonal orientations. The University of Illinois Aggregate Image Analyzer (UIAIA)

was developed by the University of Illinois which utilizes three digital cameras that are

synchronized to capture front, side, and top views of particles in motion on a conveyer belt.

According to Rao & Tutumluer (2000), this technology offers a comprehensive imaging capacity

from various perspectives, hence thus improving the analysis of particle properties. Researchers

21
 Literature review

have also directed their attention towards the development of sophisticated methodologies for

quantifying shape attributes through the utilization of half particle geometry. These methods

present a more advanced approach in contrast to two-dimensional (2D) methods and serve as an

economically feasible substitutes for three-dimensional (3D) procedures. By employing the

concept of half particle geometry, these methodologies offer supplementary insights into the

morphology of particles, while simultaneously achieving an appropriate balance between

simplicity and efficacy. These developments enable researchers to acquire vital insights into the

characterization of shape, while simultaneously maximizing the resources and lowering the

complexity. Kim et al. (2022) recently conducted a study wherein they proposed an innovative

aggregate scanning system that integrates lasers and cameras for the purpose of capturing images

of particles that are dispersed over a horizontal platform. By integrating the collected data points,

the system derives the half-particle geometry, enabling accurate shape analysis. Notably, these

methods enable the capture of the third dimension of the particles by utilizing the exposed half-

particle geometry within the camera's field of view. Hence, these methodologies can be

categorized as "2.5D measurements," as proposed by Zheng et al. (2020). Zheng et al. (2020)

explored the determination of 3D shape parameters using 2.5D methods and conducted a

comparison comparative study. They found that, except for convexity, shape parameters such as

roundness and sphericity exhibited reasonably good agreement between the 2.5D and 3D

approaches for two different geometries. While 2.5D methods offer cost-effectiveness by

eliminating the need for advanced equipment like X-ray micro-computed tomography (µCT), but

they do not provide results with the same level of accuracy and precision as 3D methods. Some

other cost-effective alternatives for characterizing 3D particle geometry have been proposed in

the literature, including include photogrammetry (Paixão et al., 2018; L. Zhao et al., 2021) and

22
 Literature review

the reconstruction of 2D planar projections obtained through a digital camera at incremental

rotations (Nadimi & Fonseca, 2017). Photogrammetry techniques require good lighting

conditions and are limited to analyzing particles of a certain size range due to their lower

resolution capabilities. On the other hand, the method proposed by Nadimi & Fonseca (2017)

only allows for scanning of only one particle at a time, making it time-consuming. Furthermore,

these methods primarily provide information about the surface characteristics of the particles and

do not offer insights into their internal structures.

The recent development of X-ray micro-computed tomography (µCT) has significantly

contributed to the advancements in 3D shape measurements. This technique enables the

visualization and precise measurement of particle morphology by capturing both the external

surface and internal structure of particles. Micro-CT is a non-destructive imaging method that

involves placing the sample on a rotating stage positioned between an X-ray source and a

detector. As the specimen rotates incrementally, the detector collects multiple projections of the

specimen. These projections are then reconstructed using a back projection algorithm to generate

a detailed 3D grayscale image. The variation in grayscale intensity within the image is indicative

of the density and material composition of the particles since X-ray energy attenuates more in

denser materials compared to less dense materials. Micro-CT has found numerous applications in

granular mechanics, including the morphological characterization and visualization of sand

grains (Fonseca et al., 2012; D. Kong & Fonseca, 2018; W. Zheng et al., 2021; B. Zhou et al.,

2018), studying the evolution of initial and shear-induced fabric (Andò et al., 2012, 2013), and

pore characterization in granular media (Al-Raoush & Alshibli, 2006; Druckrey et al., 2022). An

important development in µCT technology is the introduction of synchrotron µCT, which utilizes

intense beams to produce high-quality images with improved contrast and reduced noise between

23
 Literature review

different materials (Alshibli et al., 2015; B. Zhao & Wang, 2016). However, it is worth noting

that access to synchrotron µCT is restricted to specific research groups due to its limited

availability. In contrast, commercial µCT systems, also known as lab µCT, are more accessible

and widely used by researchers.

2.2.2 Watershed segmentation technique

To get the 3D voxel representation of particles from grayscale µCT scans, usually several steps

of image processing are needed. In the first step, the raw µCT data undergoes filtration using

methods such as Gaussian or median filters in order to reduce noise and enhance the quality of

the images. Subsequently, the filtered images undergo a transformation known as binarization

when they are turned into binary images. This stage entails the selection of a suitable threshold

value for the purpose of distinguishing the sand grains from the surrounding medium. Otsu's

thresholding (Otsu, 1979) is a widely employed technique in the field of image processing for the

purpose of binarization. This technique is favoured due to its ability to decrease the variance of

pixel intensities within each class. After acquiring the binary images, segmentation techniques

are employed to recover and separate the geometries of individual particles from the scanned

particle assembly. The utilization of segmentation methods facilitates the identification of the

boundaries of individual particles inside the voxel assembly, hence enabling subsequent analysis

and characterization of their respective shapes and properties. The accurate separation of particle

interactions while preserving the integrity of their outer edges presents a significant challenge in

the extraction of individual particle morphology.

In order to tackle this difficulty, researchers adopted alternative methodologies. One

methodology of them is the physical separation of particles and their subsequent immobilization

in a stationary state prior to the scanning process. This objective can be accomplished through

24
 Literature review

the incorporation of particles inside a resin matrix or the utilization of transparent plastic sheets.

Alternatively, the dispersion of particles in silicone grease or silicone oil can also be employed.

The variation in X-ray attenuation rate between the particles and the surrounding medium

facilitates the avoidanceeliminates the need for of complex segmentation approaches in

following subsequent stages of analysis. Nevertheless, it is important to acknowledge that these

methodologies possess certain constraints, primarily in terms of their capacity to effectively scan

only a limited quantity of particles concurrently. Moreover, the process of extracting particles

from the embedding medium, such as resin or silicone, poses significant difficulties and might

potentially lead to sample degradation or modification. As a result, the non-destructive aspect of

the approach is gets somewhat compromised.

To address the challenge of over-segmentation in watershed segmentation, researchers have

modified or improved the existing techniques. Watershed segmentation has been widely adopted

for more precise segmentation of contacting particles in various studies (Alam & Haque, 2017;

Shi & Yan, 2015). However, over-segmentation, which occurs when a single particle is

segmented as multiple particles, can lead to inaccuracies in size and shape analysis during

computational image processing. In order to mitigate over-segmentation issues, several

researchers, such as Kong & Fonseca (2018) and Sun et al. (2019), have proposed modifications

to the watershed segmentation technique. Kong & Fonseca (2018) introduced an adaptive

watershed segmentation method that utilizes iterative refinements to progressively improve

segmentation accuracy. In each iteration, under-segmentation is allowed, as under-segmented

particles can be segmented again in subsequent iterations, while it is not possible to recover an

over-segmented grain. Sun et al. (2019) developed an improved watershed algorithm based on an

extension of the 2D watershed technique developed by J. Zheng & Hryciw (2016). This method

25
 Literature review

effectively addressed the over-segmentation issue, particularly for highly irregular contacting

particles. These modified watershed segmentation approaches proposed by Kong & Fonseca

(2018) and Sun et al. (2019) have demonstrated precise results across a wide range of particle

shapes and sizes.

2.2.3 Characterization of particle morphology using 3D image analysis

Once the voxel assembly of each particle is obtained through various image processing

techniques, the shape is characterized using image analysis techniques. Simple algorithms

applied to the component voxels of particles allow for easy determination of volume, principal

dimensions, and aspect ratio. However, voxel-constructed surfaces derived from µCT images

often exhibit saw-tooth patterns and overlapping voxels due to high-order connectivity on the

surface. These irregular surfaces lead to an the overestimation of the actual surface area and

make it impossible to accurately determine parameters such as sphericity, surface curvature, and

roundness. To achieve precise quantification of particle geometry, another group of researchers

has suggested reconstructing the particle surface from the µCT images. Several smoothing

algorithms have been proposed by researchers to remove the saw-tooth patterns and preserve the

morphological features of particles in the 3D reconstructed image. For instance, Mollon & Zhao

(2014) proposed the creation of virtual sand particles using fractal geometry, generated from

three 2D closed grain contours in orthogonal directions, to interpolate and generate the 3D grain

geometry. However, this method may miss local features of highly irregular particles. Other

researchers, such as Zhao & Wang (2016) and Lin & Miller (2005), employed the Marching

Cube (MC) algorithm to reconstruct a particle surface composed of triangular surface meshes.

However, removing artificial stair-steps on the surfaces through the MC algorithm proved to be

challenging, necessitating additional mesh simplification techniques to achieve a reasonably

26
 Literature review

smooth surface. These methods can be time-consuming and computationally demanding,

especially when dealing with a large number of particles. While the triangulated surface obtained

through the Marching Cube (MC) algorithm is an improvement over voxel-based approaches, it

still has limitations. The surface is not completely continuous and differentiable, which can lead

to inaccuracies in the calculation of surface curvatures. A more sophisticated approach, based on

spherical harmonic (SH) analysis, was found to be effective in characterizing 3D particle

morphology and reconstructing a smooth and continuous particle surface. This method,

compared to others, offers better surface smoothness and accuracy in capturing the intricate

details of particle shapes.

Equation (2.1) describes how the polar radius in the SH analysis is expanded from a unit

sphere to a particle profile made up of vertices (x, y, z)

∞ n
f ( θ , ϕ ) =∑ ∑ amn Y mn (θ , ϕ ) (2.1)
n=0 m=−n

where f (θ, 𝜙) is the polar radius from the particle center, while θ ∈ [ 0 , π ] and ϕ ∈ [ 0 , 2 π ] are the

latitudinal and longitudinal coordinates, respectively; Y mn ( θ , ϕ )is the SH function deduced from

Equation (2.2); and a mn is the associated SH coefficient that has to be computed.

√
1.
( 2n+1 )( n−m ) ! m
Y mn ( θ , ϕ )= Pn (cos θ)e imϕ (2.2)
4 π ( n+ m) !

where Pmn (x) are the associated Legendre functions that may be obtained by using intrinsic

function legendre in MATLAB, and n and m are degree and order of Pmn (x), respectively. In

accordance with the precision needed for fitting, n is a non-negative integer ranging from zero to

infinity and as a result, the total number of one set of a mn is (n+1)2.

27
 Literature review

Earlier studies (Sun & Zheng, 2021; H. W. Yang et al., 2022; B. Zhou et al., 2018) have

demonstrated that the SH degree has to be more than 15 to accurately depict the regional

morphological characteristics (i.e. local roundness and surface texture). By applying the usual

least-squares estimation, the SH coefficients a mn was can be calculated, and the continuous and

smooth surface of the particle was can be reconstructed by using Equation (2.3)
nmax n
^f ( θ , ϕ ) =∑ ∑ amn Y mn (θ , ϕ ) (2.3)
n=0 m=−n

The mean spherical harmonic radius is given by Equation (2.4), which reveals that the resulting

particle shape at zero SH degrees is a sphere (Yang et al., 2022)

a00
r 00 = (2.4)
√4 π
The particle surface becomes more finely detailed when it is reconstructed at an increasing

maximum harmonicSH degree. Starting at degree 1, the reconfigured shape takes on an elliptical

shape before gradually becoming irregular as the degree increases. The particle shape has a

significant impact on the reconstruction of particles at lower harmonic degrees (Su & Yan, 2018;

Sun & Zheng, 2021; Wei et al., 2018; H. W. Yang et al., 2022; B. Zhou & Wang, 2017). The

reconstruction of the particle becomes more detailed as the maximum SH degree risesincreases,

and as a result, the reconstructed particle more closely resembles the original scanned particle.

According to Barrett (1980), particle morphology is typically described in terms of three

descriptors: form (first order approximation of shape), roundness (second order), and roughness

(third order). Sphericity is considered to be an index of form by many researchers (Clayton et al.,

2009), while some characterize particle form in terms of elongation and flatness indices

(Angelidakis et al., 2022). Fig. 2.5 illustrates the three scales of particle shape. Sphericity

(overall shape) reflects the proportions of a particle; roundness (or angularity) reflects the

28
 Literature review

sharpness of corners and roughness reflects the surface texture imposed on corners and between

corners (Barrett, 1980). Various formulae proposed by earlier researchers for quantifying

sphericity and roundness of particles are summarized in Table 2.2 and Table 2.3 respectively.

Fig. 2.5. Three characteristic scales of particle shape.

Table 2.2 Three-dimensional sphericity formulae available in literature.

Descriptor Formulae Reference, Remarks

Degree of true SA sphere Wadell (1932)
sphericity SA
SA: Surface area of the particle, SAsphere: Surface area of

a sphere having the same volume as

the particle

√
Sphericity 3 V Wadell (1933)
Vc
V: volume of the particle

Vc: Volume of the smallest circumscribing sphere

29
 Literature review

√
Krumbein 3 bc Krumbein (1941) and Masad et al. (2005)
intercept sphericity a2 a,b,c: longest, intermediate, and shortest dimension of
the particle
Sphericity Vp Alshibli et al. (2015)
Vs
Vp: Actual volume of the particle, Vs: Volume of a

sphere with diameter equals to the shortest dimension

Inscribed- D is Maroof et al. (2020)

Dcs
circumscribed Dis: Diameter of the largest inscribed sphere, Dcs:

sphere ratio Diameter of the smallest circumscribing sphere

Table 2.3 Three-dimensional roundness formulae available in literature.

Descriptor Formulae Reference, Remarks

Roundness Wadell (1932)

∑( r
R ) r: Radius of curvature of a corner, R: Radius of
N
maximum inscribed circle, N: Number of corners

Roundness (IR) Alshibli et al. (2015)

AP AP: Actual 3D surface area of the particle, dL, dI,

( d +d6 +d )
2
L I S
4π dS: Longest, intermediate, and shortest diameters
❑

of the particle

30
 Literature review

3D Roundness Zhao & Wang (2016)

index (Ri) An: Area of the nth triangular element, which is part
∑ ( ) k
An s
ki of the corners on the particle, ki: Local curvature at
∑ An this vertex, ks: The corresponding curvature of the

particle’s maximum inscribed sphere

3D Roundness Nie et al. (2018)

(R3D) Rc: Radius of spheres fitted to each corner, nc:

∑ Rc Number of corners (corners have with radius of
nc R insc
fitted sphere smaller than Rinsc), Rinsc: Radius of

the maximum inscribed sphere

3D Roundness J. Zheng et al. (2021)

n
(R3) r3, i: Radius of the ith corner or ridge sphere, N:
∑ r3 , i
i=1
N r 3 ,∈ ¿ ¿ Number of detected corner and ridge spheres, r3,in :

Radius of the maximum inscribed sphere

Angularity Yang et al. (2017)

index (AI) ∆ Di : Angle between the gradient vectors of the ith

n

∑ ∆ Di neighboring couple of the surface points on an

aggregate, in degrees, ∆ α i ∶ Angle between the

i=1
n

∑ ∆ αi
i=1

gradient vectors of the ith neighboring couple of the

surface points on an ideal sphere, in degrees

3D Roundness
R=
∑ g(k )|k max|
−1
Zhou et al. (2018)
N R ins −1
(R) 1 if |k max| < R ins
g (k )={0 if |k max|−1 ≥ Rins

31
 Literature review

g ( k ) differentiate between corners and non-

corners, kmax : Maximum principal curvature at a

vertex on the SH reconstructed surface, Rins :

Radius of the maximum inscribed sphere, N: Total

number of corners

2.2.4 Fractal dimension/surface roughness measurements

The surface irregularities of the natural sand particles can be quantified using 3D fractal analysis.

It has been proposed that the shape of irregular particles, such as natural sand grains, can be

better described using fractal geometry rather than traditional Euclidean geometry (Mandelbrot,

1977). The fractal dimension of the rough or fragmented surface of sand particles is dependent

on the degree of roughness and the type of pattern exhibited (Kaye, 1994). Researchers have

utilized fractal dimensions to measure the roughness of particles. Hyslip & Vallejo (1997)

introduced the parallel line and area perimeter method , finding the area perimeter method

simpler for quantifying particle the fractal dimensions of the particles. Guida et al. (2020)

proposed three descriptors to define particle shape at different levels of morphology using fractal

analysis on particle contours. The Triangular Prism Method (TPM), introduced by Clarke (1986),

has been commonly used to measure the fractal dimension of granular particles. Yang et al.

(2016; 2019) employed the Power Spectral Density (PSD) method, which relates fractal

dimension to surface roughness, to determine the fractal dimension of aggregate grains. Yang et

al. (2022) used the PSD method to bridge the gap between fractal dimension measurements

obtained by µCT and interferometers. They found limitations in achieving the same level of

detail during surface reconstruction using spherical harmonic (SH) expansion. To address this,

32
 Literature review

they suggested reconstructing the particle surface at the particle shape scale using a practical SH

degree and incorporating surface texture details using interferometry-derived fractal parameters.

Zhou et al. (2018) applied fractal-based methods to analyze closed 3D particle surfaces obtained

through µCT measurements. They improved the slit island method (SIM) for closed surfaces by

deriving the 3D fractal dimension from measurements of total area and perimeter of closed

regions, addressing the unsuitability of existing methods for analyzing closed surfaces. However,

to implement this method, operators need to have a strong background in image processing, and

it is difficult and computationally demanding.

2.3 Experimental analysis

Analysis of the behaviour of small soil samples in standardized laboratory experiments like the

triaxial test and the direct shear test (DST) has contributed significantly to our understanding of

granular material response. The apparatus used for the direct shear test comprises a square box

that is divided into two parts horizontally, with an upper half and a lower half. To conduct the

test, the upper portion of the shear box is moved relative to the lower portion, resulting in the

formation of a shear zone located at the middle height of the specimen. In geotechnical

engineering, DST is one of the most well-known and widely used laboratory experiments to

determine the shear strength parameters of soils. However, the test suffers several limitations

which include the non-uniformity of stresses and strains applied to the specimen as well as the

challenges in defining a failure criterion (Potts et al., 1987; Shibuya et al., 1997).

When considering the currently used Direct Shear Apparatus (DSA) worldwide, it is

possible to identify three distinct types of kinematic degrees of freedom at the top box as shown

in Fig. 2.6. The first type, known as type A, involves the independence of the top platen and the

upper part of the shear box, allowing the top platen to have vertical movement and free rotation

33
 Literature review

(Skempton & Bishop, 1950). The second type, referred to as type B, features a securely fastened

top platen to the top half of the box, enabling both parts to move vertically or rotate together

after applying the vertical stress (Jewell, 1989; Jewell & Wroth, 1987). The last type, known as

type C, restricts the vertical and rotational movement of the upper box while allowing

independent movement of the top platen. Additionally, type C configurations can also prevent

rotation of the top platen (Mikasa, 1960).

When using a free top load platen configuration (referred to as type A), the application of

horizontal shear stress results in the generation of an anticlockwise moment on the upper platen.

This leads to an uneven distribution of vertical contact stress, increasing its nonuniformity. To

maintain equilibrium, a compensating clockwise moment needs to be present (Jewell, 1989) as

illustrated in Fig. 2.7. The tilting effect also causes an undesirable fluctuation in soil density,

which further amplifies the nonuniformity of vertical stress distribution.

Fig. 2.6. Schematic diagram of conventional and modified direct shear apparatus.

34
 Literature review

Fig. 2.7. Mechanism of development of couple by shear force in conventional arrangement (a-c)
and (d-f) development of counter couple in improved arrangement (after Jewell, 1989).

Several researchers attempted to address the shortcomings by modifying the conventional direct

shear apparatus (Type A). Wernick (1979) recommended tightly fixing the upper shear box with

external support to stop both rotational and translational displacements which is referred to as

Type B. This configuration (Type B) directs the soil sample to travel exclusively vertically

during the shearing process. This arrangement proposed by Wernick (1979) was later employed

by Shibuya et al. (1997) to investigate the deformation of sand as quasi-simple shear. The

findings revealed that the friction at the vertical surfaces of the specimen significantly restricts its

dilation, resulting in significant inaccuracies when estimating the average normal stress on the

horizontal shear plane. It has been shown that to minimize the occurrence of progressive failure,

35
 Literature review

it is important to ensure that the loading platen does not rotate during testing. It was

recommended to maintain a consistent gap between the upper and lower halves of shear boxes,

approximately 10-20 times the mean particle diameter (d50). However, using a gap of this size

could have unintended effects, like sand particles leakage at the shearing plane during testing

(Simoni & Houlsby, 2006). According to Kostkanová & Herle (2012) if the vertical movement

of the upper rigid section of the shear box is restricted during the shearing process and no

adjustment is made to account for the applied vertical load, the measured shear strength can be

underestimated for contractive soils or overestimated for dilatant soils (Kostkanovà & Herle,

2012). These potential discrepancies can be avoided by employing a constant normal stiffness

Direct Shear Apparatus (DSA). The ASTM D5321 (2008) standard specifies that the horizontal

force should be corrected based on the internal resistance of the shear device. Jewell & Wroth

(1987) and Jewell (1989) introduced a direct shear test method that involves a symmetrical

arrangement (Type C). In this approach, the top load platen is securely attached to the upper

shear box, causing both the upper shear box and the load platen to move together as a unified

unit throughout the testing process and eliminates the rotation during shear. The purpose of this

approach was to achieve more consistent deformation to obtain boundary measurements that

accurately represent the behaviour of the soil. The discussion also addressed the influence of

non-uniform deformation, as well as the misalignment of principal stress axes and incremental

strain, on the interpretation of direct shear test results. It was demonstrated that using a

symmetrical loading arrangement during the peak shear resistance leads to more uniform strains

in the deforming sand. Later, Lings & Dietz (2005) adopted Jewell's symmetrical arrangement

(1989) to ensure unimpeded dilation. Additionally, they modified the method of applying

horizontal load by attaching a set of 'wings' to the upper part of the box's sides, thereby ensuring

36
 Literature review

that the load was applied closer to the center of the shear box. According to the assertion, when

the initial gap between the upper and lower halves of the box was approximately five times the

mean particle diameter (5d50), this configuration resulted in obtaining accurate data for peak

friction and dilation angles (Lings & Dietz, 2005).

2.3.1 Effect of particle morphology on direct shear behaviour of granular material

Härtl & Ooi (2008) conducted direct shear experiments on spherical and paired glass beads to

extensively investigate the influence of particle shape, stress level, and packing porosity on the

bulk friction at the limiting shear. The measured bulk friction demonstrated a significant

dependence on the initial packing porosity for both types of particles, with a stronger

dependency observed for the paired particles. Additionally, the stress dependency of the bulk

friction was found to be substantial for the paired spheres but not for the single spheres. This

suggests that the particle aspect ratio plays a crucial role in the presence of stress dependency in

bulk friction. Moreover, at the same porosity, the elongated paired spheres exhibited

significantly higher bulk friction compared to the single spheres, primarily due to a greater level

of particle interlocking.

Several researchers attempted to understand the direct shear behaviour of sands through

numerical methods considering various influencing factors (Salazar et al., 2015). In their

research, Potts et al. (1987) utilized the finite element method to analyze the direct shear test.

They employed a constitutive model that assumed linear elasticity followed by perfect plastic

behaviour. Additionally, they assumed that the dilation angle was either zero or equal to the

angle of internal shearing resistance throughout the test. The findings of the study highlighted

that the stress-strain behaviour of the material was notably affected by the assumed angle of

dilation as well as the initial stress state. Masson & Martinez (2001) presented the findings of

37
 Literature review

their two-dimensional Discrete Element Method (DEM) simulations of the direct shear test. They

utilized 1050 cylinders to create both a dense and a loose sample. By analyzing vertical profiles

of particle displacements and rotations, they demonstrated that in the dense sample, shear

deformation was concentrated within a layer located at the mid-height of the shear box. This

localized shear deformation layer had a thickness approximately five to six times the maximum

particle diameter. Interestingly, no such concentration of shear deformation was observed in the

loose sample. In the loose sample, the particle displacement profiles indicated relatively uniform

shear deformation across the entire height of the shear box, with high particle rotations randomly

distributed throughout the entire area of the box. Gu et al. (2014) conducted a study utilizing

Discrete Element Method (DEM) simulations to examine the macroscopic behaviours of

granular soil under various initial densities and confining pressures, considering both drained and

undrained conditions. They investigated how the microstructures of the soil at different initial

states, as well as their evolution during shearing, influence the overall behaviour of the sands at a

macroscopic level. Wu et al. (2021) conducted a series of biaxial tests using Discrete Element

Method (DEM) simulations to investigate how roundness affects the mechanical response of

granular materials at both macro and micro scales. The findings indicated that granular materials

with lower particle roundness exhibited higher deviatoric stress and more pronounced volumetric

dilation. Despite having higher displacement values, particles with decreased roundness

exhibited less rotation due to their better interlocking. Zhou et al. (2018) conducted numerical

direct shear tests to investigate the influence of realistic particle shapes on the macroscopic and

microscopic mechanical behaviours of granular sands. The simulations revealed that the use of

realistic particle shapes significantly enhanced the interlocking between particles rather than the

intensity of anisotropic strong contact force chains. As a result, the shear resistance and shear-

38
 Literature review

induced dilation of the sand were notably increased. Zhao et al. (2015) conducted DEM

simulations of direct shear tests to examine the mechanical behaviour of assemblies at both

macroscopic and microscopic levels. The macroscopic analysis of the specimens in the

simulations showed that assemblies with higher angularity exhibited more pronounced

interlocking, leading to increased shear strength and dilatancy. Moreover, in assemblies with

higher angularity, the distribution of normal contact force near the shear plane was observed to

be more uneven during shearing, indicating that a small fraction of particles bore the majority of

the shear force.

2.3.2 Effect of particle morphology on interface shear tests

Fig. 2.8. Schematic view of the modified interface direct shear apparatus currently in use.

Design of geosynthetic reinforced soil structures requires the knowledge of interaction behaviour

of sand-geosynthetic interfaces at multi-scale level. Geotextiles and geomembranes are widely

used in civil engineering designs, due to their desirable mechanical and hydraulic functions and

economic advantages. These geosynthetics are always in contact with soils and the soil-

geosynthetic composites are subjected to different external forces imposed by the loading

conditions of the structures. If the overall external stress exceeds the peak shear strength of the

sand-geosynthetic interface, it will result in progressive failure and intense localized shearing of

39
 Literature review

the sand-geosynthetic interfaces. To understand the sand-geosynthetic interface shear strength,

direct shear test setup can be used with suitable modifications. Various previous researchers have

modified the direct shear set-up to investigate the interaction behaviour of sand-geosynthetics

interfaces (Afzali-Nejad et al., 2017; Anubhav & Basudhar, 2010; Frost et al., 2012; Vangla &

Latha, 2015, 2016a, 2016b). Many researchers have reported the shortcomings of the

conventional direct shear setup for testing the soil-geosynthetic interfaces, which include (a)

sagging and wrinkling of geosynthetic material, (b) limited shear displacement for obtaining

residual parameters, (c) reduction in contact area between soil and geosynthetic surface during

shearing and (d) geosynthetics may slip and distort during relative shear displacement in the

absence of an appropriate support and anchoring facility. All these factors could lead to an

interface friction angle measurement that differs significantly from the actual soil-reinforcement

interface value. To overcome the shortcomings associated with conventional interface shear

apparatus, Lee & Manjunath (2000) added a rigid block to the lower half of the shear box to

support the geosynthetic, while only filling soil to the upper half of the box. This allowed for

shearing to occur between the soil and the top surface of the geosynthetic. The apparatus also

permits a constant area of soil-geosynthetic contact during shear, optimal clamping

configurations to prevent bunching of the geosynthetic owing to relative displacement, and a

significant unidirectional travel distance of up to 90 mm. Several researchers increased the size

of the lower shear box in order to have constant area between soil and geosynthetic during shear

as shown in Fig. 2.9. Interface direct shear tests can be carried out using a lower box filled with

soil that is the same size as the upper box (conventional apparatus), a bigger lower box filled

with soil (Fig. 2.9a), or a solid block (Fig. 2.9b). According to (C.-N. Liu, Ho, et al., 2009; C.-N.

Liu, Zornberg, et al., 2009) the soil support constant area test is not appropriate since the soil

40
 Literature review

outside the loading area is not contained and flows out of the lower shear box. Therefore,

investigating the interaction of the geosynthetic with soil particles is possible using rigid plate in

the lower half of the shear box. Lopes & Silvano (2010) demonstrated that the method employed

to secure the geosynthetic specimen at the DSA affects the pressure distribution along the length

of the interface at the shearing plane. It was advised to utilize an aluminium oxide abrasive sheet

that had been glued over the hard base to create an adequate connection of the geotextiles. In

reality, slippage and deformation at sand-geotextile interfaces could happen during relative shear

displacement in the absence of a strong enough anchorage system (Lee & Manjunath, 2000).

Fig. 2.9. Different modifications for lower shear box (after Liu et al. 2009).

The stress-strain response of the sand-geosynthetic interfaces is significantly influenced by the

soil particle morphology and gradation, density, effective stress, and surface characteristics of

the contact surface (Afzali-Nejad et al., 2017; Frost et al., 2012; Martinez & Frost, 2017b, 2017a;

Pillai & Latha, 2022; Uesugi & Kishida, 1986b; Vangla & Latha, 2016b, 2016a). Geotextiles

have textured surface, leading to dilative interface shear response, whereas geomembranes have

smooth surface, resulting in non-dilative shear response. Shear behaviour in dilative and non-

41
 Literature review

dilative interfaces is entirely different due to the difference in their interaction mechanisms.

