materials

Article
Concrete Cracking Prediction Including the Filling
Proportion of Strand Corrosion Products
Lei Wang 1, *, Lizhao Dai 1 , Xuhui Zhang 2 and Jianren Zhang 1
1 School of Civil Engineering and Architecture, Changsha University of Science & Technology,
Changsha 410114, China; [email protected] (L.D.); [email protected] (J.Z.)
2 College of Civil Engineering and Mechanics, Xiangtan University, Xiangtan 411105, China;
[email protected]
* Correspondence: [email protected]; Tel.: +86-731-8525-6068

Academic Editors: Erik Schlangen and Joseph F. Chiang

Received: 21 October 2016; Accepted: 20 December 2016; Published: 23 December 2016

Abstract: The filling of strand corrosion products during concrete crack propagation is investigated
experimentally in the present paper. The effects of stirrups on the filling of corrosion products and
concrete cracking are clarified. A prediction model of crack width is developed incorporating the
filling proportion of corrosion products and the twisting shape of the strand. Experimental data on
cracking angle, crack width, and corrosion loss obtained from accelerated corrosion tests of concrete
beams are presented. The proposed model is verified by experimental data. Results show that the
filling extent of corrosion products varies with crack propagation. The rust filling extent increases with
the propagating crack until a critical width. Beyond the critical width, the rust-filling extent remains
stable. Using stirrups can decrease the critical crack width. Stirrups can restrict crack propagation
and reduce the rust filling. The tangent of the cracking angle increases with increasing corrosion loss.
The prediction of corrosion-induced crack is sensitive to the rust-filling extent.

Keywords: corrosion; strand; concrete; corrosion-induced cracking; corrosion products; filling extent

1. Introduction
Steel corrosion has been identified as one of the most deteriorating factors in concrete
structures [1,2]. During the corrosion process, the metallic iron is transformed to corrosion products [3,4].
This reaction would create an expansive pressure around the concrete and lead to concrete cracking [5].
The corrosion solution can easy diffuse to steel surface through the concrete cracks, which would
further accelerate the corrosion of steel [6]. In addition, concrete cracking also weakens the bond
between steel and concrete [7,8]. These coupling effects decrease the durability and safety of concrete
structures. Cover cracking has been considered as an indicator of the service life end for the existing
concrete structures [9].
A considerable number of studies have been undertaken on corrosion-induced cracking of
reinforced concrete (RC) structures. The amount of corrosion products penetrating into cracks have
also attracted attention. In early studies, some researchers considered that corrosion products fully
filled cracks before cover cracking [10,11]. The recent detections of some existing structures, however,
indicate that corrosion products could not fully fill cracks. This consideration may overestimate the
filling effect of corrosion products. Zhao et al. [12,13] found that corrosion products exhibited the
non-uniform spatial distribution and did not fill cracks inside concrete. Lu et al. [14] reported that
the cracks were not completely filled by corrosion products and the coefficients were introduced to
quantify the filling of corrosion products. These studies focus on the filling of corrosion products before
cover cracking. After cover cracking, non-destructive studies were used to monitor the distribution of
corrosion products [15]. Cracks were being filled with corrosion products gradually over time [16].

Materials 2017, 10, 6; doi:10.3390/ma10010006 www.mdpi.com/journal/materials

Materials 2017, 10, 6 2 of 20

The composition and distribution of chloride-induced corrosion products in cracked concrete subjected
to different loading conditions were also investigated [17]. The filling of corrosion products depends
on many parameters, such as corrosion degree, steel type, and cover [18,19]. How to quantify the
filling of corrosion products is still under discussion.
Predicting crack width with corrosion loss is another important issue to investigate concrete
cracking. Torres-Acosta et al. [6,20] established an empirical relationship between crack width and
corrosion loss based on experimental data. The analytical crack width model considering the combined
effects of steel corrosion and applied load was also derived [21]. Khan et al. [22] predicted the steel
corrosion with crack width for a 26-year-old corroded reinforced concrete beam. These studies aimed
to investigate concrete cracking induced by corrosion of steel bars. The strand consists of several outer
wires twisted around a core wire and has a flower-like cross-section. Concrete cracking caused by
strand corrosion may be different from that caused by steel corrosion. For concrete structures reinforced
with strands, Dai et al. [23] assumed the filling proportion of corrosion products as a constant and
proposed a model to predict corrosion-induced cracking. A summary of studies on strand corrosion
and crack filling are given in Table 1. With further work, the present study aims to quantify the filling
proportion of corrosion products and develop a prediction model of crack width.

Table 1. Summary of studies on strand corrosion and crack filling.

References Specimens Investigation

Stress corrosion cracking on stress–strain response of
Vu et al. [24] Steel wires
steel wires
Darmawan et al. [25] Pre-stressing wires Effect of pitting corrosion on capacity of pre-stressing wires
Pre-tensioned Electrochemical characterization of early corrosion in
Vélez et al. [26]
concrete beams pre-stressed concrete
Probabilistic models for the tensile strength of
Pillai et al. [27] Strand
corroding strands
Post-tensioned Corrosion propagation of pre-stressing steel strands in
Li et al. [28]
concrete beams concrete subject to chloride attack
Post-tensioned Effect of insufficient grouting and strand corrosion on
Wang et al. [29]
concrete beams flexural behavior of pre-stressed concrete beams
Pre-tensioned Influence of strand corrosion on flexural behavior of
Rinaldi et al. [30]
concrete beams pre-stressed concrete beams
Li et al. [31] Parallel wire cables Fatigue properties of corroded parallel wire cables
Post-tensioned
Dai et al. [23] Corrosion-induced cracking induced by strand corrosion
concrete beams

The proposed study investigates the filling of strand corrosion products during concrete cracking.
A prediction model of crack width is developed to consider the filling proportion of corrosion products
and the twisting shape of the strand. The present study is organized as follows: first, the experimental
design, including material properties, geometry dimensions, accelerated corrosion, and crack width
and corrosion loss measurements are introduced; next, the filling of corrosion products and crack
widths are discussed based on the experimental results. Following this, a model is proposed to
predict crack widths incorporating the filling proportion of corrosion products and the twisting
shape of the strand; finally, some conclusions are drawn based on the experimental results and the
theoretical analysis.

2. Experimental Program
In this section, the details of specimens are given at first. Next, the accelerated corrosion test is
employed to obtain various crack widths. Following this, the measurement methods of crack widths,
corrosion products and corrosion losses are exhibited. Details are shown below.
Materials 2017, 10, 6 3 of 20
Materials 2017, 10, 6 3 of 19

2.1. Specimens Details

2.1. Specimens Details
Twelve
Twelve specimens
specimenswere
weredesigned
designedwith
witha square cross-section
a square of 150
cross-section mmmm
of 150 × 150 mm,mm,
× 150 and and
12001200
mm
in length. The specimens were divided into two groups: group S and group RS. Each
mm in length. The specimens were divided into two groups: group S and group RS. Each group group consists of
six beams.
consists ofInsix
thebeams.
group In
RS,the
stirrups
groupwith
RS,8stirrups
mm diameter
with 8and
mm150diameter
mm spacingandwere arranged.
150 mm Group
spacing were S
has no stirrups.
arranged. GroupThe details
S has of beams
no stirrups. aredetails
The shownofinbeams
Figureare
1. shown in Figure 1.

A slicesD slicesC slicesB slicesA

Φ 10 bar

75
PVC
Φ 15.2 strand
Φ 8 stirrups

75
100 A 150
Φ 10 bar
150 150 150 150
1200/2 1200/2 A-A
Beam with stirrups Beam without stirrups

Figure 1.
Figure 1. Details
Details of
of the
the beam
beam (unit:
(unit: mm).
mm).

The specimens were arranged with a 15.2 mm diameter, seven-wire steel strand. Four deformed
The specimens were arranged with a 15.2 mm diameter, seven-wire steel strand. Four deformed
bars with 10 mm diameters were used as the hanger bars at the corners of the beams. The covers of
bars with 10 mm diameters were used as the hanger bars at the corners of the beams. The covers
the strand and reinforcement were 67.4 and 30 mm, respectively. A 100 mm polyvinyl chloride (PVC)
of the strand and reinforcement were 67.4 and 30 mm, respectively. A 100 mm polyvinyl chloride
drive pipe was used to prevent the corrosion solution from flowing out from the beam end. Tables 2
(PVC) drive pipe was used to prevent the corrosion solution from flowing out from the beam end.
and 3 show the chemical compositions and mechanical characteristics of the steel. These data in
Tables 2 and 3 show the chemical compositions and mechanical characteristics of the steel. These data
Tables 2 are adapted from [28], with permission from © 2011 Elesevier. The cement used in concrete
in Table 2 are adapted from [28], with permission from © 2011 Elesevier. The cement used in concrete
was the Type 32.5 Portland cement. The Portland cement contains mainly CaO, SiO2, Fe2O3, and Al2O3.
was the Type 32.5 Portland cement. The Portland cement contains mainly CaO, SiO2 , Fe2 O3 , and
The mix proportion and the 28-day compressive strength of concrete are given in Tables 4 and 5. These
Al2 O3 . The mix proportion and the 28-day compressive strength of concrete are given in Tables 4 and 5.
data in Tables 3 and 5 are tested or obtained based on the method recommended in [32].
These data in Tables 3 and 5 are tested or obtained based on the method recommended in [32].

Table 2. Chemical
Table 2. Chemical compositions
compositions (wt
(wt %)
%) of
of steel
steel [28].
[28].

