Engineering Structures 75 (2014) 113–125

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Finite element investigation of the influence of corrosion pattern on

inelastic buckling and cyclic response of corroded reinforcing bars
Mohammad M. Kashani a,⇑, Laura N. Lowes b, Adam J. Crewe a, Nicholas A. Alexander a
a
University of Bristol, Dept. of Civil Engineering, Bristol BS8 1TR, United Kingdom
b
University of Washington, Dept. of Civil and Environmental Engineering, Seattle, WA 98195-2700, United States

a r t i c l e i n f o a b s t r a c t

Article history: An optical surface measurement technique was used to characterise three-dimensional corrosion pattern
Received 19 November 2013 of reinforcing bars subjected to accelerated corrosion. After the optical measurement process was per-
Revised 3 April 2014 formed, the corroded bar specimens were tested under monotonic and cyclic axial loading. The optical
Accepted 20 May 2014
measurement data were used to develop a 3D micro-fibre finite element model developed for simulation
Available online 19 June 2014
of the physical testing and parametric study of the influence of corrosion pattern on stress–strain
response of corroded bars. It was observed that the irregular cross sectional shape of pitted sections
Keywords:
has a significant influence on the inelastic buckling and nonlinear cyclic response.
Reinforcing steel
Corrosion
Ó 2014 Elsevier Ltd. All rights reserved.
Inelastic buckling
Stress–strain relations
Nonlinear analysis
Cyclic behaviour
Low-cycle high amplitude fatigue

1. Introduction Recent experimental testing of corroded RC elements subject to

Corrosion of reinforcing bars in aging reinforced concrete (RC)
structures and bridges is a critical issue in developed countries
[1–3]. Billions of dollars are spent annually by government and
other agencies monitoring, repairing, and replacing RC structures
with corrosion damage [1–3]. There have been incidences in which
corrosion damage resulted in the complete collapse of in-service
structures [3]. Because of this, a substantial body of research in
the past decades has focussed on corrosion in RC structures. This
research has addressed a number of topics including the corrosion
process, the behaviour of corroded reinforcement, performance
assessment and repair of corroded structures, and maintenance
optimisation for transportation structures [4,5]. A relatively small
portion of this research has addressed the impact of corrosion on
the seismic performance of RC structures, despite the fact that
there are many RC structures located in regions of high seismicity
and exposed to corrosive environments.
For RC bridges subjected to earthquake loading, the desired
inelastic response mechanism is typically flexural yielding of the
columns or piers, and drift capacity (i.e. the drift at which
significant lateral strength loss occurs) is typically determined by
buckling followed by fracture of longitudinal reinforcing bars [6].
⇑ Corresponding author.
E-mail address: mehdi.kashani@bristol.ac.uk (M.M. Kashani).
cyclic loading showed that corrosion significantly reduces plastic
rotation capacity, and thus drift capacity; this is due to premature
buckling and/or fracture of the longitudinal reinforcing bars as well
as premature fracture of horizontal hoop reinforcement that con-
fines core concrete and restrains longitudinal bar buckling [7,8].
In the recent years researchers have investigated the influence
of corrosion on the stress–strain behaviour of corroded bars in ten-
sion [9–14]. The outcome of theses researches showed that corro-
sion has significant influence on the ductility and plastic
deformation capacity of reinforcing bars owing to the level of
non-uniform pitting. To improve understanding of the impact of
corrosion on inelastic buckling and cyclic response, Kashani et al.
[15,16] conducted a comprehensive experimental investigation of
the influence of corrosion on inelastic buckling and nonlinear cyc-
lic behaviour of corroded reinforcing bars. The experimental
results show that the non-uniform distribution of pitting along
the length of corroded bars changes the buckling mechanism and
results in more rapid strength loss under cyclic loading. Kashani
et al. report that when the mass loss ratio of corroded bars is more
than approximately 25%, pitting corrosion becomes more localised.
Since the smallest cross section of the bar determines the yield
strength; localisation of pitting corrosion results in premature
yielding of the reinforcement at a relatively low tension demand.
Additionally, Kashani et al. [17] show that pitting corrosion results
in variation in the cross sectional shape of the reinforcement along