Interlocking between sand particles and surface asperities of the continuum material offers shear

resistance in dilative interfaces (Indraratna et al., 2015). However, the governing mechanisms in

case of non-dilative interfaces are rolling, sliding, and ploughing (Dove & Frost, 1999). Han et

al. (2018) performed direct shear and direct interface shear tests on steel interfaces with four

various levels of corrosion, 10 different silica sands, and interfaces with three different levels of

roughness created by attaching sandpaper to the interfaces. They demonstrated how surface

roughness affects the strength of the interface shear. When the uniform sands were tested against

mildly rusted steel surfaces, a 50% increase in the interface friction angle was seen in

comparison to the smooth steel surface, however this increase was only 6-13% for the graded

sands. For a given contact, sands with smaller particle sizes and more angular or elongated

particle shapes can mobilize a higher critical-state interface friction angle. An extensive set of

interface studies between sands in contact with geosynthetic with varied surface roughness were

carried out by Lashkari & Jamali (2021). They claimed that as initial relative density and overall

particle regularity decrease, the peak friction angle also decreases. Peak friction efficiencies were

found to be smaller for smooth interfaces like geomembrane and greater for rough interfaces like

geotextile. Additionally, they noticed that after peak friction is attained, sand-geosynthetic

interfaces start to slide. The percentage of interface slip rises with particle regularity and falls

with increased normal stress. Afzali-Nejad et al. (2017) studied the impact of particle form on

sand-geotextile interfaces under a broad range of initial relative density and applied normal

stress, and they came up with a unified stress-dilation rule for sand-geotextile and sand-rough

steel interfaces. They noticed that the rounded glass-beads and woven geotextile interfaces

exhibited similar peak and dilation angles, despite having varying levels of normal stresses. This

42
 Literature review

similarity could be attributed to the interlocking mechanism between the glass beads and the

rough surface of the geotextiles, as well as the flexible characteristics of the geotextiles.

There is a general consensus that the surface roughness of the contact material affects the

interface friction and shear behaviour between the geosynthetics and adjacent materials in a

geotechnical application (Afzali-Nejad et al., 2017; DeJong & Westgate, 2009; Lashkari &

Jamali, 2021; Martinez et al., 2015; Martinez & Frost, 2017b; Uesugi & Kishida, 1986b; Vangla

& Latha, 2016a, 2016b). The significance of evaluating surface topological properties of

continuum material has resulted in the suggestion of several indices, among which average

roughness (Ra) is studied very often which is defined as the mean value of the absolute profile

height deviations within a specified evaluation length. It is calculated by measuring the

deviations from the mean line, as indicated in Equation (2.5)

√
l

∫ z ( x ) dx (2.5)
0
Ra =
l

wherein z ( x ) is the profile calculated from the mean line at x and l is the sampling length.

(Uesugi & Kishida, 1986b) conducted a series of experiments in a laboratory using a simple

shear apparatus to examine the interaction between steel and air-dried sands. They investigated

several influential factors and discovered that the type of sand and the surface roughness of the

steel had a greater impact compared to the applied normal stress and the average grain size. In

subsequent studies (Uesugi & Kishida, 1986a) they further explored the significance of particle

size and shape in determining the friction coefficient at the sand-steel interface. They identified a

strong correlation between the average grain size and the surface roughness of the steel. To

quantify this relationship, they introduced a parameter called "normalized roughness" (Rn), which

they associated with the interface friction coefficient between sand and steel surfaces. Equation

43
 Literature review

(2.6) gives the normalized roughness (Rn) of geosynthetic materials interfacing with sands with a

given mean particle size.

Rmax (l=d50 )
Rn = (2.6)
d 50

here Rmax represents the maximum difference in height between the highest peak and lowest

valley within a profile, measured over a sampling length of l = d50. Several researchers (DeJong

& Westgate, 2009; Lashkari & Jamali, 2021; Martinez et al., 2015; Uesugi & Kishida, 1986b;

Vangla & Latha, 2015, 2016a) presented substantial experimental evidence demonstrating the

impact of normalized roughness (Rn) on the frictional and volume change characteristics of soil-

structure interfaces, Tovar-Valencia et al. (2017) proposed a new definition for calculating the

normalized roughness as shown in Fig. 2.10, where Rmax ,avg is the average of all values of

maximum roughness Rmax ,i measured within a moving window (with window size Lm, and

moving step size Δx; Δx is the horizontal distance between two consecutive data points and n is

the number of measurements for Rmax ,i over the entire measurement travel length L
n

∑ R max , i (2.7)
i
Rmax ,avg =
n

Lm was set equal to d50 of contact sand in order to calculate the normalized roughness. Ra is

merely a surface attribute; hence it is independent of testing sand; in contrast, the values for

Rmax ,avg and Rn depend on the size of the sand particle.

44
 Literature review

Fig. 2.10. Definition of normalized roughness (after Tovar-Valencia et al. 2017).

Han et al. (2018) conducted direct shear tests at the interface between sands and steel

surfaces with varying degrees of roughness. Their findings revealed that when sand sheared

against lightly rusted steel surfaces, there was a 50% increase in the shear strength compared to

smooth steel surfaces. However, as the level of roughness on the steel surfaces increased further,

the improvement in shear strength became insignificant. Uesugi & Kisidia (1986b) provided an

explanation for this phenomenon by introducing the concept of "critical roughness." They

proposed that the friction coefficient increases linearly with roughness up to a certain threshold

value (critical roughness), beyond which the friction coefficient remains constant. When the

roughness of the interface material exceeds this critical roughness, the failure shifts from the

interface to within the soil mass, and the interface friction coefficient aligns with the friction

coefficient of the sand. Vangla & Latha (2016a) conducted a study to examine the impact of

normalized roughness on three types of wire meshes, with a range of values for the normalized

roughness (Rn) from 0.13 to 6. They also considered different particle sizes where d50 ranging

45
 Literature review

from 0.2 mm to 3 mm. The researchers observed that there was a linear correlation between the

interface friction coefficient and the normalized roughness of the interfaces for coarse sands.

However, for fine and medium sands, they discovered a bilinear relationship. This finding

suggests that the shearing plane shifted from the sand-wire mesh interfaces to the soil mass,

indicating that the normalized roughness for fine and medium sand particles exceeded the critical

roughness value.

At mesoscale level, the formation of intense localized shear zone adjacent to the contact material

is a critical state of deformation, which results in progressive failure at soil-structure interfaces

(Afzali-Nejad et al., 2017; Dejong et al., 2006; Frost et al., 2012; Lashkari & Jamali, 2021;

Martinez & Frost, 2017a). Such failure can be critical for many soil-reinforcement interfaces,

including piles, nails and geosynthetic reinforcement. Previous studies have introduced several

techniques for measuring localized shear deformation in sands and interfaces between sand and

continuum materials. These techniques include stereophotogrammetry methods (Butterfield et

al., 1970; Butterfield & Andrawes, 1972; Desrues & Viggiani, 2004). X-ray methods have been

used in geotechnical research, as demonstrated by Roscoe et al. (1963). However, in recent

times, computer-based image processing techniques such as Digital Image Correlation (DIC) and

Particle Image Velocimetry (PIV) have gained popularity for assessing deformation in

geotechnical models and element tests. Researchers, including (Arshad et al., 2014; DeJong et

al., 2003; Lei et al., 2018; Shen et al., 2020; White et al., 2003), have successfully utilized these

methods. In their study, Kong et al. (2014) performed meso-direct shear experiments on sands

and analyzed the results using Digital Image Correlation (DIC). The researchers made an

observation stating that the interface between the shear boxes demonstrated the highest degree of

shear strain. Furthermore, it was noticed that the particles exhibited a tendency to align their

46
 Literature review

largest axis parallel to the shearing plane once the shearing process had concluded. The

application of digital image-based methodologies for quantifying the thickness of shear bands

under conditions of plain strain was initially introduced by Alshibli & Sture (1999). The study

undertaken by the researchers involved an examination of many factors that influence the

thickness of shear bands, encompassing particle size, effective normal stress, and specimen

density. The researchers' findings demonstrated a clear and direct relationship between the

thickness of shear bands and the angle of dilatation shown by the sands. Furthermore, it was

observed that an increase in particle dimensions and a decrease in the density of the specimen

resulted in a decrease in the normalized thickness of shear bands. In their study, Andò et al.

(2012) presented a new approach called "ID-Track" for analyzing the deformation characteristics

of granular materials at the grain scale. This novel technique allows for the investigation of

particle-level mechanics and the development of shear banding events. This technology involves

the monitoring of individual grains by comparing images taken at different stages of

deformation, using their geometric properties to provide clear identification and correspondence

between these images. The researchers conducted small-scale triaxial experiments utilizing µCT

scanning and noticed that the localized shear bands had a somewhat larger thickness in angular

sands as compared to rounded sands. The thickness of the shear zone (ts) reported by Tehrani et

al. (2016) ranges from 3.9 to 5.2 times the mean particle size (d50) in case of non-displacement

piles embedded in dense and medium-dense sand. Lashkari & Jamali (2021) performed an

extensive series of interface direct tests on sand-geosynthetic interfaces to show that ts/d50 varies

from 3.85 to 10.62 for sand-woven geotextile interfaces, 4.15 to 10.94 for sand-nonwoven

geotextile interfaces, and 3.85 to 11.87 for sand-geomembrane interfaces. Three dimensional

(3D) discrete element simulations on sand-geotextile interfaces carried out by Feng et al. (2020)

47
 Literature review

showed that the thickness of localized shear zone is around 2.4 d50 to 3 d50. Amirrahmat et al.

(2019) employed the concept of relative particle translation gradient (RPTG) to identify and

reveal the initiation of strain localization prior to reaching the peak principal stress ratio (PSR).

The morphology of particles plays a crucial role in determining the thickness and delineation of

micro-shear bands as well as the ultimate shear band. Specifically, the roundness of particles has

a significant influence on particle interlocking, relative particle translation, and the tendency of

the specimen to develop a shear band. Wang et al. (2007) performed numerical simulations using

discrete element methods to investigate shear banding between densely packed spherical

particles and rough manufactured surfaces. They introduced a novel method for calculating strain

in the simulated direct interface shear box. Their findings indicated that the thickness of the shear

band was significantly greater for surfaces with higher roughness compared to smooth surfaces.

The reported thickness of the shear band ranged from 8 to 10 times the mean grain size of the

contacting sand.

At the soil-structure interface, particle breakage frequently takes place, which has a

significant impact on the characteristics of soil particles and associated interface behaviour.

Particle breakage can significantly reduce the shear strength and dilation at the interface, as seen

through laboratory interface shear tests and numerical simulations reported in literature (Jo et al.,

2011; J. W. Liu et al., 2019; Peng et al., 2021; Saberi et al., 2017; P. Wang & Yin, 2022) .

Particle breakage produces smaller fragments that may fill spaces between nearby particles and

create a more compacted condition. The volumetric dilation and related normal and shear

stresses are consequently reduced. At the soil-structure interface, particle breakage reduces the

shear zone thickness. Particle breakage has no effect on the overall distribution of particle

displacement, but it can cause a slight reduction in shear zone thickness (Wang & Yin, 2022).

48
 Literature review

2.4 Quantification of shear induced surface changes of geosynthetics

Gaining insight into the failure mechanism of sand-geosynthetic interfaces is of utmost

significance. While it may not be feasible to directly observe the mechanism during shearing, it

is possible to understand it by examining the deformed specimens, specifically the shear-induced

geosynthetic surfaces, through micro-level investigations after the test. Moreover, measurement

of these shear induced changes on geosynthetic surfaces gives important clues to understand the

macroscopic shear behaviour of sand-geosynthetic interfaces. While such changes are very

complex to understand on textured dilative interfaces like geotextiles, they can be measured

relatively easily on smooth non-dilative interfaces like geomembranes. The shear induced

changes on geomembrane surfaces caused by traversing sand particles can be measured

accurately using profilometry. Compared to contact profilometry, digital profilometry provides

much more accurate and comprehensive quantifications of these surface changes. The

macroscopic behaviour of sand-geomembrane interfaces can be correlated with the micro level

surface changes of contact surface through visual observations and roughness quantifications

(Vangla & Latha, 2016b; Vangla & Latha, 2017).

There is a limited number of studies that have employed non-contact-based profilometry to

quantify the microstructural damage on contact surfaces following shearing, as demonstrated by

Vangla & Latha (2026b). The literature suggests the need for more advanced instrumentation to

accurately measure micro-level changes on contact surfaces. By examining the observed changes

in microstructure, researchers are able to better comprehend the interpretation of macroscopic

phenomena with the use of cutting-edge technology. Punetha et al. (2017) conducted a study to

investigate the deformation of geotextiles and geomembranes following direct shear tests. The

microstructure of the specimens was analyzed using a Field Emission Scanning Electron

49
 Literature review

Microscope (FESEM) subsequent to the interface tests conducted by the researchers. The

findings of the investigation demonstrated that the observed shear mechanisms displayed

differences depending on the particular type of interaction. As the applied force gradually

increased, a significant change in the shear behavior occurred at the interfaces between sand and

smooth geomembranes, transitioning from a sliding mechanism to a plowing mechanism. On the

other hand, the interfaces between sand and geotextile demonstrate a shearing mechanism that

involves the interlocking of sand particles with the fiber structure of the geotextile, as well as the

stretching and tearing of the geotextile fibers.

In another study, Vangla & Latha (2016b) used a non-contact profilometer to examine the

changes induced by shearing on smooth geomembrane surfaces after conducting interface direct

shear tests. Their research showed that angular particles caused deeper grooves on the

geomembrane surfaces compared to rounded particles, which explained the higher interface

shear strength observed for angular sands. Furthermore, when investigating the effect of particle

size, they found that coarser sand particles did not consistently result in higher interface shear

strength. This was attributed to the fact that coarser sand particles made fewer contacts with the

geomembranes, leading to a decrease in contact stress per particle as particle size increased.

Upon thorough review of the relevant literature, the authors have recognized the paramount

significance of accurately characterizing particle morphology to comprehensively understand the

multi-scale behaviour of granular particles and their interactions with various continuum

surfaces. The precise quantification of localized deformation in sands and sand-continuum

interfaces is crucial for comprehending the underlying failure mechanisms and serves as a vital

input for the design of diverse geotechnical structures. Furthermore, conducting microstructural

analyses to measure surface changes in the planner continuum materials provides crucial insights

50
 Literature review

into comprehending the macroscopic interface behaviour of sands and their interactions with

geosynthetics.

51
 Chapter 3. Materials

3.1 Introduction

This study focuses on analyzing the behaviour of interfaces between sand and geosynthetic

materials at different scales, by quantifying the 3D characteristics of sand particles and their

impact on the interfaces. The study uses various types of sand, each with different sizes and

shapes, in direct shear experiments to gain a better understanding of the multi-scale behaviour of

sand-geosynthetic interfaces. In total, four types of sand and two types of geosynthetic materials

were employed in the study. This chapter provides detailed information about these materials,

including their properties, to help readers understand the scope and methodology of the study.

3.2 Granular materials

The three types of sand—fine sand (FS), medium sand (MS), and coarse sand (CS)—were

separated into different size fractions by sieving river sand, which exhibits sub-rounded particle

morphology as a result of its geological transportation processes. A local quarry supplied the

angular sand (AS). The angular sand, which is essentially a type of manufactured sand, is created

by crushing big, hard stones, mostly rocks or granite. It is frequently used as a river sand

substitute in the construction industry, primarily in the creation of concrete and mortar mixtures.

The Unified Soil Classification System defined in ASTM D2487-17e1 (2020) classed all sands

as poorly graded sands (SP). Microscopic images and three-dimensional visualization of the

particles are shown in Fig. 3.11 and Fig. 3.12, respectively. Particle size distribution curves are

illustrated in Fig. 3.13. Medium and angular sands used in this study have almost similar particle

size distribution curves as shown in Fig. 3.13. Table 3.4 lists the physical properties of various

52
 Materials

sands. The 3D morphological characteristics of these sands are described and measured using X-

ray micro-Computed Tomography (µCT) and various computational techniques implemented in

MATLAB. The process used to obtain these measurements is explained in detail in the upcoming

chapter.

Fig. 3.11. Microscopic images of sands used in this study: (a) FS (b) MS (c) CS (d) AS

53
 Materials

Fig. 3.12. Scanning samples and 3D visualization of sand particles obtained using µCT.

54
 Materials

Fig. 3.13. Particle size distribution curves of tested sands.

Table 3.4 Physical properties of tested sands.

Property FS MS CS AS
Specific gravity, Gs 2.67 2.67 2.67 2.62
Effective particle size, d10: mm 0.63 1.25 2.3 1.43
Mean particle size, d50: mm 0.78 1.58 2.93 1.76
Coefficient of uniformity, Cu 1.28 1.31 1.32 1.23
Coefficient of curvature, Cc 1.02 1.03 1.09 1.12
Maximum void ratio, emax 0.92 0.84 0.77 0.96
Minimum void ratio, emin 0.66 0.65 0.63 0.76

3.3 Geosynthetics

This study employed a woven geotextile (GTX) and a smooth high-density polyethylene

geomembrane (GMB), which possess distinct differences in their surface characteristics. In this

55
 Materials

work, a geotextile made using a plain weave technique is employed. This type of geotextile is

commonly used in engineering projects because of its versatility in performing different

functions such as separation, filtration, reinforcement, and drainage. For purposes like waste

containment, water conservation, and mining, geomembranes are impermeable liners that are

utilized in geotechnical engineering projects to regulate fluid or gas flow. Given its high tensile

strength, longevity, and superior chemical and UV resistance, high-density polyethylene (HDPE)

is a common material choice for geomembranes. Properties of geosynthetics according to the

manufacture are listed in Table 3.5 and their SEM images are illustrated in Fig. 3.14.

(a) (b)

Fig. 3.14. SEM images of the geosynthetics: (a) woven geotextile; (b) HDPE geomembrane.

Table 3.5 Properties of geosynthetics.

Geosynthetic Thickness, Mass per unit Tensile strength,

Elongation, %
material mm area, g/m2 kN/m
Warp Weft Yield Break
Not
Woven geotextile 0.5 220 50 40 30
applicable
HDPE
1.0 940 15 27 12 700
geomembrane

56
 Materials

3.4 Summary

The current chapter provides a comprehensive overview of the different materials utilized in this

study. This includes particle size distribution curves, physical characteristics, microscopic

images, and 3D visualization of the four types of sand employed in the experiments.

Furthermore, the properties of the geotextile and geomembranes, along with their SEM

(Scanning Electron Microscope) images, are also presented here.

57
 Chapter 4. 3D characterization of particle
morphology

4.1 Introduction

In many different types of engineering construction, including slopes, building foundations,

embankments, tunnels, etc., sand is probably the most prevalent naturally occurring granular

material. For a very long time, geologists and sedimentologists have focused on the analysis of

particle morphology and its effects on the mechanical behaviour of granular materials. It should

be noted that the terms shape and morphology are sometimes used synonymously in literature, in

this thesis we define morphology as the combination of particle size and shape. Numerous

studies have shown a strong correlation between particle morphology and the macroscopic

mechanical properties of sand, including stiffness, shear strength, and shear-induced dilatancy

and interface shear strength with geosynthetics (Afzali-Nejad et al., 2017; Jo et al., 2011;

Lashkari & Jamali, 2021; Tian et al., 2020; Vangla & Latha, 2015, 2016a, 2016b). The discrete

element method (DEM), proposed by Cundall & Strack (1979), has proven to be an effective

method to comprehend how granular material behaviour is affected by particle shape (Ai et al.,

2011; Danesh et al., 2020; Höhner et al., 2015; Sweijen et al., 2020; S. Zhao et al., 2015; Y.

Zhou et al., 2018).

Due to the difficulty in defining and measuring the size and shape descriptors for a wide

range of complex particle morphologies, most earlier studies either completely ignored or only

evaluated the morphological descriptors in a 2D framework. Traditionally, two-dimensional (2D)

projected images obtained by scanning electron microscope (SEM), optical microscopy, or laser

beam systems have been used to evaluate the particle morphology (Alshibli & Alsaleh, 2004;

58
 3D characterization of particle morphology

Altuhafi et al., 2013; Mora & Kwan, 2000; Pillai & Latha, 2022; Sukumaran & Ashmawy, 2001;

Vangla et al., 2018). The 2D shape parameters are derived from the images by particle projection

at random, which yields unreliable values. The majority of 2D shape characterization studies

available in literature have used images taken while the particle is resting on its plane of

maximum stability or dimension. All other orientations for characterizing particle shape are not

taken into account by such measurements. The mean shape parameters that are determined

through the analysis of these images, even when multiple orientations are used for image capture,

do not sufficiently capture the particle's overall shape. Therefore, accurate quantification of the

morphology of particles depends on the measurement of shape parameters in three dimensions.

X-ray micro computed tomography (μCT) or 3D laser scanner facilitates the 3D

visualization and quantification of the shape features of granular materials (Alshibli et al., 2015;

Fonseca et al., 2012; Kong & Fonseca, 2018; Zhao & Wang, 2016; Zhou et al., 2015, 2018; Zhou

& Wang, 2017). The identification of the individual grains through image segmentation is a

crucial step for obtaining the pertinent grain scale measurements from a 3D image. By applying

image analysis and computational geometry to the µCT images of individual grains, the 3D

morphological factors can be determined. By using simple algorithms on the component voxels

of the particles, the volume, principal dimensions, and aspect ratio of the particles can be easily

determined. However, these voxels constructed surfaces are invariably characterized by saw-

tooth patterns, and overlapping voxels brought on by high order connectivity on the surface of

µCT images lead to an overestimation of the actual surface. When using this type of uneven

surface, it is impossible to determine the particle's precise measured sphericity, correct surface

curvature, and roundness. For these reasons, a different group of researchers suggests

reconstructing the particle surface from the µCT images in order to obtain the grain's geometrical

59
 3D characterization of particle morphology

characteristics more precisely (Alshibli et al., 2015; Zhao & Wang, 2016; Zhou et al., 2015,

2018). To reconstruct the particle surface for computing the shape parameters, several smoothing

algorithms have been put forth by researchers (Field, 1988; Kobbelt et al., 2001). For example,

the creation of virtual sand particles was suggested by Mollon & Zhao (2013, 2014) using the 3D

interpolation of three orthogonal 2D cross-sections simulated by Fourier descriptors and fractal

geometry. Although they claimed that the created particles had smooth surfaces, it is likely that

the artificial selection of these cross-sections led to the loss of the local characteristics of the

particle morphology, particularly for highly irregular particles. In order to reconstruct the particle

surface, Zhao & Wang (2016) applied the Marching Cube algorithm on voxel assembly to

produce a smooth surface which is composed of triangular surface meshes (Zhao & Wang,

2016). The stair-steps on the reconstructed surfaces were found to be difficult for the MC

algorithm to completely remove, so additional mesh simplification techniques were used to

create the reasonably smooth surface. An edge collapse algorithm was implemented for

controlling the minimum number of surface mesh elements required for smoothening the surface

while preserving the overall shape of particle. This method is time consuming and

computationally expansive, especially for large number of particles. A more sophisticated

method based on spherical harmonic (SH) analysis was found to be more effective in

characterizing 3D particle morphology and reconstructing its smooth and continuous particle

surface.

This chapter presents a systematic and comprehensive 3D quantification of the morphology,

including size and shape, and 3D fractal dimension of real sand particles. The first step was to

perform high resolution X-ray µCT scanning on four different types of sands with various size

and shape characteristics. The individual particle was then extracted and separated using a

60
 3D characterization of particle morphology

variety of image processing techniques. The highly irregular grains in the µCT images are

segmented using an adaptive watershed segmentation technique. The smooth and continuous

particle surface is recreated using a sophisticated approach based on spherical harmonic (SH)

analysis while the overall particle morphology is preserved. To obtain the geometrical

parameters for calculating the size and shape of sand particles, a robust MATLAB algorithm is

written and implemented in this work. First, the particle's three primary dimensions from µCT

images are obtained to determine its size, and its particle size distribution (PSD) is compared to

sieving. Then, quantification of the shape descriptors was carried out using computational

geometry and image analysis methods based on the SH reconstructed particle surface.

Additionally, a fractal dimension for the 3D closed surface of sand particles was discussed and

quantified using spherical harmonic-based fractal analysis.

4.2 X-ray micro-Computed Tomography (µCT)

X-ray micro-computed tomography, also known as µCT, is a non-invasive imaging technique

that uses X-rays to produce high-resolution 3D images of an object's internal structure. It works

by collecting several projections from different angles when an object rotates between an X-ray

source and detector. Slices from captured µCT angular projections across the object are

reconstructed into a 3D image using specialized reconstruction software. Micro-CT is commonly

used in various fields, including materials science, biology, and engineering, to visualize and

study the internal structure of objects at a micro-scale level. In recent years, there has been an

increased use of µCT by researchers to study the behaviour of granular media. Specifically, µCT

has been utilized to quantify the particle shape and shear behaviour of particles in granular media

through in-situ experiments. µCT allows researchers to non-destructively image the internal

61
 3D characterization of particle morphology

structure of granular media with high resolution, which enables them to accurately measure the

particle shapes and positions, as well as the forces acting on individual particles during shear

tests. By utilizing µCT, researchers are able to gain a better understanding of the fundamental

behavior of granular media, which has significant applications in fields such as geotechnical

engineering, materials science, and environmental science.

4.2.1 Fundamentals of Computed Tomography

Computed Tomography (CT) works on the principle of X-ray attenuation, which is the ability of

an object to absorb or scatter X-ray energy as it passes through. Materials with higher atomic

numbers (denser materials) attenuate more X-ray energy than materials with lower atomic

numbers (less dense materials). This difference in attenuation results in variations in the intensity

of the X-rays that reach the detector, providing the contrast necessary for image formation. The

resulting image shows denser materials as brighter areas, while less dense materials appear

darker. However, the ability of X-rays to penetrate an object is highly dependent on the material

properties of the object being imaged. X-ray interaction with matter is a function of the material

density, the material thickness, and the X-ray energy (Fonseca, 2011).

Computed tomography has a relationship between special resolution and object size where

smaller object sizes can result in higher resolutions. In medical CT, objects larger than 100 mm

are typically imaged with a few millimeters of resolution using a macro-focus. For particle-scale

analysis, higher special resolutions are needed to identify individual grains. Micro-CT can

achieve resolutions in the order of a few microns using micro-focus. In this study, small

specimens of a few millimeters in size were used to achieve high resolutions of a few microns,

which is necessary for identifying individual grains.

62
 3D characterization of particle morphology

4.2.2 Micro-CT set up

There are two primary types of X-ray beams used in modern micro-tomography: parallel beam

(obtained in synchrotrons) and cone beam (typically used in laboratory sources). This study

utilized the cone beam µCT (SkyScan 1272) from Bruker, which is available at the National

Centre for Biological Sciences in Bangalore, India. Fig. 4.15 presents a photograph and Fig. 4.16

presents a schematic diagram of the µCT.

Fig. 4.15. Photograph of the Micro-CT (SkyScan 1272) used in this study.

63
 3D characterization of particle morphology

Fig. 4.16. Schematic diagram of the µCT (after S. C. Lee et al., 2003)

Micro-CT scanners consist of three main components: the X-ray source, the detector, and the

rotating platform. The X-ray source emits radiation that passes through the object placed on a

rotating stage. The detector measures how much the X-ray signal is attenuated by the object. The

object is enlarged in the detector, and the degree of enlargement, known as magnification (M), is

determined by the ratio of the source-detector distance (SDD) to the source-object distance

(SOD). Geometric magnification is influenced by the sample diameter and the size of the

effective detector surface. The image resolution corresponds to the voxel size of the

reconstructed volume dataset and can be determined by dividing the detector pixel size by

geometric magnification.

64
 3D characterization of particle morphology

4.2.3 Description of the scanning process

The initial stage in conducting µCT scanning of sand samples involves preparing and mounting

the samples on a rotating stage. In this study, the sand samples were tightly packed within a thin,

transparent acrylic tube, and the ends were sealed with wax. The purpose of compacting the sand

particles was to prevent their movement during rotation, as this could result in blurred images

that would pose difficulties during segmentation. Each sample was then scanned by securely

placing it on the pedestal at the top of the rotary stage, ensuring that the object to be imaged

remained within the field of view throughout all rotational positions.

The quality of reconstructed images in µCT scanning is greatly influenced by the number of

projections acquired and the exposure time. Exposure time refers to the duration over which the

input signal is sampled on the detector. Manufacturers, such as phoenix|x-ray, recommend using

approximately 1.2 times the number of pixels in the image perpendicular to the rotation axis to

achieve good quality images for high-resolution systems. Increasing the number of projections is

often more effective than extending the exposure time as it helps prevent detector saturation.

The X-ray source settings include the tube voltage and tube current. The tube voltage determines

the maximum energy of the X-ray spectrum, controlling X-ray penetration and the signal-to-

noise ratio. Higher energy X-rays penetrate more effectively but are less sensitive to changes in

material density and composition, resulting in images with less noise. On the other hand, lower

energy X-rays provide better contrast, allowing for discrimination between different materials.

Ideally, the object being scanned should have a cylindrical shape to ensure a constant

amount of material that the X-rays pass through for each angular interval. This is important

because variations in the lengths of X-ray paths can affect the measured attenuation values.

65
 3D characterization of particle morphology

4.2.4 Scanning Artefacts

Artefacts refer to imperfections or defects that can occur during the scanning process in CT. The

most common artefact in CT is the beam hardening artefact, which is characterized by brighter

edges compared to the center of the image. One potential solution to mitigate this artefact is to

use an X-ray beam with sufficient energy to minimize the effect. However, when scanning low-

density materials like geomaterials, increasing the X-ray energy can result in poor contrast

between different features in the images. Another approach to reduce beam hardening artefacts is

to employ filters that absorb the low-energy portion of the X-ray beam.

Ring artefacts are also prevalent and arise due to defective detector pixels. These artefacts

appear as ring patterns with higher intensity levels in the center. Prior to reconstruction, these

artefacts can be detected and removed. Star artefacts occur when the scanned object contains

features with significantly higher density than the surrounding material.

Other sources of error include motion artefacts caused by movement of the X-ray source and

the object due to heat generated during X-ray generation. Motion artefacts degrade image quality

by blurring feature edges or increasing noise. Recent advancements in mechanical systems offer

improved stability against thermally induced drift effects, addressing this issue.

4.3 Particle morphology of sand particles using 3D image analysis

4.3.1 Micro-CT Image acquisition and processing

With the current advancements in X-ray technology for investigating laboratory specimens at

high resolution, µCT makes it easier to scan objects and produce 3D images with the grey value

representing the material composition and density. Each sample was scanned using a Bruker

cone-beam µCT (SkyScan 1272) available at the National Centre for Biological Sciences

66
 3D characterization of particle morphology

Bangalore, India. Micro-CT scan gathers several projections when an object rotates between an

X-ray source and detector. Slices from captured µCT angular projections across the object are

reconstructed using SkyScan's volumetric NRecon reconstruction software.

