Engineering Structures 191 (2019) 460–468

Contents lists available at ScienceDirect

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

Structural performance of deteriorating reinforced concrete columns under T

multiple earthquake events
Zhen Cui, Alice Alipour , Behrouz Shafei
⁎

Department of Civil, Construction and Environmental Engineering, Iowa State University, Ames, IA 50011, United States

ARTICLE INFO ABSTRACT

Keywords: This study investigates deteriorating reinforced concrete (RC) columns subjected to multiple earthquake events,
Reinforced concrete columns taking into consideration how a combination of environmental stressors and extreme loads can adversely affect
Multiple earthquake events the performance of RC columns during their expected service life. For this purpose, nonlinear finite-element
Corrosion-induced deterioration models are developed with all the necessary details, validated with a set of experimental test results, and then
Multi-hazard analysis
employed to predict the extent of structural degradation as a function of age and prior earthquake-induced
Residual capacity assessment
damage. This is through a systematic approach to capture the consequences of corrosion-induced deterioration,
in terms of rebar cross section loss, drop of steel yield and ultimate strength, and degradation of bond. The
consequences of earthquake events are evaluated with not only structural damage caused by lateral loads, but
also the corrosion process expedited as a result of residual damage. Such a holistic assessment strategy paves the
way to obtain realistic predictions for the performance of RC columns under future earthquake events. To study
this critical aspect, a set of representative scenarios that consist of two earthquake events of different intensities
are defined. Considering that RC columns in service commonly experience more than one earthquake during
their lifetime, the outcome of this study contributes to transforming the accuracy of predictions made on the
residual capacity and seismic response of structures subjected to multiple stressors and hazards.

1. Introduction corroded structures. Such studies can be categorized to those with ex-
perimental tests (e.g., [34,29,24]) and those with analytical investiga-
Reinforced concrete (RC) structures are commonly exposed to var- tions (e.g., [2,3,34,5–7,46]). While the findings from the cited studies
ious environmental stressors and extreme events. As a result of dete- have shed light on the residual capacity of corroded structures under
rioration due to environmental stressors, such as airborne sea salts, earthquake events, they are primarily focused on one individual
structural capacity continuously degrades. This is in addition to the earthquake event, disregarding the consequences of earthquake-in-
consequences of extreme events, which cause immediate damage often duced damage on the progress of corrosion and the performance of
in the form of concrete spalling and cracks. Depending on their loca- further corroded structures under future earthquakes.
tions, RC structures can undergo multiple extreme events. In seismic- To address this gap, the current study aims to investigate the seismic
prone regions, which are the primary focus of the current study, each performance of corroded RC columns subjected to more than one
structure witnesses several earthquakes of different intensities during earthquake event. This is achieved through developing a set of simu-
its service life. This creates a situation, in which damage induced by the lation scenarios that include a chain of initial corrosion, seismic ana-
past earthquakes expedites the continuous corrosion process. Such an lysis under the first earthquake, expedited corrosion due to damage
adverse effect not only reduces the remaining structural capacity below from the first earthquake, and seismic analysis under the second
what was predicted from single-stressor/hazard analyses, but also raises earthquake of different intensity. The time gap between the first and
concerns about the increased vulnerability of the structure to sub- second earthquake is included to capture the progress of corrosion in
sequent earthquake events. the damaged structure prior to the second earthquake. In this in-
Extensive research has been conducted on corrosion induced by vestigation, corrosion is assumed to be uniform, which is a common
chloride ions (e.g., [17,40,39,31,38,36,25]), and effects of corrosion on assumption to perform large-scale structural analyses. To capture the
the performance of structures (e.g., [35,18,19]). On the other hand, change in the extent of corrosion along the height of a column, a
there have been a number of studies on the seismic behavior of probabilistic analysis becomes necessary to evaluate pitting corrosion.

⁎
Corresponding author.
E-mail address: [email protected] (A. Alipour).

https://doi.org/10.1016/j.engstruct.2019.04.073
Received 22 December 2017; Received in revised form 18 July 2018; Accepted 23 April 2019
0141-0296/ © 2019 Elsevier Ltd. All rights reserved.
Z. Cui, et al. Engineering Structures 191 (2019) 460–468

