Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

Structure and Infrastructure Engineering

Maintenance, Management, Life-Cycle Design and Performance

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/nsie20

Probabilistic seismic loss estimation of aging


highway bridges subjected to multiple earthquake
events

Bhaskar Panchireddi & Jayadipta Ghosh

To cite this article: Bhaskar Panchireddi & Jayadipta Ghosh (2021) Probabilistic seismic loss
estimation of aging highway bridges subjected to multiple earthquake events, Structure and
Infrastructure Engineering, 17:9, 1155-1174, DOI: 10.1080/15732479.2020.1801765

To link to this article: https://doi.org/10.1080/15732479.2020.1801765

Published online: 17 Aug 2020.

Submit your article to this journal

Article views: 220

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=nsie20
STRUCTURE AND INFRASTRUCTURE ENGINEERING
2021, VOL. 17, NO. 9, 1155–1174
https://doi.org/10.1080/15732479.2020.1801765

Probabilistic seismic loss estimation of aging highway bridges


subjected to multiple earthquake events
Bhaskar Panchireddi and Jayadipta Ghosh
Civil Engineering, Indian Institute of Technology Bombay, Mumbai, India

ABSTRACT ARTICLE HISTORY


This study investigates the seismic life-cycle cost of aging highway bridges under chloride exposure Received 1 November 2019
when subjected to multiple main shock earthquake events along the design lifetime of the structure. Revised 6 March 2020
While chloride induced corrosion manifests as a continuous deterioration process, earthquake occur- Accepted 5 April 2020
rences are typically intermittent. Although relatively weak seismic events that induce none-to-minor
KEYWORDS
bridge damage may not instigate immediate intervention and repair, resulting structural cracks may Aging highway bridges;
further accelerate the corrosion process. This study presents a novel Monte Carlo based methodology corrosion deterioration;
that explicitly considers potentially altered deterioration pattern between seismic shocks as well as Monte Carlo simulations;
uncertainty in earthquake occurrences, record-to-record variability, repair decisions and repair proc- multiple earthquakes;
esses for seismic life-cycle cost computations. The proposed framework also eliminates the need for seismic life-cycle cost;
the often improbable assumption in past literature on seismic life-cycle cost assessment that necessi- uncertainty analysis
tates structural rehabilitation to pre-hazard states regardless of the seismic damage level. Case-study
application of the proposed framework is demonstrated on a single column integral bridge located in
the marine splash zone in California, United States. Results from multiple main shock analysis of both
non-deteriorating and aging case-study bridge reveals a significant contribution of cumulative damage
from corrosion deterioration and shock sequences towards seismic loss and life-cycle cost assessment.

1. Introduction As a consequence of increased structural vulnerability


due to corrosion deterioration, structural damage to aging
Transportation infrastructure comprising of highway bridges
highway bridges during earthquake events typically leads to
are critical for multiple services and benefits to the economy
increased direct losses compared to non-deteriorating struc-
and society of a nation. Typically, a decline in structural
tures in the form of repair and replacement costs, or indir-
performance of these lifeline systems are expected along
ect losses incurred by communities (longer travel time, and
their design lifetime due to natural hazard occurrences or excess fuel consumption, among others) due to the unavail-
under the action of various environmental stressors such as ability of bridges. While direct losses are primarily borne by
corrosion, erosion, and chemical deterioration (Akiyama, bridge owners or stakeholders, indirect losses are incurred
Frangopol, & Matsuzaki, 2011; Alipour, Shafei, & by regional communities through longer travel time, excess
Shinozuka, 2011a; Ghosh & Padgett, 2011; Akiyama & fuel consumption, among other secondary effects. This
Frangopol, 2014; Shekhar, Ghosh, & Padgett, 2018; paper focuses only on direct losses estimation and provides
Panchireddi & Ghosh, 2019; Yanweerasak, Pansuk, novel contributions to research through a framework for
Akiyama, & Frangopol, 2018; Cui, Alipour, & Shafei, 2019). realistic seismic life-cycle cost (SLCC) assessment of aging
While continued environmental actions and intermittent bridge structures. The present study does not include costs
natural hazard occurrences (such as earthquakes) are statis- stemming from initial construction, non-seismic repairs,
tically independent, the structural performance of degrading maintenance cost, among others, as well as the consequence
bridge components during such hazard events are typically of such repairs on the seismic performance of critical bridge
compromised when compared to pristine conditions. For components. Extensive treatise on traditional life-cycle cost
instance, under continued corrosion deterioration, that assessment can be found in Hawk (2003) and Gervasio and
accounts for a primary form of environmental degradation, da Silva (2013). The proposed framework considers uncer-
researchers over the past decade have underlined the tainty in earthquake occurrences along the design life, mul-
increase in fragility of aging highway bridges when located tiple repair decisions and rehabilitation strategies, potential
in moderate to high seismic zones (Choe, Gardoni, alterations in deterioration rates due to earthquake induced
Rosowsky, & Haukaas, 2008, 2009; Ghosh & Padgett, 2010; cracking, and is devoid of assumptions which are typically
Zhong, Gardoni, & Rosowsky, 2012; Panchireddi & resorted to in past literature for lifetime seismic loss assess-
Ghosh, 2019). ment. Some of these assumptions are discussed below.

CONTACT Jayadipta Ghosh jghosh@iitb.ac.in Civil Engineering, Indian Institute of Technology Bombay, Mumbai 400076, India
ß 2020 Informa UK Limited, trading as Taylor & Francis Group
1156 B. PANCHIREDDI AND J. GHOSH

A prevalent, yet unrealistic premise in past literature on Repair measures towards localized earthquake-induced crack
seismic life-cycle cost of non-deteriorating or aging highway regions (typically occurring near plastic hinges) may be
bridges relates to immediate restoration of the damaged oblivious to the corrosion deterioration already underway
structures to the pre-hazard state to calculate the expected elsewhere within undamaged column sections. Consequently,
annual losses due to earthquake damage (Lee, Cho, Lim, & following such localized repair actions, the bridge column
Park, 2004; Mackie, Wong, & Stojadinovic, 2010; Kumar, may now consist of dissimilar structural sections that
Gardoni, & Sanchez-Silva, 2009; Padgett, Dennemann, & behaves differently than a pristine or a thoroughly corroding
Ghosh, 2010; Alipour et al., 2011a; Ghosh & Padgett, 2011; column during a future earthquake.
Kameshwar & Padgett, 2017; Yanweerasak et al., 2018; Additionally, as discussed earlier, for a corroding bridge
Shekhar et al., 2018). These losses, after appropriate dis- column under a minor earthquake that induces relatively
counting, when aggregated along the design lifetime help inconsequential visible damage and consequently a no-repair
assess the seismic life-cycle cost. Repairing a bridge struc- decision, the deterioration pattern may accelerate due to
ture to pre-hazard conditions necessitates the assumption of further chloride ingress through earthquake-induced minor
thorough repair intervention following any extent of earth- crack formations (Scott & Alexander, 2007; Otieno,
quake-induced damage. This assumption is often improb- Alexander, & Beushausen, 2010). While such effects have
able, especially for bridge structures located in moderate to been traditionally ignored in literature, only recently Cui
high seismic zones that may experience multiple not-so- et al. (2019) have investigated the detrimental effects of
severe earthquakes along the design life. expedited corrosion due to crack formations on the seismic
Occasionally, a relatively weaker earthquake that induces performance of deteriorating columns under two-shock
none-to-minor cracks may not warrant a substantial repair sequences. While based on equivalent static analysis to sim-
to the pre-hazard state, although the bridge may be ren- plify seismic effects, the above study underlines the critical-
dered marginally weaker to resist future earthquakes ity of considering potential alterations of deterioration
(FEMA, 2012; Cui et al., 2019). The above assumption also pattern within cracked column sections. From a seismic life-
loses fundamentality with respect to seismic life-cycle cost cycle cost perspective, such practical circumstances stem-
assessment of aging bridge structures using time-dependent ming from uncertainty in repair process, selection of global
fragility curves, as often done in past literature (Padgett (complete replacement) or local intervention, alteration of
et al., 2010). For these structures, following an earthquake deterioration pattern and subsequent bridge behavior under
damage, a pre-hazard condition at year t is indicative of a future earthquakes need explicit consideration.
bridge repaired to its corroded state at year t þ 1, which is Addressing existing drawbacks and augmenting past
contrary to the actual field conditions. Additionally, corre- research on the consideration of multiple earthquake occur-
sponding to highway bridges (pristine or aging) with visible rences and aging for lifetime performance prediction
damage following an earthquake that necessitates interven- (Kumar & Gardoni, 2012; Faisal, Majid, & Hatzigeorgiou,
tion, multiple repair strategies may exist. The adoption of 2013; Di Sarno, 2013; Ghosh, Padgett, & Sanchez-Silva,
one repair strategy over another depends on the discretion 2015; Panchireddi & Ghosh, 2019) this study proposes a
of the engineer, availability of resources and other factors. novel framework for seismic life-cycle cost assessment of
Assumption of deterministic repair cost ratios for different aging highway bridges. Key aspects of this framework
damage states as often done in past literature is therefore include: i) consideration of intermittent earthquake (main
unrealistic and does not reflect the uncertainty in repair shock events) and continuous aging hazard along the bridge
decision, repair process, as well as the associated repair costs design life, ii) uncertainty in repair decisions and associated
(Basoz & Mander, 1999; Padgett et al., 2010; Alipour et al., costs depending on bridge damage level, iii) potential alter-
2011a). While few recent studies have considered the uncer- ation in corrosion deterioration pattern following seismic
tainty in repair process (Ghosh & Padgett, 2011; Kameshwar damage and crack width geometry, and iv) nonlinear time-
& Padgett, 2017; Shekhar et al., 2018), a critical drawback history analysis of finite element models with sequential
still exists as elaborated next. model updating to reflect dissimilar section characteristics
Notwithstanding replacement of a bridge component or along the pier geometry stemming from localized repair,
bridge system in entirety following seismic damage, bridge section replacement, and aging.
repair measures are often local in nature. For instance, as The next section of the paper elaborates possible repair
Padgett and DesRoches (2007) report from a survey of strategies and associated costs based on different damage
bridge engineers and field practitioners, damage to bridge levels distinguished as per their physical appearances after
piers following earthquake events may instigate repair strat- each ongoing main shock earthquake event. The cost
egies such as localized patching with concrete to seal cracks, corresponding to each repair strategy is also discussed.
epoxy grouting, lining with concrete, among others. While Time-dependent corrosion deterioration modeling and the
for non-deteriorating bridge columns such repair measures alterations of corrosion process due to the presence of
for earthquake-induced damage may reinstate the structural cracks (from earthquake-induced damage) are presented
capacity to the original pristine (as-built) condition, the subsequently. The proposed methodology is then demon-
same may not be true for corroding bridge columns when strated on a representative case-study two-span box-girder
located near sources of chloride exposure, for instance near semi-integral bridge located in the marine-splash zone in
marine environment or regions of deicing salt application. California within the United States. The probable losses
STRUCTURE AND INFRASTRUCTURE ENGINEERING 1157