Fig. 4.17. Flowchart illustrating the steps of image processing in X-ray µCT.

The software makes use of a modified Feldkamp algorithm with automated scan geometry

adaptation. To increase the scanning efficiency, particles were packed tightly into acrylic tubes

that ranged in diameter from 6 to 16 mm and height from 10 to 17 mm. The scanning parameters

used for all the samples included a special resolution of 10 microns per voxel. Additionally, the

X-ray source was set at a voltage of approximately 100 kV and a current of around 110 µA. In

this study, the average time required for data collection of a single sample was approximately 2

hours, while the reconstruction process took around 10 minutes. The scanned images in this

67
 3D characterization of particle morphology

study had dimensions of 1108 ×1108 × 1504 voxels for FS, 1180 × 1180 × 1488 voxels for MS,

1756 × 1756 × 1625 for CS, and 1212 ×1212 × 1559 for AS respectively. Due to computational

constraints, the image sizes were reduced by a factor of two for the analysis. The binning

technique involved combining 2×2×2 voxels and applying the group's mean intensity value to the

associated voxel in the reduced image.

68
 3D characterization of particle morphology

Fig. 4.18. Illustration of image processing steps: (a) typical slice of a 3D tomographic image (b)
image after applying 3D median filter (c) image after thresholding segmentation (d) image after
watershed segmentation.

The reconstructed µCT images were subjected to numerous image processing operations in

order to minimize the noise and recover and separate the scanned particles. The flowchart of

different image processing steps utilized in this study is shown in Fig. 4.17. To begin with, the

µCT images were pre-processed with a 3D median filter with an intensity level of three pixels to

69
 3D characterization of particle morphology

reduce the noise in the raw µCT data. Thresholding segmentation, based on Otsu’s algorithm

Otsu (1979), was used to distinguish sand particles from voids and the acrylic container. The

binary 3D µCT images were obtained with voxel values of 1 and 0, which correspond to

foreground (i.e. sand particles) and the background (i.e. remaining material), respectively. A

modified 3D watershed segmentation technique, proposed by Kong & Fonseca (2018), was

utilized to separate the particles in contact and label each particle with a sequential number. Most

of the watershed segmentation techniques used in earlier studies do not address the issue of over-

segmentation while separating the contacts of particles during image processing. Some of the

recent studies have specifically addressed the issue of over-segmentation while separating the

contacts of particles. Accurate segmentation requires precise separation of grains that are in

contact with each other while preserving their boundaries carefully, to alleviate the issue of

interpreting a single particle as multiple particles. While simpler segmentation techniques work

well for point contacts in spherical grains, sophisticated methods are needed for separating grains

of irregular shapes and contact topologies (Kong & Fonseca, 2018). The adaptive watershed

segmentation technique used in the current study works through identification of regions, which

are the groups of connected solid voxels in a binary image that are unconnected with any other

solid voxels. The technique was implemented through a series of iterations that involve region

identification and separation. In each iteration, identified regions were split into several new

regions, which were smaller than the regions in the previous iteration. Watershed segmentation

was then applied to separate the newly identified regions, thus making the process of grain

separation progressively more precise. These iterations were continued until no more new

regions are identified, thus ensuring that all contact boundaries are separated and none of the

grains are over-segmented. A MATLAB algorithm was written to implement these steps of

70
 3D characterization of particle morphology

adaptive 3D watershed technique on the topographic image to separate the grains for shape

analysis. This method produced exceptionally precise segmentation results for test sands that

contained several peanut-shaped particles. Fig. 4.18 displays an illustration of the outcomes of

image processing on a slice of MS particles.

Fig. 4.19. Comparison of watershed segmentation technique: (a) input binary image; (b)
conventional watershed segmentation (MATLAB 2019); (c) modified watershed segmentation
technique.

71
 3D characterization of particle morphology

Fig. 4.20. 3D visualization of sand samples after watershed segmentation.

4.3.2 Reconstruction of 3D particle surface

To reconstruct the smooth and continuous particle surface while preserving the overall particle

shape, the authors employed a more advanced technique based on spherical harmonic (SH)

analysis (Sun & Zheng, 2021; Zhou et al., 2015, 2018; Zhou & Wang, 2015, 2017). The SH

approach is briefly discussed here for completeness. The intrinsic MATLAB function bwperim

makes it simple to determine

72
 3D characterization of particle morphology

Fig. 4.21. Three-dimensional images of typical sand particles of FS sample.

the spatial locations of the boundary voxels (x, y, z) of a given particle (Mathworks, 2019).

Equation (4.8) describes how the polar radius in the SH analysis is expanded from a unit sphere

to a particle profile made up of vertices (x, y, z)

∞ n
f ( θ , ϕ ) =∑ ∑ amn Y mn (θ , ϕ ) (4.8)
n=0 m=−n

where f (θ, 𝜙) is the polar radius from the particle center, while θ ∈ [ 0 , π ] and ϕ ∈ [ 0 , 2 π ] are the

latitudinal and longitudinal coordinates, respectively; Y mn ( θ , ϕ )is the SH function deduced from

Equation (4.9); and a mn is the associated SH coefficient that has to be computed.

√
1. ( 2n+1 )( n−m ) ! m
Y mn ( θ , ϕ )= Pn (cos θ)e imϕ (4.9)
4 π ( n+ m) !
where Pmn (x) are the associated Legendre functions that may be obtained by using intrinsic

function legendre in MATLAB, and n and m are degree and order of Pmn (x), respectively. In

73
 3D characterization of particle morphology

accordance with the precision needed for fitting, n is a non-negative integer ranging from zero to

infinity and as a result, the total number of one set of a mn is (n+1)2.

Earlier studies (Sun & Zheng, 2021; Zhou et al., 2018; Zhou & Wang, 2017) have

demonstrated that the SH degree has to be more than 15 to accurately depict the regional

morphological characteristics (i.e. local roundness and surface texture). By applying the usual

least-squares estimation, the SH coefficients a mn was calculated, and the continuous and smooth

surface of the particle was reconstructed by using Equation (4.10)

nmax n (4.1
^f ( θ , ϕ ) =∑ ∑ amn Y mn (θ , ϕ )
n=0 m=−n 0)

For the purpose of reconstructing 3D particle surfaces, the maximum SH degree ( n max) was set as

18. The particle morphology represented by the voxel assembly is shown in Fig. 4.22 along with

the surface that corresponds to it that was reconstructed using SH analysis. Fig. 4.23 shows the

particle shape reconstruction at different spherical harmonic degrees. The mean spherical

harmonic radius is given by Equation (4.11), which reveals that the resulting particle shape at

zero SH degrees is a sphere (Yang et al., 2022).

a00 (4.1
0
r =
0
√4 π 1)

74
 3D characterization of particle morphology

Fig. 4.22. 3D particle surface reconstruction of a typical AS particle (a) particle morphology
represented by voxel assembly (b) particle morphology represented by reconstructed surface
generated from SH analysis (c-d) close-up views.

The particle surface becomes more finely detailed when it is reconstructed at an increasing

maximum harmonic degree. Starting at degree 1, the reconfigured shape takes on an elliptical

shape before gradually becoming irregular as degree increases. The particle shape has a

significant impact on the reconstruction of particles at lower harmonic degrees (Su & Yan, 2018;

Sun & Zheng, 2021; Yang et al., 2022; Zhou et al., 2018; Zhou & Wang, 2017). The

reconstruction of the particle becomes more detailed as the maximum degree rises, and as a

result, the reconstructed particle more closely resembles the original scanned particle.

75
 3D characterization of particle morphology

Fig. 4.23. 3D surface reconstruction at different spherical harmonic degrees.

4.3.3 Three-dimensional morphological descriptors of sand particles

According to Barrett (1980), particle morphology is typically described in terms of three

descriptors: form (first order approximation of shape), roundness (second order), and roughness

(third order). Sphericity is considered to be an index of form by many researchers (Clayton et al.,

2009), while some characterize particle form in terms of elongation and flatness indices

(Angelidakis et al., 2022). Fig. 4.24 illustrates the three scales of particle shape. Sphericity

(overall shape) reflects the proportions of a particle; roundness (or angularity) reflects the

sharpness of corners and roughness reflects the surface texture imposed on corners and between

76
 3D characterization of particle morphology

corners (Barrett, 1980). Fig. 4.25 shows the flowchart of 3D shape characterization of sand

particles used in this study.

Fig. 4.24. Three characteristic scales of particle shape.

77
 3D characterization of particle morphology

Fig. 4.25. Flowchart of 3D shape characterization of sand particles.

4.3.3.1 Particle size

In geotechnical practice, soil type classification is done based on the particle size. For obtaining

the distribution of grain sizes, standard sieves are typically used. Particle size characterization

based on sieve analysis suffers from many limitations, which include missing data between sieve

intervals and 1D circular representation of particles based on their passage or retention through

square openings of the sieves. Methods that analyze particle images to determine the grain size

distributions help in overcoming these limitations. The distance between two parallel tangents to

the particle outline on opposite sides, known as the Feret diameter, is typically representative of

the size of particles in a 2D framework (Altuhafi et al., 2013). The analysis based on 2D images

suffers from the drawback that the particle's minor principal axis is vertical when the particle is

78
 3D characterization of particle morphology

at rest in its most stable position (Altuhafi et al., 2013). Accurate grain size measurements are

possible through advanced imaging techniques like µCT or optical microscopy. The primary axis

lengths of µCT images, which can be computed easily using Principal Component Analysis

(PCA), are the 3D equivalents of particle dimensions.

Fig. 4.26. Rotation of particle to make its principal axes parallel to the cartesian axes for
calculating the principal dimensions.

In this study, PCA was used to identify the major, intermediate, and minor axes of the particle.

This technique is extensively being used in numerous fields to analyze the tomographic data

(Fonseca et al., 2012; Zhao & Wang, 2016). The particles were then orthogonally rotated till

their principal axes of inertia lined up with the cartesian axes. The major (a), intermediate (b),

and minor (c) principal dimensions were computed using the measurement data for the

coordination dimensions. The particle size distribution curves obtained from image analysis and

sieve analysis are compared and discussed in latter sections.

79
 3D characterization of particle morphology

4.3.3.2 Particle form

Most often, the particle form for three-dimensional grains is characterized using two independent

indices: elongation and flatness. These indices are evaluated using the three principal dimensions

at major axis(a), intermediate axis (b), and minor axis(c), where a ≥ b ≥ c (Zhao & Wang, 2016).

A particle's flatness refers to how flat (or platy) it is, whereas its elongation describes how

elongated (or rod-like) it is. One of the most widely used systems for classifying form was put

forth by Zingg (1935), who used two aspect ratios to describe a particle's form as expressed by

Equation(4.12) in which b/a is the elongation index and c/b is the flatness index.

EI =b /a (4.1

FI =c /b 2)

Zingg (1935) plotted the two aspect ratios b/a and c/b on a bivariate graph (widely referred to as

Zingg plot) and defined particle morphology in four classes: oblate if c/b < 2/3 and b/a > 2/3;

compact if c/b > 2/3 and b/a > 2/3; blade-like if c/b < 2/3 and b/a < 2/3 or prolate if c/b > 2/3 and

b/a < 2/3, as shown in Fig. 4.27 Zingg (1935) did not use the terms "flatness" or "elongation,"

but subsequent literature has referred to b/a and c/b as elongation index and flatness index,

respectively (Zhao & Wang, 2016; Angelidakis et al., 2022; Blott & Pye, 2008). Even though

form is primarily a 3D concept, many researchers chose to estimate form in 2D through Feret's

diameters or visual comparison charts, due to their relative simplicity. All 2D estimates of form

have the drawback of being unable to distinguish between spheres and discs with similar

outlines. Several researchers had proposed mathematical formulae for computing the elongation

and flatness to characterize the particle form for real three-dimensional particles, as summarized

in Table 4.6.

80
 3D characterization of particle morphology

Table 4.6 Available formulae for computing particle form

Reference Elongation index Flatness index

b c
Zingg (1935)
a b

Kong & Fonseca b c

1− 1−
a b
(2018)

Bagi & Orosz

c c c
(2020) − 1−
b a b

Potticary et al.
a−b 2(b−c)
(2015)
a+b+ c a+b+ c

Angelidakis et al.
2
ac c b c
(2022) − −
ac+ b a+c
2 2
ac+ b a+c

81
 3D characterization of particle morphology

Fig. 4.27. The shape classification systems for characterizing particle form: (a) proposed by
Zingg (1935) (b) proposed by Angelidakis et al. (2022).

Fig. 4.27 compares the shape classification systems proposed by Zingg (1935) and Angelidakis

et al. (2022) for characterizing the particle form. The limitations of the indices for elongation and

flatness proposed by earlier researchers like Zingg (1935) and Kong & Fonseca (2018) were

thoroughly reviewed by Angelidakis et al. (2022) and new indices were proposed, as given in

Table 1. Additionally, a new particle classification system derived from the suggested indices

was illustrated. This system corrects a range of particles that were incorrectly classified as bladed

by the Zingg (1935), as shown in Fig. 4.27(b). In the new classification system, region II of the

compact particles is preserved, and the regions of both flat and elongated particles is enlarged at

the expense of the region of bladed particles. The suggested indices and the new classification

system work well for particles with all conceivable aspect ratios (c/b >0 and b/a > 0).

82
 3D characterization of particle morphology

Fig. 4.28. Comparing the two classification systems for characterizing the particle form.

In the current study, to characterize the particle form for the tested sands, the two classification

schemes proposed by Zingg (1935) and Angelidakis et al. (2022) were used, and the results are

compared in Fig. 4.28. The b/a ratios are plotted against c/b ratios for both the classification

systems, as shown in Fig. 4.28. Since AS and FS particles have relatively lower and more

dispersed values of elongation and flatness compared to other particles, these particles were

selected for the comparison. It can be seen from Fig. 4.28 that no significant difference was

observed between the two classification systems for the tested sands. It should be noted that the

majority of naturally occurring materials, like sands and silts, do not possess extreme particle

shapes because they tend to break naturally and evolve into smaller, more regular fragments

(Buscarnera & Einav, 2021). Hence, Zingg’s classification system can be used for characterizing

the particle form for natural geomaterials. Hence the particle form characterizations and the

83
 3D characterization of particle morphology

elongation and flatness indices in the present study were evaluated using Equation (5) proposed

by Zingg (1935).

4.3.3.3 Three-dimensional sphericity

Sphericity (S), which also serves as a measure of compactness, defines how much a particle

resembles a sphere. The most popular definition of sphericity first proposed by Wadell (1935) is

that it is the ratio of the surface area of a sphere with the same volume as the particle to the

actual surface area of the particle, stated as

S=
√3 36 πV 2 (4.1
SA 3)

where S represents the given particle's 3D sphericity, and V and SA are the volume and the

surface area of the given particle, respectively. To accurately assess the morphological

descriptors, particle volume and surface area are measured from the SH reconstructed surface.

Many researchers have used complex mathematical computations to obtain the geometrical

parameters, such as volume, surface area, moment of inertia, and surface curvature in SH

analysis (Zhou et al., 2018; Zhou & Wang, 2015; Garboczi, 2002). The volume, surface area, and

local surface curvature of the particles in this study were computed using a simple and practical

approach based on the SH-reconstructed surface mesh, as illustrated in Fig. 4.29.

84
 3D characterization of particle morphology

Fig. 4.29. Calculation of geometric parameters of the particle: (a) typical sand particle with its
SH reconstructed surface mesh (b) diagram used for computations.

If the voxel resolution and SH degree are high enough, this method has been shown to be

effective for precisely computing the surface area and volume of the particles (Zhou et al., 2018).

The surface area and volume are the sums of the micro-surface areas of all the faces and micro-

volumes of all the tetrahedrons, respectively, which can be expressed by

1 (4.1
SA=∑ |A B × Bi Ci|
i 2 i i
4)

1 (4.1
V =∑ ¿
i 6
5)

O represents the particle's center and Ai, Bi, and Ci are the component vertices that make up the

ith face of the surface mesh. Equation (4.13) makes it simple to determine the precise 3D

sphericity of the given particle once the volume and surface area of the particle have been

accurately measured.

85
 3D characterization of particle morphology

4.3.3.4 Three-dimensional convexity

Convexity is a measure of particle’s compactness that describes how closely a particle resembles

a convex hull. Convexity is defined as the ratio of the particle's volume (V) to the volume of the

convex hull (V hull ) containing the particle, as given in Equation (4.16).

V (4.1
C=
V hull
6)

A convex hull (V hull ) is defined as the smallest convex surface containing all of the component's

voxels which may be readily obtained using the intrinsic function convhull in MATLAB.

Convex hull

(a) (b)

Fig. 4.30. Illustration of finding the convex hull: (a) 2D convex hull; (b) 3D convex hull.

4.3.3.5 Three-dimensional roundness

The roundness of a particle indicates how sharp its corners are. According to Wadell (1935) and

Blott & Pye (2008), 2D roundness is the average ratio of the curvature radius of all corners to the

radius of maximum inscribed circle of a given particle. However, there can be more than one

curvature at a given point on a regular 3D surface because the curvature at a given point depends

86
 3D characterization of particle morphology

on the path through that point. This made it necessary for this study to provide a brief

explanation of 3D surface curvature to define the corners on the particle surface.

Fig. 4.31. Illustration of finding the maximum inscribed circle of a given particle: (a) typical 2D
image of a particle; (b) particle boundary; (c) Euclidean distance transform; (d) maximum
inscribed circle.

In this study, the local principal curvatures were evaluated at each vertex on the SH-

reconstructed surface. The idea is to generate the local micro-surface around a given vertex using

a single-ring or higher-ring neighbourhood (see Fig. 4.29(b)). By using a higher-ring

neighbourhood, the effects of noise and mesh irregularity could be reduced. In this study, a two-

87
 3D characterization of particle morphology

ring neighbourhood was used to establish a polynomial function to approximate this micro-

surface, as expressed by Equation (4.17).

2 2
z ( x , y ) ≈ e1 ( x−x 0 ) +e 2 ( y − y 0 ) +e 3 ( x−x 0 ) ( y − y 0 )
(4.17)
+e 4 ( x−x 0 ) + e5 ( y− y 0 ) + e6

where e1, e2, e3, e4, e5, and e6 are the unknown coefficients which are determined using the least-

squares fitting method. A 2×2 Hessian matrix expressed by Equation (4.18) can describe the

shape characteristics of the micro-surface at the specified vertex (Do Carmo, 1976).

H=
[ 2 e1 e3
e3 2e 2 ] (4.18)

The eigenvectors and eigenvalues of this Hessian matrix represent the principal directions and

the principal curvatures, respectively. Generally, two principal curvatures, namely the

maximum and minimum of the normal curvatures kmax and kmin, respectively, are frequently

used to measure the maximum and minimum bending of the surface at a given point (Do

Carmo, 1976). It was found that kmax overestimates the number of corners while kmin

underestimates the number of corners as shown in Fig. 4.32. To account for this, the current

study implements the mean curvature as given in Equation (4.19) to identify the potential

corners at a given vertex.

k max + k min
k m= (4.19)
2

88
 3D characterization of particle morphology

(a) (b) (c)

Fig. 4.32. Corners identified by different curvatures on an AS particle surface: (a) maximum
principal curvature (kmax) (b) minimum principal curvature (kmin) (c) mean curvature (km). unit:
mm-1

Based on the mean curvature, a corner on the particle surface is defined by Equation (4.20) in

which c (k) =1 represents an acceptable corner and c (k) =0 represents an unacceptable corner.
−1
1 if |k m| <r ins
c ( k )={0 if |k m|−1 ≥ r ins
(4.20)

where rins is the radius of the largest inscribed sphere of the particle. The 3D roundness (R) is

then computed from Equation (4.21)

R=
∑ c(k )|k m|
−1

(4.21)
N r ins

where N represents the total number of acceptable corners.

89
 3D characterization of particle morphology

Fig. 4.33. Computational geometry techniques for determining particle roundness: (a) maximum
inscribed sphere; (b) mean curvature contours for corners identification.

The surface irregularities of the natural sand particles can be quantified using 3D fractal analysis.

The Triangular Prism Method (TPM), first proposed by Clarke (1986), and the Slit Island

Method (SIM), first proposed by Mandelbrot et al. (1984), have both been widely used to assess

the 3D fractal dimension for 3D open surfaces. The TPM and SIM, however, are unable to

determine the fractal dimension of a 3D closed surface like a sand particle surface. Recently,

Zhou et al. (2018) made notable progress towards quantifying the 3D fractal dimension of sand

particles. They developed a novel method to assess the 3D fractal dimension of the 3D closed

surface of sand particles based on the SIM concept. However, to implement this method,

operators need to have a strong background in image processing, and it is difficult and

computationally demanding. The empirical approach suggested by Russ (1994), Chan (1995) and

Quevedo et al. (2008), which measures the fractal dimension using the Fourier power spectrum

of 2D grey level images, is developed, and discussed to further measure the fractal dimension for

3D images of the sand particles.

90
 3D characterization of particle morphology

According to Equation (1), a spherical harmonic (SH) function representing a particle’s

morphology can be described by different spherical harmonic frequencies and their amplitudes

control the intensity of the morphological features at each respective frequency space (Kazhdan

et al., 2003). The amplitude at each spherical harmonic frequency (ln) can be calculated using

Equation (4.22).

√
n

∑
¿

l n= amn amn (4.22)

m=−n

where * denotes the conjugate transpose. It has been found that l 1 has no effect on the particle

morphology and it merely represents the shift of the spherical harmonic-reconstructed particle

profile with respect to the location of the original particle center (Zhou & Wang, 2017; Sun &

Zheng, 2021. l 0 is a measurement of a particle's volume. Each l n was divided by l 0 to eliminate

the effect of particle volume. Additionally, l 1 was not considered because it has no effect on the

morphology of the spherical harmonic-reconstructed particle. Finally, it is possible to define the

spherical harmonic descriptors ( Dn ¿ that describe the particle morphology, as shown by

ln
Dn= (where n=2 … .18) (4.23)
l0

It is possible to express the exponential relationship between the spherical harmonic degree n and

the spherical harmonic descriptor Dn as follows:

β
Dn ∝n (4.24)

where β = -2H and H is the Hurst coefficient (Chan, 1994; Quevedo et al., 2008). If a linear

relation is established from the log ( Dn) versus log (n) graph, then fractal dimension (FD) can be

calculated as:

91
 3D characterization of particle morphology

6+ β
FD= (4.25)
2

Fig. 4.34. Spherical harmonic descriptor (Dn) as a function of the spherical harmonic degree n in
log-log scales for MS particles.

The relationship between the spherical harmonic degree n and the mean spherical harmonic

descriptor Dn plotted for the MS particles used in the present study is shown in Fig. 4.34.

4.4 Results and Discussions

Sieve analysis is commonly used in geotechnical engineering to measure the particle size

distributions. As previously explained, sieving provides no information about the particle size

between the sieve intervals. However, it serves as a helpful reference to determine the viability

of image-based particle size measurement techniques. Fig. 4.35 compares the particle size

distribution data for the tested sands obtained through µCT data and sieve analysis. It must be

noted that the sample used for the sieve analysis contained more grains than the sample used for

92
 3D characterization of particle morphology

µCT scanning. Table 4.7 compares the particle-size distributions for the tested sands, considering

the major (a), intermediate (b), and minor (c) principal dimensions data to calculate the mean

particle sizes (a50, b50, c50) and their corresponding coefficients of uniformity (Cua, Cub, Cuc). For

FS sample, the distribution of b values and the sieve analysis data closely resemble each other

which validates the µCT data since the intermediate principal dimension determines which sieve

a given particle will pass through if adequate energy is imparted so that particles can pass

through sieve apertures with their major axis oriented vertically. However, for AS sample, the

sieving measurements are in good agreement with a. The findings are thought to be connected to

the prevalence of flat and elongated grains in AS samples that tend to lie with the major axis

along the horizontal plane (more stable position). For the grains to pass through the sieve in this

case, it should have an aperture that is larger than a unless intense vibration is used (which can

lead to particle breakage). Similar observations were reported in literature for sands (Fonseca et

al., 2012; Kong & Fonseca, 2018). It could be observed from Table 2 that d50 of the MS and CS

particles, is 1.41 and 1.69 times that of a50, respectively. The number of particles used for

imaging is considerably less than that used for the sieve analysis, which could have partially

contributed to the deviation in grain size distributions based on these two approaches.

93
 3D characterization of particle morphology

Fig. 4.35. Comparison of particle size distributions obtained using µCT images and sieve
analysis: (a) FS (b) MS (c) CS (d) AS

Table 4.7 Particle size distribution parameters obtained through µCT image analysis and sieving.

Image analysis Sieve analysis

Sand a50 b50 c50 d50
Cua Cub Cuc Cu
(mm) (mm) (mm) (mm)
FS 1.04 0.79 0.55 1.53 1.46 1.48 0.78 1.28
MS 1.12 0.88 0.67 1.30 1.39 1.38 1.58 1.31
CS 1.73 1.41 1.08 1.25 1.28 1.27 2.93 1.32

94
 3D characterization of particle morphology

AS 1.54 1.07 0.61 1.45 1.48 1.44 1.76 1.23

Two aspect ratios: elongation, EI and Flatness, FI, are plotted for all the tested sand particles in

Fig. 4.36. As per Zingg classification system, most MS and CS particles lie in compact region as

both their EI and FI values are greater than 2/3. These particles have more regular geometry in

terms of particle form. The FS and AS particles have lower values and a more disperse

distribution of aspect ratios. Among all the tested particles, AS particles have the highest

dispersion as they comprise a wide range of aspect ratios which makes them highly irregular.

The AS particles, which are obtained by crushing of the rock, have irregular form and sharp

corners due to the intersection of fracture planes, whereas river sand samples have regular form

and smooth corners due to the geological transportation process. Fig. 4.36(a) and Fig. 4.36(b)

compare the particle form with respect to particle size and particle shape, respectively. In terms

of particle size, smaller particles have wider distributions and lower aspect ratios than coarser

particles, indicating that the regularity of the particles increases with grain size.

95
 3D characterization of particle morphology

Fig. 4.36. Particle form characterization using Zingg plot: (a) different particle sizes (b) different
particle shapes.

Sphericity is a crucial shape parameter to consider when characterizing the 3D particle

shape. Although sphericity and convexity both refer to the compactness of particles, they give a

different perspective on how they are described. Firstly, sphericity is a fraction of the surface

area whereas convexity is a fraction of the volume. Secondly, sphericity compares the particle to

its volume-equivalent sphere while convexity compares it to its convex hull.

96
3D characterization of particle morphology

97
 3D characterization of particle morphology

Fig. 4.37. Distributions of shape parameters of tested sand particles: (a) Elongation index (b)
Flatness index (c) Sphericity (d) Convexity (e) Roundness (f) Regularity.

Fig. 4.37 shows the cumulative distribution of shape parameters for the particles of four

different test sands. It is found that AS particles have lower values of sphericity and convexity

among all the tested samples, as depicted in Fig. 4.37, indicating that the AS particles have more

complex morphology in terms of compactness. Strong correlation is observed between sphericity

and convexity values of sand particles as shown in Fig. 4.38, suggesting interdependency of

these two shape parameters, and reinforcing the findings of some of the earlier researchers that

these are not independent measures of particle shape.

Fig. 4.38. Correlation between shape parameters.

The form of the soil particles directly influences their sphericity values. Particles with lower

aspect ratios have lower sphericity values. For example, AS particles have lower aspect ratios as

well as lower sphericity and can be characterized as flat, elongated, and non-spherical in this

study. The values of sphericity and convexity increase with grain size. However, this increment

is not significant in the present study, as the samples were taken from the same origin.

98
 3D characterization of particle morphology

As expected, AS particles have lower mean values and higher standard deviation of roundness

among all the tested sands, which further proves that particle morphology of AS is more irregular

and angular than that of samples of river sand. Roundness values were found to be unaffected

with increasing particle size (see Fig. 4.37(e)).

All surfaces are rough even on an atomic scale. For the same surface, instruments with

different resolutions and scan lengths produce different roughness values (Majumdar &

Bhushan, 1990). The fractal dimesion, which is a non-interger parameter, is best suited to

characterize the fractal nature of sand particle surfaces (Zhou et al., 2018; Yang et al., 2022). In

this study, the fractal dimension was measured by spherical harmonic based fractal analysis. The

histograms and their corresponding normal fitting distributions of fractal dimension of test sands

are shown in Fig. 4.39. The mean values and standard deviations of fractal dimension are

summarized in Table 4.8. It is found that fractal dimension increases with inreasing partcile size.

Coarser particles have larger mean values and wider distributions of fractal dimension, as shown

in Fig. 4.39(a). These results are in agreement with Yang et al. (2016) where fractal dimesnion is

measured using power spectral density method. When comparing the particles with different

shape and similar gradation, higher irregularity leads to higher fractal dimension, as illustrated in

Fig. 4.39(b). Table 4.8 shows that mean fractal dimension of AS particles and MS particles are

2.300 and 2.283, respectively.

99
 3D characterization of particle morphology

Fig. 4.39. Histograms of fractal dimension of tested sands with their fitted normal distributions.

Table 4.8 Statistics of 3D shape descriptors of sand particles.

FS MS CS AS
3D
descriptors Mean Standard Mean Standard Mean Standard Mean Standard
value deviation value deviation value deviation value deviation
Sphericity 0.81 0.048 0.84 0.039 0.85 0.033 0.76 0.055
Convexity 0.86 0.043 0.88 0.040 0.88 0.033 0.85 0.031
Roundness 0.57 0.037 0.58 0.033 0.57 0.027 0.53 0.048
Elongation 0.75 0.118 0.80 0.094 0.83 0.079 0.68 0.144
Index
Flatness 0.73 0.123 0.77 0.106 0.80 0.109 0.62 0.149
Index
Regularity 0.74 0.074 0.77 0.062 0.79 0.056 0.69 0.085
Fractal 2.281 0.053 2.283 0.062 2.324 0.064 2.300 0.051
dimension

Wadell (1935) claimed that sphericity and roundness are two distinct characteristics of

particle shape, but in reality, they are interdependent. It is found that the shape descriptors at

100
 3D characterization of particle morphology

different characteristic scales have some correlation with one another. The correlations heavily

rely on the distance between the characteristic scales of the morphological parameters (Zhou et

al., 2018).