This, however, requires an effort beyond the scope of the current study. steel per unit length of a rebar for a time step after corrosion initiation,
To perform the planned simulations, three-dimensional, nonlinear fi- Δt (sec), is calculated as:
nite-element (FE) models are developed and validated with the ex- m
perimental test results. The FE models capture the main corrosion ef- Mst (t ) = i corr dr (t ) t = 2.894 × 10 4i corr d r (t ) t
zF (2)
fects and update the required material properties and structural details
as a function of age. This provides a proper computational platform to where is the remaining rebar diameter (cm) as a function of time.
dr (t )
predict the extent of damage in the deteriorating RC columns under The density of steel is taken as 7.8 g/cm3. Thus, the consumed volume
consideration. Furthermore, this platform is capable of utilizing the of steel per unit length can be predicted as:
residual damage due to an initial earthquake and continuous corrosion Mst (t )
Ast (t) = = 3.709 × 10 5i corr d r (t ) t
for improving the structural performance assessment of RC columns st (3)
under a subsequent earthquake. This is a key feature that makes a
holistic vulnerability assessment possible. The remaining diameter of the rebar after each time step can be
determined as:
2. Time-dependent effects of corrosion on material and structural 4Ast (t )
dr (t ) = d2
properties (4)

Corrosion of steel reinforcement is one of the main causes of dete- Once the corrosion rate is decided, Eqs. (3) and (4) predict the steel
rioration of RC structures [16,42,11]. In the coastal structures that are mass loss due to corrosion at any analysis time step.
exposed to airborne sea salts and in the structures that experience harsh The residual yield and ultimate strength of corroded reinforcement
winter conditions and multiple applications of deicing salts, chloride- can be estimated using the following empirical formulas that are re-
induced corrosion is the dominant deterioration mechanism. Through ported by Cairns et al. [16]:
this mechanism, chloride ions gradually penetrate into concrete to- f yr = (1 0.015 0
100 ) f y (5)
wards embedded reinforcement. When the chloride concentration at
the steel surface reaches a critical level, corrosion initiates, steel is f ur = (1 0.015 0
100 ) f u (6)
consumed, and rust is produced at the interface of steel and concrete.
Since the volume of the rust product is higher than that of the con- where f yr and f ur are the residual yield and ultimate strength of corroded
sumed steel, an internal pressure is created on the surrounding con- reinforcement, respectively. f y0 and f u0 are the yield and ultimate
crete, causing the formation of cracks and spalling of the concrete cover strength of original reinforcement, respectively, and η100 is the per-
(Fig. 1). centage of steel mass loss over time.
The extent of structural steel mass loss, Mst , due to chloride-induced The ultimate strain of corroded reinforcement is calculated as fol-
corrosion can be calculated using the Faraday’s Law: lows:
r 0
mIt su = (1 0.039 100 ) su (7)
Mst (t ) =
zF (1)
where andr
su are the ultimate strain of corroded and original re-
0
su
where m is the atomic mass of iron (56 g for Fe), F is the Faraday’s inforcement, respectively. The provided equations are consistent with
constant, i.e., 96,500 A·sec, z is the ionic charge (for example, 2 for those from the experimental tests performed by Kashani et al. [26].
Fe → Fe2+ + 2e−), t is the time after corrosion initiation (sec), and I is Fig. 2(a) illustrates the constitutive models of pristine and corroded
the current (A), which can be calculated from I = asicorr, where as is the steel considered for the current study.
surface area of the steel bar (cm2) and icorr is the current density (A/ The corrosion effects on concrete can be captured by considering the
cm2). Noting that for a unit length of a rebar, as = πd, the mass loss of formation of cracks and degradation of compressive strength in the

Fig. 1. Schematic illustration of chloride-induced corrosion.

461
Z. Cui, et al. Engineering Structures 191 (2019) 460–468

Fig. 2. Constitutive models of (a) steel and (b) concrete in pristine and degraded conditions.