http://dx.doi.org/10.1016/j.engstruct.2014.05.026
0141-0296/Ó 2014 Elsevier Ltd. All rights reserved.
114 M.M. Kashani et al. / Engineering Structures 75 (2014) 113–125
the length of the corroded bar. This results in a change in the
moment of inertia of the reinforcing bar and potentially affects
the buckling response of the bar under cyclic loading.
This paper explores the influence of the pattern of corrosion
along the length of a reinforcing bar on inelastic buckling and cyc-
lic response. Data from a previously presented experimental study
[15,16] are reviewed with the objectives of establishing the
response of corroded bars under monotonic and cyclic loading
and characterising the pattern of corrosion along the length of
the reinforcing bar. A computational model, employing nonlinear
beam–column elements with fibre-type cross-section models and
nonlinear material models, is proposed for simulating the response
of a corroded reinforcing bar. The variation in the geometrical
properties of corroded bars is modelled using the optical measure-
ment data provided in [17]. Moreover, for the first time, the varia-
tion in load eccentricity due to pitting corrosion along the buckling
length of bars is also included in the finite element model. The
model is validated through comparison with experimental data.
The model and experimental data are used to investigate the
impact of the corrosion pattern on the buckling strength, tensile
yield strength and cyclic response of a reinforcing bar. Finally,
experimental data are used to evaluate existing one-dimensional
constitutive models for reinforcing steel, which simulate compres-
sive strength loss due to buckling and the impact of buckling and
low-cycle fatigue on cyclic response.
2. Corrosion induced mechanical–geometrical degradation of
reinforcing bars
Kashani et al. [15] tested a series of corroded reinforcing bars
under monotonic and cyclic loading. Experimental data character-
ise the average stress–strain response of the bars under monotonic
and cyclic loading; such data can be used to develop and validate
the 1D constitutive models for reinforcing steel used typically in
fibre-type models of reinforced concrete flexural elements. Exper-
imental data characterise also the pattern of corrosion along the
length of the bars, from which variation in the bar area and first
and second moments of area are computed.
2.1. Summary of experimental programme
To develop the corroded reinforcement specimens for testing, a
total of eight reinforced concrete prisms were cast. Each specimen
was dimensioned 250  250  700 mm and incorporated 8 No.
12 mm diameter reinforcing bars. An accelerated corrosion simula-
tion technique (known as anodic corrosion) was employed to accel-
erate the corrosion time in the laboratory environment. Different
prisms were subjected to different corrosion times to achieve differ-
ent levels of reinforcement mass loss due to corrosion. After corro-
sion simulation, the concrete specimens were broken open and the
corroded bars were carefully removed from the concrete and
cleaned in accordance with ASTM G1-03 [18]. A total of 120 corroded
bars were prepared. Twenty-three (23) randomly selected corroded
bars with different mass loss ratios were selected for optical mea-
surement and statistical analysis of the corrosion patterns. Fifty-
seven (57) bars were used for monotonic buckling tests and 40 were
used for cyclic tests. A detailed discussion of specimen preparation,
optical measurement and testing is provided in [15–17].
Mechanical testing was conducted for a series of different slen-
derness ratios. The slenderness ratios were chosen based on the
range of horizontal tie spacings commonly used in RC column con-
struction. The slenderness ratio (L/D) is defined as the unsupported
length of the bar (L) divided by the diameter of the reinforcing bar
(D). The L/D ratios tested were 5, 8, 10, 15, and 20 for the monotonic
buckling tests and 5, 10 and 15 for the cyclic tests. For each slender-
ness ratio, three control (uncorroded) specimens were also tested.
2.2. 3D optical measurement of corrosion pattern
Twenty-three (23) randomly selected corroded reinforcing bars
were taken out of the total of 120 samples for more refined geo-
metrical surface analysis of corrosion patterns. The reinforcing bars
varied in length (L/D = 5 to L/D = 20) and had a range of mass loss
ratios. The surface pitting pattern of the corroded bars was mea-
sured using a white light scanner followed by a stochastic analysis
of the results in MATLAB [19]. Fig. 1 shows an example contour
plot of scanned bar. The surface area of the bar is represented by
the length along the bar (0–200 mm) and the angle of rotation
(p to p) from an arbitrary plane; the impact of corrosion is repre-
sented by the radius from the original centroid of the bar to the
surface of the corroded bar (2.0–6.0 mm). Fig. 2 also shows an
example of cross sections through the solid model of a corroded
bar with 54.23% mass loss. A detailed discussion of the optical
measurement process and analysis of optical data is provided in
[17]. Once the corrosion pattern analysis completed the 23
scanned bars have been tested under monotonic and cyclic loading.
2.3. Observed response under monotonic loading
The experimental results show that the non-uniform pitting
corrosion changes the buckling mechanism of corroded bars. The
pitted sections have irregular shapes that vary over the entire
length of the bar [15]. This irregularity creates strong and weak
axes, for which the second moments of area is reduced from the
uncorroded state, and creates load eccentricity that results in a sig-
nificant reduction in buckling capacity for corroded bars. Fig. 3
shows examples of mean stress versus strain data for corroded bars
tested in the laboratory under monotonic compression loading.
The mean stress presented in Fig. 3 was computed as the applied
load divided by the average reduced area of the bar assuming uni-
form mass loss over the unsupported length (L) of the bar.
The observed responses in Fig. 3 show that considering only the
average reduction in bar cross sectional area is not sufficient for
characterising the impact of corrosion on stress–strain response.
Other parameters such as the minimum area, centroid of the sec-
tion, and minimum second moment of area must be considered
in constitutive modelling of corroded bars.
2.4. Observed response under cyclic loading
The experimental data of the cyclic tests support the following
observations of interest to the current work:
 It was found that the cyclic degradation of the reinforcing bars
is very sensitive to the strain history. The buckling strength is
also significantly affected by the strain history and the tension
strain amplitude.
 Corroded bars subjected to cyclic loading exhibit reduced buck-
ling strength in comparison with uncorroded bars but also in
comparison with corroded bars subjected to monotonic com-
pressive loading. Strength reduction under cyclic loading is
Fig. 1. Contour plot of corrosion pattern in a corroded bar with 36.40% mass loss.
 M.M. Kashani et al. / Engineering Structures 75 (2014) 113–125
Fig. 2. Cross section through the solid model of a corroded bar with 54.23% mass
loss.
attributed to premature tensile yielding of the corroded bar.
Premature tensile yielding results from a reduction in bar area
due to corrosion (Fig. 4).
 Under cyclic loading, corroded bars exhibit more rapid strength
loss with increasing strain demand than do bars without
corrosion.
 Corrosion reduces the strain at which bars fracture in tension
(Fig. 5(a)). It was observed that for bars with higher slenderness
ratios the amount of corrosion needed to cause tension fracture
under cyclic loading is reduced (Fig. 5(a)). This is due to the
effect of inelastic buckling.
 Corrosion acts to reduce the effective slenderness ratio, L/D, of
the bar. For example, uncorroded bars with L/D = 5 displayed
a symmetric hysteresis response. However, as the level of corro-