101
 3D characterization of particle morphology

Fig. 4.40. Correlation between the 3D shape descriptors: (a-b) sphericity against roundness (c-d)
roundness against fractal dimension (e-f) sphericity against fractal dimension.

102
 3D characterization of particle morphology

Fig. 4.40 shows the correlation between these 3D shape parameters at different characteristic

scales. Any of these plots in which a clear correlation can be seen between the two shape

descriptors demonstrates that these two morphological descriptors are not independent. With

respect to the correlations between sphericity and roundness and between roundness and fractal

dimension, respectively, two relatively high correlation coefficients of 0.622 and 0.509 were

obtained. This result agrees well with the results of Fonseca et al. (2012) and Zhou et al. (2018)

who found a strong correlation between sphericity and convexity, sphericity and roundness and

roundness and fractal dimension. The correlation between sphericity and fractal dimension, in

contrast, is found to be quite weak, with a low correlation coefficient of 0.254. This is most

likely because sphericity describes the gross shape on a large scale whereas fractal dimension

describes the surface texture on a fine scale. According to these findings, the correlation between

any two morphological descriptors is dependent on the separation between their characteristic

scales.

Most significant contribution of this chapter is the development and demonstration of a

simplified computation framework to accurately characterize the size and shape of sand grains at

multiple scales through μCT scanning. This framework facilitates 3D reconstruction of the

particle surfaces through image-based computational analyses relatively easily compared to

many other existing methods. Interdependency of morphological descriptors is a topic of

potential interest in this domain of research. Findings from the present study enhanced the

understanding on the statistical interdependency of shape parameters and provided a clear basis

of dependency between any two of them. These results are valuable for future research that can

formulate a fewer statistically independent shape descriptors for enhancing the accuracy and

reducing the computational effort in 3D morphological analysis. The findings of this research

103
 3D characterization of particle morphology

can be utilized for the generation of virtual sand particles using spherical harmonic-based fractal

analysis. These virtual particles can be 3D printed for experimental studies, aiming to explore the

influence of particle morphology on the hydraulic conductivity and shear behaviour under

different conditions. The Kozeny-Carman equation, which uses the specific surface (Ss) of the

sand as an input parameter, is frequently used to estimate hydraulic conductivity (Zheng et al.,

2020; Zheng et al., 2019, 2021). The Ss of sand can be obtained using the measured surface area

and volume of sand grains. By using the mean particle size and particle regularity as the key

parameters, Lashkari & Jamali (2021) developed the empirical equations based on sand-alone

direct shear tests and sand-geosynthetic interface experiments to determine the friction angle.

Therefore, the results of this study can be used to estimate grain-size distributions and sand

properties such as friction angle and permeability. Additionally, the 3D information of particle

geometry can be employed for particle modelling in Discrete Element Method (DEM)

simulations. This integration of realistic particle shapes in DEM simulations will enable a better

understanding of the micro-mechanisms governing the macroscopic behaviours of granular

materials and facilitate the exploration of micro-to-macro correlations.

4.5 Summary

The main aim of this chapter is to develop a simplified framework to precisely characterize the

size and shape of sand particles using advanced image-based techniques and understand the

relation between different morphological parameters. The framework presented in this chapter

can be used to reconstruct the particle surface using micro-computed tomography (μCT) and

spherical harmonic (SH) analysis and then quantify the 3D morphological descriptors using

various computational techniques.

104
 3D characterization of particle morphology

Precise quantification of morphological shape descriptors of particles was done in several

simplified steps. With the help of principal component analysis (PCA), particle size was

measured in terms of three principal dimensions. The particle size distribution curves obtained

by sieve analysis and image analysis were compared for each sand sample. While the major

dimension agrees well with sieving for coarser sands, the intermediate dimension was shown to

provide the best approximation to sieve method for fine particles. The four measures—particle

form, sphericity, convexity, and roundness—were employed to describe the complete particle

shape. The two independent indices, namely, elongation index and flatness index proposed by

several researchers were used to calculate the particle form of three-dimensional grains and

compared. It was found that the simpler of these indices proposed in Zingg’s classification can

be very well used for natural geomaterials such as sands and silts as they do not have extreme

particle shapes because they typically break naturally and develop into smaller, more regular

fragments. To calculate 3D sphericity and convexity, the surface area and volume of the particles

were precisely obtained through SH analysis. To calculate roundness, the potential corners were

identified by measuring the mean curvatures using the SH-reconstructed mesh of the particle

surface. 3D fractal dimension of real sand grains was quantified using spherical harmonics-based

fractal analysis.

In this study, sands with significant variation in size and shape of particles were chosen.

Successful demonstration of the morphological quantifications through various computational

techniques is demonstrated on four distinct types of particles taken from samples of Fine Sand

(FS), Medium Sand (MS), Coarse Sand (CS), and Angular Sand (AS). FS, MS, and CS particles

obtained from river have almost similar shape and different grain sizes. AS particles, which were

produced by crushing of rocks have lower mean values and wider dispersion of all the shape

105
 3D characterization of particle morphology

parameters, indicating that AS particles have greater irregularity, angularity, and rougher surface

texture. The fractal dimension was found to be increasing with grain size. Further, by examining

the correlation between different shape descriptors, it was found that the shape descriptors of the

particle are not independent. The correlation between any two shape descriptors relies mainly on

the distance between the characteristic scales of these parameters.

106
 Chapter 5. Effects of 3D Morphological Contrasts on
Intergranular Interactions and Shear Strain Fields
in Sands

5.1 Introduction

Sand is the primary substance utilized in a variety of construction projects, including those

involving buildings, roads, bridges, and even for land regeneration. The shear behaviour of

granular materials has also been shown to be significantly influenced by particle size and shape.

Several studies have demonstrated a significant relationship between particle morphology and

the macroscopic mechanical characteristics of sand, including as stiffness, shear strength, shear-

induced dilatancy, and interface shear strength with geosynthetics (Afzali-Nejad et al., 2017;

Kandpal & Vangla, 2023; Kodicherla et al., 2020; Kozicki et al., 2012; Lashkari & Jamali, 2021;

Peña et al., 2007; Pillai & Latha, 2022; Sitharam & Nimbkar, 2000; Vangla & Latha, 2016a,

2015, 2016b; Wen et al., 2018). The discrete element method (DEM), which Cundall and Strack

first proposed in 1979, has proven to be an effective method to comprehend how granular

material behaviour is affected by particle shape (Cundall & Strack, 1979). The majority of past

studies either totally disregarded or only examined the shape descriptors in a 2D framework

when it came to measurement of particle shape. 2D shape descriptors can be measured through

image analysis and computational methods using 2D projections of particles (Alshibli & Alsaleh,

2004; Altuhafi et al., 2013; Mora & Kwan, 2000; Pillai & Latha, 2022; Sukumaran & Ashmawy,

2001; Vangla et al., 2018). The 2D shape parameters are obtained from the images taken through

random projection of particles, which results in inaccurate values. The majority of 2D shape

characterization studies published in the literature made use of images obtained when the particle

107
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

was at rest on its plane of maximal stability or dimension. The other orientations for

characterizing particle shape are not taken into account by such measurements. The mean shape

parameters that are determined from the analysis of these images, even when numerous

orientations are employed for image capture, do not sufficiently capture the overall shape of the

particle. Hence, 3D measurement of shape parameters is essential for quantifying the

morphology of particles accurately. X-ray micro-computed tomography (µCT) facilitates the 3D

visualization and measurements of the shape features of granular materials (Alshibli et al., 2015;

Fonseca et al., 2012; Su & Yan, 2018; Zhao & Wang, 2016; Zhou et al., 2018). The

morphological descriptors of tested sands are quantified using µCT and spherical harmonic

analysis which has already been discussed in Chapter 4.

Analysis of the behaviour of small soil samples in standardized laboratory experiments like

the triaxial test and the direct shear test (DST) has contributed significantly to our understanding

of granular material response. The direct shear test apparatus consists of a square box which is

split horizontally into upper and lower halves. During testing, the upper half of the shear box is

translated relative to bottom half of shear box in order to create the shear zone at the mid-height

of the specimen. In geotechnical engineering, DST is one of the most well-known and widely

used laboratory experiments to determine the shear strength parameters of soils. However, the

test suffers several limitations which include the non-uniformity of stresses and strains applied to

the specimen as well as the challenges in defining a failure criterion (Potts et al., 1987; Shibuya

et al., 1997). Several researchers attempted to address the shortcomings by modifying the

conventional direct shear apparatus. Wernick (1979) recommended tightly fixing the upper shear

box with external support to stop both rotational and translational displacements. This

configuration directs the soil sample to travel exclusively vertically during the shearing process.

108
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

This arrangement proposed by Wernick (1979) was later employed by Shibuya et al. (1997) to

investigate the deformation of sand as quasi-simple shear. The results indicated that the interface

friction that occurred along the vertical faces of the specimen constraints the dilation, which

leads to considerable error in estimating the average normal stress on the horizontal shear plane.

It was also advised to maintain the gap between the two halves of the box at a constant value of

roughly 10–20 times the mean particle diameter (d50). However, using a gap of this size could

have unintended effects, like sand particles leakage at the shearing plane during testing (Simoni

& Houlsby, 2006). Jewell & Wroth (1987) and Jewell (1989) recommended fixing the top load

platen to the shear box, which is referred to as a symmetrical loading direct shear test (SLDST)

in this study. They demonstrated that the strains in the deforming sand are sufficiently uniform

when shearing at peak shear resistance in a symmetrical loading arrangement. Later, the

symmetrical arrangement of Jewell was adopted by Lings & Dietz (2005) and (Kandpal &

Vangla, 2023; Vangla & Latha, 2015, 2016a, 2016b) with the expectation that dilation would

proceed unimpeded. It was claimed that this arrangement produced high quality data for peak

friction and dilation angles when the initial gap between the top and bottom portions of the box

was of the order of 5d50 (Lings & Dietz, 2005).

At mesoscale level, the development of strong localized shear zones or bands is a critical

state of deformation that causes progressive failure in granular materials. The macroscopic

stress-strain behaviour of granular media can be intimately related to the evolution of shear

zones. A thorough understanding of strain localization is necessary to understand the

fundamental mechanism underlying the shear behaviour of sands. Common approaches to

studying the stress and strain field and the onset and evolution of shear zone include the discrete-

element method (DEM) (Durán et al., 2010; Feng et al., 2020; Gu et al., 2014; Guo & Zhao,

109
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

2016; Jiang et al., 2010; Kuhn & Bagi, 2004; Nitka & Grabowski, 2021; Oda & Iwashita, 2000;

Zhang & Thornton, 2007) digital image correlation (DIC) techniques (Arshad et al., 2014; Hall,

Bornert, et al., 2010; Hall, Muir Wood, et al., 2010; Kong et al., 2014; Lei et al., 2018; Li & Li,

2009; Shen et al., 2020), micro-computed tomography (µCT), and numerical and analytical

approaches (Amirrahmat et al., 2019; Fu & Dafalias, 2012; Lai et al., 2020; Walker et al., 2013).

The shear zone thickness is generally normalized by d50 of the soil. Shear zone thickness ranging

from 2 to 25 times of d50 have been observed in the laboratory for interface and direct shear tests

(Afzali-Nejad et al., 2017; Dejong et al., 2006; Frost et al., 2012; L. Kong et al., 2014; Lashkari

& Jamali, 2021; Martinez & Frost, 2017a; Shen et al., 2020; Vangla & Latha, 2015, 2016a).

Owing to the shortcomings to assess simultaneously both the global (peak and residual

friction angles, dilation, etc.) and local (development of the shear zone) aspects of the shear

behaviour of granular media, the present chapter aims to investigate the influence of particle size

and shape on mechanical behaviour of sands at multi-scale level through direct shear

experiments and digital image analyses. For this purpose, 3D particle shape was quantified using

X-ray µCT and spherical harmonic analysis. Then a series of tests were carried out using a

symmetrical loading direct shear test under different magnitude of normal stresses. It is possible

to examine the development of a localized shear zone in the tested granular sands by analyzing

particle displacement using digital image correlation (DIC). Full-field shear strains were

measured to study the evolution of localized shear zone in the tested sands and to quantify its

thickness. Finally, the peak and residual friction angles, thickness of localized shear zone are

related to the applied normal stress, mean particle size and overall regularity of particles.

110
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

5.2 Testing programme

The study conducted a series of direct shear tests using a modified symmetrical loading direct

shear set-up, as suggested by Jewell (1989). In this study, the conventional direct shear set-up

was altered by substituting the closed container with a larger container featuring a steel base

plate measuring 190 mm × 160 mm as shown in Fig. 5.41. During sample preparation, the upper

and lower shear boxes, each with dimensions of 80 mm × 80 mm × 22 mm, were positioned on

this rigid base platform and secured with holders. To reduce the friction that developed during

shearing between the vertical walls of box and the sand material, the shear boxes were made

smooth through a thin layer of chromium coating. To facilitate the observation of particle

relocation during direct shear tests, the front side of the shear boxes was constructed using a

transparent acrylic sheet, which had a smooth surface and was 10 mm thick. For capturing the

movements of particles, an Olympus OMD-EM-1-Mark-III digital camera was mounted on a

tripod, positioned in front of the transparent side wall at a lens-to-specimen distance of 18 cm.

Additional lighting was provided using a monochromatic light source to enhance the image

quality and improve the accuracy of particle movement tracking.

In all experiments, the sample was prepared by filling the shear box with dry sand in three

equal layers and compacted to get a desired density, which resulted the initial void ratio ( e0) as

0.712 (Dr ≈ 80%), 0.688 (Dr ≈ 80%), and 0.661 (Dr ≈ 79%) for FS, MS, and CS, respectively. A

normal load was applied after sample preparation, and symmetrical loading conditions were

achieved by fastening the top load platen to the upper half of the box with restraining screws. A

displacement rate of 1.25 mm/min was applied to all tests, and the experiments were stopped

once the shear displacement reached 15 mm. Fig. 5.42 depicts the shearing process at the

beginning (u =0 mm) and at the end (u = 15 mm) of the experiment.

111
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

Fig. 5.41. Diagram illustrating the modified configuration of the direct shear test apparatus.

Fig. 5.42. Sand specimen during shearing process: (a) shear begin (u = 0 mm); (b) shear end (u =
15 mm).

112
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

5.2.1 Global behaviour

The impact of particle morphology on peak and residual shear strength, as well as volume

change, was examined using a series of symmetrical loading direct shear tests. The peak and

post-peak friction angle in these tests is determined using Equation (5.26)

ϕ p=tan
−1
( )
τp
σn (5.2
6
ϕ r =tan
−1
( )
τr
σn
)

where τ p represents peak (or maximum) shear stress and τ r corresponds to post-peak (or residual

shear stress) observed in direct shear tests. Equation (5) considers the normal stress ( σ n) applied

on the shear plane, and the shear strength at a displacement of 15 mm is designated as τ r. To

ensure the repeatability of the test outcomes, multiple tests were performed initially. Fig. 5.43

shows a typical plot obtained from two trials of shear tests with MS specimens under identical

test conditions. The repeated test is symbolized by the letter "R" in this plot. The consistency of

test results is confirmed by the close agreement of measured response from the repeated tests.

Fig. 5.43. Repeatability of direct shear test results of MS specimens: (a) stress-displacement
response; (b) dilation response.

113
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

Fig. 5.44. Particle size effect on direct shear test results: (a, c, e) stress-displacement response;
(b, d, f) dilation response.

Changes in particle size have a significant impact on the void space within a mass of grains,

as well as on the contact surface. These variations in particle size also affect the resistance to

rolling and sliding, as well as influencing the shear strength behavior of granular materials. The

stress, strain and strength responses vary with particle size when the granular material is

subjected to loading. Fig. 5.44 illustrates the shear and dilation responses of sand samples with

114
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

different mean particle sizes. Despite the substantial differences in particle sizes, the results

demonstrate that the shear response of the sands prior to reaching the peak is nearly identical for

all three samples. However, the peak and post-peak responses differ depending on the particle

size. The peak and residual friction angles, as shown in Table 5.9, increase with particle size.

This could be attributed to fine sands having a higher initial void ratio, resulting in lower peak

and residual friction angles. Furthermore, the small rise in the angle of internal friction for

coarser particles can be attributed to the expanded contact area at the particle. Examining the

gradation curves reveals that coarser particles (CS) have a wider gradation (higher Cu value)

compared to FS and MS particles, which contributes to higher shear strength and shear dilatancy.

The ability of particles in wider gradations to rearrange with greater flexibility may explain this

behavior. It is worth noting that the shear strength behavior of granular media is unaffected by

the fact that shape descriptors do not significantly vary with increasing particle size.

Fig. 5.45. Particle shape effect on direct shear results: (a) stress-displacement response; (b)
dilation response.

The shape characteristics determined in this study classify the AS particles as more complex

and irregular compared to the other sands tested, resulting in a higher friction angle due to their

greater irregularity and rougher surface texture. The observed higher peak shear stress and
115
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

dilation in AS particles, as shown in Fig. 5.45, can be attributed to their enhanced particle

interlocking resulting from their increased angularity and fractal dimension. Despite the

comparable particle size distribution between MS and AS, the differences in particle shape

entirely account for the increased shear strength and dilation. Interestingly, despite having a

higher initial void ratio (e0), the AS specimens exhibited greater strength due to the greater

irregularity and rougher surface of the AS particles in comparison to the MS particles. These

findings provide compelling evidence for the effect of particle shape on the shear behavior of

sands. The experimental findings are summarized in Table 5.9 in terms of peak and residual

friction angles under various normal stresses. An important finding from Table 5.9 is that the

peak friction angle exhibited a decrease as the normal stress increased, which is consistent with

the results obtained from the direct shear tests conducted in this study. This observation aligns

with the findings of a previous study by Tiwari et al. (2005), where laboratory ring shear tests

were performed on sands across a range of normal stresses from 10 to 250 kPa, demonstrating a

decrease in the peak friction angle with increasing normal stress.

Table 5.9 Peak and residual friction angles in direct shear tests.
Sand Initial void Normal stress, Peak friction angle, ϕp Residual friction angle,
Type ratio, e0 σn (kPa) (degrees) ϕr (degrees)
FS 0.712 40, 80, 120 31.85, 31.15, 30.40 25.95, 25.32, 26.22
MS 0.688 40, 80, 120 34.03, 32.71, 32.01 29.03, 28.04, 29.38
CS 0.661 40, 80, 120 38.48, 37.10, 36.18 31.00, 32.14, 30.79
AS 0.794 40, 80, 120 48.89, 46.41, 44.31 39.00, 38.38, 36.78

5.2.2 Local responses

The analysis of granular material movements has become a highly efficient non-destructive

method for investigating the thickness of localized shear zones (Dejong et al., 2006; Kong et al.,

2014; Vangla & Latha, 2016; Martinez & Frost, 2017; Shen et al., 2020). The non-contact optical

116
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

deformation measurement technique is highly attractive as it overcomes the limitations of

traditional contact techniques. The direct shear test, a widely used and simple laboratory test for

granular materials, provides insights into the macroscopic responses during shearing under a

constant normal stress, with a specified horizontal shear plane at the junction of the shear box.

To determine the thickness of the localized shear zone in this study, the digital image correlation

(DIC) technique is employed in conjunction with direct shear experiments. Fig. 5.46 illustrates

the experimental set-up during video recording of the experiment.

The DIC measurement technique relies on detecting the surface characteristics of the sample

as a means of measuring deformation. When the sample surface has a uniform light intensity, it

becomes challenging to track particle movements accurately using a digital camera for

deformation measurement. Therefore, in order to ensure precise measurements with the DIC

method, the sand surface needs to be prepared with artificial speckle patterns. These patterns

consist of randomly varying grey-level intensities that serve as carriers of deformation

information. By introducing these artificial speckle patterns, accurate measurements can be

obtained through DIC analysis.

117
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

Fig. 5.46. Experimental set-up used in this study: (a) showing the shear box and the arrangement
for video capturing; (b) providing a closer view of the transparent shear box.

In order to create a specimen with random variation in intensity level for more accurately

capturing the movement of particle, about 20% of the particle were coloured with black

permanent marker ink. To prevent any alterations to the physical properties and surface

characteristics of the sand particles, the ink was diluted with water in a ratio of 1:25.

118
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

Fig. 5.47. Localized shear zone thickness in AS specimen: (a) variation of shear strain field; (b)
typical DIC data at 15 mm displacement.

The open-source software tool 'Ncorr' (Blaber et al., 2015) with a user-friendly graphical

interface was employed for DIC analysis in MATLAB. During the shear testing, high-quality

videos were recorded using a digital camera, resulting in a magnification of approximately 0.023

mm/pixel. To enhance image quality, a monochromatic light source was used to track the particle

movements. The DIC technique compared the reference image, typically taken at the beginning

of the test, with subsequent images to track the progressive deformation of the sand specimen.

Subset-based DIC methods were utilized, which involved dividing the reference image into

smaller subsets or subwindows. These subsets were then tracked in the current image, assuming

uniform deformation within each subset. Within the ‘Ncorr’ software, subsets were initially

defined as circular group of points positioned at integer pixel locations in the reference

configuration. During the computation process, subsets consisting of 50 to 130 pixels were

selected, with a spacing of 10 pixels between adjacent measurement points. The comparison of

corresponding subsets from the reference and deformed images was performed using normalized

cross-correlation criteria, following the approach described by Blaber et al. (2015). This allowed

for the determination of full-field displacements and shear strains within the defined region of

119
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

interest (ROI). The shear strain variation with vertical distance is depicted in Fig. 5.47(b),

revealing a notable concentration of shear strains at the junction of the shear boxes. This

concentration of shear strains led to the measurement of a shear zone thickness of 15.52 mm in

this particular case.

after 1sec after 3 sec

after 5 sec after 80 sec

after 480 sec after 720 sec or 15 mm displacement

Fig. 5.48. Distribution of shear strain field in AS specimen at different time intervals (σn = 40
kPa).

Fig. 5.48 presents the relationship between shear strain and horizontal displacement at

different time intervals, providing insights into the formation and evolution of the shear zone as

well as the deformation behavior of the tested sands during shearing. Initially, the sample

undergoes compaction due to the relative movement between the upper and lower shear boxes.

120
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

As a result, sand particles continuously readjust their positions in response to external changes,

leading to localized regions with larger shear strains scattered throughout the sample. With

increasing shear displacement, the shear zone becomes more prominent as the shear strain

concentrates at the mid-height of the shear boxes and propagates from the sides towards the

middle. The development of the shear zone is more pronounced in this region due to intensified

relative motion between the sand particles. The shear zone exhibits a diagonal shape from the

lower left to the upper right, indicating a concentration of sand particles that have undergone

significant shearing and displacement deformation. This phenomenon can be attributed to the

shear process involving the fixation of the upper shear box and the movement of the lower shear

box.

(a) (b)

Fig. 5.49. (a) Distribution of horizontal displacements of AS specimen (σn = 40 kPa); (b)
distribution of vertical displacements of AS specimen (σn = 40 kPa). (unit: mm)

During the process of shearing, the particles in lower box experienced horizontal

displacement corresponding to the distance travelled by the shear box. On the other hand, the

particles in upper box did not undergo displacement, except for those near the junction. Near the

junction, the particles of the upper box were dragged by the particles in the lower box, causing

the evolution of localized shear zone (see Fig. 5.49(a)). Also, the upper shear box moves

121
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

uniformly in an upward direction during shearing, as evidenced by the relatively uniform

distribution of vertical displacements in the upper half of the shear box (see Fig. 5.49(b)). This

ensures, during shearing in the modified symmetrical loading test arrangement, the rotation of

the shear box is eliminated causing more uniform stress state on the central plane.

Fig. 5.50. Variation of shear strains as a function of vertical distance from the bottom of ROI at a
horizontal displacement of 15 mm: (a) FS; (b) MS; (c) CS; (d) AS.

The average variations in shear strains (γ) are determined by averaging the shear strain

distribution data in the horizontal direction as shown in Fig. 5.50. Table 5.10 provides the

thickness of the shear zone (ts) and the ratios of ts to the mean particle size (d50) for different

122
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

tests. Considering the variation in particle size, shape and applied normal stress, the experimental

results indicate that the thickness of the shear zone (ts) for the tested sands ranges from 12.32 mm

to 18.15 mm. Additionally, ts /d50 ratios are between 5.85 and 18.44 for the tested sand particles.

It can be observed that shear zone thickness is increasing with increase in particle size and ts/d50

is decreasing with increase in d50, which align with previous studies (Kong et al., 2014; Vangla &

Latha, 2015). As seen in Table 5.10, the shear zone thickness exhibits an opposite trend under

the application of normal stress. Because of higher normal stresses, the sample's deformation

behaviour was constrained, resulting in a thinner shear zone.

A close examination of the particle shape values obtained in Chapter 4 revealed that AS

particles have higher non-sphericity, angularity, and irregularity among all the tested sand

particles. Particle interlocking, relative particle translation, and specimen propensity to develop a

shear band appear to be primarily influenced by particle roundness and surface roughness

(Amirrahmat et al., 2019; Wu et al., 2021). It was observed from Table 5.10 that higher

irregularity and complex surface texture of AS particles lead to a reduction in shear zone

thickness. The increased irregularity of particles leads to enhanced interlocking and reduced

particle mobility, resulting in a narrower shear zone. The results of the present study for the

investigation into the influence of particle morphology are in line with the explanation provided

by (Peña et al., 2007; Liu et al., 2015; Wu et al., 2021), considering that this study provides

quantitative measurements of shape indexes using 3D images of sand grains to investigate the

influence of particle morphology.

Table 5.10 Shear zone thickness for the tested sands.

Initial void Normal stress, Shear zone thickness
Test type
ratio, e0 σn: kPa ts, mm ts/d50
FS 0.712 40, 80, 120 14.38, 14.22, 12.32 18.44, 18.23, 15.79

123
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

MS 0.688 40, 80, 120 16.74, 16.11, 15.43 10.59, 10.20, 9.76
CS 0.661 40, 80, 120 18.15, 17.91, 17.14 6.19, 6.11, 5.85
AS 0.794 40, 80, 120 15.52, 14.72, 14.41 8.82, 8.36, 8.19

Fig. 5.51 demonstrates the correlation between the 3D shape parameters of sand particles

and the macroscopic shear strength parameters, as well as the thickness of the shear zone

obtained from modified direct shear tests. The statistical analysis reveals that the shape of the

grains, as indicated by particle regularity and fractal dimension, significantly influences the shear

behaviour and the overall shear strength mobilized at the shear plane. When considering the

effect of particle shape alone, it is observed that AS particles with a mean regularity of 0.64 and

fractal dimension of 2.300 exhibit higher peak and residual friction angles compared to MS

particles with a mean regularity of 0.71 and fractal dimension of 2.283. The irregularity of shape

and rough surface hinder particle rotation, leading to greater particle interlocking and increased

tendency of dilation. Similar findings were also reported by Wu et al. (2021) in their study on

sand subjected to biaxial shearing. The thickness of the shear zone is measured to be 14.88 mm

and 16.09 mm for AS and MS particles, respectively, indicating that particle irregularity does not

necessarily lead to a greater shear zone thickness. The difficulty in particle rotation due to their

better interlocking could explain the lower shear zone thickness in AS particles. Moreover, the

peak and post-peak friction angles decrease by approximately 30% for MS particles compared to

AS particles. When considering the effect of particle size, coarser particles result in higher

macroscopic shear response and thicker shear zone compared to fine particles. The

comprehensive investigation conducted in this study utilized modified direct shear tests and

various imaging techniques to examine the particle kinematics and shear mechanisms in four

sands with different morphological characteristics. The use of μCT scanning and spherical

124
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

harmonic reconstruction allowed for accurate quantification of shape parameters at multiple

scales for representative particles in the sand samples. The thickness of the shearing zone was

measured during direct shear tests using DIC analysis, and its relationship to particle size and

shape was examined. All of the scientific data needed to precisely correlate the particle size and

shape to the shear mechanisms was provided by the combination of digital imaging techniques.

The advantage of this study over earlier studies on the subject is that it uses digital image

analysis techniques that are more sophisticated, accurate, and quicker than most other widely

used approaches.

125
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

Fig. 5.51. Correlation between particle morphology and the direct shear test results at multi-scale
level.

126
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

5.3 Summary

Direct shear tests were conducted to investigate the multi-scale aspects of the mechanical

behaviour of four types of sand with various shape characteristics. X-ray µCT and spherical

harmonic analysis were used to quantify the 3D shape of the sand particles. To achieve the

symmetrical loading conditions and capture the movement of the particles during shearing for

shear band analysis, the direct shear apparatus was modified. The following major conclusions

can be drawn from the analysis and interpretation of the experimental data.

(a) When compared to sand particles of comparable mean particle size and different shape

characteristics, the angular sand (AS) particles have lower sphericity, lower roundness,

and higher fractal dimension than the medium sand (MS) particles, according to the

quantifiable statistics of 3D shape descriptors. AS particles have higher irregularity and

roughness; their average regularity is 0.64 compared to 0.71 for MS particles, and their

average fractal dimension is 2.3 as opposed to 2.283 for MS particles.

(b) Particle irregularity has a significant impact on the macroscopic shear strength. Dilation

as well as peak and residual friction angles were reported to be higher for AS particles.

The irregularity of shape and rough surface texture resist the rotation of particles, leading

to increased interlocking among particles and enhanced dilatancy.

(c) Increase in normal stress led to a decrease in the peak friction angles and dilation as

higher normal stresses restrained the deformation behaviour of the granular material.

(d) Particle shape significantly affected the thickness of the shear zone (ts) in direct shear

tests. For AS particles and MS particles, respectively, the ts/d50 ratios ranged from 8.19 to

8.82 and 9.76 to 10.59.

127
 Effects of 3D morphological contrasts on intergranular interactions and shear strain fields in sands

(e) Shear zone thickness is smaller for irregular particles than the particles with regular

morphology. This result is explained by the difficulty in rotation that irregular particles

encounter because of their stronger interlocking.