concrete cover, as well as a brittle post-peak behavior in the compres- conducted on the correlation between the corrosion-induced crack
sive constitutive model of concrete [18]. The reduced compressive width and the degree of corrosion, which is often expressed in terms of
strength of concrete, f cr , can be calculated using the following equa- steel mass loss per unit length or corrosion penetration (i.e., rebar ra-
tions: dius reduction) [44,43,10,47,45]. It has been reported that the average
crack width due to corrosion increases with the steel section loss. An-
f c' b1
f cr = and a = 1, b1 = b0 + nw drade et al. [10] proposed an equation for predicting the corrosion-
Ra
1+ b0 induced crack width in the concrete cover based on the experimental
c (8)
tests that replicated natural exposure conditions:
where c is the concrete strain at the compressive strength of is the fc' , a
average tensile strain in the cracked concrete perpendicular to the di- x b (t )
w (t ) = k
rection of the applied compression, R is a coefficient related to the r (11)
roughness and diameter of reinforcement, b0 is the original width of the where w (t ) is the time-dependent crack width, r is the original radius of
cross section, b1 is the increased width due to corrosion cracks, n is the the reinforcement, and k is a factor, the value of which is derived from
total number of reinforcement in each layer, and w is the total crack experimental results (35.5 for the current study). According to Dur-
width, which can be calculated as follows [30]: aCrete [22], the corrosion penetration, xb, at time t after corrosion in-
w= wi = 2 x b ( 1) itiation (in years) can be calculated as:
i (9) xb (t ) = 0.0115i corr t (12)
where wi is the width of the i-th crack, xb is the corrosion penetration, 2
where icorr is the corrosion rate (µA/cm ). Substituting Eq. (12) into Eq.
and λ is the volumetric ratio, which varies depending on the compo- (11), the time-dependent crack width is obtained.
sition of corrosion products. Fig. 2(b) shows the constitutive models of Further to corrosion effects, occurrence of extreme events, such as
pristine and degraded concrete. earthquakes, can induce cracks in the concrete cover. Noting that such
Experimental studies show that the reinforcement corrosion has an cracks result in an increase in the corrosion rate, Otieno et al. [32,33]
adverse effect on the bond strength (fib 2000). Many studies have in- proposed a predictive model that takes into account the effect of total
vestigated the residual bond strength variation with respect to the ex- crack width (due to corrosion and earthquake), wtot , on the corrosion
tent of corrosion (e.g., [35,18,42]). In this study, the bond deterioration rate:
model provided by Val and Chernin [42] has been employed:
0.96Dc 0.35
x c
(1 + K1 ) x b , xb x cr i corr = 5.18e 0.01Dc for specimens in the laboratory
r
b,max = 0
b,max
cr wtot
max[K1 K2 (xb x cr ), 0.15] xb > x cr (10) (13)
where r
and
b,max are the residual and initial bond strength (MPa),
0
b,max 0.21e 0.02Dc
c
xcr is the corrosion penetration corresponding to crack initiation in the i corr = 0.64e 0.06Dc for specimens in the field
concrete cover, and K1 is the initial increase of the bond strength after
wtot (14)
corrosion initiation, which depends on the confinement level provided where Dc is the chloride diffusion coefficient, and c is the concrete cover
by the concrete cover and stirrups. The former is characterized by the thickness.
ratio of concrete cover to reinforcing bar diameter, while the latter is Implementing such detailed calculations is deemed necessary be-
represented by the ratio of total area of stirrups within the development cause although the cracks caused by extreme events can remain un-
length to that of longitudinal reinforcement enclosed by the stirrups. K2 noticed for an extended period of time, their existence results in ex-
is the rate of bond strength degradation after crack formation. Further pedited corrosion. There are requirements to make a decision on
details of the provided equations can be found in Al-Sulaimani et al. whether a crack is working or non-working, which leads to using sealers
[12], Cabrera [15], Mangat and Elgarf [28], and Auyeung et al. [14]. or fillers, respectively. Such repair actions are, however, initiated only
if the crack width exceeds 3.0–6.0 mm (1/8–1/4 in.). It is widely ac-
3. Crack width and correlation with corrosion rate cepted that a crack width of 0.3–0.4 mm can represent the durability
limit state, while a crack width of 0.8–1.0 mm reflects the serviceability
Corrosion of reinforcement leads to steel section loss and rust pro- limit state [8,13,37,41,4]. For the structural elements of RC bridges,
duction, which can initiate cracks in the concrete cover. The widening such as decks, slabs, beams, and columns, a crack width of up to 1.6 mm
and propagation of cracks in the concrete cover facilitate the intrusion shows the condition state 1 (i.e., good condition); a crack width be-
of chloride ions into concrete. Several research projects have been tween 1.6 and 3.2 mm reflects the condition state 2 (i.e., fair condition);

462
Z. Cui, et al. Engineering Structures 191 (2019) 460–468

Fig. 3. (a) Geometry of the RC column (dimensions are in mm), (b) details of the FE model generated for validation, and (c) comparison between the experimental
test and FE simulation results.