sion increased the hysteresis response of this group of bars
showed a pinching effect in a similar way to the bars with
higher slenderness ratios (Fig. 5(b)).
Further discussion of the experimental results is provided in [16].
3. Micro-fibre nonlinear finite element analysis of reinforcing
bars
Nonlinear finite element analysis has been used by other
researchers to study the behaviour, including buckling, of
reinforcing bars without corrosion under monotonic and cyclic
loading. Mau and El-Mabsout [20] developed a special beam–
column element for nonlinear finite element analysis of inelastic
buckling of reinforcing bars. Using a fibre-type model of the bar
cross section, Dhakal and Maekawa [21,22] conducted a nonlinear
Fig. 3. Observed buckling response of corroded bars under monotonic loading: (a) compression response with L/D = 8 and (b) compression response with L/D = 15.
115
Fig. 4. View of the critical pitted section in solid model and experimental testing of
a corroded bar with L/D = 15 and 34.26% mass.
finite element analysis of individual reinforcing bars, with different
L/D and yield strengths, to investigate their buckling response.
Nakamura and Higai [23] performed nonlinear finite element anal-
yses studying the stress–strain behaviour of reinforcing bars under
monotonic and cyclic loading. In all cases, analyses and experi-
ments considered isolated bars (i.e. bars were not embedded in
concrete); in all cases, good agreement between the experimental
data and finite element results was observed. Here nonlinear finite
element analysis was employed to investigate the impact of corro-
sion on the behaviour of reinforcement under monotonic and cyc-
lic loading. An approach similar to that employed by Dhakal and
Maekawa [21] for analysis of bars without corrosion was
employed.
3.1. The finite element model
The open-source three-dimensional nonlinear finite element
analysis platform OpenSees (Open System for Earthquake
Engineering Simulation) [24] and the force-based distributed
plasticity beam–column (FBBC) element formulation available in
the OpenSees platform were used for the analyses [25,26]. This
element formulation assumes a linear moment distribution and
constant axial force distribution along the length of the element.
Nonlinear material response under flexural-type loading is simu-
lated at the section level. For this study, a fibre-type model of
the reinforcing bar cross section was used; fibre response was
defined using a nonlinear constitutive model. Moment and axial
load at a section of the beam–column element are determined by
integrating the fibre stresses over the section. Using the FBBC
element formulation, generalised nodal displacements are
determined from section deformations using a Gauss-Labotto inte-
gration scheme. Previous research conducted by [27,28] showed
that co-rotational formulation is computationally very efficient
when used in beam–column elements to model the geometrical
116 M.M. Kashani et al. / Engineering Structures 75 (2014) 113–125
Fig. 5. Observed nonlinear cyclic response of corroded bars: (a) L/D = 15 and 36.35% mass loss and (b) L/D = 5 and 40.75% mass loss.
nonlinearity. Therefore, co-rotational formulation is implemented
nonlinear beam–column element in the OpenSees and is used in
this study. Further detail is available in [27,28].
A displacement control with an adaptive solution algorithm
code is developed using the Tcl code in the OpenSees to run the
nonlinear analysis. To check the solution convergence, the norm
of displacement increment is check against a defined tolerance.
This command is implemented in the OpenSees to construct a con-
vergence test which uses the norm of the left hand side solution
vector of the matrix equation to determine if convergence has been
reached. The tolerance considered is 108 over the maximum of 35
iterations. The analysis starts with Newton–Raphson solution algo-
rithm and if the convergence is not achieved the displacement
increment is cut by a factor of 0.1 and if the convergence is not
achieved again the displacement increment is cut by another factor
of 0.1. If the convergence is not achieved by cutting the step sizes
twice, then the solution algorithm is changed. The solution algo-
rithms used in the adaptive solution strategy in this study are
including Newton–Raphson, modified Newton–Raphson, New-
ton–Raphson with Line Search and Krylov–Newton algorithms.
Further details of algorithm commands are available in [24].
A series of preliminary analyses were conducted to determine
model parameters for use in the subsequent analyses. A mesh
refinement study was conducted at the section and component
levels. It was found that a section mesh comprising 6 fibres in
the radial direction and 35 fibres in the circumferential direction
resulted in a converged section response. It was found that use of
10 mm long elements with three Gauss–Lobatto integration points
per element resulted in a converged component response. An ini-
tial imperfection of 0.001 mm (1.0  105 to 4.16  106 of L)
was introduced at mid height of the bar to initiate buckling (the
initial imperfect shape was linear). To emulate the boundary con-
ditions during the laboratory tests, the rotations and displace-
ments of the bottom node of the bars were fully restrained in the
model. The axial displacement history applied in the laboratory
experiments was applied in the analyses.
3.2. Steel material model
Dhakal and Maekawa [22] modelled the path-dependent
behaviour of isolated reinforcing bars exhibiting buckling. They con-
cluded that the average stress–strain and buckling behaviour of bars
in compression and the unloading/reloading behaviour of cyclic
loops are greatly influenced by the accuracy of the basic steel model
used for the analysis. Therefore, to model the cyclic behaviour of
bars correctly, an accurate steel model must be used that is able to
simulate the unloading/reloading behaviour of reinforcing steel
including the Bauschinger effect. In this study the Giuffre–Menegot-
to–Pinto (GMP) model [29] implemented in OpenSees as the steel02
material model is used to model the uniaxial stress–strain response
of steel. The GMP model employs a smooth transition curve that
asymptotically approaches bilinear tension and compression
response envelopes (Fig. 5); this enables simulation of the Bauschin-
ger effect. Variables in Fig. 6 are defined in Table 1. Table 1 provides
also the values used in the analyses, which were determined from
coupon tests conducted per ASTM E8/E8M-09. It should be noted
that the ultimate strain (eu) is the strain at the ultimate strength (ru).
3.3. Model validation
For validation of the computational model, computed and
experimental average stress versus strain results for uncorroded
bars were compared. Fig. 7(a–d) shows a comparison between
the computed and observed experimental response of monotonic
tests for uncorroded bars with L/D = 8, 10, 15 and 20 respectively.
A very small deviation between the computed and experimental
results is seen in Fig. 7(a and b) which is due to the strain harden-
ing and spread of plasticity at the plastic hinge location of the bars
with L/D < 10 which may not be captured in the fibre model. How-
ever, as the buckling length of bars increases the computed and
experimental results are almost identical (Fig. 7(c and d)).
Fig. 8(1-c) shows a comparison between the computed and
observed experimental response of cyclic tests for uncorroded bars
with L/D = 5, 10 and 15 respectively. As shown in Figs. 7 and 8 the
numerical model accurately simulates the monotonic and cyclic
response of reinforcing bars out to relatively large strain demands.
The simulation results show that the computational model is able
to capture the pinching effect seen in the bars with L/D > 10 under
the large displacement demand. Furthermore, the compressive
σ
Eh
Es
εy ε
Fig. 6. The Giuffre–Menegotto–Pinto model.
 M.M. Kashani et al. / Engineering Structures 75 (2014) 113–125 117