corresponding to single and multiple shock scenarios using lm Eh


D¼ þb (1)
a Monte Carlo sampling approach for both non-deteriorat- lu My h y l u
ing and aging bridge conditions are outlined. The computed
In the above equation, lm is the maximum ductility demand
losses, dependent on the number of seismic main shocks,
caused by an earthquake event (Chai et al., 1994; Ghosh
are then utilized to compute the seismic life-cycle cost. The
et al., 2015) or under a sequence of events (Hatzigeorgiou &
paper ends with key conclusions and recommendations for
future explorations. Liolios, 2010; Huang et al., 2012; Ghosh et al., 2015;
Panchireddi & Ghosh, 2019), lu is the ultimate ductility
capacity under static monotonic loading, Eh is the total hys-
2. Damage quantification, classification, and teretic energy dissipated by the structure, My is the yield
associated repair decisions moment capacity, hy is the yield rotation angle and b is a
dimensionless constant usually assumed to be 0.05 for RC
Repair strategies and associated repair costs, that contrib-
structures. Additionally, following n earthquake events, the
ute to lifetime seismic losses of a civil engineering struc-
damage index Dn is a strictly increasing quantity with
ture are usually the consequence of in-situ decisions from
Dn >Dn1 >:::>D2 >D1 representing the cumulative nature of
visual inspections by bridge engineers. Consequently, past
seismic damage under shock sequences. Following a certain
several decades have witnessed several seismic damage
number of seismic main shocks, while a Park and Ang dam-
quantification models to bridge the gap between a quanti-
age index estimate near to zero indicates negligible damage
tative measure of damage (usually obtained from structural
and when more than unity represents collapse, intermediate
analysis) and corresponding physical descriptions that
ranges and corresponding physical appearance and damage
instigate a repair decision upon visual inspection. Such
levels, as typically adopted in literature, are included in
damage models typically depend on the structural response
Table 1 (Bruno & Valente, 2002; Hatzigeorgiou & Liolios,
parameters that may include the measurement of drifts or
2010; Ghosh et al., 2015; Panchireddi & Ghosh, 2019).
displacements or curvatures, among other similar measures
While prognosis on the physical appearance based on
(Park & Ang, 1985; Wang & Shah, 1987; Roufaiel &
damage index computations from computer simulations
Meyer, 1987; Mehanny & Deierlein, 2001; Colombo &
Negro, 2005; Jiang, Chen, & Chen, 2011). This section of may help bridge engineers to forecast a probable repair
the paper focuses on identifying an adequate damage meas- strategy to compute repair costs and associated lifetime seis-
ure, classifying different levels of structural damage based mic losses, such efforts are best restricted for pristine or as-
on physical descriptors, and elucidating repair decisions built structures with no prior history of residual damage.
and probable repair strategies following earthquake- Earthquake-induced damage to bridge columns invariably
induced disruptions. results in cracking of cover concrete and when located near
marine sources and left unrepaired, may lead to accelerated
ingress of chlorides (elaborated later) with potential altera-
2.1. Damage measurement and correlation with tions to the corrosion rates (Scott & Alexander, 2007;
physical descriptors Otieno et al., 2010) in aging highway bridges. Hence, in
addition to the qualitative descriptors on the physical
For reinforced concrete (RC) columns within buildings or appearance, information on expected crack width from
highway bridge structures that may experience single or seismic damage aids in predicting the altered rate of
multiple earthquake occurrences along the service life, the deterioration.
Park and Ang damage index (Park & Ang, 1985) has been This information is adopted from Nakano, Maeda,
widely adopted for quantitative prediction of structural Kuramoto, and Murakami (2004) based on a survey for
damage (Chai, Priestley, & Seible, 1994; Kunnath & Jenne, post-earthquake damage evaluation and rehabilitation of RC
1994; Williams & Sexsmith, 1997; Hatzigeorgiou & Liolios, buildings after the 1995 Hyogoken-Nambu (Kobe) earth-
2010; Huang, Qian, Zhuang, & Fu, 2012; Ghosh, Padgett, & quake in Japan that correlates damage levels similar to those
Sanchez-Silva, 2015; Panchireddi & Ghosh, 2019). indicated in Park, Ang, and Wen (1985) with crack widths
Combining information on the maximum ductility demand in RC columns as indicated in Table 1. For seismically
and dissipated hysteretic energy under shock sequences as designed ductile bridge columns, as is the focus of this
demonstrated in Equation 1, the Park and Ang damage study, such cracks are typically expected to occur in the
index D is also found to correlate well with physical damage plastic hinge regions towards column ends with progressive
observations: reduction in damage at distances further away from the

Table 1. Physical appearance and corresponding crack width descriptions of various damage levels classified based on different ranges of damage index (D).
Damage level Damage index range (D) Physical appearance Crack width description (wcr)†
Slight D < 0.1 Localized minor cracking Visible narrow cracks (wcr < 0.2 mm)
Minor 0.1  D < 0.25 Light cracking throughout Visible clear cracks (0.2 < wcr < 1 mm)
Moderate 0.25  D < 0.40 Severe cracking; localized spalling Remarkable wide cracks (1.0 < wcr < 2.0 mm)
Severe 0.40  D < 1 Crushing of concrete; reinforcement exposed Remarkable wide cracks (wcr > 2.0 mm)
Collapse D  1.0 Loss of element load resistance Structural failure
Park et al. (1985)

Nakano, Maeda, Kuramoto, and Murakami (2004)
1158 B. PANCHIREDDI AND J. GHOSH

hinge locations along the length of the column. These infer- towards column mid-height. Consequently, the plastic hinge
ences and assumptions included in Table 1 will be corrobo- regions are assumed to undergo restoration to as-built con-
rated further through a case-study section later in the paper. ditions while relatively inconsiderable repairs (such as
With damage levels and corresponding physical descriptions patching, grouting, among others) are resorted to along the
based on visual observations and crack width information mid-height depending on the damage level (Ma & Xiao,
now available, the next section briefly elaborates on deci- 1999; Cho, Kim, Feo, & Hui, 2012). For columns under
sions regarding repair or restoration of seismically damaged severe damage state that experience crushing of concrete
bridge columns and associated repair strategies. and reinforcement exposure following seismic damage, the
damage index encompasses a wide range (0.4  D < 1.0) that
may introduce substantial uncertainty in repair decision and
2.2. Repair decisions and associated restoration
subsequent seismic loss estimation. For practicality, this
strategies
range is further divided to two sub classes as: i) Class A:
Seismic damage to highway bridge components instigates 0.4  D < 0.7, and ii) Class B: 0.7  D < 1.0.
necessary repair actions, wherein the extent or type of While the repair strategies for Class A of the severe dam-
repair depends on a multitude of uncertain factors. Such age state is assumed to follow an approach similar to that
factors include the discretion of the field engineers on the for moderate damage, rehabilitation decisions for Class B
repair method, importance of the bridge, availability of with damage range 0.7  D < 1.0 depend on the residual
funds, and post-earthquake availability of resources for res- drift of the structure. According to Mangalathu, Jeon, and
toration, among others. Often, a slight to minor damage Jiang (2019), residual drift (RD) typically exceeding 1.75% is
following seismic events does not require or instigate a indicative of structural failure that needs complete replace-
repair action, particularly when monetary resources are ment to as-built conditions, while RD less than 1.75% indi-
limited (FEMA, 2012; Cui et al., 2019). Such damages, that cates possible structural rehabilitation using conventional
go unrepaired, nevertheless render a bridge more suscep- repair techniques (discussed later). Lastly, structures experi-
tible to the ill-effects of corrosion deterioration and may encing collapse damage level with D  1.0 after an earth-
lead to further reduction in structural capacity to resist quake undergoes replacement to as-built condition.
future earthquakes. Additionally, unlike past assumptions The repair strategies, and the probabilities of adopting
in seismic life-cycle cost assessment, a chosen repair strat- one repair method over another, closely follow the work by
egy may not restore a bridge to pre-hazard state in terms Padgett and DesRoches (2007) and are presented in Table 2.
of functionality and structural capacity. Incorporating these The information contained within this table stems from a
effects, the flowchart outlined in Figure 1 presents the survey of field practitioners and bridge engineers and have
assumptions on repair decisions based on the level of been used in the past for seismic life-cycle cost assessment
damage following the nth seismic event at time tn along the of highway bridge structures (Ghosh & Padgett, 2011;
design life of the bridge. Kameshwar & Padgett, 2017; Shekhar & Ghosh, 2020). For a
Structures experiencing different damage levels such as particular damage level, the wide ranging available repair
slight, minor, moderate, severe and collapse after each earth- options and their associated probabilities of implementation
quake event are treated separately based on their respective (Table 2) are representative of the uncertainties stemming
qualitative and quantitative descriptions. For the structures from the discretion of the bridge owners or field engineers
experiencing a slight damage level that have no visible and the availability of material or economic resources. The
cracks, no repair actions would be needed and hence a “Do Monte Carlo simulation based framework for seismic life-
Nothing” strategy is adopted. Also, earthquake events caus- cycle cost assessment presented in this paper explicitly
ing minor damage in structures that result in barely percep- accounts for such uncertainties, unlike past literature that
tible cracks (wcr < 0.2 mm) are assumed to be left relies on damage state dependent repair cost ratios alone
unrepaired. However, predictive models proposed by (Basoz & Mander, 1999). It is worthwhile to mention that
Otieno, Beushausen, and Alexander (2016a, 2016b) based on sometimes due to lack of resources or other constraints, a
field investigations and laboratory tests note that the pres- bridge may also be “posted” for repair. While this option
ence of cracks may result in an increase in the corrosion should ideally be included in the list of decision strategies, a
rate and hence an expedited corrosion process following preliminary analysis reveals that a delay in rehabilitation by
seismic damage. The effect of cracks (due to structural dam- a few years negligibly affects the seismic life-cycle costs and
age from earthquake events) on the corrosion rate is elabo- is hence excluded from Table 2.
rated in the next section of the paper. Although, this study directly adopts the repair strategy
Qualitative description of the structures experiencing recommendations from Padgett and DesRoches (2007), it is
moderate damage state highlights that the column undergoes worthwhile to note that the cost-effective option repair
localized spalling and severe cracking along the length with selections should ideally depend on the residual design life
remarkable crack widths ranging from 1 mm to 2 mm. and life-cycle cost analysis. Future investigations and survey
Preliminary analysis through analytical modeling reveals data will further help refine the adopted cost models for
that such damage is typically concentrated at the plastic seismic loss assessment. Until now, this paper has elaborated
hinge regions (top and bottom) of ductile columns in inte- on physical appearance based damage levels and the possible
gral bridges with successive reduction in damage levels repair strategies along with their cost estimates. The next
STRUCTURE AND INFRASTRUCTURE ENGINEERING 1159

section of the paper demonstrates the impact of aging and the corrosion deterioration process due to the formation of
deterioration process in RC bridge columns. Alterations in cracks after each ongoing earthquake event is also discussed.

Figure 1. Flowchart outlining the assumptions on repair decisions following the nth seismic event at time tn along the design lifetime of the bridge based on the
qualitative and quantitative descriptions of various damage levels.