Type C Mn Si P S Cr Cu Ni Ti Al
Type C Mn Si P S Cr Cu Ni Ti Al
Strand 0.82 0.74 0.21 0.012 0.006 0.17 0.09 0.03 0.03 0.03
Strand
Deformed bars0.820.2 0.74
1.34 0.21
0.55 0.012 0.006
0.033 0.028 0.17
/ 0.09
/ /0.03 / 0.03 / 0.03
Deformed bars 0.2 1.34 0.55 0.033 0.028 / / / / /

Table 3. Mechanical characteristics of steel [32].

Table 3. Mechanical characteristics of steel [32].
Type Diameter (mm) Yield Strength (Mpa) Elastic Modulus (Gpa)
Type
Strand Diameter
15.2(mm) Yield Strength
1830 (Mpa) Elastic Modulus
195 (Gpa)
DeformedStrand
bars (HRB335) 15.2
10 1830
335 195
200
Deformed bars
Deformed bars(HRB335)
(HRB335) 108 335
335 200
200
Deformed bars (HRB335) 8 335 200

Table 4. Concrete mix proportion.

Table 4. Concrete mix proportion.
Water to Cement Ratio Cement (kg/m3) Water (kg/m3) Sand (kg/m3) Stone (kg/m3)
Water to Cement
0.43 Ratio 417 3 )
Cement (kg/m 179 3 )
Water (kg/m (kg/m3 )
Sand676 1026(kg/m3 )
Stone
0.43 417 179 676 1026
Table 5. Twenty-eight-day compressive strength of concrete [32].
Table 5. Twenty-eight-day compressive strength of concrete [32].
Beams S6, S9, S10, S11, S12, S13 RS3, RS7, RS9, RS10, RS11, RS12
Concrete strength (MPa)
Beams
32.5
S6, S9, S10, S11, S12, S13
35.5
RS3, RS7, RS9, RS10, RS11, RS12
Concrete strength (MPa) 32.5 35.5
2.2. Accelerated Corrosion of the Strand
The artificially-accelerated
2.2. Accelerated corrosion method was employed to obtain various crack widths in
Corrosion of the Strand
the beams [33]. To clarify the effect of strand corrosion on concrete cracking independently,
The artificially-accelerated corrosion method was employed to obtain various crack widths in the
reinforcement was protected with epoxy resin to prevent it from corrosion. The specimens were
beams [33]. To clarify the effect of strand corrosion on concrete cracking independently, reinforcement
immersed in the 5% sodium chloride (NaCl) solution in a designed tank. The corrosion system
consisted of a direct current potentiostat and a stainless steel plate. The strand acted as the anode,
Materials 2017, 10, 6 4 of 20

was protected
Materials 2017, 10, 6with
epoxy resin to prevent it from corrosion. The specimens were immersed in the 4 of5%
19
sodium chloride (NaCl) solution in a designed tank. The corrosion system consisted of a direct current
and the stainless
potentiostat and asteel platesteel
stainless served as the
plate. The cathode. The as
strand acted direct currentand
the anode, flowed from thesteel
the stainless positive
plate
terminals
served as of
thethe potentiostat
cathode. to the
The direct strand,
current and then
flowed fromthrough saturated
the positive concrete
terminals and
of the saline solution
potentiostat to the
to the stainless
strand, and thensteel plate,saturated
through and finally back and
concrete to the negative
saline terminals
solution of the potentiostat.
to the stainless Figure
steel plate, and 2
finally
shows
back tothe
theaccelerated corrosionofsystem.
negative terminals the potentiostat. Figure 2 shows the accelerated corrosion system.

Direct current
Direct current
- 0 0. 3 0A +
5%NaCl
Corrosion tank
Cathode steel

(a) (b)

Figure
Figure2.2.Accelerated
Acceleratedcorrosion
corrosiondevice:
device:(a)
(a)Schematic
Schematicdiagram;
diagram;and
and(b)
(b)photo.
photo.

The test specimens were immersed in the saline solution for three days before the accelerated
The test specimens were immersed in the saline solution for three days before the accelerated
corrosion. The corrosion rate was determined by the current density. The corrosion current in the
corrosion. The corrosion rate was determined by the current density. The corrosion current in the total
total process was controlled at a constant 0.3 A. The corresponding current density was about 2702
process was controlled at a constant 0.3 A. The corresponding current density was about 270 µA/cm .
μA/cm2. The theoretical mass loss was roughly estimated based on Faraday's law. The corrosion times
The theoretical mass loss was roughly estimated based on Faraday’s law. The corrosion times were
were referred based on the relationship between the current intensity and mass loss. The accelerated
referred based on the relationship between the current intensity and mass loss. The accelerated
corrosion time for all the specimens were shown in Table 6.
corrosion time for all the specimens were shown in Table 6.
Table 6. Accelerated time of specimens.
Table 6. Accelerated time of specimens.
Beams Corrosion Time (Days)
Beams
S6 Corrosion
2 Time (Days)
S6
S9 9 2
S10S9 7 9
S10 7
S11
S11 3 3
S12
S12 5 5
S13
S13 6 6
RS3
RS3 7 7
RS7 8
RS7 8
RS9 14
RS9
RS10 14 9
RS10
RS11 9 3
RS12
RS11 3 3
RS12 3
2.3. Crack Width and Corrosion Loss Measurement
2.3. Crack Width and Corrosion Loss Measurement
Microcracks form, firstly, in the cross-section when tangential stress exceeds the concrete tensile
Microcracks
strength. form, firstly,
In the present in theinternal
study, these cross-section
cracks when
in the tangential stresszones
fracture process exceeds
arethe concrete
defined tensile
as “cracks”.
strength. In the corrosion
With increasing present study, these
loss, the internal
internal cracks
cracks couldin propagate
the fracture process
to the zones
concrete are defined
surface. as
The outer
“cracks”.
cracks onWith increasing
the concrete corrosion
surface loss,extend
usually the internal
and joincracks couldto
together propagate to the concrete
be a continuous surface.
crack along the
The outer cracks
specimen, which on the concrete
is named surface usually
a longitudinal crack inextend and join
the present together to be a continuous crack
study.
alongAfter
the specimen, whichcorrosion
the accelerated is namedtest,
a longitudinal crackcracks
the longitudinal in the were
present study. on the concrete surface.
observed
The After the accelerated
longitudinal cracks have corrosion
differenttest, the in
widths longitudinal cracksdue
various regions were observed
to the on the
uncertainty concrete
of corrosion
surface. The longitudinal
and material properties. Acracks have
portable different widths
microscope with the inresolution
various regions
of 0.01 due
mm to theused
was uncertainty
to measureof
corrosion and material properties. A portable microscope with the resolution of 0.01 mm was used
crack widths.
to measure crack widths.
To investigate the crack patterns in the radial direction and the filling of corrosion products in
cracks, four 15 mm-thick cross-sectional slices were cut out from each beam after the accelerated
corrosion. The location of the four slices is shown in Figure 1 and labeled as A, B, C, and D,
Materials 2017, 10, 6 5 of 20

To investigate the crack patterns in the radial direction and the filling of corrosion products in
cracks, four 15 mm-thick cross-sectional slices were cut out from each beam after the accelerated
corrosion. The location of the four slices is shown in Figure 1 and labeled as A, B, C, and D, respectively.
Materials 2017, 10, 6 5 of 19
For example, the four slices of S6 are named as S6A, S6B, S6C, and S6D, respectively. The total number
of slices was 48.For
respectively. Theexample,
crackingthe angle
fourwas used
slices to describe
of S6 are named theascrack
S6A, distribution
S6B, S6C, and in S6D,
the radial direction.
respectively.
Since
Thethe filling
total number of corrosion
of slices wasproducts
48. Theincracking
cracks varied
angle was at different positions,
used to describe thethe average
crack rust-filling
distribution in
depth was used to reflect the filling of corrosion products in cracks.
the radial direction. Since the filling of corrosion products in cracks varied at different positions, the
The cracking
average rust-fillingangle waswas
depth measured using the
used to reflect a contour
filling ofgauge. In the
corrosion presentintesting,
products cracks. the maximum
crack was Theselected
crackingtoangle
calculate
was the cracking
measured angle.
using The measurement
a contour gauge. In theprogram
present was as follows:
testing, the maximumfirst, the
contour shapes
crack was of cracks
selected in the radial
to calculate direction
the cracking were
angle. Thepainted to graph
measurement paper;was
program next,asthe sketch
follows: maps
first, the of
contour
cracks were shapes
scanned of cracks
into thein the radial direction
computer. were painted
The cracking angle was to graph paper;
defined as thenext, the of
angle sketch maps of
two sides
theofcrack;
cracksfinally,
were scanned
the crackinginto the computer.
angles The cracking
were calculated angle
by the was drafting
aided defined as the angleMore
program. of two sideson
details
theofcontour
the crack; finally,
gauge canthebe cracking
seen in [34].angles Thewere calculated
rust-filling by the
depth aided
was alsodrafting
measured program.
usingMore details
similar on
methods.
theStrand
contour gauge can be seen in [34]. The rust-filling depth was also measured
corrosion exhibited variability in various regions. Local area loss and average mass loss using similar methods.
Strand corrosion
were commonly used to exhibited
evaluate variability in various
the corrosion regions.
degree. Some Local area loss andstudies
experimental averageshowed
mass loss that
were commonly used to evaluate the corrosion degree. Some experimental
the average mass loss correlated well with the corrosion-induced crack widths for slightly corroded studies showed that the
average mass loss correlated well with the corrosion-induced crack widths for slightly corroded
reinforcement [35,36]. In the present experimental testing, slight corrosion loss was found to induce
reinforcement [35,36]. In the present experimental testing, slight corrosion loss was found to induce
cover cracking due to the large diameter of the strand. Therefore, the average mass loss of the strand
cover cracking due to the large diameter of the strand. Therefore, the average mass loss of the strand
in 10 mm lengths was also employed to evaluate the corrosion degree.
in 10 mm lengths was also employed to evaluate the corrosion degree.
The mass loss was measured after the accelerated corrosion, and the program was as follows.
The mass loss was measured after the accelerated corrosion, and the program was as follows.
First, concrete cover was removed by the destructive method. Next, the strand was taken out and the
First, concrete cover was removed by the destructive method. Next, the strand was taken out and the
concrete
concrete onon itsits
surface
surfacewas wasremoved
removedby byslightly
slightly knocking. Followingthis,
knocking. Following this,the
thecorroded
corroded strand
strand waswas
cleaned by 12% hydrochloric acid solution and then neutralized with alkali
cleaned by 12% hydrochloric acid solution and then neutralized with alkali [37]. The strand was kept [37]. The strand was kept
in the dry
in the environment
dry environment(relative
(relativehumidity
humidityless less than
than 25%). Finally,the
25%). Finally, theaverage
averagemass massloss
loss
ofof
thethe strand
strand
in 10 mm length was measured.
in 10 mm length was measured.