and a crack width beyond 3.2 mm represents the condition state 3 (i.e., Eight-node solid elements are used to model the concrete and two-node
poor condition) [9]. truss elements are employed to include the steel reinforcement. The bond-
slip interaction between concrete and steel is modeled with nonlinear spring
4. Development and validation of finite-element models elements at the interface of solid and truss elements [20,21]. The bilinear
kinematic hardening model is selected for the steel reinforcement. To model
Prior to performing the main numerical simulations planned for the the concrete, the damage plasticity model is employed to properly capture
current study, the FE models developed for RC columns are validated with the number and location of cracked and/or crushed elements, as well as the
the laboratory test data reported by Ma et al. [27]. In the experimental tests, extent/location of tensile and compressive damage. This model also takes
the circular RC columns that had experienced a range of corrosion were into account the degradation of elastic stiffness induced by plastic straining
tested under a constant axial load combined with cyclic lateral loads. Each in both tension and compression [1]. The reduced elastic modulus, E r , is
tested column was 260 mm in diameter and 1000 mm in height with a expressed in terms of a degradation variable, dd:
1300 mm × 360 mm × 400 mm foundation system (Fig. 3a). The clear E r = (1 dd ) E 0 (15)
concrete cover for the spiral reinforcement was 30 mm. The longitudinal
reinforcement in both column and foundation was 16 mm in diameter. The where E 0 is the initial modulus of concrete and dd is the damage variable (dt
spirals were 8 mm in diameter with a spacing of 100 mm. The reinforce- under tension and dc under compression). Upon completing a full com-
ment used in the foundation was 8 mm in diameter as well. The 28-day parison, the FE simulation results are found in a close agreement with the
compressive strength of concrete, fc’, was reported to be 32.4 MPa. The yield data obtained from the experimental tests (Fig. 3c). More detailed in-
and ultimate strength of longitudinal reinforcement were 373.2 and formation on the validation effort can be found in Cui and Alipour [20].
572.3 MPa, respectively. The yield and ultimate strength of transverse re- With the assumptions and modeling details developed based on the
inforcement in the column and foundation were 327.0 and 510.7 MPa, re- validation effort, full-scale RC column models are investigated in the
spectively. The column was subjected to an axial load calculated with nfc’A, current study. The column details are extracted from a RC bridge lo-
where A is the cross section of the column and n varies between 0.15 and cated in the Los Angeles area, which satisfies the current seismic design
0.90. The FE model generated for validation purposes is shown in Fig. 3(b). requirements [2]. The column models are 10.0 m in height and 1.6 m in

463
Z. Cui, et al. Engineering Structures 191 (2019) 460–468

Fig. 5. Results of pushover analysis for the RC columns at 10-year time inter-
vals.

To evaluate the seismic performance of deteriorating RC columns

under multiple earthquake events, a set of representative scenarios are
defined. In each scenario, the column experiences two earthquakes with
a given time gap. From the first earthquake, the location, type, and
extent of damage in the column are recorded. The residual damage,
Fig. 4. Side view of (a) full-scale RC column, and (b) embedded steel rebars.
including plastic strain and deformation, as well as concrete cracking
and spalling, are particularly stored to update the rate of corrosion,
diameter with a concrete cover of 50 mm. They are supported with a capturing the fact that the formation and propagation of cracks ex-
3.0 m × 3.0 m × 2.0 m foundation. The compressive strength of con- pedite the intrusion of chloride ions into concrete. The progress of
crete is 32.4 MPa. The diameter of longitudinal reinforcement is 62 mm. corrosion is then estimated in a recursive manner as the column ages.
The stirrups are 19 mm in diameter with a spacing of 100 mm. Square This is used as an initial condition for the RC column when it is sub-
hoops used in both directions of the foundation are 57 mm in diameter jected to the second earthquake. The damage after the second earth-
with a spacing of 150 mm. The yield and ultimate strength of steel re- quake is recorded and compared with the damage after the first
inforcement are 475 MPa and 655 MPa, respectively. The mass of the earthquake. Fig. 6 outlines the procedure for this analysis.
superstructure is simplified as a lumped mass on top of the column, There are three different approaches for the seismic analysis of struc-
including P-delta effects. While the foundation is fixed, the top of the tures: Equivalent Static Analysis (ESA), Elastic Dynamic Analysis (EDA), and
column is allowed to move in the horizontal direction to capture the Nonlinear Time History Analysis (NTHA). The ESA and EDA are appropriate
lateral loading effects. Fig. 4 shows the details of the full-scale RC for estimating the displacement demand on conventional bridges and their
column models used for the current study. individual components. The ESA can be used for short- to medium-span
bridges that have regular configurations. For long-span bridges or where the
5. Exposure to continuous corrosion and earthquake events ESA is unable to evaluate the dynamic behavior, the EDA can be used as an
alternative. The NTHA is the most detailed but time-consuming analysis
To capture corrosion-induced structural degradation, the column’s approach. It is best suited for bridges with the Seismic Design Category of D
material and structural properties, such as cross sectional area, yield or seismically isolated bridges with long periods or large damping ratios.
and ultimate strength, and bond-slip relationship, are updated in the FE Since the RC columns considered for the current study belong to a con-
model as a function of time following the equations provided in Section ventional bridge that represents a range of short- to medium-span bridges in
2. Table 1 reports the steel mass loss for the service life of the RC the United States, the ESA is found satisfactory to investigate the seismic
columns under consideration. Although the corrosion rate is significant performance of the intact and corroded RC column. This analysis approach
at early age, it starts to decrease over time, approaching a point, beyond also significantly reduces the computational cost, which makes it practical
which the corrosion rate remains almost constant. Considering that the
current study evaluates the performance of RC columns in 10-year time
intervals, a constant corrosion rate of 5.0 μA/cm2 is assumed. This has
been selected after a review of corrosion rates in similar exposure
conditions [21]. With this input, the main properties of the column
models are updated at regular time intervals. To quantify how the
structural performance of the RC columns degrades over time, a series
of pushover analyses are conducted at the intact condition and after 10,
20, 30, 40, 50, 60, and 70 years (Fig. 5). As the extent of corrosion
increases with age, both capacity and stiffness of the column are found
to drop. The base shear capacity reduces by 7.9–31.5% from 10 to
70 years, respectively.