Table 2. Probability of adopting various repair strategies based on different damage levels along with their unit cost estimates for the bridge
column (Padgett & DesRoches, 2007; Ghosh & Padgett, 2011).
Probability of adopting different repair actions
Severe
Repair strategy Moderate Class A Class B Collapse Cost estimate/unit
Epoxy injection 5% 5% 0% 0% $191/m
Patching with concrete 23% 23% 0% 0% $56/m2
Grouting 0% 0% 0% 0% $143/m3
Concrete lining 9% 9% 0% 0% $93/column
Wrapping 36% 36% 0% 0% $215/m2
Replace column 23% 23% 72% 0% Based on unit cost of steel, concrete,
form work, column dimensions,
and reinforcement ratios
Reinforce and recast 5% 5% 20% 0% Cost of replacing column plus shoring cost
Demolish and replace bridge 0% 0% 8% 100% Cost of bridge construction
1160 B. PANCHIREDDI AND J. GHOSH

3. Chloride induced deterioration modeling in Following the initiation time, the next subsection elaborates
undamaged and cracked concrete on the propagation phase of corrosion deterioration in rein-
forcing steel.
This section of the paper focuses on the chloride induced cor-
rosion process and the expedited deterioration modeling as a
consequence of earthquake induced cracks. Corrosion deteri- 3.2. Corrosion propagation phase
oration stemming from chloride ion sources has become a
major concern for bridge owners and stakeholders as it may Following the initiation phase, corrosion rate during the
lead to unexpected reduction of useful design life (Liu & propagation phase is critical to determine the extent of deteri-
Weyers, 1998; Choe et al., 2009; Zhong et al., 2012; Rao, oration primarily in the form of steel area loss and other
Lepech, Kiremidjian, & Sun, 2017). The deterioration process effects including reduction in cover and core concrete
initiates with the ingress of chloride ions through the concrete strength. While several researchers have reported a constant
cover that depassivates the reinforcing steel and thereby ini- corrosion rate during the propagation phase (Andrade,
tiates corrosion. Following the initiation phase, the propaga- Alonso, & Molina, 1993; Stewart & Rosowsky, 1998;
tion phase primarily leads to a progressive reduction in the DuraCrete, 2000; Alipour, Shafei, & Shinozuka, 2011b), few
reinforcing steel area, loss of confinement and cover concrete others assume the corrosion rate to be time-dependent
strength, cracking and spalling of the cover, among other sec- (Enright & Frangopol, 1998; Vu & Stewart, 2000; Choe et al.,
ondary effects. This situation may further be exacerbated with 2008). This study adopts the corrosion deterioration model
the presence of cracks (induced due to earthquake loads) in proposed by DuraCrete (2000) with the parametric equation
concrete. The following subsections briefly explains the initi- as a function of factors such as temperature, relative humidity
ation and propagation phases and the expedited corrosion and chloride content. These factors have been underlined by
deterioration process due to the presence of concrete cracks. multiple authors to explicitly affect the corrosion rate during
the propagation phase (El Hassan, Bressolette, Chateauneuf,
& El Tawil, 2010; Stewart, Wang, & Nguyen, 2011).
3.1. Corrosion initiation phase Adoption of this corrosion deterioration model also
When located near marine sources, aggressive chemical spe- maintains consistency with the estimated initiation time pre-
cies such as chlorides often ingress into cover concrete dicted using the model proposed by DuraCrete (2000). The
through a diffusion controlled mechanism (Bertolini, time-dependent steel area reduction during the propagation
Elsener, Pedeferri, & Polder, 2004). Over time, along the phase of corrosion deterioration is computed as (Alipour
design life of the bridge, this diffusive process leads to cor- et al., 2011b):
8
rosion deterioration in reinforced concrete structures. > 2p
>
< nDi 4 for t  Ti
Typically the initiation of this deterioration mechanism
AðtÞ ¼ n½D  ðr  tÞ 2 p
occurs when the threshold limit of the diffused chloride ion > for Ti < t < Ti þ Di =rcorr
>
:
i corr
4
concentration around the surface of the steel reinforcement 0 for t  Ti þ Di =rcorr
is met. To determine the corrosion initiation time (Ti), this
(3)
study adopts the widely popular Fick’s second law of diffu-
sion. In addition to the diffusion coefficient, investigations where, n is the number of reinforcing steel bars with initial
by DuraCrete (2000) augments the basic formulation of Ti diameter Di, t is the elapsed time in years after initiation of
based on Fick’s law with coefficients related to the depth of corrosion and rcorr is the rate of corrosion.
the cover concrete, critical and equilibrium chloride concen-
tration surrounding the steel reinforcement, type of cement,
and length of curing, among others. 3.3. Secondary effects of corrosion deterioration
Equation (2) presents the improved equation based on In addition to the primary effects of corrosion deterioration
calibration to compute Ti as: in the form of steel area reduction, this study also models
(   2 )ð1n1 d Þ well-documented secondary effects such as loss of core and
x2 C C
erf 1 cover concrete strength, and alterations in the mechanical
s cr
Ti ¼ (2)
4ke kc kt D0 ðt0 Þnd Cs properties of reinforcing steel. Reduction in confined core
concrete strength in RC columns predominantly manifests
where, x represents the concrete cover depth, ke is the envir-
as a consequence of corrosion deterioration of transverse
onmental factor, kc is the curing parameter, kt is the correc-
ties. The theoretical stress-strain model developed by
tion factor for tests to determine the reference diffusion
Mander, Priestley, and Park (1988) and later modified by
coefficient estimated from the compliance tests, D0 at t0 ¼
Zhang, Cao, and Fu (2016) to include time-dependence as
28 days, nd is the age exponent considering the densification
shown in Equation 4 is adopted in this study.
of cement paste due to hydration chloride, Cs is the equilib- 0 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
rium chloride concentration at the concrete surface, Ccr is 7:94fl ðtÞ fl ðtÞA
the critical chloride concentration and erf is the Gaussian fcc0 ðtÞ ¼ fc @1:254 þ 2:254 1 þ 2 (4)
fc fc
error function. The probability distributions along with the
uncertainties involved in the above parameters are presented In the above equation, fcc0 ðtÞ is the time-dependent compres-
in DuraCrete (2000) and are reproduced in Table 3. sive strength of the confined core concrete, fc is the
STRUCTURE AND INFRASTRUCTURE ENGINEERING 1161

Table 3. Probabilistic distribution of deterioration parameters corresponding to marine splash environmental condition used to determine corrosion initiation
time adopted from DuraCrete (2000) along with the material properties adopted from Mangalathu et al. (2016).
D0: Reference chloride diffusion coefficient at t0 ¼ 28 days
Condition Distribution Mean (mm2/year) St. Dev. (10-12m2/sec)
w/c ¼ 0.5 Normal 473 43.2
nd : Aging factor
Condition Distribution Mean St. Dev. A† B†
Splash Beta 0.367 0.07 0 1
ke: Environmental correction factor
Condition Distribution Mean St. Dev.
Splash Gamma 0.265 0.16
kc: Curing time correction factor
Condition Distribution Mean St. Dev. A† B†
Curing 7 days Beta 1.49 0.29 1 4
x: Cover thickness
Units Distribution Mean! St. Dev.!
mm Lognormal 38.10 7.62
ACl and eCl: Parameters to compute Cs (Cs ¼ ACl  w/c þ eCl)
Condition Distribution ACl‡ eCl‡
Mean St. Dev. Mean St. Dev.
Splash Normal 7.758 1.360 0 1.105
Ccr: Critical chloride concentration
Condition Distribution w/c ratio Mean St. Dev.
Splash Normal 0.50 0.50 0.10
Cs: Surface chloride concentration
Condition Distribution w/c ratio Mean St. Dev.
Splash Normal 0.50 0.50 0.10
fck: Concrete compressive strength
Units Distribution Mean! St. Dev.!
MPa Lognormal 29.03 3.59
fy: Reinforcing steel strength
Units Distribution Mean St. Dev.
MPa Normal 6.14 0.08
icorr: Corrosion current density (c/wcr is the cover depth/crack width
ratio)
Condition Distribution Mean (lA/cm2) COV
Uncraked concrete Uniform 2.01 0.2
20 < c/wcr < 80 Uniform Estimated using Equation 7 0.2
St. Dev. indicates the Standard Deviation, COV is the coefficient of variation, A† and B† are the shape parameters of the Beta distribution, ACl‡ and eCl‡ are the
regression parameters and error term for calculating the surface chloride concentration Cs, Mean! and St. Dev.! are the mean and standard deviation of the log-
normal distribution.

compressive strength of unconfined concrete, and fl (t) is where, fs(t) is the yield or ultimate strengths of corroded
the effective lateral confining stress distributed over the sur- reinforcing steel bar at time t, fs0 is the yield or ultimate
face of the core concrete. strengths of pristine reinforcing steel bar, Qcorr(t) is percent-
Voluminous rust formation due to corrosion deterior- age area loss of reinforcing steel due to corrosion at time t.
ation around the steel reinforcement leads to formation of While the above equation is widely adopted for the RC
micro-cracks in concrete resulting in the reduction of structures experiencing local pitting corrosion, it is also
unconfined cover concrete strength. These alterations are assumed to be valid for the structures undergoing uniform
captured using the model proposed by Vecchio and Collins corrosion deterioration as indicated in Alipour et al.
(1986) that are later modified by Coronelli and Gambarova (2011a, 2011b).
(2004) and are estimated as: Figures 2(a) and (b) present the time-dependent variation
fc of mean reinforcing steel area and cover concrete strength
fc ðtÞ ¼ (5) along with the 5th and 95th confidence bounds after incor-
1 þ j½e1 ðtÞ=ec0 
porating the associated parameter uncertainties. While the
where, fc(t) is the time-dependent unconfined cover concrete corrosion initiation time for the case-study highway bridge
strength, j is the coefficient related to the diameter of the columns in marine splash exposure (elaborated later) is esti-
steel bar, e1 is the transversal strain at time t and ec0 is the mated to be 10 years, the material properties and corrosion
strain at peak stress in the compression mode. rate estimates are presented earlier in Table 3. Key observa-
Corrosion deterioration also results in the reduction of tions reveal that after an exposure period of 100 years, the
reinforcing steel strength estimated using the experimental mean reinforcing steel area is reduced by 25% while that of
tests performed by Du, Clark, and Chan (2005). The time- the cover concrete strength drops significantly to approxi-
dependent reduction in the reinforcing steel strength as a mately 90% of its pristine value. Following the discussions
function of residual steel area is evaluated as: on the potential secondary effects of corrosion deterioration,
the next section of the paper discusses on likely alteration of
corrosion rate in the presence of structural cracks due to
fs ðtÞ ¼ ½1:0  0:005Qcorr ðtÞfs0 (6)
earthquake events.
1162 B. PANCHIREDDI AND J. GHOSH

Figure 2. Time-dependent variation of (a) reinforcing steel area, and (b) unconfined cover concrete strength along with the lower and upper bounds of uncertainty
representing 5th and 95th confidence bounds.