(f) Peak and residual friction angle as well as shear zone thickness increase with increase in

particle size. The ratio of the thickness of the shear to the mean particle size (ts/d50)

decreases as the particle size increases. For fine, medium, and coarse sands, this ratio

ranged from 15.79 to 18.44, 9.76 to 10.59, and 5.85 to 6.19, respectively.

128
 Chapter 6. Effect of Particle Size of Sand on Sand-
Geosynthetic Interfaces

6.1 Introduction

The interfaces between the soil and geosynthetic materials are identified as the most

vulnerable areas in geosynthetic reinforced structures. Therefore, understanding the sand-

geosynthetic interactions at the elementary level is crucial for designing reinforced earth

structures (Afzali-Nejad et al., 2017; Lashkari & Jamali, 2021; Pillai & Latha, 2022; Vangla &

Latha, 2015, 2016b, 2016a). The interlocking mechanism formed during shearing between the

sand and the contact material is also a significant factor in interface shear strength. Furthermore,

the degree of interlocking is dependent upon the size and shape of sand particles, as well as the

surface roughness of the contact material (Afzali-Nejad et al., 2017; Dove & Frost, 1999; Frost et

al., 2012; Jewell & Wroth, 1987; Lashkari & Jamali, 2021; Uesugi & Kishida, 1986; Vangla &

Latha, 2015, 2016b, 2016a). The significance of grain morphology and size concerning the

behavior of sand-geosynthetic interfaces has been highlighted by multiple researchers (Afzali-

Nejad et al., 2017; Lashkari & Jamali, 2021; Pillai & Latha, 2022; Vangla & Latha, 2015,

2016a). Yang et al. (2016, 2022) reported that changes in particle size have an impact on the

shape structure of sand grains. Consequently, when studying the mechanical properties of sands

in relation to particle size changes, it becomes crucial to analyze particle shape characteristics.

Regarding characterizing particle shape, a significant portion of previous research completely

ignored or only evaluated shape attributes using a two-dimensional approach when measuring

particle shape. (Mora & Kwan, 2000; Sukumaran & Ashmawy, 2001; Vangla et al., 2018).

However, the accuracy of these measurements is compromised because the two-dimensional

shape parameters are derived from images obtained through arbitrary particle projections.

129
 Effect of particle size of sand on sand-geosynthetic interfaces

Therefore, it is crucial to employ 3D measurement techniques, such as X-ray micro-computed

tomography (µCT), to accurately quantify particle morphology and visualize shape features

(Fonseca et al., 2012; Zhao & Wang, 2016; Zhou et al., 2018). Several researchers advised for

reconstructing the particle surface of µCT images in order to determine the geometrical

properties of grains more precisely (Zhao & Wang, 2016; Zhou et al., 2015, 2018; Zhou &

Wang, 2015, 2017).

The direct shear test seems the preferred method for studying the engineering characteristics

of interfaces between sand and geosynthetics. Several researchers modified conventional direct

shear apparatus suitable to investigate the sand-geosynthetic interfaces (Anubhav & Basudhar,

2013; Lee & Manjunath, 2000; Vangla & Latha, 2015, 2016b, 2016a). The particle morphology,

the initial density of the specimen, applied normal stress, and surface roughness of the contact

material has a substantial impact on the load-deformation behavior of sand-geosynthetic

interfaces (Afzali-Nejad et al., 2017; Arshad et al., 2014; Dejong et al., 2006; Feng et al., 2020;

Frost et al., 2012; Han et al., 2018; Lashkari & Jamali, 2021; Tehrani et al., 2016; Vangla &

Latha, 2016b, 2016a, 2015; Zhao et al., 2015). In a comprehensive study by Han et al. (2018), a

series of interface shear tests were conducted to examine the impacts of particle shape, sand

gradations, and surface roughness of steel interfaces on four distinct levels of rusting. When

examining uniform sands, they discovered that lightly rusted steel interfaces exhibited a 50%

higher interface friction angle than smooth steel surfaces, while for graded sands, this increase

was approximately 10%. Notably, the observed increase in interface friction angle was minimal

as the surface roughness was further increased, progressing from a lightly rusted to a rusted or

heavily rusted steel surface. At a local level, the concentration of shear strains at the interfaces

between soil and geosynthetic materials is a crucial phase of deformation that ultimately results

130
 Effect of particle size of sand on sand-geosynthetic interfaces

in progressive failure (Afzali-Nejad et al., 2017; Dejong et al., 2006; Lashkari & Jamali, 2021;

Martinez & Frost, 2017a). Numerous interfaces, such as piles, geosynthetic reinforcement, and

nails, can suffer from this failure, which can be extremely detrimental. Discrete element

simulations by Feng et al. (2020) on sand-geotextile interfaces concluded that shear zone

thickness varies from approximately 2.4 to 3 times of mean particle size (d50). Uesugi et al.

(1988) state that the average shear zone thickness varies between 3 to 4 times d50 considering

very dense-rough steel interfaces. In the case of dense sand and smooth steel interfaces, the sand

particles can slide smoothly on the contact surface without forming a shear band. In their

extensive experiments on sand-steel interfaces, Martinez et al. (2015) found that when subjected

to axial and torsional axisymmetric conditions, the shear band thickness for medium sands

ranged from 6 to 7d50, while for fine sands, it was in the range of 12 to 14d50.

Due to the challenges involved in evaluating the interface behaviour of granular media from

both the global (peak and residual friction angles, dilation, etc.) and local (development of the

shear zone) perspectives, this chapter aims to investigate the role of particle size on sand-

geosynthetic interfaces at the multi-scale level through modified interface shear tests and

advanced digital image techniques. As the particle size varies, it seems that the shape

characteristics of the particles also change to some degree. Because of this, the 3D particle shape

was first measured using X-ray micro-computed tomography and spherical harmonic analysis.

Subsequently, a series of interface shear tests were conducted between three graded sands of

different particle size contacting with a plain weave woven geotextile and a smooth high-density

polyethylene (HDPE) geomembrane. By analyzing particle displacement using digital image

correlation (DIC), it is possible to investigate the localized deformation of sands near the sand-

geosynthetic interfaces. Full-field displacements and shear strains were measured to quantify the

131
 Effect of particle size of sand on sand-geosynthetic interfaces

shear band's thickness. Finally, the peak and residual friction angles and shear band thickness are

related to the applied normal stress, geosynthetic surface asperities, mean particle size, overall

regularity, and fractal dimension of particles.

6.2 Experiments

To gain insights into the multi-scale behavior of sand-geosynthetic interfaces, a series of

interface shear tests were conducted, accompanied by measurements of localized particle

movements. Several researchers have pointed out the limitations of the traditional direct shear

apparatus in evaluating soil-geosynthetic interfaces. These drawbacks encompass the sagging

and wrinkling of the geosynthetic layer, the challenge of obtaining residual parameters with a

limited shear displacement and reducing the contact between the soil and geosynthetic surfaces

while undergoing shearing. A few of these limitations were addressed in earlier studies (Afzali-

Nejad et al., 2017; Anubhav & Basudhar, 2013; Lashkari & Jamali, 2021; Lee & Manjunath,

2000; Vangla & Latha, 2015, 2016b) by substituting a rigid platform for the lower shear box of

the conventional interface shear apparatus where the geosynthetic specimen is securely affixed

either through adhesive bonding or by firmly attaching it using friction clamps. In this study, the

lower shear box is replaced by an L shape rigid steel platform which includes a steel plate with

dimensions of 190 mm × 160 mm, and the geosynthetic layer is firmly affixed to it to prevent

any wrinkling during tests. The upper shear box, measuring 80 mm × 80 mm × 44 mm, was

filled with dry sand and positioned atop the rigid platform for shearing. A thin layer of chromium

coating was applied to make the box smooth to decrease the friction that occurred during the

shearing process between the vertical walls of the shear box and the sand material. To facilitate

the observation of particle movement during direct shear tests, one side of the upper shear box

was made transparent using a 10 mm thick acrylic sheet with a smooth surface. To capture the

132
 Effect of particle size of sand on sand-geosynthetic interfaces

particle motion, a tripod-mounted Olympus digital camera was positioned 18 cm away from the

transparent wall. The use of monochromatic light provided additional illumination, enhancing

the image quality for tracking the particles. The schematic diagram of the experimental set-up

modified for conducting DIC analysis is shown in Fig. 6.52. Each sample was prepared by

pouring the dry sand into the shear box in three layers and each layer was compacted to achieve

the desired density. This resulted in initial void ratios (e0) of 0.712 (Dr ≈ 80%) for FS, 0.688 (Dr

≈ 80%) for MS, and 0.661 (Dr ≈ 79%) for CS. After sample preparation, a normal load was

applied, and symmetrical loading conditions were achieved by securing the top load platen to the

upper half of the box with restraining screws. The experiments were conducted with a

displacement rate of 1.25 mm/min, and the tests were terminated when the shear displacement

reached a value of 15 mm.

133
 Effect of particle size of sand on sand-geosynthetic interfaces

Fig. 6.52. Schematic diagram of the modified interface test apparatus for DIC analysis.

Initially, a few preliminary experiments were conducted to ensure that the interface shear

tests could be replicated consistently. Subsequently, a series of interface tests were carried out,

involving a woven geotextile and a HDPE geomembrane in conjunction with three graded sands

of varying morphology and mean particle size. Many researchers like Dove & Frost (1999) and

Dove & Jarrett (2002) have classified geosynthetic interfaces as “dilative” and “non-dilative”,

based on the surface roughness of the geosynthetic material and the volume change response of

the soil-geosynthetic interface system. According to these studies, dilative interfaces are the

interfaces in which soil grains shear against a rough geosynthetic material like geotextile,

causing significant volume changes at the interface. In contrast, non-dilative interfaces are the

interfaces in which soil grains shear against a smooth geosynthetic material like geomembrane

with small volume changes. Hence the geotextile and geomembrane interfaces in the present

study are classified as dilative and non-dilative interfaces, respectively. The terms “dilative” and

“non-dilative” interfaces in this thesis are defined within the geotribology framework, in which

volume changes at the interfaces are related to the smoothness of the geosynthetic material.

However, dilation or contraction at the interface depends on many factors such as relative

density, particle morphology and particle strength. Hence the classification of interfaces into

dilative and non-dilative interfaces is limited to the current study, and it cannot be generalized.

The results of the interface shear tests between sand and geotextile are presented in Fig. 6.53.

Table 6.11 summarizes the peak and residual friction angles data obtained from direct shear tests

on sand-alone and interface shear tests at the three different normal stresses. In contrast to the

sand-alone direct shear tests, where dilation is caused by particle interlocking, the interface shear

tests between sand and geotextile show dilation as a result of particle interlocking with the

134
 Effect of particle size of sand on sand-geosynthetic interfaces

surface irregularities of the geotextile. The stress-displacement curves plotted in Fig. 6.53 show

that the FS-GTX interfaces produced higher friction angles than the MS-GTX and CS-GTX

interfaces. One of the reasons could be the better interlocking of fine sands with surface

asperities of geotextile, causing higher shear strength and dilation. Because coarse sand particles

are larger than the concavity of the geotextile's asperities, they were unable to properly interlock

with them during shearing, leading to poor interlocking and reduced dilation. This resulted in

fine sands having higher interface friction angles than medium and coarse sands, highlighting the

impact of dilation on peak shear strength.

135
 Effect of particle size of sand on sand-geosynthetic interfaces

Fig. 6.53. Particle size effect on sand-GTX interfaces: (a, c, e) stress-displacement response; (b,
d, f) dilation response.

During shear tests involving sand and geomembrane (GMB), the sand particles slide on the

surface of the GMB without significant dilation due to the smoother surface of the GMB.

Consequently, the sand-GMB interfaces exhibit reduced mobilization of shear strength and

dilation compared to the sand-woven geotextile (GTX) interfaces. This is attributed to the lower

friction or entanglement between the sand particles and the surface characteristics of the GMB.

136
 Effect of particle size of sand on sand-geosynthetic interfaces

Fig. 6.54 shows the pictorial representation to illustrate this difference in mobilization between

the two interfaces. The effect of particle size on sand-GMB interfaces is exemplified through

stress displacement and dilation response in Fig. 6.55. It can be observed that FS-GMB

interfaces resulted in higher peak stress as compared to MS-GMB and CS-GMB interfaces. This

may be due to the number of contacts increasing with the decreasing particle size, which may

result in a large number of grooves with less spacing leading to higher interface shear strength.

The experimental findings in this study suggest that the interlocking mechanism has a greater

influence on the interface shear strength and shear dilatancy. The sand-GMB interfaces

experience a maximum dilation of only 0.1 mm, whereas the sand-GTX interfaces experience a

maximum dilation of about 0.46 mm. Moreover, the interfaces between sand and geotextile

(GTX) demonstrate maximum shear stress exceeding 75 kPa, and they exhibit noticeable strain

softening beyond the peak, as illustrated in Fig. 6.53. In contrast, the interfaces between sand and

geomembrane (GMB) only reach a maximum shear stress of approximately 50 kPa, without a

distinct peak evident in Fig. 6.55.

137
 Effect of particle size of sand on sand-geosynthetic interfaces

Fig. 6.54. Interlocking mechanism between sand particles with different sizes and the surface
texture of geosynthetic materials: (a-c) interfaces between sand and geotextile (GTX); (d-f)
interfaces between sand and geomembrane (GMB).

Fig. 6.55. Particle size effect on sand-GMB interfaces: (a, c, e) stress-displacement response; (b,
d, f) dilation response.

There is a consensus that the surface roughness of the contact material affects the interface

friction and shear behavior between the geosynthetics and adjacent materials in a geotechnical

application (Afzali-Nejad et al., 2017, 2018; DeJong & Westgate, 2009; Lashkari & Jamali,

138
 Effect of particle size of sand on sand-geosynthetic interfaces

2021; Martinez et al., 2015; Martinez & Frost, 2017b; Uesugi & Kishida, 1986a; Vangla &

Latha, 2016b). The significance of evaluating surface topological properties of continuum

material has resulted in the suggestion of several indices, among which average roughness (Ra) is

studied here, which is defined as the mean value of the absolute profile height deviations within

a specified evaluation length. It is calculated by measuring the deviations from the mean line, as

indicated in Equation (6.27)

√
l
(6.2
∫ z ( x ) dx
0
Ra = 7)
l

wherein z ( x ) is the profile calculated from the mean line at x, and l is the sampling length. In

this study, a stylus profilometer was used to measure the texture profile of tested geosynthetics

whose surface profiles are given in Fig. 6.56 and yielded an average roughness ( Ra ¿ of 58.85 µm

for geotextile and 4.16 µm for geomembrane.

Fig. 6.56. Surface profile of geosynthetics: (a) Woven geotextile (GTX); (b) Smooth
geomembrane (GMB).

Several researchers (DeJong & Westgate, 2009; Lashkari & Jamali, 2021; Martinez et al.,

2015; Uesugi & Kishida, 1986b; Vangla & Latha, 2015, 2016a) presented substantial

139
 Effect of particle size of sand on sand-geosynthetic interfaces

experimental evidence demonstrating the impact of normalized roughness (Rn) on the frictional

and volume change characteristics of soil-structure interfaces.

Fig. 6.57. Definition of Rmax and Rt (redrawn from Tovar-Valencia et al., 2018)

Equation (6.28) gives the normalized roughness (Rn) of geosynthetic materials interfacing with

sands with a given mean particle size.

Rmax (l=d50 ) (6.2

Rn =
d 50
8)

here Rmax represents the difference in height between the highest peak and lowest valley within a

profile, measured over a sampling length of l = d50. The definition of calculating the Rmax is

explained in Fig. 6.57. In this study, the range of Rn values is between 0.068 and 0.294 for sand-

GTX interfaces, while for sand-GMB interfaces, the range is between 0.009 and 0.028.

According to Martinez & Frost (2017b), when the normalized roughness (Rn) of rough sand-

structure interfaces exceeds a value of 0.10 to 0.15, the height of the contact surface asperities is

140
 Effect of particle size of sand on sand-geosynthetic interfaces

significant enough to induce sand dilation, resulting in shear failure. Conversely, when sand

encounters smooth structures characterized by low values of normalized roughness (Rn < 0.03),

the sand particles are much larger than the surface asperities of the structure. Consequently, the

primary interaction between the sand particles and the smooth contact surface involves sliding

without significant dilation occurring.

The interaction between geosynthetic materials and sand particles can be assessed by

evaluating the frictional efficiency. The frictional efficiency (Ef) of these interfaces can be

calculated using Equation (6.29) as given by Lashkari & Jamali (2021)

tanδ (6.2
Ef=
tanϕ
9)

here the mobilized friction angle (ϕ) is obtained from sand-alone direct shear tests carried out

under identical conditions. For the interfaces between FS, MS, and CS particles in contact with

GTX, the average peak frictional efficiencies are, respectively, 1.03, 0.82, and 0.67. In terms of

residual friction angles, the average frictional efficiencies observed for sand-GTX interfaces are

0.91, 0.74, and 0.68 for fine sand (FS), medium sand (MS), and coarse sand (CS), respectively.

The improved interlocking of FS particles with GTX surfaces, which increases shear strength, is

further demonstrated by the higher frictional efficiencies for FS-GTX interfaces. The average

peak frictional efficiencies for non-dilative interfaces are 0.61, 0.56, 0.40 for FS, MS, and CS,

respectively. A notable observation from Table 6.11 is the decrease in the peak friction angle

with increasing normal stress. Tiwari et al. (2005) confirmed this trend through laboratory ring

shear tests on sands with normal stresses ranging from 10 to 250 kPa, demonstrating a decrease

in the peak friction angle as the normal stress increases. Roberts (2001) provides a

141
 Effect of particle size of sand on sand-geosynthetic interfaces

comprehensive explanation for this phenomenon, attributing it to the convex shape of the yield

locus, where an increase in normal stress leads to a decrease in the friction angle. The interface

test results in this study align with these observations.

Table 6.11 Peak and residual friction angles.

Test type Initial void Normal Peak friction Residual friction

ratio, e0 stress, σn: angle: degrees angle: degrees
kPa
Sand-alone FS 0.712 40 31.85 25.95
80 31.15 25.32
120 30.42 26.22
MS 0.688 40 34.03 29.03
80 32.71 28.04
120 32.01 29.38
CS 0.661 40 38.48 31.01
80 37.11 32.14
120 36.18 30.79
Sand-geotextile FS- 0.712 40 33.92 23.14
GTX 80 31.02 22.54
120 30.96 25.22
MS- 0.688 40 29.29 18.58
GTX 80 27.67 22.97
120 27.36 24.18
CS- 0.661 40 27.36 20.32
GTX 80 26.54 23.26
120 26.15 23.29
Sand- FS- 0.712 40 18.69 15.26
geomembrane GMB 80 19.62 17.31
120 21.98 18.86
MS- 0.688 40 17.48 13.89
GMB 80 19.58 16.75
120 21.43 16.94
CS- 0.661 40 14.50 11.44
GMB 80 17.21 16.20
120 17.33 16.87

142
 Effect of particle size of sand on sand-geosynthetic interfaces

6.3 DIC analysis of sand-geosynthetic interfaces

An emerging trend in geotechnical engineering is the use of non-destructive techniques to

measure the movement of individual particles at a local level to quantify the thickness of

localized shear zones between granular soils and contacting surfaces. Several researchers

(Arshad et al., 2014; Martinez et al., 2015; Shen et al., 2020; Tehrani et al., 2016; Uesugi &

Kishida, 1986a; Vangla & Latha, 2016a) have utilized this technique to examine the local

deformation and shear band thickness in granular materials. The non-contact optical deformation

measurement method, particularly the Digital Image Correlation (DIC) technique employed in

this study, offers advantages over conventional contact techniques. Fig. 6.58 presents a

photograph of the experimental setup and the selection of the region of interest for image

processing. The basis of the DIC measurement technique is the identification of the sample's

surface characteristics as a deformation carrier. It is challenging to track the movement of the

particles to measure deformation using a digital camera if the light intensity of the surface of the

sample is uniform. Therefore, the sand surface must be produced with artificial speckle patterns

that randomly vary in grey level intensity to ensure accurate measurements when applying the

DIC method. In the present study, about 20% of the particles were colored with black permanent

marker ink to produce a specimen with random variation in intensity level to capture the

particles’ movement better accurately. The ink was diluted with water in a 1:25 ratio to prevent

any changes to the physical properties and surface characteristics of the sand particles.

For DIC analysis in MATLAB, ‘Ncorr’ (Blaber et al., 2015) was employed as an open-

source tool with a graphical user interface. During interface shear experiments, the Olympus

OMD-EM-1-Mark-III mirrorless digital camera was used to record 4k resolution videos with 30

frames per second (movie image size: 3840 x 2160 pixels). These settings result in a

143
 Effect of particle size of sand on sand-geosynthetic interfaces

magnification of approximately 0.023 mm/pixel. An extra light source was employed for

tracking particle displacements, as it offers improved illumination and enhances the quality of

captured images. The fundamental concept of DIC method has been previously described in

Chapter 5. In Ncorr, circular regions known as subsets are employed, and they are selected for

computation with sizes ranging from 50 to 130 pixels, while maintaining a 10-pixel gap between

consecutive measurement points. By comparing the corresponding subsets of the two images

before and after deformation through cross correlation criteria (Blaber et al., 2015), the full field

shear strain can be obtained. By plotting the measured height of the shear zone from the interface

against the measured shear strain, thickness of the localized shear zone can be obtained. These

computations are explained in Fig. 6.59 for an FS-GTX interface tested at a normal stress of 40

kPa until a horizontal displacement of 15 mm is reached. Variation of the shear strain field with

vertical distance for this test is shown in Fig. 6.59(a). An intensity scale is provided to read the

shear strain in different zones and understand the evolution of the shear strain field. Fig. 6.59(b)

shows the shear strain variation with vertical distance. As seen from Fig. 6.7(b), significant

concentration of shear strains is measured at vertical distances between 0 and 8.9 mm from the

GTX contact plane. Hence the thickness of the shear zone for this case is obtained as 8.9 mm, as

depicted in Fig. 6.59(b).

144
 Effect of particle size of sand on sand-geosynthetic interfaces

Fig. 6.58. Experimental setup: a) video capturing arrangement; (b) modified shear box with
transparent wall.

145
 Effect of particle size of sand on sand-geosynthetic interfaces

Fig. 6.59. Measurement of shear zone thickness in FS-GTX interface: (a) shear strain field
contours; (b) typical DIC data for measuring shear zone thickness.

The shear strain distribution data is averaged along a horizontal direction to determine the

average variation of shear strains (γ) as a function of vertical distance measured from the

geosynthetic surface. Curves in Fig. 6.60 show the evolution of shear strain (γ) as a function of

vertical distance (z) at 15 mm horizontal displacement. Table 6.12 summarizes the shear zone

thickness (ts) and the ratios of ts to the mean particle size (d50) for different tests. The

experimental findings demonstrate that for sand-GTX interfaces, the ts values range from 7.8 mm

to 15.2 mm, while for sand-GMB interfaces, the ts values range from 4.8 mm to 7.9 mm. The ts

values vary based on factors such as the mean particle size range, particle regularity, and the

magnitude of applied normal stresses. Additionally, ts/d50 ratios are between 4.88 and 11.41 for

sand-GTX interfaces and 2.22 to 7.43 for sand-GMB interfaces. In accordance with earlier

researchers (Kong et al., 2014; Vangla & Latha, 2015), it can be seen that shear band thickness

increases with increasing particle size and that the normalized thickness of shear bands (ts/d50)

decreases with increasing particle size as also observed through Fig. 6.61. Furthermore, the shear

band thickness decreases as applied normal stress increases as observed in Table 6.12 and Fig.

146
 Effect of particle size of sand on sand-geosynthetic interfaces

6.61. The deformation behaviour of the sample was constrained by the high normal stresses,

producing a thinner shear band.

Fig. 6.60. Shear strains with vertical distance from the geosynthetic plane (u = 15 mm): (a) FS-
GTX; (b) MS-GTX; (c) CS-GTX; (d) FS-GMB; (e) MS-GMB; (f) CS-GMB.

147
 Effect of particle size of sand on sand-geosynthetic interfaces

Fig. 6.61. Effect of particle size on shear band thickness: (a) variation of shear band thickness
with d50; (b) variation of ts/d50 ratios.

Table 6.12 Shear zone thickness for the tested sand-geosynthetic interfaces.

Test type Normal stress, σn: kPa Shear zone thickness

ts: mm ts/d50
FS-GTX 40, 80, 120 8.90, 8.60, 7.80 11.41, 11.02, 10.00
MS-GTX 40, 80, 120 10.10, 9.80, 9.60 6.38, 6.20, 6.08
CS-GTX 40, 80, 120 15.20, 14.60, 5.19, 4.98, 4.88
14.30
FS-GMB 40, 80, 120 5.20, 5.80, 4.80 6.67, 7.43, 6.15
MS- 40, 80, 120 6.50, 6.30, 5.80 4.11, 3.99, 3.67
GMB
CS-GMB 40, 80, 120 7.90, 7.10, 6.50 2.70, 2.43, 2.22

148
 Effect of particle size of sand on sand-geosynthetic interfaces

6.4 Micro-topographical analysis of sheared geomembrane

Several researchers have reported that the mechanisms of interaction between soil and

geosynthetics can be correlated with the measures of surface roughness of these continuum

construction materials ((Dove et al., 2006; Kandpal & Vangla, 2023; Vangla & Latha, 2017).

The literature reveals the existence of multiple surface measurement techniques, including

atomic force microscopy (AFM), stylus profilometry, non-contact optical profilometry, and field

emission scanning electron microscopy (Dove et al., 2006; Kandpal & Vangla, 2023; Punetha et

al., 2017; Vangla & Latha, 2016b). Applying a coating of high-reflectance materials such as gold

or aluminum for measurements through an optical profilometer is necessary. The application of

the coating will inevitably alter the characteristics of the surfaces and yield inaccurate

measurements of roughness. Dove et al. (2006) and Kandpal & Vangla (2023) employed the

stylus profilometer as a tool for assessing the surface topography of manufactured materials,

such as geomembranes and transparent acrylic sheets to measure the extent of surface changes

and wear. The stylus profilometer (Bruker’s DektakXT) was used in this study to trace a

significant number of profiles to obtain representative roughness values. This profilometer is

enabled with a 3D mapping option to measure the surface roughness of sheared geomembranes.

Quantification of surface changes to geotextile surfaces was not possible because of their

inherent surface texture, which makes the identification of indents very difficult.

Following the interface shear testing, the geomembrane samples with dimensions of 30 mm

× 15 mm were cut from the center of the geomembrane, where the most wear was seen. After

testing, the samples were carefully cleaned with isopropyl alcohol (IPA) to remove dust and

impurities from the surface. The stylus tip, which had a radius of 12.5 µm, was used to trace the

149
 Effect of particle size of sand on sand-geosynthetic interfaces

surface. One hundred sixty profiles were traced and recorded on a map with a resolution of 12

µm per trace. This resulted in a scan area measuring 6 mm by 2 mm, with the 6 mm dimension

being perpendicular to the direction of shearing. The characterization of surface roughness in this

study is accomplished by utilizing three different surface roughness parameters, as outlined in

Equations (6.30(6.32).

1 (6.3
Sa =
A
∬|h(x , y )|dxdy
0)

(6.3
Sq =
√
1
A
∬ h2 (x , y)dxdy
1)

(6.3
S z=|max(h ( x , y ))|+|min ⁡(h ( x , y ))|
2)

The parameter known as an average surface roughness (Sa) is characterized by the average

variation in height of a surface from its mean plane within a specified evaluation area. The root

mean square roughness (Sq) quantifies the root mean square of the height deviations from the

mean plane across the evaluation area. The variable Sz quantifies the vertical displacement

between the highest peak and lowest valley within the evaluation area.

150
Effect of particle size of sand on sand-geosynthetic interfaces

151
 Effect of particle size of sand on sand-geosynthetic interfaces

Fig. 6.62. Microscopic images of sheared geomembranes: (a) sheared by FS; (b) sheared by MS;
(c) sheared by CS.

The microscopic images of sheared geomembranes are presented in Fig. 6.62, visualizing

the different formations of grooves resulting from the shearing of sands. It is worth noting that

the fine sand (FS) particles exhibit more grooves with narrower spacing due to their higher

contact density per unit area compared to the coarse particles. In contrast, spacing between the

grooves is wider for coarser particles due to their lower contacts per unit area. The 3D maps of

the surface topography of sheared geomembranes by FS particles with varying normal stresses

are shown in Fig. 6.63, where the roughness of the surface increases as the normal stress rises,

correlating with the deeper grooves. As the normal stress increases, the geomembrane material

undergoes plowing behavior. This leads to increased shear resistance at the interface. Initially,

shearing occurs through sliding at low normal stresses, but shifts to plowing at higher normal

stresses. This shift, known as the critical normal stress, marks the transition from sliding to

plowing. Experimental tests were conducted on MS-GMB interfaces at normal stresses of 40, 60,

80, and 120 kPa, measuring the interface friction coefficient (µ = tan(δp)). The findings revealed

an initial decrease in the interface friction coefficient as the normal stress increased, followed by

an eventual increase. This transition was particularly noticeable at 60 kPa, representing the

critical normal stress. Similar critical normal stress values have been reported in previous studies

by Dove et al. (2006) at 60 kPa and Vangla & Latha (2016a), who identified a value of 53 kPa.

These results closely align with the values obtained in the current study.

152
 Effect of particle size of sand on sand-geosynthetic interfaces

Fig. 6.63. 3D surface topography of sheared geomembranes at different normal stresses.

Fig. 6.64. Zoomed view of the sheared geomembranes (a) Enlarged view of GMB; (b) groove
geometry.

153
 Effect of particle size of sand on sand-geosynthetic interfaces

Fig. 6.64 depicts a magnified representation of the surface topography of the sheared

geomembrane and the geometrical features of the grooves. The illustration is intended to focus

on the specific segment of interest in close proximity to the groove. The cross-sectional area of

the groove is comparable to the area of the adjacent shoulders, suggesting that the material is not

lost from the surface but instead undergoes plastic redistribution. Dove et al. (2006) presented a

comprehensive explanation based on the tribometer experiments conducted to investigate the

interaction between granular materials and polymers. The process of plowing has been explained

as the shearing mechanism, wherein there is no loss of material attributed to abrasive wear.