Table 1 of area, an equivalent rectangular cross section was used in the

Mechanical properties of uncorroded reinforcement. fibre model of a corroded bar. The equivalent section had the same
Reinforcement type 12 mm (B12) area and the minimum second moment of area as calculated from
Yield strength ry (MPa) 540 optical measurement data. To validate this assumption, initially,
Modulus of elasticity Es (MPa) 210,000 equivalent circular and rectangular cross section shapes were
Hardening ratio Eh/Es 0.008 considered for an uncorroded bar. It was observed that the cross
Ultimate strength ru (MPa) 616 section shape has no effect on the computational response (Fig. 9).
Ultimate strain at ultimate strength eu 0.06
Strength ratio ru/ry 1.14
As shown in Fig. 2 the cross section shape of corroded bars can
be considered as s a simple a non-self intersecting polygon. There-
strength degradation of bars under cyclic loading due to the influ- fore, using Green’s Theorem the area of a simple polygon with n
ence of tension history observed in the experiment is also captured vertices can be calculated as defined in Eq. (1).
accurately in the computational simulation. For bars with L/D = 10
1Xn1
 
and 15 subjected to cyclic loading, crack propagation caused by A¼ y zi1  yiþ1 zi ð1Þ
2 i¼0 i
low-cycle fatigue in bars resulted in degradation of the tension
envelope and ultimate tensile strength; the numerical model was where y and z are the coordinate of the vertices.
not able to capture this response as strength degradation due to Subsequently by taking moments about the centre of the
low-cycle fatigue is not incorporated in the steel02 material model. polygon, the centroid and second moment of area about y and z
axis can be calculated using Eqs. (2)–(6).
3.4. Modelling the effect of corrosion pattern
1 X
n1
 