3.4. Impact of concrete cracks on the corrosion rate unrepaired, earthquake induced cracks would alter the cor-
prediction models rosion deterioration pattern. As elaborated earlier in
Section 2, bridge columns as a whole or in-parts along the
Laboratory tests and field observations by past researchers
height that experience minor damage levels and do not
indicate that formation of structural cracks lead to an
necessitate repair intervention, the crack formations are
increase in the corrosion rate but to an extent depending on
likely to alter the corrosion pattern with expedited deteri-
the concrete quality, resistivity and external repair works
oration rates. Figure 3 demonstrates the future path of
(Raupach, 1996; Scott & Alexander, 2007; Otieno et al.,
structural degradation following a single seismic shock
2010). The rate of corrosion initially increases steadily with
while accounting for the accelerated corrosion deterior-
the crack width and then saturates after the crack width
ation. Figure 3(a) exemplifies a scenario wherein the bridge
reaches a threshold value (Otieno et al., 2016a). Extreme
structure undergoing corrosion deterioration experiences
events such as earthquakes that may induce structural cracks
an earthquake at 30 years of design life that induces a
in RC columns could result in the alteration of corrosion
minor damage level resulting in structural cracks with
rate, especially when located in a chloride-laden environ-
crack widths ranging between 0.2 to 1 mm (Table 1). Also,
ment. Based on the predictive models proposed by Otieno
shown in the figure is the original expected mean path of
et al. (2016a, 2016b), for a given concrete type, the crack
corrosion deterioration and the mean altered path due to
width dependent corrosion current density can be computed
increased corrosion rate along with the uncertainties asso-
as:
ciated with the deterioration process. Corresponding to the
0:036ðwccr Þ 0:002ðwccr Þ
icorr ¼ 1:35e þ 2:69e (7) mean estimate of altered corrosion rate, formation of
minor cracks along the length of the column following an
where, icorr is the corrosion current density in lA/cm2, c is earthquake damage has expedited the rate of corrosion by
cover depth of the concrete in mm and wcr is the crack 42% (0.023 mm/year to 0.033 mm/year). Subsequently, at
width in mm. the end of lifetime of the structure, provided no further
Note that the above equation is valid for cracked con- earthquakes occur and there is a lack of periodic bridge
crete alone with c/wcr ratio ranging between 20 and 100. For maintenance, the rebar area reduces by 55% considering
an uncracked concrete, the corrosion current density icorr is the mean expedited rate as compared to 76% when influ-
estimated using the equation proposed by DuraCrete (2000) ence of cracks are disregarded. Influence of alteration of
as mentioned earlier. The corrosion current density icorr in corrosion rate on other secondary effects such as core and
lA/cm2 can be converted to a more convenient form of the cover concrete strength, and steel strength characteristics,
corrosion rate rcorr in mm/year as: among others can be computed accordingly.
Unlike the previous example, Figure 3(b) on the other
rcorr ¼ 0:0116icorr (8) hand represents a bridge structure that experiences the col-
The computed rcorr can now be used to estimate the time- lapse damage level at 65 years of design life due to seismic
dependent area loss of steel using Equation 3 as elaborated damage and then subsequently replaced to as-built condi-
earlier in Section 3.2. It is worthwhile to note that, while tions within a year. This instance does not require consid-
the above Equations 7 and 8 provide an estimate of the eration of altered corrosion rate since the repair strategy
mean, the corrosion rate is assumed to follow a uniform (complete replacement) alone restores bridge column to
distribution with coefficient of variation (COV) of 0.2 as as-built or pristine conditions. After restoration, however,
mentioned in Vu and Stewart (2000). corrosion deterioration of the reinforcing steel may ensue
For an aging highway bridge structure that experiences following the corrosion initiation of the new rehabili-
a certain damage level after an earthquake event, if left tated structure.
STRUCTURE AND INFRASTRUCTURE ENGINEERING 1163

Figure 3. Variation of reinforcing steel area with time for the structures experiencing an earthquake event that has induced (a) minor damage level, and (b) col-
lapse damage level. The uncertainty band in Figure 3(a) represents 5th and 95th percentile estimates of the altered mean rebar area.

3.5. Scope of deterioration modeling and seismic life- determining seismic vulnerability of aging structures (Val,
cycle cost estimation Chernin, & Stewart, 2009; Ghosh & Padgett, 2010; Alipour
et al., 2011b), few studies have considered deep intermittent
Among the various secondary effects, this study explicitly
pit formations or pitting corrosion while modeling the
models the cover concrete strength reduction due to corro- chloride induced corrosion deterioration (Dizaj,
sion induced micro-crack formation. Although this strategy Madandoust, & Kashani, 2018; Ghosh & Sood, 2016;
may emulate physically observable damage such as cracking Kashani, Crewe, & Alexander, 2013; Shekhar et al., 2018).
and spalling of cover concrete, non-observable degradation Although the framework elaborated in this study could be
in the form of delamination is not considered. This limita- modeled using pitting form of corrosion deterioration in
tion primarily stems from a lack of existing analytical mod- determining the seismic life-cycle cost, these effects are
els in literature to simulate this effect. A recent study by excluded in this study for simplicity and thereby providing
Zhang and Su (2019) mentions the rarity of analytical stud- future opportunities.
ies on cover delamination failure in RC structures and high- Additionally, the seismic life-cycle cost estimation
light the need for future work in this area. Consequently, approach presented in this study focuses solely on the costs
the authors have left the consideration of delamination in stemming from post-earthquake repair or rehabilitation
corroding RC structures for future exploration. works. Other costs and their influence on structural
Another key assumption within the present study upgrades or repairs stemming from intermediate routine
includes crack formation within RC columns emanating maintenance work are not incorporated. Furthermore, indir-
from seismic actions alone. While corrosion deterioration ect costs stemming from increased travel time due to traffic
itself can result in cover cracking due to accumulation of rerouting, operating costs, cost due to loss of business or
expansive rust products, the alteration of corrosion rate due human lives costs, and injury costs, among others are left
to such effects are presently left out in this study. Corrosion for future inclusion with the seismic loss estimation frame-
can also lead to reduced bond strength, reduced inelastic work. For a comprehensive treatise on indirect costs
buckling, low-cycle fatigue degradation, among others (Ou, incurred due to highway bridge closure, the reader is
Tsai, & Chen, 2012; Meda, Mostosi, Rinaldi, & Riva, 2014; referred to Stein, Young, Trent, and Pearson (1999) and
Rao et al., 2017). However, the effect of bond strength in Deco and Frangopol (2013).
RC structures with modest tie reinforcements is negligible,
particularly on those which fail under cyclic loading (Fang,
Lundgren, Chen, & Zhu, 2004; Wang & Liu, 2004). 4. Seismic loss estimation under multiple
Consequently, this study neglects the effect of loss of bond earthquake events: representative case-study
strength within deteriorating structures. Additionally, example
although previous studies such as Kashani, Alagheband, This section of the paper presents a representative case-
Khan, and Davis (2015a) and Kashani, Lowes, Crewe, and study example of a highway bridge with deteriorating RC
Alexander (2015b) indicate a substantial impact of inelastic column that incorporates the probabilistic framework of
buckling strength and low-cycle fatigue life on RC struc- repair or replacement decisions along with subsequent cost
tures, these studies underline the need for further experi- estimates to compute probable lifetime seismic losses. The
mental investigations and hence these effects are following subsection first elaborates on the proposed frame-
presently excluded. work and highlights the uncertainties that affect seismic
Furthermore, the progressive reinforcing steel area loss losses and life-cycle costs. The bridge description along with
during the corrosion propagation phase is assumed to be details on the finite element modeling is presented next.
reducing uniformly along the length of the column. While Lastly, the proposed methodology is applied on the case-
several researchers have adopted this approximation in study bridge to evaluate and compare monetary seismic loss
1164 B. PANCHIREDDI AND J. GHOSH

and life-cycle costs for the non-deteriorating and aging the plastic hinge region, and c) completely replaced, and
bridge structure. thereby returning to as-built pristine conditions. Also, the
choice of repair strategy for a particular level of damage after
each shock may vary from one Monte Carlo trial to another as
4.1. Uncertainty considerations in seismic loss and life-
demonstrated in Figure 4. Consequently, the post-earthquake
cycle cost assessment
structural condition to resist a future earthquake is uncertain
In order to capture various sources of uncertainties, this and is captured using the Monte Carlo trials. Following these
study adopts a Monte Carlo simulation based seismic loss trials one can eventually derive the seismic loss distribution
and life-cycle cost assessment framework that relies on con- given n earthquake events along the bridge design life.
ducting sufficient number of trials to arrive at confident loss The following subsections will first elaborate on the finite
estimates. This proposed strategy helps in capturing uncer- element modeling of the case-study bridge structure and
tainties associated with: a) ground motion characteristics then proceeds to estimate the probabilistic seismic loss
that affect the level of damage, b) time along the design life under single shock and multiple shock scenarios for both
or planning horizon for shock occurrence, c) extent of cor- the non-deteriorating and corroding structures.
rosion deterioration affecting the column section and mater-
ial characteristics, and d) repair decisions and the adopted
4.2. Finite element modeling of case-study bridge and
repair strategies. Additionally, as demonstrated later, the
seismic loss evaluation
seismic loss histograms do not conform to standard statis-
tical distributions given the random nature of input varia- The case-study bridge adopted in this study is assumed to
bles. Consequently, Monte Carlo trials emerge as the be located in the state of California, United States wherein
method of choice for seismic loss analysis for non-deterio- the highway bridges are expected to experience multiple
rating as well as aging bridge structures. earthquakes along their design life due to regional seismi-
A schematic representation of this framework is demon- city. Additionally, proximity to marine sources render these
strated in Figure 4. For a selected number of seismic events bridges susceptible to chloride induced corrosion deterior-
(n), the jth trial of NMC Monte Carlo runs comprises of n seis- ation. The following subsection elaborates on the represen-
mic sequences applied to the structure (non-deteriorating or tative case-study bridge description along with the finite
aging) as main shocks at different points in time. Following element modeling of the non-deteriorating and aging condi-
each main shock damage, the deteriorating structure may be: tions. The methodology to determine probabilistic seismic
a) left unrepaired (for minor damages) but with potential losses under single and multiple earthquake events is dem-
acceleration of corrosion, b) partially repaired, especially near onstrated in the later subsections.

Figure 4. Seismic loss estimation based on various repair strategies adopted after every ongoing earthquake event until n shocks repeated for NMC number of
Monte Carlo trials. cj,i represents the cost of the adopted repair strategy corresponding to jth trial of the ith earthquake event.
STRUCTURE AND INFRASTRUCTURE ENGINEERING 1165

Figure 5. General layout of case-study two-span box-girder bridge along with the fiber discretization of reinforced concrete column section for non-deteriorating
and aging bridge conditions.