3. Experimental Results
3. Experimental Resultsand
andDiscussion
Discussion

3.1.3.1.
Corrosion Morphology,
Corrosion Cracking
Morphology, CrackingPropagation,
Propagation,and
andCorrosion
Corrosion Loss
Loss
3.1.1. Corrosion Morphology of the Strand
3.1.1. Corrosion Morphology of the Strand
Strand used
Strand ininthe
used thepresent
presentstudy
studyincludes
includes the
the core wire and
core wire andsix
sixouter
outerwires.
wires.Figure
Figure 3 shows
3 shows thethe
corrosion
corrosionmorphology
morphologyofofthe thestrand.
strand.The
The strand
strand showed pitting and
showed pitting andcrevice
crevicecorrosion.
corrosion.Some
Some small
small
corrosion pits were observed on the strand surface. These corrosion pits exhibited oval
corrosion pits were observed on the strand surface. These corrosion pits exhibited oval or circle andor circle and
their depths were small. Additionally, the gaps existed between the core wire and
their depths were small. Additionally, the gaps existed between the core wire and outer wires andouter wires and
could provide
could providea path
a pathforforthe
theflow
flowofofaggressive
aggressiveliquid,
liquid, resulting in crevice
resulting in crevicecorrosion.
corrosion.

Crevice corrosion

(a) (b)

Figure 3. Cont.
Materials 2017, 10, 6 6 of 20
Materials 2017, 10, 6 6 of 19

Materials 2017, 10, 6 6 of 19

(c)

Figure
Figure 3. 3. Corrosionmorphology
Corrosion morphologyofofstrand:
strand: (a)
(a) Pitting
Pitting and
and crevice
crevicecorrosion;
corrosion;(b)
(b)crevice corrosion;
crevice and
corrosion; and
(c) wire corrosion.
(c) wire corrosion.
(c)
The movement of corrosive liquid along the crevices can lead to the range extension of corrosion along
The movement of corrosive
Figure 3. Corrosion morphologyliquid along(a)the
of strand: crevices
Pitting can lead
and crevice to the(b)range
corrosion; creviceextension of corrosion
corrosion; and
the strand, which will accelerate the corrosion rate of the strand. Corrosion loss in the strand can be higher
along the (c) wire corrosion.
strand, which will accelerate the corrosion rate of the strand. Corrosion loss in the strand can
than in steel reinforcement due to crevice effects, resulting in a larger corroding area per unit length.
be higherThe than in steelrate reinforcement steel due to crevice
with effects, resulting in a larger corroding area per
Thecorrosion of the liquid
movement of corrosive increases
along the crevices the increase
can lead to the of current
range density.
extension The uniform
of corrosion along
unit length.
corrosion occurred with a low current density.
the strand, which will accelerate the corrosion rateFor a high
of the current
strand. density,
Corrosion pitting
loss in corrosion
the strand can be occurred
higher
The corrosion
extensively rate of the steel increases with the increase of current density. The uniform
than in steel reinforcement due to crevice effects, resulting in a larger corroding area per unit length. as the
on the steel surface [38]. In the present test, the current density was designed corrosion
occurred
constantwith
The a low
value. current
More
corrosion studies
rate density.
of on various
the Forincreases
steel acurrent
high current density,
densities
with the pitting
are needed
increase incorrosion
the future.
of current occurred
density. extensively
The uniform
on thecorrosion occurred with a low current density. For a high current density, pitting corrosion occurredvalue.
steel surface [38]. In the present test, the current density was designed as the constant
3.1.2.
More Crackon
studies
extensively Width
on and
various
the Corrosion
current
steel surface Loss In are
densities
[38]. the needed in the
present test, thefuture.
current density was designed as the
constant value. More studies on various current densities are needed in the future.
With corrosion propagation, the first visible crack was found through the portable microscope.
3.1.2. Crack Width and Corrosion Loss
The crack then widened and extended along the corroded strand. Some corrosion products were
3.1.2. Crack Width and Corrosion Loss
Withtocorrosion
found flow out propagation, the firstcracks.
from the longitudinal visibleFigure
crack was found
4 shows throughproducts
corrosion the portable
on themicroscope.
concrete
crackWith
Thesurface fromcorrosion
then 10 mm topropagation,
widened and mm for the
110 extended first
S11, visible
S6,along S9,crack
the corroded
and was foundThe
strand.
respectively. through
Some thecrack
portable
corrosion
average microscope.
products
widths were
of found
S6, S11,
The
and S9
to flow outcrack
are
fromthen widened
0.13,the
0.48, and extended
and 0.83 mm,
longitudinal along
respectively.
cracks. the corroded
Figure 4Scarce strand.
showscorrosion Some
corrosionproducts corrosion
products were products
found
on the were
to flow
concrete out
surface
found
from to flow out from the longitudinal cracks. Figure 4 shows corrosion productsproducts
on the concrete
from 10 the
mmnarrow
to 110 longitudinal
mm for S6, S11, cracks.
andWith cracking propagation,
S9, respectively. morecrack
The average corrosion appeared
widths of S6, S11, and S9
surface from 10 mm to 110 mm for S6, S11, and S9, respectively. The average crack widths of S6, S11,
areon the 0.48,
0.13, concrete
andsurface.
0.83 mm, Therespectively.
filling of corrosion
Scarceproducts
corrosion propagates
productswith
werethefound
widening crack.
to flow out from the
and S9 are 0.13, 0.48, and 0.83 mm, respectively. Scarce corrosion products were found to flow out
narrow longitudinal cracks. With cracking propagation, more corrosion products
from the narrow longitudinal cracks. With cracking propagation, more corrosion products appeared appeared on the
concrete surface. The filling of corrosion products propagates with the widening
on the concrete surface. The filling of corrosion products propagates with the widening crack. crack.
S6 0.13 mm crack width

S6 0.13 mm crack width

(a)

S11 0.48 mm(a)

crack width

S11 0.48 mm crack width

(b)

(b)
S9 0.83 mm crack width

S9 0.83 mm crack width

(c)

Figure 4. Longitudinal crack and corrosion products(c) on concrete surface: (a) Slight cracking; (b)
modest cracking; and (c) severe cracking
Figure
Figure 4. Longitudinal
4. Longitudinal crackcrack
and and corrosion
corrosion products
products on concrete
on concrete surface:
surface: (a) Slight
(a) Slight cracking;
cracking; (b) (b)
modest
modest cracking; and (c) severe
cracking; and (c) severe cracking. cracking
Materials 2017,
Materials 2017, 10,
10, 66 77 of
of 20
19

Materials 2017, 10, 6 7 of 19

The beam ends were not immersed in the saline solution. Some radial cracks were still found at
The beam
the specimen ends
ends
The beam were
due
ends tonot
were the immersed
not immersedinin
movement ofthe saline
corrosive
the solution. Some
liquid.Some
saline solution. Figure radial cracks
5 shows
radial cracksthe were
still still
radial
were crack
found found at
atat the
the specimen
beamthe
end. ends due
The radial
specimen to
endscrack the
due toisthemovement
vertically
movement of corrosive
inclined liquid.
to theliquid.
of corrosive concreteFigure
surface.
Figure 5 shows the radial crack
5 shows the radial crack at the at the
beambeam
end. end.
The The
radial crack
radial is vertically
crack is verticallyinclined
inclined toto the concretesurface.
the concrete surface.

(a) (b) (c)

(a) (b) (c)
Figure
Figure 5. Radial
5. Radial cracks
cracks at at beamend:
beam end:(a)
(a)One
One crack;
crack; (b)
(b)two
twocracks; and
cracks; (c) (c)
and three cracks
three cracks.
Figure 5. Radial cracks at beam end: (a) One crack; (b) two cracks; and (c) three cracks
Predicting strand corrosion is one of the most important procedures for structural degradation
Predicting strand corrosion
evaluation. Corrosion is onedifficult
loss is usually of the
the tomost
most important
important
measure in terms procedures
procedures for
for structural
structural
of strands embedded degradation
degradation
into concrete.
evaluation. Corrosioncrack
Correspondingly, loss is usually
widths difficult
on the to measure
concrete surface areineasy
termsto of strands
obtain. Khanembedded
embedded into concrete.
into concrete.
et al. [22] indicated
Correspondingly, crack
that crack width crack widths
could widths on the
correlate the concrete
well concrete surface
with surface are easy
corrosion loss. In easy to
the to obtain.
obtain. Khan et al. [22]
present study, crack widths [22] indicated
are also
indicated
employed
that crack width to identify the corrosion
could correlate welldegree of strand. Crack
with corrosion loss. Inwidth and corrosion
the present study,loss werewidths
crack measuredare also
in 10 mm
employed
employed to length. the
to identify
identify Figures 6 and 7degree
thecorrosion
corrosion show the
degree relationship
ofofstrand.
strand.Crackbetween
Crack width
width crack
and
and width
corrosionandloss
corrosion corrosion
were
loss were loss for in
measured
measured
both groups, respectively.
10
in mm
10 mm length.
length.Figures
Figures6 and 7 show
6 and 7 showthe relationship
the relationshipbetween crackcrack
between width and corrosion
width and corrosionloss for both
loss for
groups, respectively.
both groups, respectively.