The effect of particle size on the surface topography of the sheared geomembranes is

investigated, and it found that surface roughness resulted due to shearing by FS particles is more

than the coarser particles, as seen in Fig. 6.65. This can be due to the increase in effective

contacts per unit area for FS particles and confirms that effective contacts per unit area govern

the interface shear behaviour rather than particle size alone. Moreover, the 3D shape analysis

characterized the FS particles as less spherical and more irregular among all tested samples,

which have the coupling effect on enhanced surface roughness changes of geomembranes and

yield higher interface friction. In contrast, coarser particles make wider grooves due to their

larger size resulting in lower interface friction because the number of contacts is less. The Sz

parameter represents the distance between the highest peak and lowest valley of the surface,

revealing that the grooves formed by FS particles are deeper than MS and CS particles,

increasing the overall surface roughness of sheared geomembranes. In conclusion, the higher

effective contacts per unit area and the irregular morphology of FS particles result in the rougher

surface of geomembranes after shearing, which can be correlated with the higher interface shear

154
 Effect of particle size of sand on sand-geosynthetic interfaces

strengths. Table 6.13 summarizes the surface roughness parameters of sheared geomembranes

tested with different particle sizes.

Fig. 6.65. Effect of particle size on the surface roughness of sheared geomembrane (a) tested
with FS; (b) tested with MS; (c) tested with CS.

Table 6.13 Surface roughness parameters of sheared geomembranes.

Interface type Normal Sa Sq Sz

stress, (µm) (µm) (µm)
(kPa)
Untested - 2.04 2.85 24.11
FS-GMB 40, 80, 120 3.14, 3.49, 3.76 4.02, 4.54, 5.09 45.75, 50.02, 74.95
MS-GMB 40, 80, 120 2.98, 3.06, 3.18 3.81, 3.95, 4.12 41.05, 49.77, 51.16
CS-GMB 40, 80, 120 2.88, 3.00, 3.02 3.69, 3.82, 3.89 33.39, 34.5, 39.46

155
 Effect of particle size of sand on sand-geosynthetic interfaces

6.5 Summary

The following conclusions have been made after conducting interface shear tests between three

graded sands with different particle size distributions, in contact with a woven geotextile and a

smooth geomembrane, with the support of various image-based analyses.

a) Results of 3D shape analysis indicate that the irregularity of particles tends to decrease as

their size increases. Specifically, the average regularity of FS particles is 0.74, which is

lower than the regularity of MS and CS particles, which have average regularities of 0.77

and 0.79, respectively. The fractal dimension, however, tends to increase with particle size.

b) Higher normal stress levels caused a decrease in the peak friction angles observed in sand-

geotextile interfaces, whereas the opposite trend was observed in sand-geomembrane

interfaces, attributed to transition in the shearing mechanisms. This variation can be

attributed to the particles engaging in plowing behavior on the smoother surface of the

geomembrane at higher normal stresses, resulting in an increased interface friction angle.

c) When fine sand particles interacted with the geotextile surfaces, it resulted in higher peak

friction angles due to their ability to interlock better with the asperities of the geotextile

surfaces. However, in the case of medium and coarse sand particles, their size was larger

than the concavity of the geotextile's asperities, which made it difficult for them to interlock

properly during shearing. This resulted in low peak friction angles and reduced dilation.

d) In sand-geomembrane interfaces, it was observed that fine particles led to higher peak

friction angles compared to medium and coarse particles. Micro-topographical analysis of

sheared geomembranes revealed that fine particles tend to make a more significant number

of effective contacts per unit area with the geomembrane surfaces during shearing, which

156
 Effect of particle size of sand on sand-geosynthetic interfaces

results in more grooves with less spacing. Also, higher irregularity of FS particles resulted

in deeper grooves, enhancing their interface shear response.

e) The average peak frictional efficiency for sand-geotextile (sand-GTX) and sand-

geomembrane (sand-GMB) interfaces was approximately 0.84 and 0.52, respectively. It

was observed that the interface frictional efficiency decreased significantly for coarse sand

specimens due to poor interlocking with geotextile surfaces and a low number of contacts

with geomembrane surfaces.

f) The ratio of the thickness of the shear zone (ts) to the mean particle diameter (d50) varied

significantly in interface tests depending on the particle size and the texture of the

continuum surface. The ts/d50 ratios ranged from 4.88 to 11.41 for dilative interfaces, while

for non-dilative interfaces, the ts/d50 ratios ranged from 2.22 to 7.43. It has been observed

that the ts/d50 ratio tends to decrease as the mean particle size and applied normal stress

levels increase.

The outcomes of this study validate the influence of particle size on interface shear

characteristics, as demonstrated by the comprehensive multi-scale analysis. In scenarios

involving dilative interfaces, the dominant mechanism is the interlocking between sand particles

and surface irregularities. In contrast, the primary mechanisms for non-dilative systems are

sliding and plowing. The particle morphology changes according to the particle size,

consequently influencing interface shear behavior.

157
 Chapter 7. Effect of Particle Shape on Sand-
Geosynthetic Interface Shear Response

7.1 Introduction

Design of geosynthetic reinforced soil structures requires the knowledge of interaction behaviour

of sand-geosynthetic interfaces at multi-scale level. This can be accomplished by

characterization of particle shape and quantifying shear induced damages to the geosynthetic

surface. Particle shape highly influences the macroscopic properties of sands, such as

compressibility, shear strength, attainable density, crushability, and critical state behaviour

(Fonseca et al., 2012; Sun & Zheng, 2021; Vangla et al., 2018; Vangla & Latha, 2016a; Vangla

& Latha, 2015; Zhao & Wang, 2016; Zhou et al., 2018). Most of the earlier studies either

completely ignored the quantification of particle morphology or evaluated the morphological

descriptors in a 2D framework (Alshibli & Alsaleh, 2004; Altuhafi et al., 2013; Mora & Kwan,

2000; Pillai & Latha, 2022; Sukumaran & Ashmawy, 2001; Vangla et al., 2018). The 2D shape

parameters are obtained from the images taken through random projection of particles, which

results in inaccurate values. Chapter 4 provides a thorough explanation of the drawbacks

associated with the utilization of 2D shape descriptors. Hence, 3D measurement of shape

parameters is essential for quantifying the morphology of particles accurately. X-ray micro-

computed tomography (µCT) and 3D laser scanner facilitates the 3D visualization and

measurements of the shape features of granular materials (Alshibli et al., 2015; Fonseca et al.,

2012; Su & Yan, 2018; Sun et al., 2014; Yang et al., 2019; Zhao & Wang, 2016; Zhou et al.,

2018).

158
 Effect of particle shape on sand on sand-geosynthetic interface shear response

Geotextiles and geomembranes are widely used in civil engineering designs, due to their

desirable mechanical and hydraulic functions and economic advantages. These geosynthetics are

always in contact with soils and the soil-geosynthetic composites are subjected to different

external forces imposed by the loading conditions of the structures. If the overall external stress

exceeds the peak shear strength of the sand-geosynthetic interface, it will result in progressive

failure and intense localized shearing of the sand-geosynthetic interfaces. To understand the

sand-geosynthetic interface shear strength, direct shear test setup can be used with suitable

modifications. Various previous researchers have modified the direct shear set-up to investigate

the interaction behaviour of sand-geosynthetics interfaces (Afzali-Nejad et al., 2017; Anubhav &

Basudhar, 2010; Frost et al., 2012; Vangla & Latha, 2015, 2016a). The stress-strain response of

the sand-geosynthetic interfaces is significantly influenced by the soil particle morphology and

gradation, density, effective stress, and surface characteristics of the contact surface (Afzali-

Nejad et al., 2017; Frost et al., 2012; Martinez & Frost, 2017a, 2017b; Pillai & Latha, 2022;

Uesugi & Kishida, 1986b; Vangla & Latha, 2016a, 2016b). Geotextiles have textured surface,

leading to dilative interface shear response, whereas geomembranes have smooth surface,

resulting in non-dilative shear response. Shear behaviour in dilative and non-dilative interfaces is

entirely different due to the difference in their interaction mechanisms. Interlocking between

sand particles and surface asperities of the continuum material offers shear resistance in dilative

interfaces (Indraratna et al., 2015). However, the governing mechanisms in case of non-dilative

interfaces are rolling, sliding, and plowing (Dove & Frost, 1999).

At mesoscale level, the formation of intense localized shear zone adjacent to the contact

material is a critical state of deformation, which results in progressive failure at soil-structure

interfaces (Afzali-Nejad et al., 2017; Dejong et al., 2006; Frost et al., 2012; Lashkari & Jamali,

159
 Effect of particle shape on sand on sand-geosynthetic interface shear response

2021; Martinez & Frost, 2017a). Such failure can be critical for many soil-reinforcement

interfaces, including piles, nails and geosynthetic reinforcement. The thickness of the shear zone

(ts) reported by Tehrani et al. (2016) ranges from 3.9 to 5.2 times the mean particle size (d50) in

case of non-displacement piles embedded in dense and medium-dense sand. Lashkari & Jamali

(2021) performed an extensive series of interface direct tests on sand-geosynthetic interfaces to

show that ts/d50 varies from 3.85 to 10.62 for sand-woven geotextile interfaces, 4.15 to 10.94 for

sand-nonwoven geotextile interfaces, and 3.85 to 11.87 for sand-geomembrane interfaces. Three

dimensional (3D) discrete element simulations on sand-geotextile interfaces carried out by Feng

et al. (2020) showed that the thickness of localized shear zone is around 2.4 d50 to 3 d50.

At the soil-structure interface, particle breakage frequently takes place, which has a

significant impact on the characteristics of soil particles and associated interface behaviour.

Particle breakage can significantly reduce the shear strength and dilation at the interface, as seen

through laboratory interface shear tests and numerical simulations reported in literature (Jo et al.,

2011; Liu et al., 2019; Peng et al., 2021; Saberi et al., 2017; Wang & Yin, 2022). Particle

breakage produces smaller fragments that may fill spaces between nearby particles and create a

more compacted condition. The volumetric dilation and related normal and shear stresses are

consequently reduced. At the soil-structure interface, particle breakage reduces the shear zone

thickness. Particle breakage has no effect on the overall distribution of particle displacement, but

it can cause a slight reduction in shear zone thickness (Wang & Yin, 2022).

Measurement of shear induced changes on geosynthetic surfaces gives important clues to

understand the microscopic shear mechanisms at sand-geosynthetic interfaces. While such

changes are very complex to understand on textured dilative interfaces like geotextiles, they can

be measured relatively easily on smooth non-dilative interfaces like geomembranes. The shear

160
 Effect of particle shape on sand on sand-geosynthetic interface shear response

induced changes on geomembrane surfaces caused by traversing sand particles can be measured

accurately using profilometry. Compared to contact profilometry, digital profilometry provides

much more accurate and comprehensive quantifications of these surface changes. The

macroscopic behaviour of sand-geomembrane interfaces can be correlated with the micro level

surface changes of contact surface through visual observations and roughness quantifications

(Vangla & Latha, 2016b, 2017).

The aim of this chapter is to investigate the effect of particle shape at multi-scale level on

sand-geosynthetic interface shear behaviour through interface shear tests and digital image

analyses. Digital image-based methods were used for various key steps in achieving this goal,

which include particle shape characterizations, shear band analysis of interface shear tests and

surface change quantifications of sheared geosynthetics. A new practical approach to precisely

characterize the 3D particle morphology using X-ray µCT images is developed which is already

discussed in Chapter 4 of this thesis. Then, a series of interface direct shear tests were carried out

between two types of sands in contact with a woven geotextile and a geomembrane. Digital

Image correlation (DIC) technique is used to analyze the particle displacement on the tested

sand-geosynthetic interfaces. Full-field shear strains were measured to study the evolution of

localized shear zone adjacent to the outer surface of the contact material and to quantify its

thickness. Finally, microscopic shear induced changes on the geomembrane surfaces were

visualized and quantified by means of stylus profilometry. Microscopic characterizations are

related to understand the macroscopic shear response of the sand-geosynthetic interfaces.

7.2 Interface shear tests

A series of interface direct shear tests were conducted on dilative and non-dilative interfaces to

understand the interaction behaviour of sand-geosynthetic interfaces at multi-scale level. As

161
 Effect of particle shape on sand on sand-geosynthetic interface shear response

explained in Chapter 6, many researchers have reported the shortcomings of the conventional

direct shear setup for testing the soil-geosynthetic interfaces, which include (a) sagging and

wrinkling of geosynthetic material, (b) limited shear displacement for obtaining residual

parameters and (c) reduction in contact area between soil and geosynthetic surface during

shearing. Some of the earlier studies (Lee & Manjunath, 2000; Vangla & Latha, 2015, 2016b)

addressed few of these limitations by replacing the lower shear box of the conventional direct

shear setup with a rigid platform and fixing the geosynthetic layer to it with high friction gripper

plates. In the present study, the existing conventional direct shear setup is modified by replacing

the lower shear box with a rigid steel plate of dimension 190 mm × 160 mm. Sandpaper is glued

to this platform to make the surface rough before fixing the geosynthetic material to avoid any

slippage of geosynthetic material as shearing progresses. Chapter 6 provides a comprehensive

description of the modifications made to the interface shear apparatus.

In all the tests, the sample was prepared by filling the shear box with dry sand in three equal

layers, each layer tamped to the desired relative density (Dr) with the help of a wooden hammer.

To ensure uniform density, the number of blows gradually increased from the sample's bottom

layer to its top layer. Care has been taken not to over-compact the samples as it may lead to

particle crushing, which may affect the shear strength. Prior to shearing, the initial void ratio (e0)

of each sample was calculated through direct measurement of the soil mass and the height of the

specimen after applying the normal stress (Afzali-Nejad et al., 2017). This resulted in e0 = 0.794

(Dr ≈ 81%) for all AS samples and e0 = 0.688 (Dr ≈80%) for all MS samples. All the tests were

performed at a deformation rate of 1.25 mm/min and stopped when the shear displacement

reached 15 mm. All the interface tests were conducted as per the guidance available in ASTM

D5321/D5321M-14. During shearing, the movement of particles was captured through the

162
 Effect of particle shape on sand on sand-geosynthetic interface shear response

transparent side of the shear box using Olympus OMD-EM-1-Mark-III mirrorless digital camera.

High resolution images at various displacement intervals were extracted from the recorded

videos for shear band analysis. Chapter 6 provides a comprehensive explanation of the modified

interface shear apparatus, including a schematic diagram.

7.2.1 Results of interface shear tests

To ensure the repeatability of interface shear test results, a few test trials were conducted

initially. Then, a series of interfaces direct shear tests were performed at three normal stresses of

40, 80 and 120 kPa. In one set of tests, woven geotextile (GTX) was used as the counterface and

in another set, geomembrane (GMB) was used as the counterface. Chapter 3 provides

information about the properties of the tested geosynthetic material, while Chapter 6 delves into

the surface characteristics of the same material. Results of the interface shear tests are presented

in Fig. 7.66. Table 7.14 Peak and residual friction angles in direct and interface shear

tests.provides a summary of the data for the peak and residual friction angles in sand-alone direct

shear tests (results of sand-alone tests are presented in Chapter 5) and interface shear tests on

sand-GTX and sand-GMB interfaces at three different normal stresses.

163
 Effect of particle shape on sand on sand-geosynthetic interface shear response

Fig. 7.66. Shear behaviour of sand-geosynthetic interfaces (a) Stress-displacement response of

sand-GTX interfaces (b) Dilation response of sand-GTX interfaces (c) Stress-displacement
response of sand-GMB interfaces (d) Dilation response of sand-GMB interfaces.

Table 7.14 exemplifies the global (macroscopic) shear behaviours of sands in contact with

woven geotextile and smooth geomembrane under three different magnitudes of normal stresses.

The stress-displacement response of sand-GTX interfaces is shown in Table 7.14(a) and the

dilation response for these interfaces is shown in Table 7.14(b). The stress-displacement

response of sand-GMB interfaces is shown in Table 7.14(c) and the dilation response for these

interfaces is shown in Table 7.14(d). For the interfaces between sands and GTX, a clear peak is

seen in the shear stress and strong dilation occurred. Peak shear strength and dilation are reduced

with an increase in overall regularity of particles, according to a comparison of the behaviour of

164
 Effect of particle shape on sand on sand-geosynthetic interface shear response

AS-geosynthetic interfaces with that of interfaces between MS particles in contact with selected

geosynthetics.

It is widely acknowledged that the surface roughness of the contact material has a major role

on the mobilization of the peak and residual shear strengths, as well as dilation in the interfaces

between granular soils and other construction materials (Afzali-Nejad et al., 2017; Lashkari &

Jamali, 2021; Uesugi & Kishida, 1986b; Vangla & Latha, 2016a, 2016b). In sand-GMB shear

tests, sand particles move on the geomembrane surface without experiencing significant dilation

because of the lower roughness features of these surfaces. As a result, the mobilization of the

shear strength and dilation in sand-GMB interfaces is less in comparison to sand-GTX interfaces

(Fig. 7.66) because there is less apparent friction or entanglement of sand particles with the

surface features of the geomembrane. The overall height of the contact surface asperities at the

sand-GTX interfaces on the other hand is such that it can cause sand dilation and eventually

shear failure. The maximum dilation experienced by sand-GMB interfaces is only 0.15 mm (see

Fig. 7.66(d)), whereas it is about 1.2 mm (see Fig. 7.66(b)) in the case of sand-GTX interfaces.

The maximum shear stress is more than 100 kPa in case of sand-GTX interfaces with a clear

strain softening beyond the peak (see Fig. 7.66(a)), whereas the maximum shear stress is about

70 kPa in sand-GMB interfaces, without any clear peak (see Fig. 7.66(c)).

The mobilized peak friction angle (δ p) and the post-peak residual friction angle (δ r) are

estimated for sand-geosynthetic interfaces. Also, the frictional efficiency (Ef) of sand-

geosynthetic interfaces is calculated and the mean Ef values considering peak friction angles

were computed as 0.90 and 0.66, respectively for AS-GTX and AS-GMB surfaces and 0.85 and

0.59, respectively for MS-GTX and MS-GMB surfaces. Lashkari and Jamali (2021) showed that

the average frictional efficiency of sand is around 0.95 when interfacing with geotextile and 0.79

165
 Effect of particle shape on sand on sand-geosynthetic interface shear response

when interfacing with geomembrane. Results from the present study are not too different from

these values. The friction efficiency reduced significantly for river sand specimens due to their

regular shape and non-dilative interfaces reduced the efficiency further due to absence of particle

entrapment.

Table 7.14 Peak and residual friction angles in direct and interface shear tests.

Normal stress, Peak friction angle: Residual friction

Test type
σn: kPa degrees angle: degrees
40 48.89 39.00
AS 80 46.41 38.38
120 44.31 36.78
Sand-alone
40 34.03 29.03
MS 80 32.71 28.04
120 32.01 29.38
40 42.37 32.53
AS-
80 41.53 36.19
GTX
120 41.22 35.94
Sand-geotextile
40 29.29 18.58
MS-
80 27.67 22.97
GTX
120 27.36 24.18
40 29.79 25.45
AS-
80 30.58 28.02
GMB
Sand- 120 31.01 29.10
geomembrane 40 17.48 13.89
MS-
80 19.58 16.75
GMB
120 21.43 16.94

One of the important observations from Table 7.14 is that the peak friction angle decreased

with the increase in normal stress for sand-alone tests and sand-GTX interface tests. This

behaviour is also observed for sand-geosynthetic interfaces explained in Chapter 6. Through

laboratory ring shear tests on sands in normal stress range of 10-250 kPa, Tiwari et al. (2005)

showed that the peak friction angle of the sand decreases with the increase in normal stress. A

clear explanation for the reduction in interface friction angle of bulk solids with increase in

166
 Effect of particle shape on sand on sand-geosynthetic interface shear response

normal stress is provided by Roberts (2001). Since the yield locus of solids is convex upward in

shape, its intersection with the shear stress axis corresponds to the internal friction angle of sand

and increase in the normal stress causes reduction in friction angle because of the shape of the

yield locus. Hence the friction at the interface is generated within the soil fabric at lower normal

stresses and at higher normal stresses, sliding of the particles against the counterface governs the

friction, which is usually lesser than the internal friction. Also, the interface friction angle ( )

cannot be larger than the effective angle of internal friction ( ) for normal solids. Results from

the interface shear tests carried out in this study conform to these observations, with an exception

with sand-GMB interfaces. The surface asperities of the geomembrane are significantly smaller

in comparison to the size of the sand particles, resulting in a smooth interface, which causes the

peak and residual shear stresses to be almost identical at all normal stresses. These interfaces are

non-dilative since negligible volume change was seen during shearing. The peak shear strength is

noted to be higher for AS-GMB interfaces because of the higher irregularity and surface texture

of AS particles. The peak friction angle observed for AS-GMB interfaces was approximately

33% more than that observed for MS-GMB interfaces. Also, increase in normal stress caused

deeper indentations on the geomembrane surface, transforming the mechanism of sliding to

plowing along the indents, thus increasing the interface friction. As per Dove & Frost (1999),

when applied normal stress is more than roughly 60 kPa, plowing mechanism takes over sliding.

However, hardness of the geomembrane surface can affect these interaction mechanisms at the

interface. The dominant mechanism may be sliding even at higher normal stresses if the

geomembrane surfaces are very hard. The normal stress at which the mechanism changes from

primarily sliding to primarily plowing can be called critical normal stress. Fig. 7.67 illustrates the

variations in the friction coefficient of the interface as the normal load is increased. The friction

167
 Effect of particle shape on sand on sand-geosynthetic interface shear response

coefficient at the interface exhibits an initial decrease as a result of the non-linear increase in the

contact area in response to an increase in normal stress (Archard, 1957; Johnson, 1982), thus

contradicts Amonton's law (Bowden et al., 1943). The friction coefficient has increased further

because of the frictional process switching from sliding to plowing at critical normal stress. For

this study, the critical normal stress levels were found to be 50 kPa at the AS-GMB interface and

60 kPa at the MS-GMB interface, given that the hardness of the geomembrane maintains the

same. It is worth noting that the critical normal stress will increase as the hardness of the

geomembrane surface increases. This is because a harder surface necessitates a greater amount of

normal stress to initiate surface plowing during shearing. In the present study, slight change in

particle gradation was observed after the interface shear tests. However, the change is not

significant and hence particle breakage effects are ignored while explaining the interface shear

response.

Fig. 7.67. Variation of interface friction coefficient with normal stress: (a) AS-GMB interface;
(b) MS-GMB interface.

168
 Effect of particle shape on sand on sand-geosynthetic interface shear response

7.3 Digital Image Correlation (DIC) analysis

Analysis of movements, particularly for the particles near the contact surface, has demonstrated

its exceptional capacity as a non-destructive measurement technique in the study of the localized

shear zone thickness between granular soils and construction materials (Dejong et al., 2006;

Martinez & Frost, 2017a, 2017b; Tehrani et al., 2016; Uesugi &Kishida, 1986b; Vangla & Latha,

2015, 2016a). The study of the shear band, or local deformation of granular materials, is of

tremendous relevance to many domains including mechanics, materials, industrial, civil, and

hydraulic engineering. In this study, DIC technique is used to compute the localized shear zone

thickness in the sand-geosynthetic interface tests (Arshad et al., 2014; Lei et al., 2018; Shen et

al., 2020). For conducting DIC analysis in MATLAB, an open-source application ‘Ncorr’ with a

graphical user interface developed by Blaber et al. (2015) was used. Movement of sand particles

at the interface was captured through the transparent side of the top shear box using Olympus

OMD-EM-1-Mark-III mirrorless digital camera. During shear, high-quality videos with 4k

resolution (movie image size 3840 pixels × 2160 pixels) and 30 frames per second were

captured. These settings correspond to a magnification of about 0.023 mm/pixel. For tracing

particle relocations, a monochromatic light source was used to provide more illumination to suit

the increased image quality. A thorough explanation of the DIC technique can be found in

Chapter 5.

7.3.1 Results of DIC analysis of the sand-geosynthetic interfaces

The average variation of shear strains (γ) with vertical distance from the geosynthetic plane is

obtained by averaging the shear strain distribution in horizontal direction. In sand-geosynthetic

interface testing, plots showing the evolution of γ over the height of the sand specimens for 15

mm horizontal displacement are shown in Fig. 7.68. Table 7.15 presents the thickness of shear

169
 Effect of particle shape on sand on sand-geosynthetic interface shear response

zone (ts) for different tests. The ratios of ts and the mean particle size (d50) for all interface shear

tests are also presented in Table 7.15.

Fig. 7.68. Shear strains as a function of vertical distance from the geosynthetic outer surface at a
horizontal displacement of 15 mm for interfaces (a) AS-GTX; (b) MS-GTX; (c) AS-GMB; and
(d) MS-GMB.

Table 7.15 Shear zone thickness for the selected sand-geosynthetic interfaces.

Test type Normal stress, Shear zone thickness

(σn): kPa ts, mm ts/d50
AS-GTX 40, 80, 120 9.10, 8.50, 7.80 5.17, 4.83, 4.43
MS-GTX 40, 80, 120 10.10, 9.80, 9.60 6.38, 6.20, 6.08
AS-GMB 40, 80, 120 8.20, 7.90, 7.50 4.66, 4.49, 4.26
MS-GMB 40, 80, 120 6.50, 6.30, 5.80 4.11, 3.99, 3.67

170
 Effect of particle shape on sand on sand-geosynthetic interface shear response

The experimental findings suggest that ts for the tested sand-GTX interfaces ranges from 7.8

mm to 10.1 mm, based on the range of the mean particle size and overall shape of the particles

and the normal stresses used. Additionally, normalizing ts values with their corresponding d50

values revealed that the tested sand-GTX interfaces have ts/d50 ratios that range from 4.43 to

6.38. Similar investigation on geomembrane interfaces shows that ts ranges from 5.8 and 8.2 mm,

and as a result, ts/d50 ratios are between 3.67 and 4.66 for the sand-GMB interfaces. It is noted

that shear zone thickness is not significantly impacted by the normal stress for all the interface

types evaluated here, which can be attributed to the intrinsic deformability of the tested

geosynthetics. Geotextile being a dilative interface, sand-GTX interfaces developed high shear

strains, leading to relatively higher thickness of the shear zone. The non-dilative geomembrane

surface showed relatively lesser shear strains at the interface, leading to relatively smaller shear

zone thickness. According to Lashkari & Jamali (2021), an increase in particle sliding reduces

the shear strains in the zone of the soil-structure interface. In the present study, the smooth non-

dilative geomembrane caused sliding of particles at the interface and therefore, the shear strains

recorded for GMB interfaces are lower than the corresponding GTX interfaces. AS particles

could indent the geosynthetic surfaces easily because of their sharp corners and hence

progressive shearing needed higher shear stresses to move these particles to the same distance

compared to MS particles, which roll on the surface with lesser capacity to indent. To understand

these mechanisms further, surface changes to geomembranes were quantified after the shear test.

Quantification of surface changes to geotextile surfaces was not possible because of their

inherent surface texture, which makes the identification of indents very difficult.

171
 Effect of particle shape on sand on sand-geosynthetic interface shear response

7.4 Quantification of shear induced surface changes to geomembrane

Measuring the surface topographical characteristics of geomembranes and sand particles in order

to quantify their surface roughness and comprehend their interface shear behaviour has recently

become an interest of many researchers (Alshibli & Alsaleh, 2004; Araújo et al., 2022; Dove et

al., 2006; Dove & Frost, 1996; Frost et al., 2012; Frost & Karademir, 2016; Vangla & Latha,

2016b, 2017). Stylus profilometer, Atomic Force Microscopy (AFM), image-based optical

profile microscopy (OPM), and optical profilometry are widely used to assess the roughness of

the sheared geomembrane surfaces. Stylus profilometers use a contact-based method that only

provides 2D information about the surface profile measured through the traversing stylus tip on

the surface. OPM method measures the roughness of a geomembrane sample along three cross

sections that are oriented at an angle of 120 degrees to one another, which is time consuming and

incomplete. Dove & Frost (1996) employed Atomic Force Microscopy (AFM), which has a

restricted scan area and is only suitable for a limited range of geomembranes. Optical

profilometer is relatively sophisticated among these techniques, which works on non-contact

image based profilometry and provides comprehensive information of small areas, which can be

stitched together to get the overall surface roughness. However, this method needs highly

reflecting surfaces for obtaining images with sufficient contrast. Vangla & Latha (2016b, 2017)

have recently utilized optical profilometry to detect the shear-induced changes on geomembrane

surfaces. To improve the reflectance, geomembrane samples were coated with a thin non-

reflective gold layer. Gold sputtering used in optical profilometry alters the real roughness,

causing errors in the measurements.

After the interface shear test, geomembrane samples were exhumed from the shear box and

representative sample pieces were cut from the middle of the sample, where the most surface

172
 Effect of particle shape on sand on sand-geosynthetic interface shear response

wear was seen. The preparation of the geomembrane samples and the profilometer used in this

study are both covered in great detail in Chapter 6. Fig. 7.69 displays the surface height (h(x,y))

maps of the virgin geomembrane and the geomembranes after shearing at a normal stress of 80

kPa. In line with expectations, AS particles caused deeper grooves to the geomembrane surfaces

than MS particles during shear because of their higher irregularity and complex surface texture.

The microscopic changes observed on the geomembrane surface offer insights into the higher

interface shear strength observed with AS particles. The roughness parameters of the

geomembrane surfaces calculated from the surface height profiles before and after shearing are

given in Table 7.16.

Fig. 7.69. Micro-topographical study of typical geomembrane samples (a) virgin geomembrane;
(b) after shearing with MS particles at 80 kPa; and (c) after shearing with AS particles at 80 kPa.

In Table 7.16, Sa is the average roughness, Sq is the root mean square roughness and Sz is the

greatest peak to valley height. The definitions for these parameters can be found in Chapter 6. As

173
 Effect of particle shape on sand on sand-geosynthetic interface shear response

seen from Table 7.16, S z increased with the applied normal stress and the irregularity of the sand

particles. In other words, the sharp edges of the AS particles abraded the geomembrane surface

more, resulting in deeper indentations. Additionally, compared to the geomembrane samples

sheared by MS particles, those sheared by AS particles showed higher root-mean square

roughness and average surface roughness.