4.2.1. Description of representative case-study bridge As typical of integral bridge types, this study assumes the
Based on the review by Ramanathan (2012) on highway structural response and vulnerability of the case-study bridge
bridge inventory in California, more than 21% of all under seismic events to be primarily dominated by the bridge
bridges are comprised of multi-span continuous concrete column. This structural element is modeled using the finite elem-
box-girder type. This study adopts a two-span concrete ent software package OpenSees (McKenna, Fenves, & Scott,
box-girder bridge as shown in Figure 5 for the representa- 2000) while assuming the superstructure mass lumped at the col-
tive case-study example. The bridge is assumed to be con- umn top. The column cross-section is modeled as a fiber-element
structed in pre-1971 seismic design era with span length of with distributed plasticity using RCSection2d module available
31.78 m, span width of 9.78 m, and a single column bent of within OpenSees. While the inner confined and outer uncon-
6.63 m height. The adopted measurements of the case- fined concrete is modeled using Concrete04 material, the rein-
study bridge structure represent the median bridge dimen- forcing steel is modeled using uniaxialMaterial Hysteretic
sions of two-span concrete box-girder bridges in the region material that is capable of accounting strength and stiffness deg-
and hence deemed representative of general bridge geome- radation during load cycles. The analytical challenge associated
tries. This study assumes the structure is in proximity to with sequential updating of section properties and material char-
marine splash environmental condition with the probabil- acteristics to reflect the effect of corrosion deterioration in
ity distribution of deterioration parameters adopted from between seismic shocks is addressed by altering the OpenSees
DuraCrete (2000) and reproduced in Table 3. The material source code and incorporating new material definitions. While
properties, such as concrete compressive strength (fck), and the procedure to modify OpenSees source code is briefly elabo-
reinforcing steel yield strength (fy) are adopted form the rated in Appendix, more details can be found in Panchireddi and
in-house database of the California Department of Ghosh (2019).
Transportation (Caltrans) as reported in Mangalathu, Jeon, These material definitions have similar base line properties
Padgett, and DesRoches (2016). The column cross-section as Concrete04 and uniaxialMaterial Hysteretic material, but
is square having 1.29 m width with 36 #11 bars (35.81 mm with an additional flexibility of time-dependent updating. The
diameter) that constitute longitudinal steel reinforcement representative case-study bridge is utilized next to demon-
ratio (q) of 2.25% (Mangalathu et al., 2016). For the time- strate the probabilistic framework to compute lifetime seismic
history seismic analysis, a suite of 100 earthquake records losses towards column repair. These losses are computed for
of the California region developed by Gupta and non-deteriorating as well as for corroding bridge columns to
Krawinkler (2000) and Krawinkler, Medina, and Alavi further highlight the impact of degradation on seismic
(2003) are adopted in this study. While the spectral accel- life-cycle cost. The next subsections will first highlight the
eration of the ground motion suite at the fundamental probabilistic seismic loss computation under a single shock
period of the structure ranges from 0.03 g to 1.57 g, the scenario (for a non-deteriorating and corroding structure),
peak ground accelerations vary between 0.03 g to 1.3 g, and then under multiple shock scenario, followed by probabilistic
duration from 28.7 s to 100 s. aggregation of losses to compute seismic life-cycle.
1166 B. PANCHIREDDI AND J. GHOSH

Figure 6. (a) Single earthquake pulse of PGA ¼ 0.36 g, and (b) moment-curvature relationship depicting the maximum curvature ductility corresponding to bridge
column section (pristine and deteriorated condition at 50 and 100 years).

4.2.2. Seismic losses under single shock scenario calculate the present day seismic losses corresponding to a
This section demonstrates the estimation of probabilistic single shock alone. It is worthwhile to note that, while the
seismic loss for the representative case-study bridge when first two damage states (slight and minor) do not necessitate
experiencing only one earthquake along the design life that a repair action, multiple repair decisions and corresponding
is assumed to be 100 years. Although AASHTO (2017) strategies for column rehabilitation exist for other damage
recommends a design life period as 75 years for highway levels (Table 2) that adds an additional layer of uncertainty
bridge structures, preliminary review of the NBI database in seismic loss calculations.
(FHWA, 2018) reveals the existence of several highway For instance, corresponding to the example at hand (say
bridges in California that are nearing or even exceed Trial # j), a damage index of 0.26 for the 50 year old bridge
100 years of service. Consequently, this study adopts that falls in the moderate damage level comprises of 5%
100 years as the design life period of the bridge structure. chance of being repaired through epoxy injection, 23%
While the time t1 of this single shock occurrence is chance of getting patched with concrete, 9% chance of con-
irrelevant for the non-deteriorating structure, an explicit crete lining, 36% chance of wrapping, 23% chance of replac-
accounting is required for the corroding bridge since the ing column, and 5% chance of reinforcing and recasting the
degree of damage depends on the scale of deterioration. column. The adopted repair strategy of this trial leads to a
This is demonstrated in Figure 6 for the bridge in pris- seismic loss of cj,1 as represented earlier using Figure 4. On
tine condition and under corrosion deterioration at 50 and the other hand, the damage index of 0.24 for the non-dete-
100 years of design life when subjected to the same ground riorating bridge falls in the minor damage level and conse-
motion time-history with peak ground acceleration (PGA) quently does not necessitate a repair (cj,1 ¼ 0).
of 0.36 g [Figure 6(a)], chosen from the suite of accelero- Consequently, the seismic losses from repair corresponding
grams. Figure 6(b) shows the increase in column curvature to pristine and aging bridge column may vary. In instances,
ductility for the deteriorating bridge column (lm) which in however, when the damage indices for the as-built and
conjunction with the dissipated hysteretic energy (Eh) aids aging bridge fall within the same range, the repair probabil-
in the computation of the Park and Ang damage index D1 ities are identical.
after the first (and only, in this case) seismic shock using While a preliminary analysis on the Monte Carlo simula-
Equation 1. For this ground motion time-history in particu- tion strategy conducted by the authors indicate that the seis-
lar, the damage index increases from 0.24 to 0.26 and 0.29 mic loss computations incorporating the uncertainties tend
for the 50 year old and 100 year old bridge reflecting an to stabilize following 2000 trials, a total of 5000 Monte
increase in damage level due to deterioration. In this Carlo simulations are conducted for completeness. Across
example, following the range of damage indices outlined in each of the 5000 Monte Carlo trials, deterioration effects are
Table 1, the physical level of damage has shifted from minor accounted by randomly sampling the age of the bridge from
to moderate for a deteriorating bridge structure under the uniform distribution with lower and upper bounds at year 0
same earthquake pulse. and year 100, indicative of the beginning and end of design
While the extent of deterioration depending on the life respectively. Each bridge sample is then randomly paired
design life of the bridge influences the structural damage with a ground motion time-history from the suite of accel-
index, computation of seismic loss for a pristine (non-dete- erograms for nonlinear time-history analysis to compute the
riorating) and aging structure fundamentally follows a simi- damage index and associated seismic loss.
lar procedure. Based on the level of damage the structure is Figure 7(a) displays the histogram of the discounted seis-
categorized to either slight, minor, moderate, severe or col- mic losses for a deteriorating structure when subjected to a
lapse damage state as elaborated in the flowchart presented single shock scenario. As expected, a majority of the ground
in Figure 1. Subsequently, based on the damage state a motions induces none-to-minor damage and thereby incurs
repair strategy is chosen, and the repair cost discounted to zero seismic losses. The zoomed-in subplot within Figure
STRUCTURE AND INFRASTRUCTURE ENGINEERING 1167

Figure 7. (a) Histogram of seismic loss of bridge column along with the zoomed-in subplot showing the non-zero seismic losses when subjected to a single shock
for deteriorating condition, (b) comparison of non-zero seismic losses (natural logarithmic scale) of deteriorating bridges with those obtained from non-deteriorat-
ing bridges when subjected to a single shock, and (c) pie-charts representing the probability of adopting a repair decision after the first shock for both the non-
deteriorating and deteriorating bridges.

7(a) depicts the non-zero seismic losses stemming from Nr , 1 Nt , 1


pr, 1 ¼ and pt, 1 ¼ (9)
repair and replacements for the deteriorating bridge struc- NMC NMC
ture. Figure 7(b) compares these non-zero losses with those In the above equation, Nr,1 and Nt,1 are the corresponding
obtained from the non-deteriorating structure and reveals
number of trials among the total samples wherein bridge
interesting trends. As demonstrated in the deterministic
needs to be repaired or replaced after the nth shock, NMC is
example earlier, since an aging highway bridge sample is
the total number of Monte Carlo trials (NMC ¼ 5000).
likely to incur a higher damage index (and hence seismic
Consequently, the discounted seismic loss can be eval-
loss from repair) compared to a non-deteriorating bridge,
uated as:
the histogram peaks towards the right indicating higher
losses on an average for the deteriorating bridge. Cr, 1, j Ct, 1, j
C1, j ¼ t1  pr, 1 þ  pt, 1 (10)
These observations are further evident in the pie-charts ð1 þ aÞ ð1 þ aÞt1
presented in Figure 7(c) that indicate an aging bridge would
require more number of “actions” along the lifetime where, C1,j corresponds to the seismic loss of bridge struc-
amounting to a higher percentage of repair or replacement ture after the first earthquake event occurring at time t1 for
cases (25%) compared to the cases for a non-deteriorating the jth trial, Cr,1,j and Ct,1,j are the respective costs of jth
structure (22%). The percentage difference between median repair and replacement action that is sampled from the
losses of non-zero seismic cost between non-deteriorating probability of adopting different repair actions as indicated
and aging bridge samples is observed to be around 10%. earlier in Table 2, a is the adjusted discounted ratio
While the above numbers correspond to a bridge under sin- (assumed to be 0.03 in this study as mentioned by Beck
gle shock scenario, the percentage differences are likely to et al., 2002). Cr,i,j values can be directly sampled in such a
increase under multiple shocks, as demonstrated later. manner that the selection of any repair action strictly fol-
While the above histograms are derived using the Monte lows the probability of adopting that particular strategy as
Carlo simulations, an easy replicability of the loss results provided in Table 2.
can be attained after computing the probabilities for repair Using this closed form solution, Figure 8(a) and (b) pre-
pr,1 and replacement pt,1 (subscript “1” referring to the sin- sent the overlay of empirical cumulative density function
gle shock events) as: (CDF) of seismic losses obtained through Monte Carlo trials
and the fitted equation (Equation 10) for a single shock
1168 B. PANCHIREDDI AND J. GHOSH

Figure 8. Overlay of empirical cumulative distribution functions for the nonzero seismic losses (natural log scale) obtained through Monte Carlo trials and the fitted
equation [Equation (10)] for a single shock scenario corresponding to (a) non-deteriorating, and (b) deteriorating bridge conditions.