(a) (b) (c)

(a) (b) (c)

(d) (e) (f)

Figure 6. Crack width and corrosion loss (group S): (a) S6; (b) S9; (c) S10; (d) S11; (e) S12; and (f) S13.

The variation
(d) of crack width along beam (e) length depends on corrosion degree. (f) Under low
corrosion loss, the variation of crack width is small. With future increasing corrosion, the variation
of crack6.
Figure
Figure 6.width
Crack increases.
width andAdditionally,
corrosion lossit(group
can beS):
also
(a)found
S6; (b)that theS10;
S9; (c)
(c) cracks in the
(d) S11;
(d) S11; (e)middle
(e) span
S12; and
S12; and are
(f) S13.
(f) S13.
usually wider than that in the beam ends. The reason for this phenomenon is that just the middle
The variation of crack width along beam length depends on corrosion degree. Under low
corrosion loss, the variation of crack width is small. With future increasing corrosion, the variation
of crack width increases. Additionally, it can be also found that the cracks in the middle span are
usually wider than that in the beam ends. The reason for this phenomenon is that just the middle
 Materials 2017, 10, 6 8 of 19

span of the beam has been immersed in the 5% sodium chloride (NaCl) solution during the
accelerated corrosion. The corrosion degrees of the strand at the end regions are smaller than that in
Materials 2017, 10, 6 8 of 20
the middle span. Therefore, the crack in the middle span is wider than at the ends of the samples.
Materials 2017, 10, 6 8 of 19

span of the beam has been immersed in the 5% sodium chloride (NaCl) solution during the
accelerated corrosion. The corrosion degrees of the strand at the end regions are smaller than that in
the middle span. Therefore, the crack in the middle span is wider than at the ends of the samples.

(a) (b) (c)

(a) (b) (c)

(d) (e) (f)

Figure
Figure 7.
7. Crack
Crack width
width and
and corrosion
corrosion loss
loss (group
(group RS):
RS): (a)
(a) RS3;
RS3; (b)
(b) RS7;
RS7; (c)
(c) RS9;
RS9; (d)
(d) RS10;
RS10; (e)
(e) RS11;
RS11; and
and
(f) RS12.
(f) RS12.

To analyze the effect of stirrups on crack width, the linear regression was used to describe the
The variation of crack width along beam length depends on corrosion degree. Under low corrosion
relation between crack width and corrosion loss and given in Figure 8.
loss, the variation of crack width is small. With future increasing corrosion, the variation of crack
(d) Additionally, it can be also found
width increases. (e) that the cracks in the middle span (f) are usually
wider than that in the beam ends. The reason for this phenomenon is that just the middle span
Figure
of 7. Crack
the beam has width and corrosion
been immersed loss
in the 5%(group
sodiumRS):chloride
(a) RS3;(NaCl)
(b) RS7;solution
(c) RS9; during
(d) RS10;
the(e)accelerated
RS11; and
(f) RS12. The corrosion degrees of the strand at the end regions are smaller than that in the middle
corrosion.
span. Therefore, the crack in the middle span is wider than at the ends of the samples.
To analyze thethe
To analyze effect ofofstirrups
effect stirrupsonon crack width,the
crack width, thelinear
linearregression
regressionwaswas used
used to describe
to describe the the
relation between
relation betweencrack
crackwidth
widthandandcorrosion
corrosion loss andgiven
loss and givenininFigure
Figure
8. 8.

Figure 8. Summary of crack width and corrosion loss.

As Figure 8 shows, stirrups exhibit a significant restraint effect on corrosion-induced cracking.

The cracks of Group S are wider than that of Group RS in the similar corrosion loss. The stirrups can
bear the tangential stress induced by expansive pressure. Using stirrups decreases the corrosion-
induced crack width. In practical engineering, increasing the amount of stirrups can restrain the crack
extension.
Strand corrosion easily leads to concrete cover cracking. Some studies have been performed to
improve the cracking resistance behavior of concrete. It has been reported that the supplementary
cementing materials can significantly
Figure
Figure improve
8.8.Summary
Summary of the
of crack
crack concrete
width
width resistance
andcorrosion
and corrosion against chloride ingress,
loss.
loss.

As Figure 8 shows, stirrups exhibit a significant restraint effect on corrosion-induced cracking.

The cracks of Group S are wider than that of Group RS in the similar corrosion loss. The stirrups can
bear the tangential stress induced by expansive pressure. Using stirrups decreases the corrosion-
induced crack width. In practical engineering, increasing the amount of stirrups can restrain the crack
extension.
Strand corrosion easily leads to concrete cover cracking. Some studies have been performed to
3.1.3. Cracking Propagation
The crack feature inside concrete is an important issue to investigate cover cracking. Crack
propagation inside concrete is usually difficult to observe. To analyze the crack feature in the radial
Materials 2017, 10, 6
direction, specimens were cut into 15 mm-thick slices. Figure 9a shows three cracks in a slice9 of 20
and
named as: crack A, crack B, and crack C.
As Figure
As Figure 89 shows,
shows, stirrups
crack A and crack
exhibit C are the two
a significant forks effect
restraint of oneon crack in the cross-section.
corrosion-induced This
cracking.
separation of the crack could be attributed to the existing aggregate near the bifurcation
The cracks of Group S are wider than that of Group RS in the similar corrosion loss. The stirrups point. Crack
A is bear
can locatedtheintangential
the inner concrete and did by
stress induced not expansive
extend to the concreteUsing
pressure. surface. Both cracks
stirrups B andthe
decreases C
propagated to the concrete surface. Crack B varied small along the radial direction. Crack
corrosion-induced crack width. In practical engineering, increasing the amount of stirrups can restrain C was the
widest
the crack inextension.
the three cracks and it widened with the radius.
Cracks corrosion
Strand exhibit various
easily width under
leads to different
concrete covercorrosion
cracking.degrees. Figurehave
Some studies 9b gives
been the schematic
performed to
diagram of crack propagation. Cracks exhibit three types of shapes at the different
improve the cracking resistance behavior of concrete. It has been reported that the supplementary stages: the triangle,
the rectangle,
cementing and thecan
materials trapezoid. The similar
significantly improvecrack shapes
the wereresistance
concrete also observed in the
against literature
chloride [13].
ingress,
Before cover cracking, the crack inside concrete seems like a triangle,
lengthening the corrosion initiation time and cracking time of concrete structures under shown as crack A. With
increasing corrosion,
chloride-affected the crack [39,40].
environment propagates to the concrete surface, shown as crack B. This crack shape
can still be considered as a triangle. After the crack appears on the concrete surface, it widens and
exhibits
3.1.3. the similar
Cracking width in the radial direction. In this case, the crack shape is taken as a rectangle,
Propagation
shown as crack C. With the future increase of corrosion degree, the crack becomes wider along the
The crack feature inside concrete is an important issue to investigate cover cracking.
radial direction. The crack shape is simplified as a trapezoid, shown as crack D. Corrosion products
Crack propagation inside concrete is usually difficult to observe. To analyze the crack feature in
would accumulate at the strand-concrete interface and migrate from the interface to the concrete
the radial direction, specimens were cut into 15 mm-thick slices. Figure 9a shows three cracks in a slice
surface, which induce the crack shapes to transform from the triangular to the rectangle.
and named as: crack A, crack B, and crack C.

Corrosion products
Reinforcement
Crack A

Crack A Crack D Puce rust

Crack B

θ Strand
Brownish-red rust
Black rust
Crack C Crack B

Concrete Crack C

(a) (b)

Figure 9. Schematic
Figure Schematicdiagrams
diagramsofof
the crack
the propagation:
crack (a) Crack
propagation: distribution;
(a) Crack and (b)
distribution; andsimplified crack
(b) simplified
propagation.
crack propagation.