Table 1
Mass loss of steel rebars in the RC columns at 10-year time intervals.
Time (year) 0 10 20 30 40 50 60 70
Fig. 6. Overview of two-earthquake simulation scenarios considered in this
Steel mass loss 0 6.0% 9.6% 12.6% 15.3% 17.6% 19.9% 21.9%
study.

464
Z. Cui, et al. Engineering Structures 191 (2019) 460–468

been symbolized as Y0 and Y70, respectively. Three scenarios are discussed

in this section to illustrate how the developed FE framework investigates the
lifetime performance of RC columns under two earthquake events while
they are continuously subjected to chloride-induced corrosion with a cor-
rosion rate of icorr,1 and icorr,2 prior to and after the occurrence of the first
earthquake, respectively. The three scenarios are 0Y30Y, 20Y50Y, and
40Y70Y. In Scenario 20Y50Y, for example, the RC column experiences
corrosion for 20 years from the time of construction with a corrosion rate of
5.0 μA/cm2 (icorr,1). The corrosion-induced crack width (w1) in the column
after 20 years is predicted to be 0.7 mm. Upon the occurrence of the first
earthquake, the crack width (w2) increases to 1.6 mm. This magnifies the
corrosion rate to 7.3 μA/cm2 (icorr,2) until the second earthquake occurs
30 years later when the column is 50-year old. To make direct comparisons
possible, the time gap between the first and second earthquake event has
maintained constant (i.e., 30 years) in this study. It has also been assumed
that the columns first experience a less severe earthquake (i.e., 10% PE in
50 years) and then a more severe earthquake (i.e., 2% PE in 50 years). Such
a sequence results in capturing the effect of minor damage, which is defined
as small cracking or spalling of concrete, requiring no more than cosmetic
repair [23]. This is deemed an important contribution, as the former studies
Fig. 7. Calculation procedure to estimate the equivalent static load.
have disregarded how the presence of residual damage can affect the cor-
rosion process and eventually the vulnerability of the entire structural
to perform various simulation scenarios. The ESA utilizes a lateral load to system to extreme events.
replicate the effect of an earthquake event. In this study, the equivalent
static load is calculated following the steps provided in Fig. 7 and then 6. Damage prediction and vulnerability assessment
applied to the top of the column for different hazard levels.
The earthquake hazard levels considered for this study are 2% and 10% The extent of damage to the RC columns is investigated under the
probability of exceedance (PE) in 50 years, which are equivalent to an an- three scenarios introduced in the previous section. Based on the out-
nual probability of occurrence of 4.0 × 10−4 and 2.1 × 10−3, respectively. come of the ESA, Fig. 8 summarizes the percentage of damage in con-
The column’s age considered for the ESA spans from 0 to 70 years. This has crete elements and the number of steel rebars yielded, as two important

Fig. 8. Extent of damage to the RC columns under the scenarios with (a) variable and (b) constant corrosion rates.

465
Z. Cui, et al. Engineering Structures 191 (2019) 460–468

Fig. 9. Comparison of steel mass loss percentage for the variable and constant corrosion rates: (a) Scenario 20Y50Y, and (b) Scenario 40Y70Y.