Fig. 7.70. 3D surface topography of sheared geomembranes at different normal stresses.

This is due to the fact that when normal load increases, the shearing mechanism switches from

sliding to plowing, causing deeper grooves and a corresponding rise in the peak friction angle.

Fig. 7.70 and Table 7.16illustrate how the application of increasing normal stress leads to shear

induced surface changes on the geomembrane.

Table 7.16 Surface roughness parameters of virgin and sheared geomembrane specimens.

174
 Effect of particle shape on sand on sand-geosynthetic interface shear response

Normal Sa Sq Sz
Interface type
stress, σn: kPa (µm) (µm) (µm)
Virgin - 2.04 2.85 24.11
AS-GMB 40, 80, 120 3.75, 4.69, 7.6 5.45, 6.61, 13.75 64.18, 92.49, 122.18
MS-GMB 40, 80, 120 2.98, 3.06, 3.18 3.81, 3.95, 4.12 41.05, 49.77, 51.16

175
 Effect of particle shape on sand on sand-geosynthetic interface shear response

Fig. 7.71. Correlation between the 3D shape descriptors and the tested sand-geosynthetic
interfaces at multi-scale level.

Fig. 7.71 correlates the 3D shape parameters of the sand particles with the thickness of the

shear zone and the macroscopic shear strength parameters measured in interface shear tests. As

seen from the statistical data, the overall shape of the grains expressed in terms of particle

regularity and roughness has a significant effect on the interface shear mechanism and the overall

shear strength mobilized at the interface. AS particles with an average regularity of 0.69 and

roughness of 1.3 could induce higher degree of surface changes on non-dilative geomembrane

surfaces compared to MS particles with average regularity of 0.77 and roughness of 0.96. The

interface friction angles are higher for AS particles due to their higher irregularity and surface

roughness. Shape irregularity and rough surface texture obstruct the particle rotation, causing

higher particle interlocking and enhanced dilatancy. Similar results were also reported by Wu et

al. (2021) through biaxial shearing of sand. Thickness of the shear zone is 8.47 mm and 9.83

mm, respectively for AS-GTX and MS-GTX interfaces, indicating that particle irregularity need

not always cause higher shear zone thickness. Difficulty in particle rotation during shearing on

the geotextile interface with rough surface texture could be the reason for lower shear zone

thickness in AS. However, in case of sand-GMB interfaces, shear zone thickness is 7.87 mm

with AS and 6.2 mm in MS. The sharp edges of AS particles indent the GMB surfaces to deeper

depths compared to MS particles during shearing (shown by Sz parameter in Fig. 7.71), which

resulted in higher shear zone thickness. Comparison of measured peak and residual friction

angles shown in Fig. 7.71 reveals that the interface friction angles of AS-GTX and AS-GMB

interfaces are much higher compared to the MS-GTX and MS-GMB interfaces, indicating that

particle irregularity has a profound effect of interface friction. Within AS or MS specimens,

dilative geotextile interfaces showed much higher peak and residual friction angles compared to
176
 Effect of particle shape on sand on sand-geosynthetic interface shear response

non-dilative geomembrane interfaces, highlighting the importance of surface texture for

mobilizing higher friction at the interface. Peak and residual friction angles reduced by about 20-

30% in case of sand-GMB interfaces compared to sand-GTX interfaces.

The comprehensive study carried out on sand-geosynthetic interfaces through interface

direct shear tests complemented by different imaging techniques provided important clues to

explore the particle kinematics and microscopic shear mechanisms at a deeper level. Precise

quantification of multiple scale shape parameters of several representative particles of sand

samples was possible through -CT scanning and spherical harmonic reconstruction of particle

shape. DIC analysis of shear tests measured the thickness of the shearing zone and related it to

the particle shape and size. Laser microscopy of sheared geomembrane surfaces quantified the

surface changes. Together, the digital imaging techniques provided all scientific information to

precisely correlate the particle shape to the interface shear mechanisms. The merit of the present

study over the previously published research on this topic lies in its utilization of specific digital

image analysis methods, which are much advanced, more precise, yet faster compared to most of

the other widely used methods.

7.5 Summary

Based on the interface shear tests on two sands with similar grain size and different grain shapes

supported by various image-based analyses, the following major conclusions are drawn.

(a) Increase in normal stress reduced the peak friction angles of sand-geotextile interfaces

but increased the same for sand-geomembrane interfaces, due to change in shearing

mechanisms. Particles attempt to plough the smooth geomembrane surface at greater normal

stresses, which results in higher interface friction angle.

177
 Effect of particle shape on sand on sand-geosynthetic interface shear response

(b) Particle irregularity and surface texture of the geosynthetics have a coupled effect on the

friction efficiency of sand-geosynthetic interfaces. The mean friction efficiency values

considering peak friction angles were computed as 0.90 and 0.66, respectively for AS-GTX

and AS-GMB surfaces and 0.85 and 0.59, respectively for MS-GTX and MS-GMB surfaces.

Friction efficiency reduced significantly for river sand specimens due to the regular shape of

grains and non-dilative interfaces reduced the efficiency further because of lack of surface

texture to entrap the grains.

(c) Thickness of the shear zone (ts) in interface tests varied significantly with particle shape

and counterface surface texture. The ts/d50 ratios varied from 4.43 to 10.10 for dilative

interfaces and 3.67 to 8.20 for non-dilative interfaces.

(d) Shear zone thickness at the interface is smaller for irregular particles in case of geotextile

interfaces, due to the difficulty in particle rotation during shearing. However, for

geomembrane interfaces, irregular particles could deeply abrade the surface, resulting in a

thicker shear zone.

178
 Chapter 8. Summary and Conclusions

8.1 Introduction

This thesis aims to comprehensively explore the behaviour of sands and sand-geosynthetic

interfaces across multiple scales, considering particle morphology effects, using direct shear

experiments and digital image analysis. Digital image-based methods were used for various key

steps in achieving this goal, which include particle shape characterizations, shear band analysis,

and surface change quantifications of sheared geosynthetics. A new practical approach to

precisely characterize the 3D particle morphology using X-ray µCT images is developed in this

study. In this study, four different types of sands varying in mean particle size and shape were

examined, and for conducting interface shear tests, two distinct types of geosynthetics with

significantly contrasting surface characteristics were employed to investigate the effects of

surface roughness of geosynthetics on interface shear behaviour across various scales. High

resolution images obtained at different stages of shearing were analyzed for quantifying the shear

band thickness using digital image correlation (DIC) technique. Finally, microscopic shear

induced changes on the geosynthetic surfaces were visualized and quantified by means of

profilometry. The microscopic characterizations are related to understand the macroscopic shear

response of the sands and sand-geosynthetic interfaces.

8.2 Conclusions

 This study provides valuable insights into the multi-scale shear behaviour of sands and

the interfaces between sands and geosynthetics, focusing on the influence of particle

morphology and surface roughness of geosynthetics. Based on the results obtained in this

study, the following major conclusions can be inferred.

179
 Summary and Conclusions

 A new practical approach to precisely characterize the 3D particle morphology using X-

ray µCT images and spherical harmonic analysis is developed in this study. This

approach is successfully implemented to quantify the particle size, sphericity, convexity,

aspect ratios, roundness, and fractal dimension.

 The statistical analysis of the particle descriptors revealed that they are not independent.

The correlation between any two shape descriptors relies mainly on the distance between

the characteristic scales of these parameters.

 The conventional direct shear apparatus was modified in order to address the issue of

non-uniform stresses and deformations within the shear box. These modifications also

allowed for easier observation of particle displacement during shearing, which is crucial

for shear band analysis.

 Initially, sand-alone direct shear tests were performed to investigate the particle

morphology effects on particle interactions and shear strain fields in sands. The results

revealed that particle irregularity has a significant impact on the macroscopic shear

strength. Dilation as well as peak and residual friction angles were reported to be higher

for angular sand (AS) particles. The irregularity of shape and rough surface texture resist

the rotation of particles, leading to increased interlocking among particles and enhanced

dilatancy. Also, increase in normal stress led to a decrease in the peak friction angles and

dilation as higher normal stresses restrained the deformation behaviour of the granular

material.

 Particle shape affects the thickness of the shear band (ts) in direct shear tests. Shear zone

thickness is smaller for irregular particles than the particles with regular morphology.

180
 Summary and Conclusions

This result is explained by the difficulty in rotation that irregular particles encounter

because of their stronger interlocking.

 Peak and residual friction angle as well as shear band thickness increase with increase in

particle size. The ratio of the thickness of the shear to the mean particle size (ts/d50)

decreases as the particle size increases. For fine, medium, and coarse sands, this ratio

ranged from 15.79 to 18.44, 9.76 to 10.59, and 5.85 to 6.19, respectively.

 In interface shear tests, when fine sand particles interacted with the geotextile surfaces, it

resulted in higher peak friction angles due to their ability to interlock better with the

asperities of the geotextile surfaces. However, in the case of medium and coarse sand

particles, their size was larger than the concavity of the geotextile's asperities, which

made it difficult for them to interlock properly during shearing. This resulted in low peak

friction angles and reduced dilation.

 Higher normal stress levels caused a decrease in the peak friction angles observed in

sand-geotextile interfaces, whereas the opposite trend was observed in sand-

geomembrane interfaces, attributed to transition in the shearing mechanisms. This

variation can be attributed to the particles engaging in plowing behaviour on the

smoother surface of the geomembrane at higher normal stresses, resulting in an increased

interface friction angle.

 In sand-geomembrane interfaces, it was observed that fine particles led to higher peak

friction angles compared to medium and coarse particles. Micro-topographical analysis of

sheared geomembranes revealed that fine particles tend to make a more significant

number of effective contacts per unit area with the geomembrane surfaces during

shearing, which results in more grooves with less spacing. Also, higher irregularity of

181
 Summary and Conclusions

fine particles compared to coarser sands resulted in deeper grooves, enhancing their

interface shear response.

 The average peak frictional efficiency for sand-geotextile (sand-GTX) and sand-

geomembrane (sand-GMB) interfaces was approximately 0.84 and 0.52, respectively.

The increased efficiency of geotextile interfaces can be attributed to the presence of

rough surface asperities on geotextile surfaces, as opposed to the smooth surfaces of

geomembranes. It was observed that the interface frictional efficiency decreased

significantly for coarse sand specimens due to poor interlocking with geotextile surfaces

and a low number of contacts with geomembrane surfaces.

 Particle irregularity and surface texture of the geosynthetics have a coupled effect on the

friction efficiency of sand-geosynthetic interfaces. Comparing the sands with similar

mean particle size and different shape characteristics, friction efficiency reduced

significantly for regular shape of grains and non-dilative interfaces reduced the efficiency

further because of lack of surface texture to entrap the grains.

 The ratio of the thickness of the shear band (ts) to the mean particle diameter (d50) varied

significantly in interface tests depending on the particle size and the texture of the

geosynthetic surface. The ts/d50 ratios ranged from 4.88 to 11.41 for dilative interfaces,

while for non-dilative interfaces, the ts/d50 ratios ranged from 2.22 to 7.43. It has been

observed that the ts/d50 ratio tends to decrease as the mean particle size and applied

normal stress levels increase.

 Shear zone thickness at the interface is smaller for irregular particles in case of geotextile

interfaces, due to the difficulty in particle rotation during shearing. However, for

182
 Summary and Conclusions

geomembrane interfaces, irregular particles could deeply abrade the surface, resulting in

a thicker shear zone.

8.3 Few recommendations for practitioners

The comprehensive investigation involving digital image analysis and direct shear experiments

conducted on four different sands and two different geosynthetic materials has generated some

insightful findings that will enhance the understanding of both the shear behaviour of sands

independently and the interactions between sands and geosynthetics across multiple scales. Few

recommendations for practitioners based on the findings of this study are as follows:

 Particle morphology should be thoroughly characterized in geotechnical designs since it

is essential for predicting the shear behaviour accurately. When determining the particle

size and shape, visual comparison charts and 2D image-based methods should be avoided

because they produce misleading results.

 The study has shown that manufactured sand, which is produced through the crushing of

granite rocks, exhibits irregular particle shapes when compared to natural river sands.

This irregularity in particle shape is a noteworthy characteristic of manufactured sand and

has significant improvements for the shear strength of sands and their interactions with

geosynthetic materials. Given these findings, geotechnical design professionals may

consider substituting manufactured sand for natural river sand in certain applications for

improved durability, enhanced stability, and more cost-effective designs.

 An important finding of this study is that the interface friction angle is notably influenced

by both the particle morphology and the surface roughness of the geosynthetic material.

Hence, instead of considering the fraction of the internal friction angle of sands as the

interface friction angle, it is advisable for geotechnical design practitioners to prioritize

183
 Summary and Conclusions

the performance of interface shear experiments in order to accurately determine the

interface friction angle of the materials involved.

 This study conducted interface shear experiments using various particle sizes, ranging

from 0.6 mm to 4.75 mm, in contact with a woven geotextile. Dilative interfaces are

defined as the interactions between soil particles and rough continuous materials, such as

geotextiles. The results demonstrated that the peak interface shear strength is greater for

finer particles. This suggests that the surface asperities of the geotextile are similar in size

to the fine particles, resulting in enhanced interlocking and consequently improved

interface shear strength. It is important to acknowledge that bigger particles do not

necessarily result in improved interface friction. This is because the bigger particles tend

to be larger than the surface asperities of the geotextile, leading to inadequate

interlocking and thus, poorer frictional interaction. Therefore, practitioners are advised to

carefully choose the continuum material and sands in a manner that ensures the asperities

of the continuum material are of a similar scale to the size of the sands. This careful

selection can optimize interlocking and, in turn, enhance interface shear strength in the

project's design and execution.

 When dealing with non-dilative interfaces, such as the interaction between soil particles

and a smooth continuous material like a geomembrane, it has been observed that fine

particles exhibit greater shear strength at the interfaces compared to coarser particles.

This can be attributed to the larger number of effective contacts between the fine particles

and the geomembrane. The presence of coarser particles may not always lead to better

interface shear strength because they offer fewer effective contacts, which can limit their

performance. Furthermore, it is important to highlight that finer particles are slightly

184
 Summary and Conclusions

more irregular in shape compared to coarser particles, although being sourced from

similar origins. This added irregularity further improved the interfacial shear strength for

finer particles when interfacing with smooth continuous materials like geomembranes.

8.4 Scope for future work

 The present thesis does not extensively explore the effects of sand gradation, relative

density, and the hardness of the continuum interface on interface shear behaviour.

Additional investigation into these variables would facilitate a more comprehensive

analysis of the topic and enhance the overall completeness of the study.

 This study is based on limited set of sand samples and geosynthetic materials. Therefore,

there is a need for further extensive research in this area in order to establish empirical

equations that relate the friction angle to various influencing factors, such as particle

shape, relative density, applied normal stress, mean particle size, and the surface

roughness of geosynthetic materials.

 In order to enhance the validity of the findings of this research, it is imperative to carry

out particle-scale studies utilizing Micro-CT or Discrete Element Method (DEM)

simulation

185
 References

Afzali-Nejad, A., Lashkari, A., & Shourijeh, P. T. (2017). Influence of particle shape on the

shear strength and dilation of sand-woven geotextile interfaces. Geotextiles and

Geomembranes, 45(1), 54–66. https://doi.org/10.1016/j.geotexmem.2016.07.005

Ai, J., Chen, J. F., Rotter, J. M., & Ooi, J. Y. (2011). Assessment of rolling resistance models in

discrete element simulations. Powder Technology, 206(3), 269–282.

https://doi.org/10.1016/j.powtec.2010.09.030

Al-Raoush, R., & Alshibli, K. A. (2006). Distribution of local void ratio in porous media systems

from 3D X-ray microtomography images. Physica A: Statistical Mechanics and Its

Applications, 361(2), 441–456. https://doi.org/10.1016/j.physa.2005.05.043

Alam, M. F., & Haque, A. (2017). A new cluster analysis-marker-controlled watershed method

for separating particles of granular soils. Materials, 10(10).

https://doi.org/10.3390/ma10101195

Alshibli, K. A., & Alsaleh, M. I. (2004). Characterizing Surface Roughness and Shape of Sands

Using Digital Microscopy. Journal of Computing in Civil Engineering, 18(1), 36–45.

https://doi.org/10.1061/(ASCE)0887-3801(2004)18:1(36)

Alshibli, K. A., Druckrey, A. M., Al-Raoush, R. I., Weiskittel, T., & Lavrik, N. V. (2015).

Quantifying Morphology of Sands Using 3D Imaging. Journal of Materials in Civil

Engineering, 27(10), 1–10. https://doi.org/10.1061/(asce)mt.1943-5533.0001246

186
 References

Alshibli, K. A., & Sture, S. (1999). SAND SHEAR BAND THICKNESS MEASUREMENTS

BY DIGITAL IMAGING TECHNIQUES. Journal OfComputing in Civil Engineering,

13(2), 103–109.

Altuhafi, F., O’Sullivan, C., & Cavarretta, I. (2013). Analysis of an Image-Based Method to

Quantify the Size and Shape of Sand Particles. Journal of Geotechnical and

Geoenvironmental Engineering, 139(8), 1290–1307.

https://doi.org/10.1061/(ASCE)GT.1943-5606.0000855

Amirrahmat, S., Druckrey, A. M., Alshibli, K. A., & Al-Raoush, R. I. (2019). Micro shear bands:

Precursor for strain localization in sheared granular materials. Journal of Geotechnical and

Geoenvironmental Engineering, 145(2), 1–18. https://doi.org/10.1061/(ASCE)GT.1943-

5606.0001989

Andò, E., Hall, S. A., Viggiani, G., Desrues, J., & Bésuelle, P. (2012). Grain-scale experimental

investigation of localised deformation in sand: A discrete particle tracking approach. Acta

Geotechnica, 7(1), 1–13. https://doi.org/10.1007/s11440-011-0151-6

Andò, E., Viggiani, G., Hall, S. A., & Desrues, J. (2013). Experimental micro-mechanics of

granular media studied by x-ray tomography: Recent results and challenges. Geotechnique

Letters, 3(3), 142–146. https://doi.org/10.1680/geolett.13.00036

Angelidakis, V., Nadimi, S., & Utili, S. (2022). Elongation, flatness and compactness indices to

characterise particle form. Powder Technology, 396, 689–695.

https://doi.org/10.1016/j.powtec.2021.11.027

Anubhav, & Basudhar, P. K. (2010). Modeling of soil-woven geotextile interface behavior from

direct shear test results. Geotextiles and Geomembranes, 28(4), 403–408.

187
 References

https://doi.org/10.1016/j.geotexmem.2009.12.005

Arasan, S., Yenera, E., Hattatoglu, F., Hinislioglua, S., & Akbulut, S. (2011). Correlation

between Shape of Aggregate and Mechanical Properties of Asphalt Concrete: Digital image

processing approach. Road Materials and Pavement Design, 12(2), 239–262.

https://doi.org/10.1080/14680629.2011.9695245

Araújo, G. L. S., Sánchez, N. P., Palmeira, E. M., & Almeida, M. das G. G. de. (2022). Influence

of micro and macroroughness of geomembrane surfaces on soil-geomembrane and

geotextile-geomembrane interface strength. Geotextiles and Geomembranes, 50(4), 751–

763. https://doi.org/10.1016/j.geotexmem.2022.03.015

Archard, J. . (1957). Elastic deformation and the laws of friction. Proceedings of the Royal

Society of London. Series A: Mathematical and Physical Sciences, 243(1233), 190–205.

https://doi.org/10.1098/rspa.1957.0214

Arshad, M. I., Tehrani, F. S., Prezzi, M., & Salgado, R. (2014). Experimental study of cone

penetration in silica sand using digital image correlation. Geotechnique, 64(7), 551–569.

https://doi.org/10.1680/geot.13.P.179

ASTM D2488-09a. (2009). Standard Practice for Description and Identification of Soils (Visual-

Manual Procedure). ASTM International, 1–12. https://doi.org/10.1520/D2488-09A.

Bagi, K., & Orosz, Á. (2020). A new variable for characterising irregular element geometries in

experiments and DEM simulations. Proceedings - European Council for Modelling and

Simulation, ECMS, 34(1), 256–260. https://doi.org/10.7148/2020-0256

Barrett, P. J. (1980). The shape of rock particles, a critical review. Sedimentology, 27(3), 291–

188
 References

303.

Blaber, J., Adair, B., & Antoniou, A. (2015). Ncorr: Open-Source 2D Digital Image Correlation

Matlab Software. Experimental Mechanics, 55(6), 1105–1122.

https://doi.org/10.1007/s11340-015-0009-1

Blott, S. J., & Pye, K. (2008). Particle shape: A review and new methods of characterization and

classification. Sedimontology, 55(1), 31–63.

Bowden, F. P., Moore, A. J. W., & Tabor, D. (1943). The ploughing and adhesion of sliding

metals. Journal of Applied Physics, 14(2), 80–91. https://doi.org/10.1063/1.1714954

Buscarnera, G., & Einav, I. (2021). The mechanics of brittle granular materials with coevolving

grain size and shape. Proceedings of the Royal Society A: Mathematical, Physical and

Engineering Sciences, 477(2249). https://doi.org/10.1098/rspa.2020.1005

Butterfield, R., & Andrawes, K. Z. (1972). On the angles of friction betweens sand and plane

surfaces. Journal of Terramechanics, 8(4), 15–23. https://doi.org/10.1016/0022-

4898(72)90043-2

Butterfield, R., Harkness, R. M., & Andrews, K. M. (1970). A Stero-Photogrammetric Method

for Measuring Displacement Fields. Geotechnique, 20(3), 308–314.

https://doi.org/10.1680/geot.1970.20.3.308

Chan, K. L. (1995). Quantitative characterization of electron micrograph image using fractal

feature. IEEE Trans Biomed Eng, 42(10), 1033–1037.

Cho, G.-C., Dodds, J., & Santamarina, J. C. (2006). Particle Shape Effects on Packing Density,

Stiffness, and Strength: Natural and Crushed Sands. Journal of Geotechnical and

189
 References

Geoenvironmental Engineering, 132(5), 591–602. https://doi.org/10.1061/(asce)1090-

0241(2006)132:5(591)

Clarke, K. C. (1986). Computation of the fractal dimension of topographic surfaces using the

triangualr prism surface area method. Computers and Geosciences, 12(5), 713–722.

Clayton, C. R. I., Abbireddy, C. O. R., & Schiebel, R. (2009). A method of estimating the form

of coarse particulates. Geotechnique, 59(6), 493–501.

https://doi.org/10.1680/geot.2007.00195

Cundall, P. A., & Strack, O. D. L. (1979). A discrete numerical model for granular assemblies.

Geotechnique, 29(1), 47–65.

Danesh, A., Mirghasemi, A. A., & Palassi, M. (2020). Evaluation of particle shape on direct

shear mechanical behavior of ballast assembly using discrete element method (DEM).

Transportation Geotechnics, 23(March), 100357.

https://doi.org/10.1016/j.trgeo.2020.100357

DeJong, J. T., Randolph, M. F., & White, D. J. (2003). Interface load transfer degradation during

cyclic loading: A microscale investigation. Soils and Foundations, 43(4), 81–93.

https://doi.org/10.3208/sandf.43.4_81

DeJong, J. T., & Westgate, Z. J. (2009). Role of Initial State, Material Properties, and

Confinement Condition on Local and Global Soil-Structure Interface Behavior. Journal of

Geotechnical and Geoenvironmental Engineering, 135(11), 1646–1660.

https://doi.org/10.1061/(asce)1090-0241(2009)135:11(1646)

190
 References

Dejong, J. T., White, D. J., & Randolph, M. F. (2006). Microscale observation and modeling of

soil-structure interface behavior using particle image velocimetry. Soils and Foundations,

46(1), 15–28. https://doi.org/10.3208/SANDF.46.15

Desrues, J., & Viggiani, G. (2004). Strain localization in sand: An overview of the experimental

results obtained in Grenoble using stereophotogrammetry. International Journal for

Numerical and Analytical Methods in Geomechanics, 28(4), 279–321.

https://doi.org/10.1002/nag.338

Do Carmo, M. P. (1976). Differential Geometry of Curves and Surfaces. Prentice-Hall,

Englewood Cliffs. https://doi.org/10.1038/169560b0

Dove, J. E., Bents, D. D., Wang, J., & Gao, B. (2006). Particle-scale surface interactions of non-

dilative interface systems. Geotextiles and Geomembranes, 24(3), 156–168.

https://doi.org/10.1016/j.geotexmem.2006.01.002

Dove, J. E., & Frost, J. D. (1996). A method for measuring geomembrane surface roughness.

Geosynthetics International, 3(3), 369–392. https://doi.org/10.1680/gein.3.0067

Dove, J. E., & Frost, J. D. (1999). Peak Friction Behavior of Smooth Geomembrane-Particle

Interfaces. Journal of Geotechnical and Geoenvironmental Engineering, 125(7), 544–555.

https://doi.org/10.1061/(ASCE)1090-0241(1999)125:7(544)

Dove, J. E., & Jarrett, J. B. (2002). Behavior of Dilative Sand Interfaces in a Geotribology

Framework. Journal of Geotechnical and Geoenvironmental Engineering, 128(1), 25–37.

https://doi.org/10.1061/(ASCE)1090-0241(2002)128:1(25)

Druckrey, A. M., Imseeh, W. H., & Alshibli, K. A. (2022). Experimental evaluation of the

191
 References

anisotropic critical state theory for sand using 3D fabric evolution data of triaxial

experiments. Acta Geotechnica, 17(6), 2193–2211. https://doi.org/10.1007/s11440-021-

01260-3

Durán, O., Kruyt, N. P., & Luding, S. (2010). Analysis of three-dimensional micro-mechanical

strain formulations for granular materials: Evaluation of accuracy. International Journal of

Solids and Structures, 47(2), 251–260. https://doi.org/10.1016/j.ijsolstr.2009.09.035

Feng, S. J., Chen, J. N., Chen, H. X., Liu, X., Zhao, T., & Zhou, A. (2020). Analysis of sand –

woven geotextile interface shear behavior using discrete element method (DEM). Canadian

Geotechnical Journal, 57(3), 433–447. https://doi.org/10.1139/cgj-2018-0703

Field, D. A. (1988). Laplacian smoothing and Delaunay triangulations. Communications in

Applied Numerical Methods, 4(6), 709–712. https://doi.org/10.1002/cnm.1630040603

Fonseca, J. (2011). The evolution of morphology and fabric of a sand during shearing. Imperial

College London.

Fonseca, J., O’Sullivan, C., Coop, M. R., & Lee, P. D. (2012). Non-invasive characterization of

particle morphology of natural sands. Soils and Foundations, 52(4), 712–722.

https://doi.org/10.1016/j.sandf.2012.07.011

Frost, J. D., & Karademir, T. (2016). Shear-induced changes in smooth geomembrane surface

topography at different ambient temperatures. Geosynthetics International, 23(2), 113–128.

https://doi.org/10.1680/jgein.15.00036

Frost, J. D., Kim, D., & Lee, S. W. (2012). Microscale geomembrane-granular material

interactions. KSCE Journal of Civil Engineering, 16(1), 79–92.

192
 References

https://doi.org/10.1007/s12205-012-1476-x

Fu, P., & Dafalias, Y. F. (2012). Quantification of large and localized deformation in granular

materials. International Journal of Solids and Structures, 49(13), 1741–1752.

https://doi.org/10.1016/j.ijsolstr.2012.03.006

Garboczi, E. J. (2002). Three-dimensional mathematical analysis of particle shape using X-ray

tomography and spherical harmonics: Application to aggregates used in concrete. Cement

and Concrete Research, 32(10), 1621–1638. https://doi.org/10.1016/S0008-8846(02)00836-

Ghalib, A. M., & Hryciw, R. D. (1999). SOIL PARTICLE SIZE DISTRIBUTION BY MOSAIC

IMAGING AND WATERSHED ANALYSIS. Journal of Computing in Civil Engineering,

13(2), 80–87.

Gu, X., Huang, M., & Qian, J. (2014). DEM investigation on the evolution of microstructure in

granular soils under shearing. Granular Matter, 16(1), 91–106.

https://doi.org/10.1007/s10035-013-0467-z

Guida, G., Viggiani, G. M., & Casini, F. (2020). Multi-scale morphological descriptors from the

fractal analysis of particle contour. Acta Geotechnica, 15(5), 1067–1080.

https://doi.org/10.1007/s11440-019-00772-3

Guo, N., & Zhao, J. (2016). 3D multiscale modeling of strain localization in granular media.

Computers and Geotechnics, 80, 360–372. https://doi.org/10.1016/j.compgeo.2016.01.020

Hall, S. A., Bornert, M., Desrues, J., Pannier, Y., Lenoir, N., Viggiani, G., & Bésuelle, P. (2010).

Discrete and continuum analysis of localised deformation in sand using X-ray μCT and

193
 References

volumetric digital image correlation. Geotechnique, 60(5), 315–322.

https://doi.org/10.1680/geot.2010.60.5.315

Hall, S. A., Muir Wood, D., Ibraim, E., & Viggiani, G. (2010). Localised deformation patterning

in 2D granular materials revealed by digital image correlation. Granular Matter, 12(1), 1–

14. https://doi.org/10.1007/s10035-009-0155-1

Han, F., Ganju, E., Salgado, R., & Prezzi, M. (2018). Effects of Interface Roughness, Particle

Geometry, and Gradation on the Sand–Steel Interface Friction Angle. Journal of

Geotechnical and Geoenvironmental Engineering, 144(12), 10–16.

https://doi.org/10.1061/(asce)gt.1943-5606.0001990

Härtl, J., & Ooi, J. Y. (2008). Experiments and simulations of direct shear tests: Porosity, contact

friction and bulk friction. Granular Matter, 10(4), 263–271. https://doi.org/10.1007/s10035-

008-0085-3

Höhner, D., Wirtz, S., & Scherer, V. (2015). A study on the influence of particle shape on the

mechanical interactions of granular media in a hopper using the Discrete Element Method.