Cy ¼ ai yi þ yiþ1 ð2Þ
As it was shown in Figs. 2 and 4, pitting corrosion creates irreg- 6A i¼0
ular cross section shapes along the length of corroded bars. This
irregularity creates strong and weak axes, for which the second 1 X
n1

moments of area is reduced from the uncorroded state, and creates Cz ¼ ai ðzi þ ziþ1 Þ ð3Þ
6A i¼0
load eccentricity at the cross section. These factors must be
included in analysis of isolated corroded reinforcing bars.
1 X
n1
 
Iy ¼ ai z2i þ zi ziþ1 þ z2iþ1 ð4Þ
3.4.1. Modelling the cross section with equivalent rectangular 12 i¼0
As it was discussed in previous sections, nonuniform corrosion
affects both area of cross section and second moment of area due to 1 X
n1
 
the change in the cross section shape. Circular section has only one Iz ¼ ai y2i þ yi yiþ1 þ y2iþ1 ð5Þ
12 i¼0
parameter (radius), so that, it is not possible to fit two parameters
(area and second moment of area) in a circular section. Therefore,  
ai ¼ yi zi1  yiþ1 zi ð6Þ
to model the impact of corrosion on both area and second moment

Fig. 7. Comparison of fibre model simulation with monotonic test results: (a) L/D = 8, (b) L/D = 10, (c) L/D = 15 and (d) L/D = 20.
118 M.M. Kashani et al. / Engineering Structures 75 (2014) 113–125
Fig. 8. Comparison of fibre model simulation with cyclic test results: (a) L/D = 5, (b) L/D = 10 and (c) L/D = 15.
Fig. 9. Comparison of the nonlinear response of beam–column element models
with circular and equivalent rectangular fibre sections.
The above equations are used to calculate the geometrical proper-
ties of corroded bars and used in the fibre model.
3.4.2. Modelling the load eccentricity
Pitting corrosion results also in shifting of the centroid of the
section along the length of the bar. Since load is applied through
the average centroid of bar, at any location along on the length
of the bar, the axial load may be applied eccentrically. This load
eccentricity creates an imperfection in the corroded bars and sig-
nificantly affects the inelastic buckling response.
To simulate the shifting centroid of the corroded bar, the nodes
of the nonlinear beam–column elements were located at the mea-
sured centroid of the section and elastic beam–column elements
were used as rigid link elements (RGE) to connect the nonlinear
elements.
The load eccentricity is considered as the product of the cen-
troid in y and z axes as defined in Eq. (7):
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
eðxÞ ¼ cy ðxÞ2 þ cz ðxÞ2 ð7Þ
where e(x) is the coefficient of load eccentricity ratio as a function of
length x and cy(x) and cz(x) are the section centroid in the y and z
axes respectively. The load eccentricity is used together with the
minimum principal second moment of area (considering axis
rotation) and applied to the weak axis of the equivalent rectangular
section in the fibre model. Fig. 10 shows an idealised view of the
finite element model in the OpenSees [24] including the RGEs
arrangement. The elastic beam–column elements were defined to
have an axial and flexural stiffness (EA and EI) twenty times higher
than the axial and flexural stiffness of the original uncorroded
reinforcing bars (a 12 mm diameter bar).
3.4.3. Mesh sensitivity and averaging length
The optical measurement data defined the corroded bar cross
section at 0.5 mm intervals. However, the results of stochastic
corrosion pattern analysis using autocorrelation and cross
correlation functions showed that averaging the data over 10 mm
segments yields a correlation factor above 70% (the detailed dis-
cussion is available in [17]). This suggests that the variation of
the cross section within 10 mm of any given point along the bar
is relatively small and that meshing the corroded bar using
10 mm long elements with constant cross sections could be
expected to yield accurate results. Since the computational
demand of modelling with 0.5 mm is significantly greater than
modelling using 10 mm long elements, the impact of mesh refine-
 M.M. Kashani et al. / Engineering Structures 75 (2014) 113–125
P 4. Discussion of computational results and comparison with
Rigid link elements
to represent the
eccentricity
10mm fibre
beam-column
elements
Fig. 10. Schematic representation of fibre beam–column model of a corroded bar
with eccentricity.
ment was investigated. The nonlinear buckling responses of two
finite element models, one with 10 mm and one with 5 mm ele-
ments, was compared (Fig. 11). In both models, a constant cross
section was assumed for the entire element length, cross section
geometric properties (area, centroid and second moment of area)
were defined as the average over the element length, and three
Gauss Lobatto integration points per element were used. As it is
shown in Fig. 11, reducing the element length from 10 mm to
5 mm did not significantly affect the nonlinear buckling response
of a corroded bar with L/D = 20 and 34.91% mass loss. It is also find
very small element lengths results in numerical instability at post-
buckling response (Fig. 11). Therefore, using 10 mm long elements
are reasonably accurate and also computationally efficient.
3.4.4. Reference model
Preliminary analyses resulted in the development of a reference
model used for all subsequent analyses. This reference model
employed (i) geometric properties (area, centroid and second
moment of area) were averaged over a 10 mm length, (ii) 10 mm
long force-based beam–column elements with three Gauss–
Lobatto integration points, and (iii) RGE to link element end nodes
and represent eccentricity of the corroded section.
Fig. 11. Comparison of the nonlinear response of fibre models with 5 mm and
10 mm elements.
119
observed experimental results
4.1. Buckling response under monotonic loading
To investigate the factors that affect the buckling strength of
corroded bars, experimental and simulation data were compared
for specimens with different L/D ratios and mass loss ratios and
for models with and without representation of section eccentricity
resulting from corrosion. Models varied only in representation of
eccentricity; all numerical models represented the loss in section
area and second moment of area resulting from corrosion.
The data in Fig. 12(a) shows load–displacement response for a
bar with L/D = 20 and a mass loss ratio of 34.91%. These data show
that if the reduction in cross section area and second moment of
area are included in the simulation, a 46% reduction in peak buck-
ling load is predicted; if these factors as well as load eccentricity
are included in the simulation, a 73% reduction in peak buckling
strength is predicted. The data show also that even with load
eccentricity included in the model, predicted buckling strength is
18% greater than measured strength. This is attributed to irregular
cross sectional shape of pitting sections. This will result in stress
concentration at pitted sections and subsequently premature
yielding of the section. This phenomenon then results in an addi-
tional imperfection in the bar under compression which affects
the buckling load capacity. As expected, it was observed that the
influence of cross section shape is a function of mass loss and slen-
derness ratio. Finally, the data in Fig. 12(a) shows that while the
model over predicts maximum buckling strength, post-buckling
response is accurately predicted using the model with or without
simulation of load eccentricity.
Fig. 12(b and c) shows simulated and experimental response for
bars with L/D = 15 and 10, respectively; in these figures, simulation
results include consideration of the impact of corrosion on area,
second moment of area and load eccentricity. For these tests, max-
imum simulated strength exceeds measured strength by 16% for L/
D = 15 and 10% for L/D = 10. This shows that the difference
between the simulated and experimental results reduced by reduc-
ing the slenderness ratio of the corroded bars. However, as the
slenderness ratio of bars reduces the contribution of cross section
shape to the overall reduction of buckling load increases. In other
words, as the slenderness ratio of the bars increases the impact
of imperfection induced by the combined cross sectional shape
effect and second moment of area on the buckling load increases.
Comparison of Fig. 12(b and c) with Fig. 12(a) shows that bigger
mass loss ratio results in more sever localised pitting corrosion and
subsequently degrades the geometrical properties of pitted sec-
tions more significantly. This will increases the imperfection in
the bar due to effect of second moment of area and cross section
shape that has more significant effect on the bars with bigger slen-
der ratio. Whereas in the shorter bars, the influence of second
moment of area is less visible and the main parameter governing
the buckling load reduction is the premature yielding of pitted sec-
tions due to stress concentration. Finally, the data in Fig. 12(a–c)
shows that while corrosion reduces buckling strength, it does not
significantly affect the shape of the post-peak response curve and
that the numerical model, with or without simulation of eccentric-
ity, accurately simulates post-peak response.
4.2. Buckling response under cyclic loading
Experimental and computational results are compared for cor-
roded bars with different corrosion patterns along the length of
the corroded bars. Here the uniformity of the corrosion pattern is
quantified by considering the ratio of the minimum cross section
area to the maximum cross section area along the gauge length
120 M.M. Kashani et al. / Engineering Structures 75 (2014) 113–125