structural vulnerability measures. The concrete damage includes de- the natural frequency of the FE column models is evaluated prior to and
gradation in both tension (dt) and compression (dc). The percentage of after each earthquake for both variable and constant corrosion rates. It
damage reported in Fig. 8 is the ratio of the number of damaged con- can be seen in Fig. 10 that the dynamic characteristics of the intact and
crete elements to the total number of concrete elements in the RC corroded RC columns before experiencing the first earthquake are very
column. The steel damage is expressed in terms of number of elements close to each other. The natural frequencies, however, significantly
that experience a plastic strain. To further elaborate the importance of drop after the occurrence of each of the first and second earthquakes.
taking into consideration the effects of damage due to both corrosion This is more pronounced under the variable corrosion rate than the
and extreme events, a separate set of simulations are performed with constant corrosion rate (4.7% more for the Scenario Y20Y50 and 10.7%
the assumption that the corrosion rate is not updated after the first more for the Scenario Y40Y70). This highlights the importance of
earthquake. The predicted results can be found in Fig. 8(a) and (b) for a considering the history of extreme events that a structure experiences to
variable and constant corrosion rate, respectively. The bottom part of obtain realistic predictions for future risk and vulnerability assessment
each bar shows the damage after the first earthquake. This helps dis- purposes.
tinguish the additional damage induced by the second earthquake.
It can be noted in Fig. 8 that under both variable and constant 7. Conclusions
corrosion rates, the total damage witnesses an increase from the Sce-
nario 0Y30Y to 20Y50Y and from the Scenario 20Y50Y to 40Y70Y, This study investigated the lifecycle structural performance of de-
despite the fact that the time gap between the first and second earth- teriorating RC columns located in seismic-prone regions. The effects of
quakes remains the same for all the three scenarios. This indicates how chloride-induced corrosion, in terms of rebar cross section loss, drop of
the progress of corrosion makes the columns more vulnerable to ex- steel yield and ultimate strength, and degradation of bond, were iden-
treme events over time. After the first earthquake, the percentage of tified and incorporated into a nonlinear FE computational framework.
concrete damage for all the three scenarios is relatively low, ranging The extent of structural degradation was obtained as a function of age
from 14.5% to 35.1% for the tensile damage and 10.3% to 32.3% for based on the models that provided a realistic estimate of the progress of
the compressive damage, while no yielding occurs in the column’s steel corrosion. Upon validation of the FE models with experimental test
rebars. After the second earthquake, however, both concrete and steel results, a set of simulation scenarios were defined to understand how
damage significantly increase. This is where the difference between the residual damage can affect the response of the RC columns to
variable and constant corrosion rates can be clearly observed. With the subsequent extreme events. The main conclusions and findings are as
variable corrosion rate, the total percentage of tensile damage in con- follows:
crete is found 2.2% and 70.6% higher than that with the constant
corrosion rate for the Scenarios 20Y50Y and 40Y70Y, respectively. The • With pushover analyses, both strength and stiffness of the FE
total percentage of steel damage after two earthquakes under the con- column models were evaluated at different ages. The base shear
stant corrosion rate is 1.1% and 3.4% for the Scenarios 20Y50Y and capacity was found to drop to two-thirds of the initial capacity after
40Y70Y, respectively. The same percentage increases to 2.7% and 70 years of continuous corrosion.
17.6% under the variable corrosion rate for the Scenarios 20Y50Y and • The presence of residual damage due to the combination of corro-
40Y70Y, respectively. It is important to mention that for the Scenario sion and earthquake events was captured by performing a sequence
40Y70Y, both concrete and steel damage dramatically increased after of analyses, first with a less severe earthquake (i.e., 10% PE in
the second earthquake under the variable corrosion rate, leading to a 50 years) and then with a more severe earthquake (i.e., 2% PE in
structural failure. However, under the constant corrosion rate, which 50 years). Such a sequence determined how even minor damage can
neglects the contribution of residual damage to expediting the corrosion influence the vulnerability of the RC columns to extreme events.
process, the column only experiences extensive damage. This can • Assuming a 30-year time gap between the first and second earth-
greatly influence the subsequent decisions made for repair or replace- quakes, the corrosion-induced crack width in the columns was
ment of the column. predicted for two time intervals, i.e., prior to and after the first
The steel rebar’s mass loss percentage under both variable and earthquake. Upon the occurrence of the first earthquake, the cor-
constant corrosion rates are shown in Fig. 9 for the Scenarios 20Y50Y rosion rate was magnified by a factor of 1.4–1.6. This was supported
and 40Y70Y. It can be seen in this figure that considering the increase based on the correlation between the corrosion rate and crack
of corrosion rate due to the damage induced by the first earthquake width.
magnifies the mass loss by 4.0% after 30 years. Fig. 9 also shows that • The effect of corrosion rate on the predicted extent of damage was
the increase in the corrosion rate slows down as the mass loss percen- studied by considering both variable and constant corrosion rates.
tage increases. This was anticipated because of the formation and For all the types of damage measured for concrete and steel, the RC
growth of a passive layer of rust. Further to various damage measures, columns were more vulnerable under a variable corrosion rate.