As Figure
As mentioned previously,
9 shows, crack A each
andbeam was
crack cut into
C are fourforks
the two 15 mm-thick cross-sectional
of one crack slices. The
in the cross-section.
total number of slices was 48. The schematic diagram of cracking angle, , was given
This separation of the crack could be attributed to the existing aggregate near the bifurcation in Figure 9b.
point.
The cracking
Crack angleinmainly
A is located represents
the inner concretethe
andvariation of crack
did not extend towidth in the radial
the concrete direction.
surface. Both cracksWhen the
B and
crack narrows along the radius, the cracking angle is less than zero. With cracking propagation,
C propagated to the concrete surface. Crack B varied small along the radial direction. Crack C was the the
cracking
widest in angle equals
the three zeroand
cracks when the crackwith
it widened widththeisradius.
similar in the radial direction. After that, the crack
width on the concrete surface is larger than that at
Cracks exhibit various width under different corrosionthe interface. The cracking
degrees. Figure 9b angle
givesinthe
thisschematic
situation
diagram of crack propagation. Cracks exhibit three types of shapes at the different stages: the triangle,
the rectangle, and the trapezoid. The similar crack shapes were also observed in the literature [13].
Before cover cracking, the crack inside concrete seems like a triangle, shown as crack A. With increasing
corrosion, the crack propagates to the concrete surface, shown as crack B. This crack shape can still be
considered as a triangle. After the crack appears on the concrete surface, it widens and exhibits the
similar width in the radial direction. In this case, the crack shape is taken as a rectangle, shown as crack
C. With the future increase of corrosion degree, the crack becomes wider along the radial direction.
The crack shape is simplified as a trapezoid, shown as crack D. Corrosion products would accumulate
 Materials 2017, 10, 6 10 of 20

at the strand-concrete interface and migrate from the interface to the concrete surface, which induce
the crack shapes to transform from the triangular to the rectangle.
As mentioned previously, each beam was cut into four 15 mm-thick cross-sectional slices. The total
number of slices was 48. The schematic diagram of cracking angle, θ, was given in Figure 9b.
The cracking angle mainly represents the variation of crack width in the radial direction. When
the crack narrows along the radius, the cracking angle is less than zero. With cracking propagation, the
Materials 2017, 10,
cracking 6 equals zero when the crack width is similar in the radial direction. After that, the crack
angle 10 of 19
width on the concrete surface is larger than that at the interface. The cracking angle in this situation is
is larger
largerthan
thanzero.
zero. Figure 10shows
Figure 10 showsthethe linear
linear regression
regression and polynomial
and polynomial regression
regression betweenbetween
tanθ and tanθ
and corrosion
corrosionloss. loss.

Figure 10.10.Tangent
Figure Tangentof
of cracking angleand
cracking angle andcorrosion
corrosion loss.
loss.

The The
tangent of cracking
tangent of cracking angle
angleincreases
increases with increasingcorrosion
with increasing corrosion degree.
degree. TheThe discreteness
discreteness of theof the
correlation between tanθ and the corrosion
correlation between tanθ and the corrosion loss may be attributed to the measurement uncertainty
may be attributed to the measurement uncertainty
of the
of crack width
the crack widthandandthethecorrosion
corrosion loss.
loss. Concrete
Concrete isisaaheterogeneous
heterogeneous material.
material. TheThe variability
variability of of
cracking propagation is inevitable.
cracking propagation is inevitable.
As Figure
As Figure 1010shows,
shows, the
the fitting
fittingprecision of linear
precision regression
of linear and polynomial
regression regression regression
and polynomial are similar. are
In the present study, the linear regression was used to describe the relation between
similar. In the present study, the linear regression was used to describe the relation between tanθ and corrosion
tanθ and
loss and given as follows:
corrosion loss and given as follows:
tan θ = aρ − b (1)
tan = − (1)
where θ is the cracking angle; a and b are the constants, for Group S, a = 0.1309, b = 0.0048, for Group
RS, a =is0.0999,
where b = 0.0040;
the cracking and ρ isand
angle; thebcorrosion loss of the for
are the constants, strand.
Group S, a = 0.1309, b = 0.0048, for Group
RS, a = 0.0999, b = 0.0040; and is the corrosion loss of the strand.
3.2. Filling of Corrosion Products in Cracks

3.2. Filling of Corrosionof

3.2.1. Composition Products in Cracks
Corrosion Products
The compositions of corrosion products depends on the alkalinity degree, the oxygen supply, and
3.2.1. Composition of Corrosion Products
the moisture content [41]. Corrosion products also exhibit various colors at different regions in the
The compositions
present of corrosion
study. Three colors products
of corrosion depends
products on the alkalinity
were observed: degree, theand
black, brownish-red, oxygen
puce. supply,
Figure 11 shows
and the moisture contentthe black
[41]. rust at theproducts
Corrosion strand-concrete interface.
also exhibit The cover
various colorsprevents the oxygen
at different regions in
from reaching the strand-concrete interface. For the reaction with some oxygen, the main
the present study. Three colors of corrosion products were observed: black, brownish-red, and puce. compositions
of corrosion
Figure products
11 shows theare ferrous
black rustoxide (FeO)
at the and ferroferricinterface.
strand-concrete oxide (Fe3 O 4 ) [13,17].
The cover The colors the
prevents of FeO
oxygen
and Fe3 O4 are black. FeO is unstable and can easily become Fe3 O4 in air. Therefore, Fe3 O4 is considered
from reaching the strand-concrete interface. For the reaction with some oxygen, the main
as the primarily composition of black rust.
compositions of corrosion products are ferrous oxide (FeO) and ferroferric oxide (Fe3O4) [13,17]. The
colors of FeO and Fe3O4 are black. FeO is unstable and can easily become Fe3O4 in air. Therefore, Fe3O4
is considered as the primarily composition of black rust.
 Figure 11 shows the black rust at the strand-concrete interface. The cover prevents the oxygen
from reaching the strand-concrete interface. For the reaction with some oxygen, the main
compositions of corrosion products are ferrous oxide (FeO) and ferroferric oxide (Fe3O4) [13,17]. The
colors of FeO and Fe3O4 are black. FeO is unstable and can easily become Fe3O4 in air. Therefore, Fe3O4
Materials 2017, 10, 6 11 of 20
is considered as the primarily composition of black rust.

Figure 11.11.
Figure TheTheblack
blackrust
rust at
at the strand-concrete
the strand-concrete interface.
interface.
Materials 2017, 10, 6 11 of 19

Figure
Figure 1212 shows the
shows thebrownish-red
brownish-redrust
rustin
incracks.
cracks. Cracks provideaapath
Cracks provide pathforforoxygen
oxygentoto the
the inner
inner
concrete.
concrete.The The oxygen
oxygensupply
supplyisissufficient
sufficientin
incracks.
cracks. The color of iron
color of iron oxide
oxide(Fe (Fe2O
2O ) isbrownish-red,
3)3is brownish-red,
andand
Fe2Fe
O23Ois3 isregarded
regardedasasthe
themain
maincomposition
composition of the brownish-red
brownish-redrustrust[13].
[13].

(a) (b)

(c)

Figure
Figure 12.12.
TheThe filling
filling ofofcorrosion
corrosionproducts
productsin
incracks:
cracks: (a)
(a) Slight
Slightfilling
fillingofofrust;
rust;(b)
(b)partial
partialfilling of of
filling rust;
rust;
and (c) vast filling of rust.
and (c) vast filling of rust.

As Figure 3 shows, the puce rust was found in the gaps between the core wire and outer wires.
The oxygen can reach the gaps with the flow of aggressive liquid. The oxygen supply in the gaps may
be lower than those in cracks, and higher than that at the strand-concrete interface. Therefore, the
color of rust in the gaps is between the black and the brownish-red.

3.2.2. Filling of Corrosion Products

Materials 2017, 10, 6 12 of 20

As Figure 3 shows, the puce rust was found in the gaps between the core wire and outer wires.
The oxygen can reach the gaps with the flow of aggressive liquid. The oxygen supply in the gaps may
be lower than those in cracks, and higher than that at the strand-concrete interface. Therefore, the color
of rust in the gaps is between the black and the brownish-red.

3.2.2. Filling of Corrosion Products

Some corrosion products were observed to flow out from the longitudinal cracks during the
corrosion process. The slices were broken down to observe the filling of corrosion products in
cracks. Figure 12 shows the slices profiles of S6B, S9A, and S9C, respectively. Concrete slice profiles
were broken down along the widest cracks. Corrosion products filled principally in the widest
crack. In another small crack, a few corrosion products were found. The similar experimental
observations were also found by Šavija et al. [15]. In their study, the micro-computed X-ray tomography
technique (CT-scanning) was used to monitor corrosion products formation during corrosion process.
The scanning results showed that corrosion products principally penetrated into the widest crack,
and few corrosion products were observed in other small cracks. This phenomenon is similar to the
experimental observation obtained in the present study.
Figure 12 shows the profiles of concrete slices along the widest cracks. It should be noted that the
upper part of the profile is the position of the widest crack and the bottom part is the broken surface
obtained by the destruction method. As mentioned before, the corrosion products mainly filled in the
widest cracks. The salt water can also immerse into the cracks. Therefore, the upper part is overlaid by
the corrosion products and salt powders. Additionally, no aggregates can clearly be observed in that
region. The bottom part, however, is a new surface. Very few, or no, corrosion products can be found,
but the aggregates are clear in that region.
The filling of corrosion products depends on crack widths. The crack widths of the three slices in
Figure 12 are 0.08, 0.39, and 0.91 mm, respectively. The filling of corrosion products is slight in the
narrow crack. Corrosion products principally fill in the wide crack. Corrosion products propagate
with increasing crack width. The filling of corrosion products varies at different positions.
In the experimental testing, the volume of corrosion products is difficult to measure.
Correspondingly, the filling depth of corrosion products is easy to obtain. Based on the geometric
formula conversion, the volume of corrosion products can be obtained with the rust-filling depth.
Therefore, the rust-filling depth was used to describe the rust-filling ratio in the experimental testing.
The average rust-filling depth is used to represent the filling of corrosion products in the slice.
The rust-filling ratio, defined as the ratio of average rust-filling depth to cover, is employed to reflect
the filling of corrosion products:
R
f = i (2)
C
where f is the rust-filling ratio; Ri is the average rust-filling depth; and C is the concrete cover.
Before corrosion products full fill cracks, concrete cover would have cracked. From the
experimental testing, corrosion products cannot fully fill cracks, even with severe cracking.
The rust-filling ratio is defined as the ratio of the average rust-filling depth to the cover. Therefore, the
rust-filling ratio is less than 1.0.
Figure 13 shows the linear regression and polynomial regression of rust-filling ratio and crack
width. The rust-filling ratio increases with increasing crack width until a critical value. After that,
the rust-filling ratio can be taken as a constant. This constant is less than one and considered as the
maximum rust-filling ratio in the present study. The maximum rust-filling ratios of group S and group
RS are 0.85 and 0.88, respectively. The critical widths of maximum rust-filling ratio are 0.79 mm and
0.63 mm in group S and group RS, respectively.
 Figure 13 shows the linear regression and polynomial regression of rust-filling ratio and crack
width. The rust-filling ratio increases with increasing crack width until a critical value. After that, the
rust-filling ratio can be taken as a constant. This constant is less than one and considered as the
maximum rust-filling ratio in the present study. The maximum rust-filling ratios of group S and
group RS
Materials are
2017, 10,0.85
6 and 0.88, respectively. The critical widths of maximum rust-filling ratio are 0.79
13 of 20
mm and 0.63 mm in group S and group RS, respectively.