Fig. 12. Comparison of the simulated and experimental results of corroded bars: (a) corroded bar with L/D = 20 and 34.91% mass loss, (b) corroded bar with L/D = 15 and
19.93% mass loss and (c) corroded bar with L/D = 10 and 14.44% mass loss.

Fig. 13. Comparison of the uniform and localised pitting: (a) corroded bar with L/D = 10 and 38.97% mass loss, (b) corroded bar with L/D = 15 and 34.26%.

defined in Eq. (8). Given the stress–strain response is the average
strain over the buckling length, therefore in this study, the buck-
ling length is considered as the gauge length.
Minimum Cross SectionArea ðAmin Þ
f¼
Maximum Cross SectionArea ðAmax Þ
If f > 0.8, the specimen considered as being uniformly corroded and
if f < 0.8, it is considered as specimen with localised pitted section.
This assumption is in good agreement with the observed experi-
mental results and the results reported by other researchers
[10,11]. Fig. 13 shows the variation of the cross section area of
two specimens with relatively similar mass loss ratio and different
corrosion patterns. Fig. 13(a) shows a uniformly distributed area
loss with f = 0.80 and Fig. 13(b) shows a localised pitted specimen
with f = 0.67.
Fig. 14(a) shows the experimental and simulation results for a
corroded with a relatively uniform pattern of pitting corrosion,
(specimen in Fig. 13(a)); Fig. 14(b) shows results for a corroded
ð8Þ
with a relatively non-uniform pattern of pitting corrosion, (speci-
men in Fig. 13(b)). Fig. 14 shows that for a bar with a relatively uni-
form pattern of corrosion, the model provides reasonably accurate
simulation of response, with two exceptions. The numerical model
over predicts compressive strength by as much as 10%; as with
monotonic loading, this is attributed to 15%. As cyclic displacement
demands increase, the experimental data show deterioration in
tensile strength and, ultimately, complete strength loss due to
bar fracture. Strength loss is attributed to low-cycle fatigue; exper-
 M.M. Kashani et al. / Engineering Structures 75 (2014) 113–125
Fig. 14. Comparison of the simulated and experimental results of a corroded bar under cyclic loading: (a) L/D = 10 with 38.97% mass loss and (b) L/D = 15 with 34.26% mass
loss.
imental result shows that corrosion reduces the low-cycle fatigue
life of the reinforcement. Strength degradation due to low-cycle
fatigue is not captured by the numerical model; thus degradation
of tensile strength is not simulated.
When the irregularity of corrosion along the bar increases, the
differences between the computed and experimental responses
increase. Fig. 14(b) shows the computed and experimental response
for a bar with severe localised pitting corrosion (f = 0.67). Simula-
tion data are provided for the reference model described above as
well as for a model in which the cross section of the bar is constant
over the height of the bar and the minimum measured bar area and
second moment of area are used. Using the reference model, tensile
and compressive strength are overestimated. Using the minimum-
section model, computed response is in a very good agreement with
the observed experimental response. This supports the conclusion
of [16], that the nonlinear cyclic behaviour of corroded bars with
a localised pitting corrosion is mainly governed by the inelastic
response of the minimum pitted section.
A view of the solid model and fracture mechanism of the cor-
roded bar analysed in Fig. 14(b) are previously shown in Fig. 4. It
is evident from Figs. 4 and 14(b) that the plastic hinging mecha-
nism of the bar starts at the location of the minimum pitted section
which then follows by a severe inelastic buckling in compression.
This mechanism results in a significant stress concentration at
the minimum pitted section which subsequently increases the
total strain amplitude at this location during the cyclic tests. This
will cause the premature fracture of bar in tension due to reduced
low-cycle high-amplitude fatigue life.
It should be noted that the location of the minimum section is
also very important. In the test specimen shown in Figs. 4 and
14(b), the minimum section is located very close to the mid height
of the bar. This resulted in initiation of the buckling at that location.
Another example is shown in Fig. 15 where f = 0.70 and the mini-
mum section is located very close to the grip. Despite the localised
pitted section in this specimen the response is not governing by
the minimum section (Fig. 15(a)). It is clearly shown in Fig. 15(b
and c) that a plastic hinge has been formed at the location of the min-
imum section but the bar did not fracture at that location. This cor-
rosion pattern resulted in an unsymmetrical buckling mode shape as
reported previously by Kashani et al. [15,16]. Due to the increase in
the local strain amplitude at the buckling location the low-cycle fati-
gue life of this specimen is shortened. As a result the specimen frac-
tured at the buckling location instead of the minimum section. It is
evident from Fig. 15(a) that in this case the numerical model using
the minimum cross section is underestimating the capacity and
shows more pinching effect in the response. However, the detailed
model shows a very good fit to the observed experimental response.
121
5. Critical review of the existing analytical models and
comparison with the experimental and computational results
In the previous sections the influence of different parameters on
inelastic behaviour of corroded bars under monotonic and cyclic
loadings are discussed. As it was discussed in the previous sections,
the irregularity in shape and distribution of pitted sections along
the bars has a significant impact on the buckling and fracture
mechanism of corroded bars. Due to the random nature of corro-
sion phenomenon and significant uncertainties associated with
the corrosion pattern, it is almost impossible to quantify the irreg-
ularity of pitted sections with a deterministic approach. Moreover,
in the real world, it is very challenging to measure the area loss and
the locations of maximum/minimum pitted sections in inspection
of bridges using a non-destructive method; i.e. in a bridge pier.
However, with the help of recent technologies in bridge inspec-
tions, it is possible to measure the corrosion rate and any given
location in a RC component [3]. Using the corrosion rate measure-
ment data it is very simple to calculate the average mass loss of
reinforcing bars. Therefore, it is very useful to relate all the models
to a single measure parameter such as percentage mass loss. To do
this, an empirical approach using the experimental results should
be used to model the average behaviour of corroded bars. Using
this method it is possible to adopt a pseudo stress–strain curve
as a function of the average percentage mass loss of corroded bars.
Using this method the yield and ultimate strengths are adjusted
empirically to account for pitting effects. This is sufficient to inves-
tigate the global structural response of bridges/structures under
vehicle and/or earthquake loading. If further detailed modelling
is required, (for example detailed model of corroded RC bridge
pier) it is then possible to model the spatial variability of corrosion
pattern probabilistically using the models developed in [17].
In this section a comprehensive review of the existing analytical
models of reinforcing bars with the effect of buckling and low-
cycle fatigue is conducted. A comparison between the observed
experimental response, detailed FE analysis and the empirical ana-
lytical models is conducted and critical issues are reported.
5.1. Analytical modelling of post-yield buckling behaviour of corroded
bars
In recent decades the nonlinear analysis of RC framed structures
subject to seismic loading has received a lot of attention. This has
been focused on the development of the fibre element technique
[25,26]. In this approach the member cross section is decomposed
into a number of steel and concrete fibres at selected integration
122 M.M. Kashani et al. / Engineering Structures 75 (2014) 113–125