466
Z. Cui, et al. Engineering Structures 191 (2019) 460–468

Fig. 10. Change of the natural frequency in the RC columns prior to and after the first and second earthquake events.

Noting that the variable corrosion rate reflects realistic exposure [18] Coronelli D, Gambarova P. Structural assessment of corroded reinforced concrete
conditions, the predictions made with a constant corrosion rate can beams: modeling guidelines. J Struct Eng 2004;130(8):1214–24.
[19] Cui Z, Alipour A. A detailed finite-element approach for performance assessment of
underestimate the expected damage during future extreme events.
•
corroded reinforced concrete beams. Proceedings, ASCE Structures Congress,
Based on the cumulative damage recorded after the two earthquakes Boston, Massachusetts. 2014.
(in the presence of continuous corrosion), it was concluded that a [20] Cui Z, Alipour A. Performance assessment of corroded highway bridge piers under
cyclic loading. Proceedings, Transportation Research Board (TRB) 95th Annual
reliable estimate of risk and vulnerability cannot be achieved Meeting, Washington, D.C.. 2016.
without taking into account the history of the environmental stres- [21] Cui Z, Alipour A. Concrete cover cracking and service life prediction of reinforced
sors and extreme events that a structure experiences during its ser- concrete structures in corrosive environments. Constr Build Mater
2018;159:652–71.
vice life. [22] DuraCrete. Probabilistic performance based durability design of concrete structures.
The European Union-Brite EuRam III. 2000.
References [23] FEMA. HAZUS-MH MR4 technical manual, Washington, DC. 2012.
[24] Guo A, Li H, Ba X, Guan X, Li H. Experimental investigation on the cyclic perfor-
mance of reinforced concrete piers with chloride-induced corrosion in marine en-
[1] ABAQUS 6.12 Documentation. Dassault Systèmes Simulia Corp.; 2012. vironment. J Eng Struct 2015;105:1–11.
[2] Alipour A. Life-cycle performance assessment of highway bridges under multi-ha- [25] Hu N, Burgueño R, Haider S, Sun Y. Framework for estimating bridge-deck chloride-
zard conditions and environmental stressors Ph.D. Dissertation. Irvine: Department induced degradation from local modeling to global asset assessment. J Bridge Eng
of Civil and Environmental Engineering, University of California; 2010. 2016;21(9):1–7.
[3] Alipour A, Shafei B, Shinozuka M. Performance evaluation of deteriorating highway [26] Kashani MM, Crewe AJ, Alexander NA. Nonlinear stress–strain behaviour of cor-
bridges located in high seismic areas. J Bridge Eng 2011;16(5):597–611. rosion-damaged reinforcing bars including inelastic buckling. J Eng Struct
[4] Alipour A, Shafei B, Shinozuka M. Capacity loss evaluation of reinforced concrete 2013;48:417–29.
bridges located in extreme chloride-laden environments. J Struct Infrastruct Eng [27] Ma Y, Che Y, Gong JX. Behavior of corrosion damaged circular reinforced concrete
2013;9(1):8–27. columns under cyclic loading. J Construct Build Mater 2012;29:548–56.
[5] Alipour A, Shafei B. Structural vulnerability measures for assessment of deterior- [28] Mangat PS, Elgarf MS. Bond characteristics of corroding reinforcement in concrete
ating bridges in seismic prone areas. In: Frangopol Dan, Tsompanakis Yiannis, beams. Mater Struct 1999;32(216):89–97.
editors. Maintenance and safety of aging infrastructure. CRC Press; 2014 [29] Meda A, Mostosi S, Rinaldi Z, Riva P. Experimental evaluation of the corrosion
[Chapter 3]. influence on the cyclic behaviour of RC columns. J Eng Struct 2014;76:112–23.
[6] Alipour A, Shafei B. Assessment of postearthquake losses in a network of aging [30] Molina FJ, Alonso C, Andrade C. Cover cracking as a function of rebar corrosion:
bridges. J Infrastruct Syst 2016;22(2):1–12. Part 2 – Numerical model. J Mater Struct 1993;26:532–48.
[7] Alipour A, Shafei B. Seismic resilience of transportation networks with deteriorating [31] O’Connor A, Kenshel O. Experimental evaluation of the scale of fluctuation for
components. J Struct Eng 2016;142(8):1–12. spatial variability modeling of chloride-Induced reinforced concrete corrosion. J
[8] American Concrete Institute (ACI). 562–16 code requirements for assessment, re- Bridge Eng 2013;18(1):3–14.
pair, and rehabilitation of existing concrete structures and commentary. 2016. [32] Otieno M, Beushausen H, Alexander M. Chloride-induced corrosion of steel in
[9] American Association of State Highway and Transportation Officials. AASHTO cracked concrete – Part I: Experimental studies under accelerated and natural
guide specifications for LRFD seismic bridge design; 2011. marine environments. J Cem Concr Res 2016;79:373–85.
[10] Andrade C, Muñoz A, Torres-Acosta A. Relation between crack width and corrosion [33] Otieno M, Beushausen H, Alexander M. Chloride-induced corrosion of steel in
degree in corroding elements exposed to the natural atmosphere. Proceedings, cracked concrete—Part II: Corrosion rate prediction models. J Cem Concr Res
fracture mechanics of concrete and concrete structures Seoul, Korea. 2010. 2016;79:386–94.
[11] Apostolopoulos CA, Demis S, Papadakis VG. Chloride-induced corrosion of steel [34] Ou Y, Tsai L, Chen H. Cyclic performance of large-scale corroded reinforced con-
reinforcement – mechanical performance and pit depth analysis. J Constr Build crete beams. J Earthq Eng Struct Dyn 2012;41:593–604.
Mater 2013;38:139–46. [35] Rodriguez J, Ortega L, Casal J. Load carrying capacity of concrete structures with
[12] Al-Sulaimani GJ, Kaleemullah M, Basunbul IA, Rasheeduzzafar. Influence of cor- corroded reinforcement. J Constr Build Mater 1997;11(4):239–48.
rosion and cracking on bond behavior and strength of reinforced concrete members. [36] Shafei B, Alipour A. Application of large-scale non-Gaussian stochastic fields for the
ACI Struct J 1990;87(2):220–31. study of corrosion-induced structural deterioration. J Eng Struct 2015;88:262–76.
[13] Andrade C, Alonso CM, Molina FJ. Cover cracking as a function of bar corrosion: [37] Sakai K, Shimomura T, Sugiyama T. Design of concrete structures in the 21st cen-
Part I – Experimental test. Mater Struct 1993;26(8):453–64. tury. Proceedings of international conference on controlling concrete degradation,
[14] Auyeung Y, Balaguru P, Chung L. Bond behavior of corroded reinforcement bars. Dundee, Scotland. 1999.
ACI Mater J 2000;97(2):214–20. [38] Shafei B, Alipour A. A stochastic computational framework to investigate the initial
[15] Cabrera JG. Deterioration of concrete due to reinforcement steel corrosion. Cem stage of corrosion in reinforced concrete superstructures. Comput – Aided Civ
Concr Compos 1996;18(1):47–59. Infrastruct Eng 2013;28(7):482–94.
[16] Cairns J, Plizzari GA, Du Y, Law DW, Franzoni C. Mechanical properties of corro- [39] Shafei B, Alipour A, Shinozuka M. Prediction of corrosion initiation in reinforced
sion-damaged reinforcement. ACI Mater J 2005;102(4):256–64. concrete members subjected to environmental stressors: a finite-element frame-
[17] Cheung M, Zhao J, Chan Y. Service life prediction of RC bridge structures exposed work. J Cem Concr Res 2012;42(2):365–76.
to chloride environments. J Bridge Eng 2009;14(3):164–78. [40] Shafei B. Stochastic finite-element analysis of reinforced concrete structures