Figure 13.
Figure 13. Rust-filling ratio and
Rust-filling ratio and crack
crack width.
width.

As Figure 13 shows, the rust-filling ratio increases faster in the specimens with stirrups than that
As Figure 13 shows, the rust-filling ratio increases faster in the specimens with stirrups than that
in the specimens without stirrups. The volume of corrosion products can be obtained with the crack
in the specimens without stirrups. The volume of corrosion products can be obtained with the crack
width and rust-filling ratio. In the similar corrosion loss, the volumes of corrosion products in Group S
and Group RS are the same. The stirrups can bear the tangential stress and decrease the crack width,
which would lead to the large rust-filling ratio.
The discreteness of correlation between the rust-filling ratio and crack width may be attributed
to the measurement uncertainty of crack width and corrosion loss. As Figure 13 shows, the fitting
precision of polynomial regression is larger than that of linear regression. The polynomial regression
was used to describe the relation between crack width and rust-filling ratio in the present study. Two
regressed curves of the rust-filling ratio are proposed for the both groups as follows:

−0.773w2s + 1.515ws + 0.1353; ws ≤ 0.79 mm

fs = { (3a)
0.85; ws > 0.79 mm

−1.4938wr2 + 2.2011wr + 0.085; wr ≤ 0.63 mm

fr = { (3b)
0.88; wr > 0.63 mm
where f s and f r are the rust-filling ratios of group S and group RS, respectively; and ws and wr are the
crack widths of group S and group RS, respectively.
The specimens in the present study were immersed in the saline solution and accelerated by the
electrochemical corrosion. The rust-filling ratio obtained in this situation may be different from that in
natural corrosion. The longer corrosion time can induce the higher corrosion degree, which would
lead to the larger crack width and the deeper rust-filling depth.
It should be pointed out that cover depth, crack extension, corrosion rate, and corrosion
environment are size-dependent and can affect the filling of corrosion products. The different concrete
covers might induce the various filling extent of corrosion products. In the practical engineering
structure, the cover ranges from 30 to 50 mm based on the design code [32]. In the present testing, the
cover of 30 mm was used to investigate the filling of corrosion products in cracks. Different corrosion
rate may lead to the various filling extent of corrosion products. The rust-filling ratio obtained in the
electrochemical corrosion may be different from that in natural corrosion. More studies on the filling
of corrosion products in cracks are required.

4. Prediction Model of Crack Propagation

In this section, a model is proposed to predict the crack propagation based on corrosion loss.
The filling of corrosion products and geometric properties of the strand are incorporated in the model.
 Materials 2017, 10, 6 14 of 20

During the corrosion process, corrosion products first fill the pores around the strand-concrete interface
and then contribute to the expansive pressure. After that, it would fill the corrosion-induced cracks.
With the principle of volume equal to the corrosion products, the relationship between the crack width
and corrosion loss can be obtained. Details are shown below.

4.1. Corrosion Products at the Micro-Crack Formation

The accumulation of corrosion products would create expansive pressure at the strand-concrete
interface. When the tangential stress reaches the concrete tensile strength, some micro-cracks would
form. Based on the radial deformation at the strand-concrete interface, the volume of corrosion
products at the micro-crack formation can be obtained.
The strand usually has a relative large diameter and slight corrosion may induce cover
cracking [23]. The distribution of corrosion products around the strand surface relates to the corrosion
degrees. Corrosion products are relatively uniform around the strand surface when the corrosion
degree is low [7]. With increasing corrosion loss, the strand corrosion products may exhibit the
non-uniform distribution. Non-uniform corrosion, compared with uniform corrosion, would induce
the large expansive pressure and accelerate the concrete cracking process. In the present experimental
observation, the strand corrosion degree was low and the depths of corrosion pits were small. In terms
of this phenomenon, corrosion products are simplified to uniformly distribute around the strand in
the prediction model. The expansion of corrosion products would produce a uniform pressure on the
surrounding concrete. Figure 14 shows the expansive deformation at the strand-concrete interface
Materials 2017,by
caused 10,corrosion.
6 14 of 19

Reinforcement
δp
δr
δc
δs R0
6R0

Qc
C

Concrete

Figure
Figure 14.14. Expansivedeformation
Expansive deformation at
at the
thestrand-concrete
strand-concreteinterface.
interface.

The The thick-walledcylinder

thick-walled cylinder model
modelhashasbeen frequently
been employed
frequently to analyze
employed to the corrosion-induced
analyze the corrosion-
induced concrete cracking [3,8,11,21]. This model can be considered as an axisymmetrictoproblem
concrete cracking [3,8,11,21]. This model can be considered as an axisymmetric problem subject the
uniform pressure, which can be further modeled as the plane stress problem under the
subject to the uniform pressure, which can be further modeled as the plane stress problem under thesymmetric
conditions [42]. The governing stress equilibrium in the radial direction is:
symmetric conditions [42]. The governing stress equilibrium in the radial direction is:
dσr
+ − σθ = 0
σr −
(4) (4)
dr + r =0

wherewhere and
σr and σθare
are the
the radial
radial stress
stressand
andthe tangential
the stress
tangential at any
stress radius
at any r.
radius r.
For the plane stress problem under the symmetric conditions, the strain-displacement
For the plane stress problem under the symmetric conditions, the strain-displacement equation is
equation
given as:
is given as: du
εr = (5a)
dr
ε = u (5a)
εθ = (5b)
r
ε = (5b)

where , , and u are the radial strain, tangential strain, and radial displacement at any radius r.
The constitutive relationship between concrete stress and strain is:
Materials 2017, 10, 6 15 of 20

where ε r , ε θ , and u are the radial strain, tangential strain, and radial displacement at any radius r.
The constitutive relationship between concrete stress and strain is:

Ec
σr = (ε r + vc ε θ ) (6a)
(1 − v2c )

Ec
σθ = (ε + vc ε r ) (6b)
(1 − v2c ) θ
where Ec is the elastic modulus of concrete; vc is the Poisson’s ratio of concrete.
Since concrete is a heterogeneous material, a porous zone surrounds the strand-concrete interface.
Corrosion products first diffuse into the porous zone [43]. As corrosion products exceed the quantity
needed to fill the porous zone, these products generate expansive pressure. The radial pressure would
produce a concrete displacement. Combing Equations (4)–(6), the concrete displacement, δr , is:


R0 + δ p
δr = (1 + k + v c ) Q c (7)
Ec

2 
where k is a constant, k = 2 R0 + δp / C2 + 2C R0 + δp ; Qc is the expansive pressure; R0 is the



radius of the wire; C is the concrete cover; and δp is the thickness of the porous zone, δp = 10–20 µm [18].
Considering the geometric relationship, the volume of corrosion products per units of length at
the micro-crack formation, Vm , is:

4πn
Vm = [( R0 + δc − δs )2 − R20 ] (8)
n−1

where n is rust expansion ratio, n = 2–4 [18]; δs is the radial loss of wire; and δc is the thickness of
corrosion products, δc = δs + δp + δr .
The units of R0 are millimeters, and the units of (δp + δr ) are microns. The value of (δp + δr ) is
much smaller than that of R0 . The term, (δp + δr )2 , is neglected in the calculations. Equation (8) is
rewritten as:
4πnR0
Vm = (δP + δr ) (9)
n−1
Combining Equations (7) and (9), the expansive pressure, Qc , can be written as:

Ec (n − 1)Vm
Qc =
( − δp ) (10)
(1 + k + v c ) R0 + δ p 4πnR0

The micro-crack forms when the tangential stress exceeds the concrete tensile strength.
The tangential stress is derived with an elastic mechanics axisymmetric stress solution. Then, the
maximum expansive pressure at the micro-crack formation can be obtained. More details can be found
in Dai et al. [23]. The maximum expansive pressure at the micro-crack formation is expressed as:

C
Qcmax = (0.225 + 0.075 ) ft (11)
R0

where f t is the concrete tensile strength.

Combining Equations (10) and (11), the volume of corrosion products at the micro-crack formation,
Vm , is:  
πnR0 C

Vm = [ 0.9 + 0.3 f t (1 + k + vc ) R0 + δp + Ec δp ] (12)
(n − 1) Ec R0

4.2. Crack Width on the Concrete Surface

Crack widths can be observed after cover cracking. As described previously, some discrete cracks
were small. Most corrosion products were located in the widest crack. The filling of corrosion products
 ( − 1)

4.2. Crack Width on the Concrete Surface

Crack 2017,
Materials widths
10, 6 can be observed after cover cracking. As described previously, some discrete
16 of 20
cracks were small. Most corrosion products were located in the widest crack. The filling of corrosion
products in the small cracks are ignored in the present study. The widest crack in the cross-section is
in the small cracks are ignored in the present study. The widest crack in the cross-section is selected
selected as an analysis object. A simplified trapezoid model is proposed to reflect crack shape in the
as an analysis object. A simplified trapezoid model is proposed to reflect crack shape in the radial
radial direction.
direction. Figure
Figure 15 shows
15 shows the schematic
the schematic diagram
diagram of shape
of crack crack incorporating
shape incorporating theoffilling
the filling the of
the corrosion product.
corrosion product.

wc Brownish-red rust
θ

C
Ri
wb
Puce rust
δp Strand

Black rust
Concrete

Reinforcement

Figure
Figure15.
15. Simplified crackmodel.
Simplified crack model.