Figure 9. (a) Train of three earthquake pulses (PGA1 ¼ 0.06 g, PGA2 ¼ 0.30 g and PGA3 ¼ 0.27 g) occurring at different points in time during the design life, and
(b) potential alterations of bridge capacity and the discounted cost that may be accrued at the end of useful design life based on intermediate repair decisions.

scenario corresponding to non-deteriorating and deteriorat- the time-gap between any two consecutive events follow an
ing bridge conditions respectively. These results highlight exponential distribution. This corroborates with fundamen-
that the seismic losses obtained using the above equations tal probability theory, that for a Poisson process, the time to
replicate those from Monte Carlo trials reasonably well and the first occurrence of an event follows an exponential dis-
thereby can be conveniently utilized for seismic loss distri- tribution (Melchers, 1987). Additionally, in this study the
bution predictions. Following the single shock scenario, the number of earthquake shocks along the design life of
next subsection proceeds to estimate the seismic losses in 100 years are assumed to vary from one to eight, beyond
the case-study bridge structure when subjected to multiple which the occurrence probabilities of further seismic events
main shocks along the design life. are deemed negligible, as elaborated in the next section for
loss aggregation.
Adopting the Monte Carlo based framework as elabo-
4.2.3. Seismic losses under multiple shock scenarios rated earlier using Figure 4, the seismic loss assessment
Following the single shock scenario, this section of the under multiple shocks is demonstrated for a sample trial of
paper presents the seismic loss assessment for multiple three-shock sequence [Figure 9(a)] wherein the first shock
shock events along the lifetime of the structure. For a with PGA of 0.06 g is assumed to occur at year 7 while the
selected number of seismic events (n), multiple main shock bridge is still in pristine condition (before corrosion initi-
events are applied with sufficient time-gap between shocks ation time), the second with PGA 0.30 g at 43 years (on a
for the transient response of the structure (non-deteriorating deteriorating bridge) and lastly, the third shock with PGA
or aging) to dissipate between successive shocks (Figure 4) 0.27 g at 84 years. Figure 9(b) demonstrates the potential
during the simulation. This time-gap is representative of the alterations of bridge capacity and a sample cost that may be
number of years elapsed between one main shock and the accrued at the end of useful service life based on intermedi-
next, which is particularly important for a deteriorating ate repair decisions. For instance, the first accelerogram at
bridge structures. Since, this study adopts the conventional year 7 induces a damage index D1 of 0.01 (slight category)
Poisson process to model earthquake occurrences (main that does not necessitate a repair, but leads to a minor drop
shocks) during the design life period of the bridge structure, in column capacity.
STRUCTURE AND INFRASTRUCTURE ENGINEERING 1169

Figure 10. (a) Histogram of nonzero seismic losses (natural log scale) of deteriorating bridge column when subjected to three shocks, and (b) bar chart represent-
ing the probability of repair and replacement of the deteriorating bridge compared to the “Do Nothing” scenario after each earthquake event.

Table 4. Probabilities of the structure being repaired and replaced after each earthquake event (from one to eight)
for both the non-deteriorating and deteriorating bridge conditions.
Non-deteriorating scenario Deteriorating scenario
Number of Probability of Probability of Probability of Probability of
shocks (n) repair (pr,n) replacement (pt,n) repair (pr,n) replacement (pt,n)
1 11% 11% 10% 15%
2 18% 12% 16% 16%
3 22% 12% 22% 16%
4 32% 13% 27% 16%
5 38% 13% 32% 16%
6 43% 13% 36% 16%
7 48% 13% 39% 16%
8 51% 13% 42% 17%

As the structure continues to deteriorate, it experiences Compared to the single shock scenario, the median of
the second shock at 43 years that induces a damage index this histogram increases by 110% indicating the potential
D2 of 0.38 (moderate category) that needs repair. At this increase in seismic losses stemming from repair and replace-
juncture, for the moderate damage level, multiple repair ments with the number of earthquake occurrences along the
options are available (Table 2), and the choice of one repair life of the structure. Additionally, compared to the non-
strategy over another depends on the availability of resour- deteriorating case under three-shock sequences, the mean
ces, and discretion of the engineer, among other factors. losses are found to increase by 8%, further highlighting the
This particular example chooses “Patching with Concrete” importance of considering corrosion deterioration for seis-
that constitutes a 23% probability of being chosen as a mic loss assessment of aging bridges.
repair action for this damage level and incurs a cost of $96. A similar trend as the three-shock scenario presented
This repair strategy also improves the structural capacity of here can be observed for higher number of shocks with a
the column and when in place, delays the onset of corrosion gradual increase in the probability of bridge repair and
deterioration. replacement compared to the “Do Nothing” scenario as
The last seismic shock assumed to occur at 84 years indu-
shown in Figure 10(b). Similar to the single shock scenario,
ces a damage index D3 ¼ 0.34 that again leads to a moder-
these probabilities can be used to compute seismic loss sam-
ate level of damage and necessitates repair. Since this
ples for the jth shock for n earthquake sequences in closed
damage occurs towards the end of bridge design life, the
form as:
bridge engineer may opt for a cheaper repair option before
the bridge is decommissioned. This particular example X
n X
n
Cr, i, j Ct, i, j
chooses “Concrete lining” as the repair strategy following Cn, j ¼  pr , i þ  pt , i (11)
this particular seismic shock that incurs a cost of $93.1. i¼1 ð1 þ aÞti i¼1 ð1 þ aÞti
Subsequently, the discounted losses calculated in present
day dollar values for the above example that considers three where, Cn,j is the seismic loss of bridge structure after being
shocks till the end of bridge design life amounts to $38. exposed to n earthquake events for the jth sample, Cr,i,j and
While this example demonstrates just one trial scenario in Ct,i,j are the respective costs of adopted repair strategy and
the seismic loss computation under three shocks, the histo- replacement of the bridge structure after being exposed to
gram presented in Figure 10(a) captures the randomness in ith earthquake event occurring at time ti for the jth sample,
seismic shock pairing, earthquake occurrence times, deteri- a is the adjusted discounted ratio, and pr,i and pt,i are the
oration mechanism, and repair strategy selection. respective probabilities of the structure being repaired and
1170 B. PANCHIREDDI AND J. GHOSH

Figure 11. (a) Hazard curve representing the main shock hazard near Stanford University campus, California, and (b) probability of experiencing n shocks in a ser-
vice lifetime of T ¼ 100 years.

replaced after the ith event irrespective of any repair deci- years can be computed as:
sions employed during earlier events. X
m
These probabilities can be determined using the following CSLCC, j ¼ Cn, j  P½n, T  (13)
equation given as: n¼1

Nr, i Nt, i In the above equation, CSLCC,j corresponds to the seismic


pr , i ¼ and pt, i ¼ (12) life-cycle cost of the structure for the jth simulation, m is
NMC NMC
the maximum possible number of shocks the structure could
where, NMC is the total number of Monte Carlo trials, Nr,i
experience in T years (eight, in this study), Cn,j can be esti-
and Nt,i are the corresponding number of trials among the
mated from Equation 11, and P[n,T] is the probability of
total samples in which the bridge needs to be repaired or
structure experiencing n earthquake events in a design life-
replaced, respectively after the ith shock. The above proba- time of T years (determined using Figure 11(b) for a design
bilities as shown in Equation 12 computed for different lifetime of T ¼ 100 years).
number of shocks are shown in Table 4, and similar to the Figure 12(a) and (b) depict the histogram representing
single shock scenario, the empirical distribution from the the seismic life-cycle cost corresponding to the case-study
closed form equation and Monte Carlo trials are found to bridge column for both the non-deteriorating and aging
match reasonably well. The following subsection of the conditions respectively. Observations report that the low
paper presents the methodology to estimate seismic life- cost samples are higher in non-deteriorating bridges than
cycle cost using the hazard information of bridge location. that of aging bridges. Also, the samples corresponding to
higher costs are slightly dominant in aging bridges. This is
4.3. Seismic life-cycle cost (SLCC) analysis obvious as the deteriorating bridges are vulnerable to earth-
quake loads and demand repair works after an earthquake
The seismic losses estimated for single and multiple shock event. The mean seismic life-cycle cost of the deteriorating
scenarios are now used in conjunction with the regional bridges is about 9.2% greater than that of the non-deterio-
seismic hazard information of the bridge site to determine rating bridges. This highlights the significance of consider-
the seismic life-cycle cost. Assuming that the occurrence of ing corrosion deterioration during the estimation of seismic
earthquake events follows Poisson process, Figure 11(a) rep- life-cycle cost.
resents the main shock hazard curve of the Stanford Interestingly, the coefficient of variation (COV) around the
University campus region within California. This is particu- seismic life-cycle cost for the non-deteriorating bridge samples
larly chosen as the above region is in close proximity to the is slightly higher than that of the deteriorating samples. This
assumed case-study bridge location. primarily stems from the additional uncertainties involved in
On the other hand, Figure 11(b) utilizes the hazard curve the repair costs (non-deteriorating bridges are more likely to
information to determine the probability of occurrence of n get repaired than being replaced) than that of replacement
damaging earthquake events (with intensity greater than cost. While the above considered repair strategies are limited
0.1 g) within a design lifetime of 100 years (Panchireddi & to a single column among the different components of a
Ghosh, 2019).While the probability of experiencing three bridge structure, it is underlined that the seismic life-cycle
shocks in a lifetime of 100 years is highest, the probability of cost of other bridge components that are affected by corrosion
experiencing more than eight shocks is negligible. deterioration would be substantially higher.
Consequently, this study is restricted to consideration of It is worth to note that this study estimates the most rep-
maximum of eight main shock occurrences during the life- resentative seismic life-cycle cost by addressing the limita-
time for seismic loss assessment. Utilizing the NMC seismic tions of the past studies that consider the structure to be
loss samples computed to n shocks in the previous section always restored to its pre-hazard state after the repair or
(n ranging from one to eight), the seismic life-cycle cost rehabilitation works. Lastly, the results presented in this sec-
corresponding to the jth trial for a design life period of T tion underlines the significance of considering joint effects
STRUCTURE AND INFRASTRUCTURE ENGINEERING 1171

Figure 12. Histogram of seismic life-cycle cost (SLCC) along with the median and coefficient of variation (COV) values corresponding to the case-study bridge col-
umn for (a) non-deteriorating condition, and (b) deteriorating condition.

of corrosion deterioration and the multiple earthquake case under three-shock sequences, the mean losses are found
occurrences in estimating seismic life-cycle cost of the to increase by 8%, further highlighting the importance of
bridge structures. considering corrosion deterioration for seismic loss assess-
ment of aging bridges. Hence it is recommended to incorp-
orate deterioration effects in the design and analysis of
5. Conclusions
highway bridges when located in high seismic regions and
Structural performance of the highway bridges may decline are simultaneously exposed to aggressive environmen-
due to various environmental stressors or multiple natural tal conditions.
hazards during their design lifetime. While corrosion deteri- A gradual increase in the probability of bridge repair and
oration constitute a primary part of environmental degrad- replacement compared to the “Do Nothing” scenario is
ation, natural hazards such as earthquakes are predominant observed for higher number of shocks. These probabilities
in regions of moderate to high seismicity. This paper in conjunction with the corresponding repair costs aids in
presents a novel Monte Carlo based methodology in esti- reproducing the seismic cost samples directly using the pro-
mating the seismic life-cycle cost that explicitly considers posed equation. Lastly, the median seismic life-cycle cost of
the alteration in deterioration pattern between seismic the deteriorating bridges is about 9.2% greater than that of
shocks due to earthquake-induced cracks as well as the the non-deteriorating bridges. This highlights the signifi-
uncertainties in repair decisions based on different damage cance of considering corrosion deterioration during the esti-
levels and the repair processes. This study utilizes the Park mation of seismic life-cycle cost. While the considered
and Ang damage index for the determination of damage repair strategies in this study are limited to a single column
level and the corresponding repair or replacement strategies alone among the different components of a bridge structure,
are outlined based on their respective qualitative and quanti- it is underlined that the seismic life-cycle cost of other
tative descriptions. For the structures that demand no repair bridge components that are affected by corrosion deterior-
works after an earthquake event may however experience ation would be substantially higher.
visible minor cracks. Consequently, the alterations in the Future studies on this topic will investigate the seismic
corrosion deterioration process due to the formation of life-cycle cost considering intermediate maintenance meas-
these cracks are also included in the framework. ures along with the consequence of component maintenance
The proposed framework is demonstrated on a represen- on seismic performance of the bridge structures. While this
tative case-study two-span single column box-girder bridge study specifically focused on direct structural losses from
located in California region. For the structures experiencing seismic damage, indirect losses experienced by the commun-
single shock alone, an aging bridge would require more ities requires further exploration. In addition to losses stem-
number of “actions” along the design life amounting to a ming from column damage alone, economic consequences
higher percentage of repair or replacement cases (25%) from seismic damage to other critical bridge components
compared to the cases for a non-deteriorating structure such as bearings, bridge deck, and abutments needs consid-
(22%). The percentage difference between median losses of eration. Apart from the economic losses due to seismic
non-zero seismic cost for the non-deteriorating and aging events, future studies on highway bridges will also investi-
bridge samples is around 10%. gate other indicators of sustainability such as environmental
Compared to the single shock scenario, for an aging and social metrics.
structure, the median seismic loss after three shocks is
increased by 110% indicating the potential increase in seis-
mic losses stemming from repair and replacements with the Acknowledgements
number of earthquake occurrences along the life of the The authors would also like to acknowledge the anonymous Reviewer
structure. Additionally, compared to the non-deteriorating 3 for his/her review comments that has led to incorporation of several
1172 B. PANCHIREDDI AND J. GHOSH