Fig. 15. Comparison of the simulated and experimental results of a corroded bar under cyclic loading: (a) L/D = 10 with 15.48% mass loss, (b) fractured bar at the end of the
cyclic test and (c) variation of the cross section area along the bar.

points. The material nonlinearity is represented through a uniaxial

constitutive material model of steel (tension and compression) and
η = σ σy (1,1) (η 2 ,ξ 2
*
)
concrete (confined core concrete and unconfined cover concrete).
Therefore a lot of researchers around the world put efforts to
(η2 , ξ2 )
develop accurate uniaxial constitutive models for reinforcing steel
and concrete to improve the accuracy of response prediction mod-
els. Moreover, buckling of vertical reinforcing bars is one of the
most common observed collapse mechanisms of RC structures in
the past earthquakes which ultimately results in crushing of core 0.2
concrete in RC columns. Therefore, in the recent years several
researchers have put efforts to develop analytical models to cap- ξ = ε εy
ture the buckling behaviour of vertical reinforcing bars within
the RC elements [30–34]. However, most of the previous analytical Fig. 16. The proposed Dhakal and Maekawa buckling model.

models ignore the interaction of the vertical reinforcement and

where ry is the yield strength of reinforcement in MPa.
horizontal tie reinforcement. Dhakal and Maekawa [21] developed
Given non-dimensional stress g = r/ry and strain n = e/ey
a new analytical model to model the post-yield buckling behaviour
(where ry is the yield stress and ey is the yield strain) the following
of reinforcing bars with different L/D ratios. The analytical model
equations define the stress–strain envelope shown in Fig. 16.
has a base curve considering the L/D is equal to the ratio of spacing
8
of horizontal ties to diameter of the vertical reinforcement in RC > n; n61
>
> ðg2 1Þ
columns. The interaction of horizontal tie reinforcement and verti- < ðn  1Þ þ 1; 1 < n 6 n2
ðn2 1Þ
cal reinforcement is then captured by replacing the L/D ratio of the g¼ ð10Þ
>
> g  0:02; ðn  n2 Þ; n2 6 n&g P 0:2
basic curve to the actual effective buckling length considering the >
: 2
stiffness of horizontal ties. The detail of buckling length calculation 0:2; otherwise
is reported by Dhakal and Maekawa in [35].
where, the empirical relationships for (g2, n2) are given below:
In this research the Dhakal and Maekawa buckling model
(Fig. 16) is used for modelling the post-yield buckling behaviour n2 ¼ 55  2:3kp ; n2 P 7 ð11Þ
of corroded reinforcing bars.
In this model, the post-yield buckling response of reinforcement g2 ¼ að1:1  0:016kp Þg2 ; g2 P 0:2 ð12Þ
is defined as a function of a compound variable called the non-

dimensional slenderness ratio kp as defined in Eq. (9): where g 2 is the non-dimensional piecewise stress corresponding to the
rffiffiffiffiffiffiffiffiffi n2. The value of a is a softening coefficient and depends on the strain
ry L hardening of reinforcement. Dhakal and Maekawa found that for elas-
kp ¼ ð9Þ
100 D tic-perfectly plastic reinforcement a = 0.75 and for reinforcement with
 M.M. Kashani et al. / Engineering Structures 75 (2014) 113–125 123

linear hardening a = 1.0. The detailed description of the model is avail-

able in [15,21].
To model the post-yield buckling behaviour of corroded bars,
Kashani et al. [15] proposed a methodology to modify the ry and
kp using the experimental data. The detail of the model calibration
is available in [15]. In this paper the results of the calibrated model
based on the experimental data is compared with the results of
computational models and the corresponding experimental
results. It should be noted that experimental data is the result of
the physical testing of the set of corroded bars which were scanned
and used in the computational modelling and were not included in
the calibration of the buckling model. Fig. 16 shows an example
comparison of the corrosion extended Dhakal–Maekawa buckling
model with experimental and computational results. It is evident
from Fig. 17 that the proposed analytical model has a better fit
Fig. 18. Comparison of the experimental response and the proposed analytical
to the experimental results compare to computational results. As
model.
it was discussed before this is due to the influence of cross section
shape of pitted sections. and the second moment of area are included in a so called pitting
Fig. 18 shows a comparison of the analytical model and the coefficient to modify the ry and kp as a function of mass loss ratio.
experimental results of a corroded bar with L/D = 15 and 30.32% This method is much simpler and is very easy to be implemented
mass loss. Despite that the experimental data in Fig. 18 were not in finite element programme for nonlinear analysis of the RC struc-
included in the model calibration; the analytical model can still tures considering reinforcing bar buckling.
predict response accurately.
Due to the complexity of the problem, random nature of the pit- 5.2. Analytical modelling of cyclic behaviour of corroded bars including
ting corrosion and limitation in the computational models the buckling and low-cycle fatigue
influence of cross section shape may not be captured through the
fibre model analysis. The only way to model this complex phenom- There are currently a number of analytical models available in
enon is a 3D solid element finite element model which is compu- the literature for cyclic stress–strain response of uncorroded rein-
tationally very expensive. However, in the analytical model the forcing steel with and without buckling [30–34,36–38]. The effect
influence of cross section loss is considered assuming an average of corrosion on the behaviour of uncorroded and corroded reinforc-
reduced cross section. Then, the influence of cross section shape ing bars subject to low-cycle fatigue loading has also been studied