As observed in the
As observed previous
in the previousexperiment,
experiment, the fillingofofcorrosion
the filling corrosion products
products varies
varies withwith increasing
increasing
crack width.
crack Crack
width. width
Crack in in
width thethe
radial
radialdirection
directionrelates
relates to the
the cracking
crackingangle.
angle.The
Thecracking
cracking angle is
angle
employed to describe
is employed the the
to describe volume
volume of of
crack.
crack.Considering
Considering thethe filling
fillingofofcorrosion
corrosion products
products and and
the the
cracking angle, the volume of corrosion products in the cracks, V , can be
cracking angle, the volume of corrosion products in the cracks, p , can be written as:written as:

VP = [wc + C ( f − 2)tanθ ] × Ri (13)

where wc is the crack width on the concrete surface.

According to the geometric characteristic, the total volume of corrosion products is:

Vc = nρVs (14)

where Vs is the strand volume per units of length.

With the equal principle of volume, Vc = Vm + Vp . Combining Equations (1) and (12)−(14), the
crack width on concrete surface is:
h  i
C


nVs ρ πnR 0 0.9 + 0.3 R0 f t ( 1 + k + v c ) R 0 + δ p + E δ
c p
wc = C ( f − 2)(b − aρ) + − (15)
Cf (n − 1) Ec C f

The relationship between crack width and the rust-filling ratio is given in Equation (3). Combining
Equations (3) and (15), the crack width can be calculated by the corrosion loss. Results show that
the filling of corrosion products, cover, rust expansion ratio, concrete tensile strength, and geometric
properties of the strand can affect the corrosion-induced cracking. These parameters should be
incorporated in the predicted model.
The material and shape of strands are different to those of steel bars. These would induce the
corrosion mechanisms of strands differ from that of steel bars. Nowadays, no suitable model exists to
relate the corrosion degree and the corrosion time for the strand. Therefore, the prediction model in
the present study was proposed based on the corrosion loss. More studies are needed in the future to
incorporate the time factor into the corrosion-induced cracking model.
corrosion mechanisms of strands differ from that of steel bars. Nowadays, no suitable model exists
to relate the corrosion degree and the corrosion time for the strand. Therefore, the prediction model
in the present study was proposed based on the corrosion loss. More studies are needed in the future
to incorporate the time factor into the corrosion-induced cracking model.
Materials 2017, 10, 6 17 of 20

4.3. Verification of the Prediction Model

4.3. Verification of the Prediction Model
The proposed model was used to predict the corrosion-induced cracking in the present testing.
The proposed model was used to predict the corrosion-induced cracking in the present testing.
Some parameters in the present model were selected as follows: the rust expansion ratio and the
Some parameters in the present model were selected as follows: the rust expansion ratio and the
thickness of the porous zone were selected as 3 and 15 μm, respectively; the proposed rust-filling
thickness of the porous zone were selected as 3 and 15 µm, respectively; the proposed rust-filling
ratio was incorporated into the prediction. Figure 16 shows the comparison between the predicted
ratio was incorporated into the prediction. Figure 16 shows the comparison between the predicted
and and
testtest
crack widths.
crack widths.The Thepredicted
predictederrors
errors are 13.56% and
are 13.56% and11.79%
11.79%forfor group
group S and
S and groupgroup
RS, RS,
respectively. TheThe
respectively. average
average error
errorofofboth
both groups is12.68%
groups is 12.68%and
andthe
the variation
variation coefficient
coefficient is 0.175
is 0.175 with with
a a
95%95%
confidence interval.
confidence interval.

Figure
Figure16.
16. Predicted andtest
Predicted and testcrack
crack widths.
widths.

TheThepredicted errors
predicted errorsmay
mayattribute
attribute to the measurement
to the measurementuncertainty
uncertainty of crack
of crack width,
width, corrosion
corrosion
loss,loss,
andandcracking angle.
cracking Another
angle. Anotherreason
reason is
is that thefilling
that the fillingofofcorrosion
corrosion products
products in small
in the the small cracks
cracks is is
ignored in the
ignored present
in the model.
present model.Additionally, the heterogeneity
Additionally, the heterogeneity andand variability
variability of materials
of materials may may
also also
affect the prediction. These errors can be accepted in view of the complexity of the corrosion-induced
cracking process.
To clarify the filling effect of corrosion products, the theoretical crack widths under the various
rust-filling ratios are also given in Figure 17. As Figure 17 shows, P1 and P2 represent the predicted
results considering the rust-filling ratios as 1 and 0.5, respectively. The rust-filling ratio is a sensitive
parameter for the proposed model. The P1 is smaller than the test value. The P2 is larger than the
test result. The main cause of the discrepancy between the predicted values and the test results is
the filling of corrosion products into cracks. To predict crack widths, it is essential to determine the
rational amount of corrosion products penetrating into cracks.
The predicted values by the proposed method are expressed as the P3. As described previously,
the filling of corrosion products varies with the crack width. By using the proposed rust-filling ratio,
P3 agrees well with the test result. The filling of corrosion products has a significant influence on the
predicted model. The corrosion-induced cracking model should incorporate the rational filling effect
of corrosion products.
rust-filling ratios are also given in Figure 17. As Figure 17 shows, P1 and P2 represent the predicted
results considering the rust-filling ratios as 1 and 0.5, respectively. The rust-filling ratio is a sensitive
parameter for the proposed model. The P1 is smaller than the test value. The P2 is larger than the test
result. The main cause of the discrepancy between the predicted values and the test results is the
filling of corrosion products into cracks. To predict crack widths, it is essential to determine the
Materials 2017, 10, 6 18 of 20
rational amount of corrosion products penetrating into cracks.

(a) (b)

Figure 17.17.
Figure Predicted crack
Predicted crackwidths
widthsunder
under the
the various rust-fillingratios:
various rust-filling ratios:(a)
(a)Group
GroupS; S;
andand
(b)(b) Group
Group RS.RS.

The predicted values by the proposed method are expressed as the P3. As described previously,
5. Conclusions
the filling of corrosion products varies with the crack width. By using the proposed rust-filling ratio,
An experimental investigation is proposed to study the filling of strand corrosion products in
P3 agrees well with the test result. The filling of corrosion products has a significant influence on the
cracked concrete. A prediction model of crack widths is developed incorporating the filling proportion
predicted model.
of corrosion The corrosion-induced
products cracking
and the twisting shape of the model
strand. should incorporate
The following the rational
conclusions filling
are drawn effect
based
of on
corrosion products.test and theoretical analysis:
the experimental

5. •Conclusions
The filling extent of corrosion products varies with crack propagation. The rust-filling ratio
increases with the propagating crack until a critical width. Beyond the critical width, the rust-filling
An experimental investigation is proposed to study the filling of strand corrosion products in
extent remains stable. Using stirrups can decrease the critical crack width.
cracked concrete. A prediction model of crack widths is developed incorporating the filling
• Stirrups can restrict the corrosion-induced crack propagation. The tangent of cracking angle
proportion of corrosion products and the twisting shape of the strand. The following conclusions are
increases with the increasing corrosion degree. Using stirrups decreases the corrosion-induced
drawn based on the experimental test and theoretical analysis:
crack width.
 • The Thefilling extent
proposed of corrosion
model products
can provide varies prediction
a reasonable with crackfor propagation. The rust-filling
corrosion-induced ratio
crack width.
increases with the
The prediction propagating crackcracks
of corrosion-induced until are
a critical width.
sensitive to theBeyond the extent.
rust-filling criticalThe
width, the rust-
prediction
model should incorporate the rational filling effect of corrosion products.
filling extent remains stable. Using stirrups can decrease the critical crack width.
 Stirrups can restrict the corrosion-induced crack propagation. The tangent of cracking angle
It should be pointed out that the specimens in the present study were subjected to an
increases with the increasing corrosion degree. Using stirrups decreases the corrosion-induced
electrochemically-accelerated corrosion. The concrete cracking process induced by the artificial
crack width.
corrosion may be different from that induced by natural corrosion situations. The effect of pre-stressing
 The proposed model can provide a reasonable prediction for corrosion-induced crack width.
on corrosion-induced cracking is also not incorporated. The difference caused by these factors needs
The prediction of corrosion-induced cracks are sensitive to the rust-filling extent. The prediction
to be studied in the future.
model should incorporate the rational filling effect of corrosion products.
Acknowledgments: This work reported here was conducted with the financial supports from the State Key
It should Program
Development be pointed outResearch
for Basic that the specimens
of China in the
(Grant No. present study
2015CB057705), were subjected
the National to an
Natural Science
electrochemically-accelerated corrosion.theThe
Foundation of China (Grant No. 51678069), concrete
National cracking
Science process
Foundation induced by
for Distinguished theScholars
Young artificial
of Hunan Province (Grant No. 14JJ1022), the Key Discipline Civil Engineering Fund, and Safety Control and
corrosion may be different from that induced by natural corrosion situations. The effect of pre-
Durability of Bridge Structures Fund (Grant No. 15YTD06). Their supports are gratefully acknowledged.
stressing on corrosion-induced cracking is also not incorporated. The difference caused by these
Author Contributions: Lei Wang and Lizhao Dai designed the experiments and wrote the paper; Xuhui Zhang
factors needsZhang
and Jianren to beperformed
studied inthethe future.
experiments.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Bossio, A.; Tullio, M.; Francesco, B.; Lignola, G.P.; Prota, A. Modeling of concrete cracking due to corrosion
process of reinforcement bars. Cem. Concr. Res. 2015, 71, 85–92. [CrossRef]
Materials 2017, 10, 6 19 of 20