practical aspects pertaining to seismic loss estimation of deteriorating Choe, D.E., Gardoni, P., Rosowsky, D., & Haukaas, T. (2009). Seismic
highway bridges. fragility estimates for reinforced concrete bridges subject to corrosion.
Structural Safety, 31(4), 275–283. doi:10.1016/j.strusafe.2008.10.001
Colombo, A., & Negro, P. (2005). A damage index of generalised
Disclosure statement applicability. Engineering Structures, 27(8), 1164–1174. doi:10.1016/j.
engstruct.2005.02.014
No potential conflict of interest was reported by the author(s). Coronelli, D., & Gambarova, P. (2004). Structural assessment of cor-
roded reinforced concrete beams: Modeling guidelines. Journal of
Structural Engineering, 130(8), 1214–1224. doi:10.1061/(ASCE)0733-
Funding 9445(2004)130:8(1214)
Cui, Z., Alipour, A., & Shafei, B. (2019). Structural performance of
The authors acknowledge the financial assistance provided by the
deteriorating reinforced concrete columns under multiple earth-
Ministry of Human Resource Development (MHRD), Government of
quake events. Engineering Structures, 191, 460–468. doi:10.1016/j.
India (GoI), and Department of Science and Technology (Grant No.
engstruct.2019.04.073
ECR/2016/001622) for the research work at Indian Institute of
Deco, A., & Frangopol, D. M. (2013). Life-cycle risk assessment of spa-
Technology Bombay.
tially distributed aging bridges under seismic and traffic hazards.
Earthquake Spectra, 29(1), 127–153. doi:10.1193/1.4000094
Di Sarno, L. (2013). Effects of multiple earthquakes on inelastic struc-
tural response. Engineering Structures, 56, 673–681. doi:10.1016/j.
References engstruct.2013.05.041
AASHTO. (2017). LRFD bridge design specifications (8th ed.). Dizaj, E. A., Madandoust, R., & Kashani, M. M. (2018). Probabilistic
seismic vulnerability analysis of corroded reinforced concrete frames
Washington, DC: American Association of State Highway and
including spatial variability of pitting corrosion. Soil Dynamics and
Transportation Officials.
Earthquake Engineering, 114, 97–112. doi:10.1016/j.soildyn.2018.07.013
Akiyama, M., & Frangopol, D. M. (2014). Long-term seismic perform-
Du, Y.G., Clark, L.A., & Chan, A.H.C. (2005). Residual capacity of cor-
ance of RC structures in an aggressive environment: Emphasis on
roded reinforcing bars. Magazine of Concrete Research, 57(3),
bridge piers. Structure and Infrastructure Engineering, 10(7), 135–147. doi:10.1680/macr.2005.57.3.135
865–879. doi:10.1080/15732479.2012.761246 DuraCrete. (2000). Probabilistic performance based durability design of
Akiyama, M., Frangopol, D. M., & Matsuzaki, H. (2011). Life-cycle concrete structures. The European Union-Brite EuRam III. Project
reliability of RC bridge piers under seismic and airborne chloride BE95-1347
hazards. Earthquake Engineering & Structural Dynamics, 40(15), El Hassan, J., Bressolette, P., Chateauneuf, A., & El Tawil, K. (2010).
1671–1687. Reliability-based assessment of the effect of climatic conditions on
Alipour, A., Shafei, B., & Shinozuka, M. (2011a). Performance evalu- the corrosion of RC structures subject to chloride ingress.
ation of deteriorating highway bridges located in high seismic areas. Engineering Structures, 32(10), 3279–3287. doi:10.1016/j.engstruct.
Journal of Bridge Engineering, 16(5), 597–611. doi:10.1061/ 2010.07.001
(ASCE)BE.1943-5592.0000197 Enright, M.P., & Frangopol, D.M. (1998). Probabilistic analysis of
Alipour, A., Shafei, B., & Shinozuka, M. S. (2011b). Capacity loss resistance degradation of reinforced concrete bridge beams under
evaluation of reinforced concrete bridges located in extreme chlor- corrosion. Engineering Structures, 20(11), 960–971. doi:10.1016/
ide-laden environments. Structure and Infrastructure Engineering, S0141-0296(97)00190-9
9(1), 1–20. doi:10.1080/15732479.2010.525243 Faisal, A., Majid, T. A., & Hatzigeorgiou, G. D. (2013). Investigation of
Andrade, C., Alonso, C., & Molina, F. J. (1993). Cover cracking as a story ductility demands of inelastic concrete frames subjected to
function of bar corrosion: Part I- Experimental test. Materials and repeated earthquakes. Soil Dynamics and Earthquake Engineering,
Structures, 26(8), 453–464. doi:10.1007/BF02472805 44, 42–53. doi:10.1016/j.soildyn.2012.08.012
Basoz, N., & Mander, J.B. (1999). Enhancement of the highway trans- Fang, C., Lundgren, K., Chen, L., & Zhu, C. (2004). Corrosion influ-
portation module in HAZUS. National Institute of Building Sciences, ence on bond in reinforced concrete. Cement and Concrete
16(1), 31–40. Research, 34(11), 2159–2167. doi:10.1016/j.cemconres.2004.04.006
Beck, J. L., Porter, K. A., Shaikhutdinov, R. V., Au, S. K., Mizukoshi, FEMA. (2012). HAZUS-MH MR4 technical manual. Washington, DC.
K., Miyamura, M., Ishida, H., Moroi, T., Tsukada, Y., & Masuda, FHWA. (2018). National Bridge Inventory Data. Washington, DC.
M. (2002). Impact of seismic risk on lifetime property values. Final Gervasio, H., & da Silva, L. S. (2013). Life-cycle social analysis of
motorway bridges. Structure and Infrastructure Engineering, 9(10),
Report, CUREE, Richmond, CA. doi:10.1044/1059-0889(2002/er01
Bertolini, L., Elsener, B., Pedeferri, P., & Polder, R.B. (2004). Corrosion of 1019–1039.
Ghosh, J., & Padgett, J.E. (2010). Aging considerations in the develop-
steel in concrete: Prevention, diagnosis, repair. KgaA, Weinheim: John
ment of time-dependent seismic fragility curves. Journal of
Wiley and Sons.
Structural Engineering, 136(12), 1497–1511. doi:10.1061/(ASCE)ST.
Bruno, S., & Valente, C. (2002). Comparative response analysis of con-
1943-541X.0000260
ventional and innovative seismic protection strategies. Earthquake
Ghosh, J., & Padgett, J.E. (2011). Probabilistic seismic loss assessment
Engineering & Structural Dynamics, 31(5), 1067–1092. of aging bridges using a component-level cost estimation approach.
Chai, Y. H., Priestley, M. N., & Seible, F. (1994). Analytical model for
Earthquake Engineering & Structural Dynamics, 40(15), 1743–1761.
steel-jacketed RC circular bridge columns. Journal of Structural doi:10.1002/eqe.1114
Engineering, 120(8), 2358–2376. doi:10.1061/(ASCE)0733- Ghosh, J., Padgett, J. E., & Sanchez-Silva, M. (2015). Seismic damage
9445(1994)120:8(2358) accumulation in highway bridges in earthquake-prone regions.
Cho, C. G., Kim, Y. Y., Feo, L., & Hui, D. (2012). Cyclic responses of Earthquake Spectra, 31(1), 115–135. doi:10.1193/120812EQS347M
reinforced concrete composite columns strengthened in the plastic Ghosh, J., & Sood, P. (2016). Consideration of time-evolving capacity
hinge region by HPFRC mortar. Composite Structures, 94(7), distributions and improved degradation models for seismic fragility
2246–2253. doi:10.1016/j.compstruct.2012.01.025 assessment of aging highway bridges. Reliability Engineering &
Choe, D.E., Gardoni, P., Rosowsky, D., & Haukaas, T. (2008). System Safety, 154, 197–218. doi:10.1016/j.ress.2016.06.001
Probabilistic capacity models and seismic fragility estimates for RC Gupta, A., & Krawinkler, H. (2000). Behavior of ductile SMRFs at vari-
columns subject to corrosion. Reliability Engineering & System ous seismic hazard levels. Journal of Structural Engineering, 126(1),
Safety, 93(3), 383–393. doi:10.1016/j.ress.2006.12.015 98–107. doi:10.1061/(ASCE)0733-9445(2000)126:1(98)
STRUCTURE AND INFRASTRUCTURE ENGINEERING 1173