Fig. 17. Comparison of the computational and experimental response with the proposed analytical model: (a) L/D = 20 and 34.91% mass loss, (b) L/D = 15 and 19.93% mass
loss, (c) L/D = 10 and 14.44% mass loss.
124 M.M. Kashani et al. / Engineering Structures 75 (2014) 113–125
Fig. 19. Comparison of the corrosion extended Kunnath et al. model and experimentally observed response of corroded bars: (a) a corroded bar with L/D = 5 and 10.36% mass
loss and (b) a corroded bar with L/D = 10 and 21% mass loss.
by other researchers [39–41]. Kashani et al. [16] developed a meth-
odology that accounts for the influence of corrosion on combined
effect of low-cycle fatigue degradation and inelastic buckling of
corroded reinforcing bars subject to cyclic loading.
Kashani et al. used the experimental data reported by Apostolop-
oulos and Papadopoulos [41] for model calibration. To model the
cyclic behaviour of corroded bars the reinforcing steel model devel-
oped by Kunnath et al. [34] which is already available in the Open-
Sees have been used. This model consists of a low-cycle fatigue
degradation model in tension and Dhakal–Maekawa buckling
model in compression. It also includes the GMP cyclic rules to
model the Bauschinger effect. The combined effect of low-cycle fati-
gue degradation and the corrosion extended Dhakal–Maekawa
buckling model is incorporated in and simulated in OpenSees. Fur-
ther details and model calibration is available in [16]. A qualitative
comparison between the experimental and analytical cyclic mean
stress–strain response of two corroded bars with different L/D and
mass loss ratios is shown in Fig. 19. As it is shown in Fig. 19(a) the
corrosion extended Kunnath et al. model can accurately predict
the response of a corroded bar with L/D = 5 where buckling of bars
does not have a significant influence on degradation of the hyster-
etic cycles. However, as the slenderness ratio of bars increases the
accuracy of corrosion extended Kunnath et al. model reduces.
Fig. 19(b) is an example comparison of this model with the observed
experimental response of a corroded bar with L/D = 10.
As discussed before the pinching effect in the cyclic stress–
strain curve of reinforcing bars is due to the inelastic buckling
and geometrical nonlinearity. In other words as the slenderness
ratio of the reinforcing bars increases the hysteretic response
changes towards a nonlinear beam–column element type response
(e.g. concentric steel bracing). This has also been reported by other
researchers [23,42] and their conclusions are in a good agreement
with the results of this study. This phenomenon is more critical in
corroded bars where non-uniform pitting corrosion can result in a
significant change to the effective slenderness ratio of reinforcing
bars and subsequently degradation of hysteretic response.
With reference to the Fig. 19, it is evident that there is need for
further analytical research to develop a new uniaxial material
model for reinforcing bars that accounts for the influence of inelas-
tic buckling on degradation of the hysteretic cycles.
6. Conclusion
The influence of corrosion pattern on inelastic buckling and
cyclic behaviour of corroded bars has been investigated computa-
tionally. The key parameters affecting the inelastic buckling and
hysteresis behaviour of corroded bars have been identified. A
critical review of the existing analytical material models and a
comparison with the results of computational modelling and phys-
ical testing has been conducted. The key outcomes from this
research can be summarised as follows:
1. The nonlinear fibre beam–column element can accurately sim-
ulate the nonlinear behaviour of reinforcing bars subject to
monotonic and cyclic loading. It should be noted that the basic
steel02 model is not able to capture the low-cycle fatigue degra-
dation of the reinforcing bars. Apart from this the qualitative fit
of the numerical and experimental models are very good. This
method can be used as a computational tool in the future
research to generate further data for development and calibra-
tion of new analytical models
2. It was observed in both experimental and computational
responses that the cyclic degradation of compressive strength
of reinforcing bars with L/D > 8 is a function of the previous
strain history in tension. This conclusion agrees with the results
obtained by others [42,43].
3. As the slenderness ratio of bars increased the contribution of
combined cross section area and second moment of area to
the overall buckling load reduction increased. Whereas by
reducing the slenderness ratio the contribution of the cross sec-
tional shape increased.
4. It was found that the load eccentricity caused by nonuniform
pitting corrosion has a significant contribution on reduction of
the buckling load of corroded bars.
5. It was observed that the cyclic response of corroded bars with
severe localised pitting is mainly governed by the premature
yielding of the minimum pitted section.
6. The corrosion extended Dhakal–Maekawa buckling model can
accurately predict the post-yield buckling behaviour of corroded
bars under monotonic compression. However, the pinching
effect and cyclic degradation of the compressive strength of
reinforcing bars with L/D > 8 due to the influence of geometrical
nonlinearity is not included in the existing analytical models.
Therefore, there is a need for further analytical study and devel-
opment of a new phenomenological hysteretic model to account
for these effects. The computational platform explained in this
paper can be used by other researchers for model calibration.
Acknowledgements
The experimental work is funded by Earthquake Engineering
Research Centre (EERC) of the University of Bristol. The computa-
tional part of this research is conducted in collaboration with the
 M.M. Kashani et al. / Engineering Structures 75 (2014) 113–125
University of Washington while the first author was on sabbatical
leave in the US. The funding provided by the World Wide Univer-
sity Network through Research Mobility Programme to the first
author is much appreciated. Any findings, opinions and recommen-
dations provided in this paper are only based on the author’s view.
References
[1] Wallbank EJ. The performance of concrete in bridges: a survey of 200 highway
bridges, London; 1989.
[2] The economic impact of current investment trends in surface transportation
infrastructure. Economic Development Research Group Inc.: ASCE; 2011.
[3] Broomfield JP. Corrosion of steel in concrete: understanding, investigation and
repair. 2nd ed. Taylor and Francis; 2007.
[4] Rafiq MM, Chryssanthopoulos MK, Onoufriou T. Performance updating of
concrete bridges using proactive health monitoring methods. Reliab Eng Syst
Saf 2004;86:247–56.
[5] Vu KAT, Stewart MG. Structural reliability of concrete bridges including
improved chloride-induced corrosion models. Struct Saf 2000;22:313–33.
[6] Leman DE, Moehle JP. Seismic performance of well-confined concrete bridge
columns. Pacific Earthquake Engineering Research Centre; 2000.
[7] Ou Y, Tsai L, Chen H. Cyclic performance of large-scale corroded reinforced
concrete beams. Earthq Eng Struct D 2011;41:592–603.
[8] Ma Y, Che Y, Gong J. Behavior of corrosion damaged circular reinforced
concrete columns under cyclic loading. Constr Build Mater 2012;29:548–56.
[9] Almusallam AA. Effect of degree of corrosion on the properties of reinforcing
steel bars. Constr Build Mater 2001;15:361–8.
[10] Du YG, Clark LA, Chan AHC. Residual capacity of corroded reinforcing bars. Mag
Concr Res 2005;57(3):135–47.
[11] Du YG, Clark LA, Chan AHC. Effect of corrosion on ductility of reinforcing bars.
Mag Concr Res 2005;57(7):407–19.
[12] Cairns J, Plizzari GA, Du YG, Law DW, Chiara F. Mechanical properties of
corrosion-damaged reinforcement. ACI Mater J 2005;102(4):256–64.
[13] Apostolopoulos CA, Papadopoulos MP, Pantelakis SG. Tensile behavior of
corroded reinforcing steel bars BSt 500s. Constr Build Mater 2006;20:782–9.
[14] Apostolopoulos CA. Mechanical behavior of corroded reinforcing steel bars
S500s tempcore under low cycle fatigue. Constr Build Mater
2007;21:1447–56.
[15] Kashani MM, Crewe AJ, Alexander NA. Nonlinear stress–strain behaviour of
corrosion-damaged reinforcing bars including inelastic buckling. Eng Struct
2013;48:417–29.
[16] Kashani MM, Crewe AJ, Alexander NA. Nonlinear cyclic response of corrosion-
damaged reinforcing bar with the effect of buckling. Constr Build Mater
2013;41:388–400.
[17] Kashani MM, Crewe AJ, Alexander NA. Use of a 3D optical measurement
technique for stochastic corrosion pattern analysis of reinforcing bars
subjected to accelerated corrosion. Corros Sci 2013;73:208–21.
[18] ASTM G1-03 standard practice for preparing, cleaning, and evaluating
corrosion test specimens. ASTM Int’l; 2011.
[19] MTLAB R2012b. The MathWorks lnc.; 1994–2012. <www.mathworks.com>.
[20] Mau ST, El-Mabsout M. Inelastic buckling of reinforcing bars. J Eng Mech
1989;115(1):1–17.
125
[21] Dhakal RP, Maekawa K. Modeling for postyield buckling of reinforcement. J
Struct Eng 2002;128(9):1139–47.
[22] Dhakal RP, Maekawa K. Path-dependent cyclic stress–strain relationship of
reinforcing bar including buckling. Eng Struct 2002;24:1139–47.
[23] Nakamura H, Higai T. Modeling nonlinear cyclic behavior of reinforcing bars.
ACI SP 205-14; 2002, vol. 205. p. 273–92.
[24] OpenSees, the open system for earthquake engineering simulation. Pacific
Earthquake Engineering Research Centre, University of California, Berkeley;
2011.
[25] Spacone E, Filippou FC, Taucer FF. Fibre beam–column model for non-linear
analysis of R/C frames. Part I: Formulation. Earthq Eng Struct D
1996;25:711–25.
[26] Spacone E, Filippou FC, Taucer FF. Fibre beam–column model for non-linear
analysis of R/C frames. Part II: Applications. Earthq Eng Struct D
1996;25:727–42.
[27] Neuenhofer A, Filippou FC. Geometrically nonlinear flexibility-based frame
finite element. J Struct Eng 1998;124(6):704–11.
[28] Souza RM. Force-based finite element for large displacement inelastic analysis
of frames. PhD thesis, University of California, Berkeley; 2000.
[29] Menegotto M, Pinto PE. Method of analysis of cyclically loaded RC plane
frames including changes in geometry and nonelastic behavior of elements
under normal force and bending. Preliminary report IABSE, vol. 13, Zurich;
1973. p. 15–22.
[30] Monti G, Nuti C. Nonlinear cyclic behavior of reinforcing bars including
buckling. J Struct Eng 1992;118(12):3268–84.
[31] Rodriguez ME, Botero JC, Villa J. Cyclic stress–strain behavior of reinforcing
steel including the effect of buckling. J Struct Eng 1999;125(6):605–12.
[32] Gomes A, Appleton J. Nonlinear cyclic stress–strain relationship of reinforcing
bars including buckling. Eng Struct 1997;19:822–6.
[33] Bae S, Mieses A, Bayrak O. Inelastic buckling of reinforcing bars. J Struct Eng
2005;131(2):314–21.
[34] Kunnath SK, Heo Y, Mohle JF. Nonlinear uniaxial material model for reinforcing
steel bars. J Struct Eng 2009;135(4):335–43.
[35] Dhakal RP, Maekawa K. Reinforcement stability and fracture of cover concrete
in RC members. J Struct Eng 2002;128(10):1253–62.
[36] Dodd LL, Restrepo-Posada JI. Model for predicting cyclic behavior of
reinforcing steel. J Struct Eng 1995;121(3):433–45.
[37] Balan TA, Filippou FC, Popov EP. Hysteretic model of ordinary and high-
strength reinforcing steel. J Struct Eng 1998;124(3):288–97.
[38] Hoehler MS, Stanton JF. Simple phenomenological model for reinforcing steel
under arbitrary load. J Struct Eng 2006;132(7):1061–9.
[39] Chang GA, Mander JB. Seismic energy based fatigue damage analysis of bridge
columns: Part I – Evaluation of seismic capacity. Technical report NCEER-94-
0006; 1994.
[40] Higai T, Nakamura H, Saito S. Fatigue failure criterion for deformed bars
subjected to large deformation reversals, vol. 237. ACI SP 237-4; 2006. p. 37–
54.
[41] Apostolopoulos CA, Papadopoulos MP. Tensile and low cycle fatigue behavior
of corroded reinforcing steel bars S400. Constr Build Mater 2007;21:855–64.
[42] Prota A, Cicco F, Cosenza E. Cyclic behavior of smooth steel reinforcing bars:
experimental analysis and modeling issues. J Earthq Eng 2009;13:500–19.
[43] Moyer MJ, Kowalsky MJ. Influence of tension strain on buckling of
reinforcement in concrete columns. ACI Struct J 2003;100(1):75–85.