2. Gurdián, H.; García-Alcocel, E.; Baeza-Brotons, F.; Garcés, P.; Zornoza, E. Corrosion behavior of steel
reinforcement in concrete with recycled aggregates, fly ash and spent cracking catalyst. Materials 2014, 7,
3176–3197. [CrossRef]
3. Liu, Y.; Weyers, R.E. Modeling the time-to-corrosion cracking in chloride contaminated concrete structures.
ACI Mater. J. 1998, 95, 675–680.
4. Balafas, I.; Burgoyne, C.J. Modeling the structural effects of rust in concrete cover. J. Eng. Mech. 2011, 137,
175–185. [CrossRef]
5. Malumbela, G.; Alexander, M.; Moyo, P. Model for cover cracking of RC beams due to partial surface steel
corrosion. Constr. Build. Mater. 2011, 25, 987–991. [CrossRef]
6. Torres-Acosta, A.A.; Sagüés, A.A. Concrete cracking by localized steel corrosion-geometric effects.
ACI Mater. J. 2004, 101, 501–507.
7. Lin, H.; Zhao, Y. Effects of confinements on the bond strength between concrete and corroded steel bars.
Constr. Build. Mater. 2016, 118, 127–138. [CrossRef]
8. Chen, H.; Nepal, J. Analytical model for residual bond strength of corroded reinforcement in concrete
structures. J. Eng. Mech. 2016, 142, 149–158. [CrossRef]
9. Solgaard, A.O.S.; Michel, A.; Geiker, M.; Stang, H. Concrete cover cracking due to uniform reinforcement
corrosion. Mater. Struct. 2013, 46, 1781–1799. [CrossRef]
10. Zhao, Y.; Jin, W. Modeling the amount of steel corrosion at the cracking of concrete cover. Adv. Struct. Eng.
2006, 9, 687–696. [CrossRef]
11. Bažant, Z. Physical model for steel corrosion in concrete sea structures—Theory. J. Struct. Div. ASCE 1979,
105, 1137–1153.
12. Zhao, Y.; Wu, Y.; Jin, W. Distribution of millscale on corroded steel bars and penetration of steel corrosion
products in concrete. Corros. Sci. 2013, 66, 160–168. [CrossRef]
13. Zhao, Y.; Yu, J.; Hu, B.; Jin, W. Crack shape and rust distribution in corrosion-induced cracking concrete.
Corros. Sci. 2012, 55, 385–393. [CrossRef]
14. Lu, C.; Jin, W.; Liu, R. Reinforcement corrosion-induced cover cracking and its time prediction for reinforced
concrete structures. Corros. Sci. 2011, 53, 1337–1347. [CrossRef]
15. Šavija, B.; Luković, M.; Hosseini, S.; Pacheco, J.; Schlangen, E. Corrosion induced cover cracking studied
by X-ray computed tomography, nanoindentation, and energy dispersive X-ray spectrometry (EDS).
Mater. Struct. 2014, 48, 2043–2062. [CrossRef]
16. Val, V.; Chen, H.; Stewart, M.G. Experimental and numerical investigation of corrosion-induced cover
cracking in reinforced concrete structures. J. Struct. Eng. 2009, 135, 376–385. [CrossRef]
17. Jaffer, S.J.; Hansson, C.M. Chloride-induced corrosion products of steel in cracked-concrete subjected to
different loading conditions. Cem. Concr. Res. 2009, 39, 116–125. [CrossRef]
18. Jamali, A.; Angst, U.; Adey, B.; Elsener, B. Modeling of corrosion-induced concrete cover cracking: A critical
analysis. Constr. Build. Mater. 2013, 42, 225–237. [CrossRef]
19. Otieno, M.B.; Beushausen, H.D.; Alexander, M.G. Modelling corrosion propagation in reinforced concrete
structures—A critical review. Cem. Concr. Compos. 2011, 33, 240–245. [CrossRef]
20. Torres-Acosta, A.A.; Navarro-Gutierrez, S.; Terán-Guillén, J. Residual flexure capacity of corroded reinforced
concrete beams. Eng. Struct. 2007, 29, 1145–1152. [CrossRef]
21. Li, C.Q.; Yang, S.T. Prediction of concrete crack width under combined reinforcement corrosion and applied
load. J. Eng. Mech. 2011, 137, 722–731. [CrossRef]
22. Khan, I.; François, R.; Castel, A. Prediction of reinforcement corrosion using corrosion induced cracks width
in corroded reinforced concrete beams. Cem. Concr. Res. 2014, 56, 84–96. [CrossRef]
23. Dai, L.; Wang, L.; Zhang, J.; Zhang, X. A global model for corrosion-induced cracking in prestressed concrete
structures. Eng. Fail. Anal. 2016, 62, 263–275. [CrossRef]
24. Vu, N.A.; Castel, A.; François, R. Effect of stress corrosion cracking on stress–strain response of steel wires
used in prestressed concrete beams. Corros. Sci. 2009, 51, 1453–1459. [CrossRef]
25. Darmawan, M.S.; Stewart, M.G. Effect of pitting corrosion on capacity of prestressing wires. Mag. Concr. Res.
2007, 59, 131–139. [CrossRef]
26. Vélez, W.; Matta, F.; Ziehl, P. Electrochemical characterization of early corrosion in prestressed concrete
exposed to salt water. Mater. Struct. 2015, 49, 507–520. [CrossRef]
Materials 2017, 10, 6 20 of 20

27. Pillai, R.G.; Gardoni, P.; Trejo, D.; Hueste, M.B.D.; Reinschmidt, K.F. Probabilistic models for the tensile
strength of corroding strands in posttensioned segmental concrete bridges. J. Mater. Civil Eng. 2010, 22,
967–977. [CrossRef]
28. Li, F.; Yuan, Y.; Li, C.Q. Corrosion propagation of prestressing steel strands in concrete subject to chloride
attack. Constr. Build. Mater. 2011, 25, 3878–3885. [CrossRef]
29. Wang, L.; Zhang, X.; Zhang, J.; Ma, Y.; Xiang, Y.; Liu, Y. Effect of insufficient grouting and strand corrosion
on flexural behavior of PC beams. Constr. Build. Mater. 2014, 53, 213–224. [CrossRef]
30. Rinaldi, Z.; Imperatore, S.; Valente, C. Experimental evaluation of the flexural behavior of corroded P/C
beams. Constr. Build. Mater. 2010, 24, 2267–2278. [CrossRef]
31. Li, H.; Lan, C.M.; Ju, Y.; Li, D.S. Experimental and numerical study of the fatigue properties of corroded
parallel wire cables. J. Bridge Eng. 2012, 17, 211–220. [CrossRef]
32. Code for Design of Concrete Structures; GB50010-2010; Construction Ministry of PR China: Beijing, China, 2010.
33. Wang, L.; Zhang, X.; Zhang, J.; Yi, J.; Liu, Y. Simplified model for corrosion-induced bond degradation
between steel strand and concrete. J. Mater. Civil Eng. 2016. [CrossRef]
34. Farrow, W.C.; Higgins, C. Tests of reinforced concrete beams with corrosiondamaged stirrups. ACI Struct. J.
2006, 103, 133–141.
35. Yu, L.; François, R.; Dang, V.; L’Hostis, V.; Gagné, R. Distribution of corrosion and pitting factor of steel in
corroded RC beams. Constr. Build. Mater. 2015, 95, 384–392. [CrossRef]
36. Zhang, R.; Castel, A.; François, R. Concrete cover cracking with reinforcement corrosion of RC beam during
chloride-induced corrosion process. Cem. Concr. Res. 2010, 40, 415–425. [CrossRef]
37. Standard for Test Methods of Long-Term Performance and Durability of Ordinary Concrete; GB-T50082-2009;
Ministry of Housing and Urban-Rural Development of the PR China: Beijing, China, 2009.
38. Fu, A.Q.; Cheng, Y.F. Effects of alternating current on corrosion of a coated pipeline steel in a
chloride-containing carbonate/bicarbonate solution. Corros. Sci. 2010, 52, 612–619. [CrossRef]
39. Papadakis, V.G. Effect of supplementary cementing materials on concrete resistance against carbonation and
chloride ingress. Cem. Concr. Res. 2000, 30, 291–299. [CrossRef]
40. Scott, A.; Alexander, M.G. Effect of supplementary cementitious materials (binder type) on the pore solution
chemistry and the corrosion of steel in alkaline environments. Cem. Concr. Res. 2016, 89, 45–55. [CrossRef]
41. Nossoni, G.; Harichandran, R.S. Electrochemical-mechanistic model for concrete cover cracking due to
corrosion initiated by chloride diffusion. J. Mater. Civil Eng. 2014, 26, 362–376. [CrossRef]
42. Bhargava, K.; Ghosh, A.; Mori, Y.; Ramanujam, S. Modeling of time to corrosion-induced cover cracking in
reinforced concrete structures. Cem. Concr. Res. 2005, 35, 2203–2218. [CrossRef]
43. Maaddawy, T.E.; Soudki, K. A model for prediction of time from corrosion initiation. Cem. Concr. Compos.
2007, 27, 168–175. [CrossRef]

© 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).