467
Z. Cui, et al. Engineering Structures 191 (2019) 460–468

subjected to multiple environmental stressors. PhD dissertation. Irvine: Department reinforced concrete. J Cem Concr Res 2004;34(1):165–74.
of Civil and Environmental Engineering, University of California; 2011. [45] Xia J, Jin W. Prediction of corrosion-induced crack width of corroded reinforced
[41] Stewart Mark G, Val Dimitri V. Multiple limit states and expected failure costs for concrete structures. Proceedings, 4th International Conference on the Durability of
deteriorating reinforced concrete bridges. J Bridge Eng 2003;8(6):405–15. Concrete Structures, West Lafayette, IN. 2014.
[42] Val D, Chernin L. Serviceability reliability of reinforced concrete beams with cor- [46] Yalciner H, Sensoy S, Eren O. Seismic performance assessment of a corroded 50-
roded reinforcement. J Struct Eng 2009;135(8):896–905. year-old reinforce concrete building. J Struct Eng 2015;27(6):683–96.
[43] Val D, Chernin L, Stewart M. Experimental and numerical investigation of corro- [47] Zhang R, Castel A, François R. Concrete cover cracking with reinforcement corro-
sion-induced cover cracking in reinforced concrete structures. J Struct Eng sion of RC beam during chloride-induced corrosion process. J Cem Concr Res
2009;135(4):376–85. 2010;40(3):415–25.
[44] Vidal T, Castel A, François R. Analyzing crack width to predict corrosion in

468