Powder Technology, 278, 286–305. https://doi.org/10.1016/j.powtec.2015.02.046

Hyslip, J. P., & Vallejo, L. E. (1997). Fractal analysis of the roughness and size distribution of

granular materials. Engineering Geology, 48(3–4), 231–244. https://doi.org/10.1016/s0013-

7952(97)00046-x

Indraratna, B., Biabani, M. M., & Nimbalkar, S. (2015). Behavior of Geocell-Reinforced

Subballast Subjected to Cyclic Loading in Plane-Strain Condition. Journal of Geotechnical

and Geoenvironmental Engineering, 141(1), 1–16. https://doi.org/10.1061/(asce)gt.1943-

5606.0001199

194
 References

Jewell, R. A. (1989). Direct shear tests on sand. Geotechnique, 39(2), 309–322.

Jewell, R. A., & Wroth, C. P. (1987). Direct shear tests on reinforced sand. Geotechnique, 37(1),

53–68.

Jia, X., & Garboczi, E. J. (2016). Advances in shape measurement in the digital world.

Particuology, 26, 19–31. https://doi.org/10.1016/j.partic.2015.12.005

Jiang, M., Zhu, H., & Li, X. (2010). Strain localization analyses of idealized sands in biaxial

tests by distinct element method. Frontiers of Architecture and Civil Engineering in China,

4(2), 208–222. https://doi.org/10.1007/s11709-010-0025-2

Jo, S. A., Kim, E. K., Cho, G. C., & Lee, S. W. (2011). Particle shape and crushing effects on

direct shear behavior using dem. Soils and Foundations, 51(4), 701–712.

https://doi.org/10.3208/sandf.51.701

Johnson, K. L. (1982). One hundred years of Hertz contact. Proceedings of the Institution of

Mechanical Engineers, 196(1), 363–378.

https://doi.org/10.1243/PIME_PROC_1982_196_039

Kandpal, L., & Vangla, P. (2023). New insights into geotribology of non-dilative interfaces from

novel experimental studies. Geosynthetics International.

https://doi.org/10.1680/jgein.23.00013

Kaye, B. H. (1994). A random walk through fractal dimensions. John Wiley & Sons, Ltd.

Kazhdan, M., Funkhouser, T., & Rusinkiewicz, S. (2003). Rotation Invariant Spherical

Harmonic Representation of 3D Shape Descriptors. Eurographics Symposium on Geometry

Processing, 43(2), 156–164.

195
 References

http://www.ihcworld.com/_protocols/chromogen_substrates/POD_DAB_black.htm

Kim, Y., Ma, J., Lim, S. Y., Song, J. Y., & Yun, T. S. (2022). Determination of shape parameters

of sands: a deep learning approach. Acta Geotechnica, 17(4), 1521–1531.

https://doi.org/10.1007/s11440-022-01464-1

Kobbelt, L. P., Botsch, M., Schwanecke, U., & Seidel, H. P. (2001). Feature sensitive surface

extraction from volume data. Proceedings of the 28th Annual Conference on Computer

Graphics and Interactive Techniques, 57–66. https://doi.org/10.1145/383259.383265

Kodicherla, S. P. K., Gong, G., Fan, L., Wilkinson, S., & Moy, C. K. S. (2020). Investigations of

the effects of particle morphology on granular material behaviors using a multi-sphere

approach. Journal of Rock Mechanics and Geotechnical Engineering, 12(6), 1301–1312.

https://doi.org/10.1016/j.jrmge.2020.04.005

Koerner, R. M. (1998). Designing with Geosynthetics (4th ed.). Prentice Hall.

Kong, D., & Fonseca, J. (2018). Quantification of the morphology of shelly carbonate sands

using 3D images. Geotechnique, 68(3), 249–261. https://doi.org/10.1680/jgeot.16.P.278

Kong, L., Chen, F., & Hua, L. (2014). Meso-direct-shear Test of Sands Based on the Digital

Image Method. Soil Behavior and Geomechanics, 265–274.

Kostkanová, V., & Herle, I. (2012). Measurement of wall friction in direct shear tests on soft

soil. Acta Geotechnica, 7(4), 333–342. https://doi.org/10.1007/s11440-012-0167-6

Kozicki, J., Tejchman, J., & Mróz, Z. (2012). Effect of grain roughness on strength, volume

changes, elastic and dissipated energies during quasi-static homogeneous triaxial

compression using DEM. Granular Matter, 14(4), 457–468.

196
 References

https://doi.org/10.1007/s10035-012-0352-1

Krumbein, W. C. (1941). Measurement and Geological Significance of Shape and Roundness of

Sedimentary Particles. Journal of Sedimentary Research, 11(2), 64–72.

https://doi.org/10.1306/D42690F3-2B26-11D7-8648000102C1865D

Krumbein, W. C., & Sloss, L. L. (1963). Stratigraphy and sedimentation (W.H. Freeman (ed.);

2nd Editio).

Kuhn, M. R., & Bagi, K. (2004). Alternative Definition of Particle Rolling in a Granular

Assembly. Journal of Engineering Mechanics, 130(7), 826–835.

https://doi.org/10.1061/(asce)0733-9399(2004)130:7(826)

Kuo, C.-Y., Frost, J. D., Lai, J. S., & Wang, L. B. (1996). Three-Dimensional Image Analysis of

Aggregate Particles from Orthogonal Projections. Transportation Research Record,

1526(1), 98–103. https://doi.org/10.1177/0361198196152600112

Lai, Z., Chen, Q., & Huang, L. (2020). Fourier series-based discrete element method for

computational mechanics of irregular-shaped particles. Computer Methods in Applied

Mechanics and Engineering, 362, 112873. https://doi.org/10.1016/j.cma.2020.112873

Lashkari, A., & Jamali, V. (2021). Global and local sand–geosynthetic interface behaviour.

Geotechnique, 71(4), 346–367. https://doi.org/10.1680/jgeot.19.P.109

Lee, K. M., & Manjunath, V. R. (2000). Soil-geotextile interface friction by direct shear tests.

Canadian Geotechnical Journal, 37(1), 238–252. https://doi.org/10.1139/T99-124

Lee, S. C., Kim, H. K., Chun, I. K., Cho, M. H., Lee, S. Y., & Cho, M. H. (2003). A flat-panel

detector based micro-CT system: Performance evaluation for small-animal imaging.

197
 References

Physics in Medicine and Biology, 48(24), 4173–4185. https://doi.org/10.1088/0031-

9155/48/24/014

Lei, D., Huang, J., Xu, W., Wang, W., & Zhang, P. (2018). Deformation analysis of shear band

in granular materials via a robust plane shear test and numerical simulation. Powder

Technology, 323, 385–392. https://doi.org/10.1016/j.powtec.2017.10.027

Li, X., & Li, X.-S. (2009). Micro-Macro Quantification of the Internal Structure of Granular

Materials. Journal of Engineering Mechanics, 135(7), 641–656.

https://doi.org/10.1061/(asce)0733-9399(2009)135:7(641)

Lin, C. L., & Miller, J. D. (2005). 3D characterization and analysis of particle shape using X-ray

microtomography (XMT). Powder Technology, 154(1), 61–69.

https://doi.org/10.1016/j.powtec.2005.04.031

Lings, M. L., & Dietz, M. S. (2005). The peak strength of sand-steel interfaces and the role of

dilation. Soils and Foundations, 45(6), 1–14. https://doi.org/10.3208/sandf.45.1

Liu, C.-N., Ho, Y. H., & Huang, J. W. (2009). Large scale direct shear tests of soil/PET-yarn

geogrid interfaces. Geotextiles and Geomembranes, 27(1), 19–30.

https://doi.org/10.1016/j.geotexmem.2008.03.002

Liu, C.-N., Zornberg, J. G., Chen, T.-C., Ho, Y.-H., & Lin, B.-H. (2009). Behavior of Geogrid-

Sand Interface in Direct Shear Mode. Journal of Geotechnical and Geoenvironmental

Engineering, 135(12), 1863–1871. https://doi.org/10.1061/(asce)gt.1943-5606.0000150

Liu, J. W., Cui, L., Zhu, N., Han, B., & Liu, J. (2019). Investigation of cyclic pile-sand interface

weakening mechanism based on large-scale CNS cyclic direct shear tests. Ocean

198
 References

Engineering, 194(May), 106650. https://doi.org/10.1016/j.oceaneng.2019.106650

Lopes, M. L., & Silvano, R. (2010). Soil/Geotextile Interface Behaviour in Direct Shear and

Pullout Movements. Geotechnical and Geological Engineering, 28(6), 791–804.

https://doi.org/10.1007/s10706-010-9339-z

Majumdar, A., & Bhushan, B. (1990). Role of Fractal Geometry in Roughness Characterization

and Contact Mechanics of Surfaces. Journal of Tribology, 112(2), 205–216.

https://doi.org/10.1115/1.2920243

Mandelbrot, B. B. (1977). Fractals: Form, Chance and Dimension. Freeman.

Mandelbrot, B. B., Passoja, D. E., & Paullay, A. J. (1984). Fractal character of fracture surfaces

of metals. Nature, 308(19), 721–722. https://doi.org/10.1038/308721a0

Maroof, M. A., Mahboubi, A., Noorzad, A., & Safi, Y. (2020). A new approach to particle shape

classification of granular materials. Transportation Geotechnics, 22(August 2019), 100296.

https://doi.org/10.1016/j.trgeo.2019.100296

Martinez, A., & Frost, J. D. (2017a). Particle-scale effects on global axial and torsional interface

shear behavior. International Journal for Numerical and Analytical Methods in

Geomechanics, 41(3), 400–421. https://doi.org/10.1002/NAG.2564

Martinez, A., & Frost, J. D. (2017b). The influence of surface roughness form on the strength of

sand-structure interfaces. Geotechnique Letters, 7(1), 104–111.

https://doi.org/10.1680/jgele.16.00169

Martinez, A., Frost, J. D., & Hebeler, G. L. (2015). Experimental study of shear zones formed at

sand/steel interfaces in axial and torsional axisymmetric tests. Geotechnical Testing

199
 References

Journal, 38(4), 409–426. https://doi.org/10.1520/GTJ20140266

Masad, E., Saadeh, S., Al-Rousan, T., Garboczi, E., & Little, D. (2005). Computations of particle

surface characteristics using optical and X-ray CT images. Computational Materials

Science, 34(4), 406–424. https://doi.org/10.1016/j.commatsci.2005.01.010

Masson, S., & Martinez, J. (2001). MICROMECHANICAL ANALYSIS OF THE SHEAR

BEHAVIOR OF A GRANULAR MATERIAL. Journal of Engineering Mechanics,

127(10), 1007–1016. https://doi.org/10.1061/(ASCE)0733-9399(2001)127:10(1007)

Mikasa, M. (1960). New direct shear test apparatus. 15th Annual Convention Japanese Soc. Civ.

Engrs, 45–48.

Mollon, G., & Zhao, J. (2013). Generating realistic 3D sand particles using Fourier descriptors.

Granular Matter, 15(1), 95–108. https://doi.org/10.1007/s10035-012-0380-x

Mollon, G., & Zhao, J. (2014). 3D generation of realistic granular samples based on random

fields theory and Fourier shape descriptors. Computer Methods in Applied Mechanics and

Engineering, 279, 46–65. https://doi.org/10.1016/j.cma.2014.06.022

Mora, C. F., & Kwan, A. K. H. (2000). Sphericity, shape factor, and convexity measurement of

coarse aggregate for concrete using digital image processing. Cement and Concrete

Research, 30(3), 351–358. https://doi.org/10.1016/S0008-8846(99)00259-8

Nadimi, S., & Fonseca, J. (2017). Single-Grain Virtualization for Contact Behavior Analysis on

Sand. Journal of Geotechnical and Geoenvironmental Engineering, 143(9), 1–10.

https://doi.org/10.1061/(asce)gt.1943-5606.0001740

Nie, Z., Liang, Z., & Wang, X. (2018). A three-dimensional particle roundness evaluation

200
 References

method. Granular Matter, 20(2), 1–11. https://doi.org/10.1007/s10035-018-0802-5

Nitka, M., & Grabowski, A. (2021). Shear band evolution phenomena in direct shear test

modelled with DEM. Powder Technology, 391, 369–384.

https://doi.org/10.1016/j.powtec.2021.06.025

Oda, M., & Iwashita, K. (2000). Study on couple stress and shear band development in granular

media based on numerical simulation analyses. International Journal of Engineering

Science, 38(15), 1713–1740. https://doi.org/10.1016/S0020-7225(99)00132-9

Otsu, N. (1979). A threshold selection method from gray-level histograms. IEEE Transactions

on Systems, Man, and Cybernetics, 9(1), 62–66.

Paixão, A., Resende, R., & Fortunato, E. (2018). Photogrammetry for digital reconstruction of

railway ballast particles – A cost-efficient method. Construction and Building Materials,

191, 963–976. https://doi.org/10.1016/j.conbuildmat.2018.10.048

Peña, A. A., García-Rojo, R., & Herrmann, H. J. (2007). Influence of particle shape on sheared

dense granular media. Granular Matter, 9(3–4), 279–291. https://doi.org/10.1007/s10035-

007-0038-2

Peng, Y., Liu, H., Li, C., Ding, X., Deng, X., & Wang, C. (2021). The detailed particle breakage

around the pile in coral sand. Acta Geotechnica, 16(6), 1971–1981.

https://doi.org/10.1007/s11440-020-01089-2

Pillai, A. G., & Latha, G. M. (2022). Role of particle shape on the shear strength of sand-GCL

interfaces under dry and wet conditions. Geotextiles and Geomembranes, 50(2), 262–281.

https://doi.org/10.1016/j.geotexmem.2021.11.004

201
 References

Potticary, M., Zervos, A., & Harkness, J. (2015). An investigation into the effect of particle

platyness on the strength of granular materials using the discrete element method.

Proceedings of the 4th International Conference on Particle-Based Methods -

Fundamentals and Applications, PARTICLES 2015, 767–778.

Potts, D. M., Dounias, G. T., & Vaughan, P. R. (1987). Finite element analysis of the direct shear

box test. Geotechnique, 37(1), 11–23.

Powers, M. C. (1953). A new roundness scale for sedimentary particles. Journal of Sedimentary

Petrology, 23(2), 117–119.

Punetha, P., Mohanty, P., & Samanta, M. (2017). Microstructural investigation on mechanical

behavior of soil-geosynthetic interface in direct shear test. Geotextiles and Geomembranes,

45(3), 197–210. https://doi.org/10.1016/j.geotexmem.2017.02.001

Quevedo, R., Mendoza, F., Aguilera, J. M., Chanona, J., & Gutiérrez-López, G. (2008).

Determination of senescent spotting in banana (Musa cavendish) using fractal texture

Fourier image. Journal of Food Engineering, 84(4), 509–515.

https://doi.org/10.1016/J.JFOODENG.2007.06.013

Rao, C., & Tutumluer, E. (2000). Determination of Volume of Aggregates: New Image-Analysis

Approach. Transportation Research Record: Journal of the Transportation Research

Board, 1721(1), 73–80. https://doi.org/10.3141/1721-09

Roberts, A. W. (2001). CHUTE DESIGN CONSIDERATIONS FOR FEEDING AND

TRANSFER. Proceedings of BELTON, Vol.11. International Materails Handling

Conference.

202
 References

Roscoe, K. H., Arthur, J. R. F., & James, R. G. (1963). The determination of strains in soils by

an X-ray method. Civil Engrg. and Public Works Review, 58, 873–876, 1009–1012.

Russ, J. (1994). Fractal surfaces (1st ed.). Plenum Press.

Saberi, M., Annan, C.-D., & Konrad, J.-M. (2017). Constitutive Modeling of Gravelly Soil–

Structure Interface Considering Particle Breakage. Journal of Engineering Mechanics,

143(8), 1–14. https://doi.org/10.1061/(asce)em.1943-7889.0001246

Salazar, A., Sáez, E., & Pardo, G. (2015). Modeling the direct shear test of a coarse sand using

the 3D Discrete Element Method with a rolling friction model. Computers and Geotechnics,

67, 83–93. https://doi.org/10.1016/j.compgeo.2015.02.017

Shen, J., Wang, X., Liu, W., Zhang, P., Zhu, C., & Wang, X. (2020). Experimental study on

mesoscopic shear behavior of calcareous sand material with digital imaging approach.

Advances in Civil Engineering, 2020. https://doi.org/10.1155/2020/8881264

Shi, Y., & Yan, W. M. (2015). Segmentation of irregular porous particles of various sizes from

X-ray microfocus computer tomography images using a novel adaptive watershed approach.

Geotechnique Letters, 5(4), 299–305. https://doi.org/10.1680/jgele.15.00100

Shibuya, S., Mitachi, T., & Tamate, S. (1997). Interpretation of direct shear box testing of sands

as quasi-simple shear. Géotechnique, 47(4), 769–790.

Simoni, A., & Houlsby, G. T. (2006). The direct shear strength and dilatancy of sand-gravel

mixtures. Geotechnical and Geological Engineering, 24(3), 523–549.

https://doi.org/10.1007/s10706-004-5832-6

Sitharam, T. G., & Nimbkar, M. S. (2000). Micromechanical modelling of granular materials:

203
 References

Effect of particle size and gradation. Geotechnical and Geological Engineering, 18(2), 91–

117. https://doi.org/10.1023/A:1008982027109

Skempton, A. W., & Bishop, A. W. (1950). The Measurement of the Shear Strength of Soils.

Géotechnique, 2(2), 90–108. https://doi.org/10.1680/geot.1950.2.2.90

Sneed, E. D., & Folk, R. L. (1958). Pebbles in the Lower Colorado River, Texas a Study in

Particle Morphogenesis. Https://Doi.Org/10.1086/626490, 66(2), 114–150.

https://doi.org/10.1086/626490

Su, D., & Yan, W. M. (2018). 3D characterization of general-shape sand particles using

microfocus X-ray computed tomography and spherical harmonic functions, and particle

regeneration using multivariate random vector. Powder Technology, 323, 8–23.

https://doi.org/10.1016/J.POWTEC.2017.09.030

Sukumaran, B., & Ashmawy, A. K. (2001). Quantitative characterisation of the geometry of

discrete particles. Géotechnique, 51(7), 619–627.

https://doi.org/10.1680/geot.51.7.619.51393

Sun, Q., & Zheng, J. (2021). Realistic soil particle generation based on limited morphological

information by probability-based spherical harmonics. Computational Particle Mechanics,

8(2), 215–235. https://doi.org/10.1007/s40571-020-00325-6

Sun, Q., Zheng, J., & Li, C. (2019). Improved watershed analysis for segmenting contacting

particles of coarse granular soils in volumetric images. Powder Technology, 356, 295–303.

https://doi.org/10.1016/j.powtec.2019.08.028

Sweijen, T., Hassanizadeh, S. M., Aslannejad, H., & Leszczynski, S. (2020). The effect of

204
 References

particle shape on porosity of swelling granular materials: Discrete element method and the

multi-sphere approximation. Powder Technology, 360, 1295–1304.

https://doi.org/10.1016/j.powtec.2019.09.036

Sympatec. (2008). Windox-operating instructions release 5.4.1.0. Sympatec.

Tehrani, F. S., Han, F., Salgado, R., Prezzi, M., Tovar, R. D., & Castro, A. G. (2016). Effect of

surface roughness on the shaft resistance of non-displacement piles embedded in sand.

Geotechnique, 66(5), 386–400. https://doi.org/10.1680/jgeot.15.P.007

Tian, J., Liu, E., & He, C. (2020). Shear band analysis of granular materials considering effects

of particle shape. Acta Mechanica, 231(11), 4445–4461. https://doi.org/10.1007/s00707-

020-02771-y

Tiwari, B., Brandon, T. L., Marui, H., & Tuladhar, G. R. (2005). Comparison of Residual Shear

Strengths from Back Analysis and Ring Shear Tests on Undisturbed and Remolded

Specimens. Journal of Geotechnical and Geoenvironmental Engineering, 131(9), 1071–

1079. https://doi.org/10.1061/(asce)1090-0241(2005)131:9(1071)

Tovar-Valencia, R. D., Galvis-Castro, A., Salgado, R., & Prezzi, M. (2017). Effect of Surface

Roughness on the Shaft Resistance of Displacement Model Piles in Sand. Journal of

Geotechnical and Geoenvironmental Engineering, 144(3), 1–17.

https://doi.org/10.1061/(asce)gt.1943-5606.0001828

Uesugi, M., & Kishida, H. (1986a). Frictional Resistance At Yield Between. Soils and

Foundations, 26(4), 139–149. https://doi.org/10.3208/sandf1972.26.4

Uesugi, M., & Kishida, H. (1986b). Influential Factors of Friction Between Steel and Dry Sands.

205
 References

Soils and Foundations, 26(2), 33–46. https://doi.org/10.3208/SANDF1972.26.2_33

Uesugi, M., Kishida, H., & Tsubakihara, Y. (1988). Behavior of Sand Particles in Sand-Steel

Friction. Soils and Foundations, 28(1), 107–118. https://doi.org/10.3208/sandf1972.28.107

Vangla, P., & Latha, G. M. (2015). Influence of Particle Size on the Friction and Interfacial

Shear Strength of Sands of Similar Morphology. International Journal of Geosynthetic and

Ground Engineering, 1(6), 1–6. https://doi.org/10.1007/s40891-014-0008-9

Vangla, P., & Latha, G. M. (2016a). Effect of particle size of sand and surface asperities of

reinforcement on their interface shear behaviour. Geotextiles and Geomembranes, 44(3),

254–268. https://doi.org/10.1016/j.geotexmem.2015.11.002

Vangla, P., & Latha, G. M. (2016b). Shear behavior of sand-smooth geomembrane interfaces

through micro-topographical analysis. Geotextiles and Geomembranes, 44(4), 592–603.

https://doi.org/10.1016/j.geotexmem.2016.04.001

Vangla, P., & Latha, G. M. (2017). Surface topographical analysis of geomembranes and sands

using a 3D optical profilometer. Geosynthetics International, 24(2), 151–166.

Vangla, P., Roy, N., & Latha, G. M. (2018). Image based shape characterization of granular

materials and its effect on kinematics of particle motion. Granular Matter, 20(1), 1–19.

https://doi.org/10.1007/s10035-017-0776-8

Wadell, H. (1932). Volume , Shape , and Roundness of Rock Particles. The Journal of Geology,

40(5), 443–451.

Wadell, H. (1933). Sphericity and Roundness of Rock Particles A. The Journal of Geology,

41(3), 310–331.

206
 References

Wadell, H. (1935). Volume , Shape , and Roundness of Quartz Particles. Journal of Geology,

43(3), 250–280.

Walker, D. M., Tordesillas, A., & Rechenmacher, A. L. (2013). Transmission of kinematic

information in dense granular systems: Local and nonlocal network sensing. Acta

Geotechnica, 8(5), 547–560. https://doi.org/10.1007/s11440-013-0266-z

Wang, J., Gutierrez, M. S., & Dove, J. E. (2007). Numerical studies of shear banding in interface

shear tests using a new strain calculation method. International Journal for Numerical and

Analytical Methods in Geomechanics, 31(12), 1349–1366. https://doi.org/10.1002/nag.589

Wang, P., & Yin, Z. Y. (2022). Effect of particle breakage on the behavior of soil-structure

interface under constant normal stiffness condition with DEM. Computers and Geotechnics,

147(April), 104766. https://doi.org/10.1016/j.compgeo.2022.104766

Wei, D., Wang, J., Nie, J., & Zhou, B. (2018). Generation of realistic sand particles with fractal

nature using an improved spherical harmonic analysis. Computers and Geotechnics, 104, 1–

12. https://doi.org/10.1016/j.compgeo.2018.08.002

Wen, R., Tan, C., Wu, Y., & Wang, C. (2018). Grain Size Effect on the Mechanical Behavior of

Cohesionless Coarse-Grained Soils with the Discrete Element Method. Advances in Civil

Engineering, 2018. https://doi.org/10.1155/2018/4608930

Wentworth, C. K. (1919). A Laboratory and Field Study of Cobble Abrasion. The Journal of

Geology, 27(7), 507–521.

Wernick, E. (1979). A true direct shear appratus to measure soil parameters of shear bands. In

Proceedings of the 7th European Conference on Soil Mechanics and Foundation

207
 References

Engineering, 175–182.

White, D. J., Take, W. A., & Bolton, M. D. (2003). Soil deformation measurement using particle

image velocimetry (PIV) and photogrammetry. Geotechnique, 53(7), 619–631.

https://doi.org/10.1680/geot.2003.53.7.619

Wu, M., Xiong, L., & Wang, J. (2021). DEM study on effect of particle roundness on biaxial

shearing of sand. Underground Space, 6(6), 678–694.

https://doi.org/10.1016/j.undsp.2021.03.006

Yang, H., Baudet, B. A., & Yao, T. (2016). Characterization of the surface roughness of sand

particles using an advanced fractal approach. Proceedings of the Royal Society A:

Mathematical, Physical and Engineering Sciences, 472, 20160524.

https://doi.org/10.1098/rspa.2016.0524

Yang, H. W., Lourenço, S. D. N., & Baudet, B. A. (2022). 3D fractal analysis of multi-scale

morphology of sand particles with μCT and interferometer. In Geotechnique (Vol. 72, Issue

1, pp. 20–33). https://doi.org/10.1680/jgeot.19.P.120

Yang, H. W., Lourenço, S. D. N., Baudet, B. A., Choi, C. E., & Ng, C. W. W. (2019). 3D

Analysis of gravel surface texture. Powder Technology, 346, 414–424.

https://doi.org/10.1016/j.powtec.2019.01.074

Yang, X., Chen, S., & You, Z. (2017). 3D Voxel-Based Approach to Quantify Aggregate

Angularity and Surface Texture. Journal of Materials in Civil Engineering, 29(7), 1–10.

https://doi.org/10.1061/(asce)mt.1943-5533.0001872

Zhang, L., & Thornton, C. (2007). A numerical examination of the direct shear test.

208
 References

Geotechnique, 57(4), 343–354. https://doi.org/10.1680/geot.2007.57.4.343

Zhao, B., & Wang, J. (2016). 3D quantitative shape analysis on form, roundness, and

compactness with μCT. Powder Technology, 291, 262–275.

https://doi.org/10.1016/j.powtec.2015.12.029

Zhao, L., Zhang, S., Deng, M., & Wang, X. (2021). Statistical analysis and comparative study of

multi-scale 2D and 3D shape features for unbound granular geomaterials. Transportation

Geotechnics, 26(October 2019), 100377. https://doi.org/10.1016/j.trgeo.2020.100377

Zhao, S., Zhou, X., & Liu, W. (2015). Discrete element simulations of direct shear tests with

particle angularity effect. Granular Matter, 17(6), 793–806. https://doi.org/10.1007/s10035-

015-0593-x

Zheng, J., He, H., & Alimohammadi, H. (2021). Three-dimensional Wadell roundness for

particle angularity characterization of granular soils. Acta Geotechnica, 16(1), 133–149.

https://doi.org/10.1007/s11440-020-01004-9

Zheng, J., & Hryciw, R. D. (2015). Traditional soil particle sphericity, roundness and surface

roughness by computational geometry. Geotechnique, 65(6), 494–506.

https://doi.org/10.1680/geot.14.P.192

Zheng, J., & Hryciw, R. D. (2016). Segmentation of contacting soil particles in images by

modified watershed analysis. Computers and Geotechnics, 73, 142–152.

https://doi.org/10.1016/j.compgeo.2015.11.025

Zheng, J., Sun, Q., Zheng, H., Wei, D., Li, Z., & Gao, L. (2020). Three-dimensional particle

shape characterizations from half particle geometries. Powder Technology, 367, 122–132.

209
 References

https://doi.org/10.1016/j.powtec.2020.03.046

Zheng, J., Zhang, Z., Li, C., Li, Z., & Gao, L. (2022). Laboratory-on-a-smartphone for

estimating angularity of granular soils. Acta Geotechnica, 17(7), 2651–2674.

https://doi.org/10.1007/s11440-021-01259-w

Zheng, W., Hu, X., Tannant, D. D., Zhang, K., & Xu, C. (2019). Characterization of two- and

three-dimensional morphological properties of fragmented sand grains. Engineering

Geology, 263(August), 105358. https://doi.org/10.1016/j.enggeo.2019.105358

Zheng, W., Hu, X., Tannant, D. D., & Zhou, B. (2021). Quantifying the influence of grain

morphology on sand hydraulic conductivity: A detailed pore-scale study. Computers and

Geotechnics, 135, 104147. https://doi.org/10.1016/j.compgeo.2021.104147

Zhou, B., & Wang, J. (2017). Generation of a realistic 3D sand assembly using X-ray micro-

computed tomography and spherical harmonic-based principal component analysis.

International Journal for Numerical and Analytical Methods in Geomechanics, 41(1), 93–

109. https://doi.org/10.1002/nag.2548

Zhou, B., Wang, J., & Wang, H. (2018). Three-dimensional sphericity, roundness and fractal

dimension of sand particles. Geotechnique, 68(1), 18–30.

https://doi.org/10.1680/jgeot.16.P.207

Zhou, Y., Wang, H., Zhou, B., & Li, J. (2018). DEM-aided direct shear testing of granular sands

incorporating realistic particle shape. Granular Matter, 20(3), 20–55.

https://doi.org/10.1007/s10035-018-0828-8

Zingg, T. (1935). Beitrag zur Schotteranalyse. Schweiz. Mineral. Petrogr. Mitt., 15, 52–56.

210
 References

https://doi.org/10.3929/ethz-a-000103455

211
 List of Publications

1. Khan, R., and Latha, G.M. (2023). “Multiscale understanding of sand-geosynthetic

interface shear response through Micro-CT and shear band analysis”. Geotextiles and

Geomembranes, 51(3), 437-453.

DOI: https://doi.org/10.1016/j.geotexmem.2023.01.006

2. Khan, R., and Latha, G.M. (2023). “Statistical interdependence of multi-scale 3D

morphological descriptors of sand grains”. Granular Matter, (GRMA-D-23-00065R1),

under review.

3. Khan, R., and Latha, G.M. (2023). “3D analysis of surface topography of sand particles

using spectral methods”. Journal of Materials in Civil Engineering, (MTENG-

16892R1), under review.

4. Khan, R., and Latha, G.M. (2023). “Integrated digital image analyses for understanding

the particle shape effects on sand-geomembrane interface shear”. International journal

of Geosynthetics and Ground Engineering, (IGGE-D-23-00239), under review.

212