Hatzigeorgiou, G. D., & Liolios, A. A. (2010). Nonlinear behaviour of McKenna, F., Fenves, G., & Scott, M. (2000). Open system for earth-
RC frames under repeated strong ground motions. Soil Dynamics quake engineering simulation. Berkeley: University of California.
and Earthquake Engineering, 30(10), 1010–1025. doi:10.1016/j.soil- Meda, A., Mostosi, S., Rinaldi, Z., & Riva, P. (2014). Experimental
dyn.2010.04.013 evaluation of the corrosion influence on the cyclic behaviour of RC
Hawk, H. (2003). Bridge life-cycle cost analysis. Transportation columns. Engineering Structures, 76, 112–123. doi:10.1016/j.eng-
Research Board, 483. struct.2014.06.043
Huang, W., Qian, J., Zhuang, B. B., & Fu, Q. S. (2012). Damage assess- Mehanny, S. S., & Deierlein, G. G. (2001). Seismic damage and collapse
ment of RC frame structures under multi- earthquake sequences. In assessment of composite moment frames. Journal of Structural
Advanced materials research (Vol. 446, pp. 739–744).Trans Tech Engineering, 127(9), 1045–1053. doi:10.1061/(ASCE)0733-
Publications. 9445(2001)127:9(1045)
Jiang, H. J., Chen, L. Z., & Chen, Q. (2011). Seismic damage assess- Melchers, R. E. (1987). Structural reliability analysis and prediction.
ment and performance levels of reinforced concrete members. Chichester: Wiley.
Procedia Engineering, 14, 939–945. doi:10.1016/j.proeng.2011.07.118 Nakano, Y., Maeda, M., Kuramoto, H., & Murakami, M. (2004,
Kameshwar, S., & Padgett, J. E. (2017). Characterizing and Predicting August). Guideline for post-earthquake damage evaluation and
Seismic Repair Costs for Bridges. Journal of Bridge Engineering, rehabilitation of RC buildings in Japan. In 13th World Conference
22(11), 04017083. doi:10.1061/(ASCE)BE.1943-5592.0001129 on Earthquake Engineering (No. 124).
Kashani, M. M., Alagheband, P., Khan, R., & Davis, S. (2015a). Impact Otieno, M. B., Alexander, M. G., & Beushausen, H. D. (2010).
of corrosion on low-cycle fatigue degradation of reinforcing bars Corrosion in cracked and uncracked concrete– influence of crack
with the effect of inelastic buckling. International Journal of Fatigue, width, concrete quality and crack reopening. Magazine of Concrete
77, 174–185. doi:10.1016/j.ijfatigue.2015.03.013 Research, 62(6), 393–404. doi:10.1680/macr.2010.62.6.393
Kashani, M. M., Crewe, A. J., & Alexander, N. A. (2013). Nonlinear Otieno, M., Beushausen, H., & Alexander, M. (2016a). Chloride-induced
stress–strain behaviour of corrosion- damaged reinforcing bars corrosion of steel in cracked concrete– Part I: Experimental studies
including inelastic buckling. Engineering Structures, 48, 417–429. under accelerated and natural marine environments. Cement and
doi:10.1016/j.engstruct.2012.09.034 Concrete Research, 79, 373–385. doi:10.1016/j.cemconres.2015.08.009
Kashani, M.M., Lowes, L.N., Crewe, A.J., & Alexander, N.A. (2015b). Otieno, M., Beushausen, H., & Alexander, M. (2016b). Chloride-
Phenomenological hysteretic model for corroded reinforcing bars includ- induced corrosion of steel in cracked concrete— Part II: Corrosion
ing inelastic buckling and low-cycle fatigue degradation. Computers & rate prediction models. Cement and Concrete Research, 79, 386–394.
Structures, 156, 58–71. doi:10.1016/j.compstruc.2015.04.005 doi:10.1016/j.cemconres.2015.08.008
Krawinkler, H., Medina, R., & Alavi, B. (2003). Seismic drift and duc- Ou, Y. C., Tsai, L. L., & Chen, H. H. (2012). Cyclic performance of
tility demands and their dependence on ground motions. large-scale corroded reinforced concrete beams. Earthquake
Engineering Structures, 25(5), 637–653. doi:10.1016/S0141- Engineering & Structural Dynamics, 41(4), 593–604.
0296(02)00174-8 Padgett, J.E., Dennemann, K., & Ghosh, J. (2010). Risk-based seismic life-
Kumar, R., & Gardoni, P. (2012). Modeling structural degradation of cycle cost-benefit (LCC-B) analysis for bridge retrofit assessment.
RC bridge columns subjected to earthquakes and their fragility esti- Structural Safety, 32(3), 165–173. doi:10.1016/j.strusafe.2009.10.003
mates. Journal of Structural Engineering, 138(1), 42–51. doi:10.1061/ Padgett, J.E., & DesRoches, R. (2007). Bridge functionality relationships
(ASCE)ST.1943-541X.0000450 for improved seismic risk assessment of transportation networks.
Kumar, R., Gardoni, P., & Sanchez-Silva, M. (2009). Effect of cumula- Earthquake Spectra, 23(1), 115–130. doi:10.1193/1.2431209
tive seismic damage and corrosion on the life-cycle cost of rein- Panchireddi, B., & Ghosh, J. (2019). Cumulative vulnerability assess-
forced concrete bridges. Earthquake Engineering & Structural ment of highway bridges considering corrosion deterioration and
Dynamics, 38(7), 887–905. doi:10.1002/eqe.873 repeated earthquake events. Bulletin of Earthquake Engineering,
Kunnath, S. K., & Jenne, C. (1994). Seismic damage assessment of 17(3), 1603–1638. doi:10.1007/s10518-018-0509-3
inelastic RC structures. In 5th US National Conference on Park, Y. J., & Ang, A. H. S. (1985). Mechanistic seismic damage model
Earthquake Engineering (Vol. 1, pp. 55–64). for reinforced concrete. Journal of Structural Engineering, 111(4),
Lee, K. M., Cho, H. N., Lim, J. K., & Park, K. H. (2004). Life-cycle 722–739. doi:10.1061/(ASCE)0733-9445(1985)111:4(722)
cost effective optimal seismic design for continuous PSC bridges. In Park, Y. J., Ang, A. H. S., & Wen, Y. K. (1985). Seismic damage ana-
Life-cycle performance of deteriorating structures: Assessment, design lysis of reinforced concrete buildings. Journal of Structural
and management (pp. 247–262). Porto, Portugal. doi:10.1061/ Engineering, 111(4), 740–757. doi:10.1061/(ASCE)0733-
9780784407073. 9445(1985)111:4(740)
Liu, T., & Weyers, R. W. (1998). Modeling the dynamic corrosion pro- Ramanathan, K.N. (2012). Next generation seismic fragility curves for
cess in chloride contaminated concrete structures. Cement and California bridges incorporating the evolution in seismic design phil-
Concrete Research, 28(3), 365–379. doi:10.1016/S0008- osophy (Ph.D. dissertation). Atlanta, GA: Georgia Institute of
8846(98)00259-2 Technology.
Ma, R., & Xiao, Y. (1999). Seismic retrofit and repair of circular bridge Rao, A. S., Lepech, M. D., Kiremidjian, A. S., & Sun, X. Y. (2017).
columns with advanced composite materials. Earthquake Spectra, Simplified structural deterioration model for reinforced concrete
15(4), 747–764. doi:10.1193/1.1586070 bridge piers under cyclic loading. Structure and Infrastructure
Mackie, K. R., Wong, J. M., & Stojadinovic, B. (2010). Post-earthquake Engineering, 13(1), 55–66. doi:10.1080/15732479.2016.1198402
bridge repair cost and repair time estimation methodology. Raupach, M. (1996). Chloride-induced macrocell corrosion of steel in
Earthquake Engineering & Structural Dynamics, 39(3), 281–301. concrete—Theoretical background and practical consequences.
Mander, J. B., Priestley, M. J., & Park, R. (1988). Theoretical stress- Construction and Building Materials, 10(5), 329–338. doi:10.1016/
strain model for confined concrete. Journal of Structural 0950-0618(95)00018-6
Engineering, 114(8), 1804–1826. doi:10.1061/(ASCE)0733- Roufaiel, M. S., & Meyer, C. (1987). Reliability of concrete frames
9445(1988)114:8(1804) damaged by earthquakes. Journal of Structural Engineering, 113(3),
Mangalathu, S., Jeon, J. S., & Jiang, J. (2019). Skew Adjustment Factors 445–457. doi:10.1061/(ASCE)0733-9445(1987)113:3(445)
for Fragilities of California Box-Girder Bridges Subjected to near- Scott, A., & Alexander, M. G. (2007). The influence of binder type,
Fault and Far-Field Ground Motions. Journal of Bridge Engineering, cracking and cover on corrosion rates of steel in chloride-contami-
24(1), 04018109. doi:10.1061/(ASCE)BE.1943-5592.0001338 nated concrete. Magazine of Concrete Research, 59(7), 495–505. doi:
Mangalathu, S., Jeon, J. S., Padgett, J. E., & DesRoches, R. (2016). 10.1680/macr.2007.59.7.495
ANCOVA-based grouping of bridge classes for seismic fragility Shekhar, S., & Ghosh, J. (2020). A metamodeling based seismic life-
assessment. Engineering Structures, 123, 379–394. doi:10.1016/j.eng- cycle cost assessment framework for highway bridge structures.
struct.2016.05.054 Reliability Engineering & System Safety, 195, 106724.
1174 B. PANCHIREDDI AND J. GHOSH

Shekhar, S., Ghosh, J., & Padgett, J. E. (2018). Seismic life-cycle cost Zhang, G., Cao, X., & Fu, Q. (2016). Experimental study on residual
analysis of ageing highway bridges under chloride exposure condi- strength of concrete confined with corroded stirrups. Canadian Journal
tions: Modelling and recommendations. Structure and Infrastructure of Civil Engineering, 43(6), 583–590. doi:10.1139/cjce-2016-0138
Engineering, 14(7), 941–966. doi:10.1080/15732479.2018.1437639 Zhang, Y., & Su, R. K. L. (2019). Concrete cover delamination model for
Stein, S. M., Young, G. K., Trent, R. E., & Pearson, D. R. (1999). non-uniform corrosion of reinforcements. Construction and Building
Prioritizing scour vulnerable bridges using risk. Journal of Materials, 223, 329–340. doi:10.1016/j.conbuildmat.2019.06.199
Infrastructure Systems, 5(3), 95–101. doi:10.1061/(ASCE)1076- Zhong, J., Gardoni, P., & Rosowsky, D. (2012). Seismic fragility estimates
0342(1999)5:3(95) for corroding reinforced concrete bridges. Structure and Infrastructure
Stewart, M. G., & Rosowsky, D. V. (1998). Time-dependent reliability Engineering, 8(1), 55–69. doi:10.1080/15732470903241881
of deteriorating reinforced concrete bridge decks. Structural Safety,
20(1), 91–109. doi:10.1016/S0167-4730(97)00021-0
Stewart, M. G., Wang, X., & Nguyen, M. N. (2011). Climate change Appendix. Addition of new material definitions
impact and risks of concrete infrastructure deterioration. Engineering in OpenSees
Structures, 33(4), 1326–1337. doi:10.1016/j.engstruct.2011.01.010
Val, D. V., Chernin, L., & Stewart, M. G. (2009). Experimental and Figure A1 provides the flowchart elaborating the procedure of adding
numerical investigation of corrosion-induced cover cracking in rein- new material definitions to the OpenSees source code. Firstly, two new
materials that are similar in the lines of already existing Concrete04
forced concrete structures. Journal of Structural Engineering, 135(4),
and HystereticMaterial files with names Concrete04Update and
376–385. doi:10.1061/(ASCE)0733-9445(2009)135:4(376)
HystereticUpdate are added to the existing OpenSees (McKenna et al.,
Vecchio, F. J., & Collins, M. P. (1986). The modified compression-field
2000) source code. Review on the updatable parameters within
theory for reinforced concrete elements subjected to shear. ACI
OpenSees source code reveals that the material class containing defined
Journal, 83(2), 219–231.
MovableObject class as the virtual functions in its associated header
Vu, K. A. T., & Stewart, M. G. (2000). Structural reliability of concrete
file (“MovableObject.h”) would be possibly updated. Any updatable
bridges including improved chloride- induced corrosion models. parameter in OpenSees is defined by a unique parameterID within the
Structural Safety, 22(4), 313–333. doi:10.1016/S0167-4730(00)00018-7 finite element domain.
Wang, X., & Liu, X. (2004). Modeling bond strength of corroded Importantly, these parameters contain three critical functions namely
reinforcement without stirrups. Cement and Concrete Research, setParameter, updateParameter and activateParameter that are identified
34(8), 1331–1339. doi:10.1016/j.cemconres.2003.12.028 as the objectives responsible for setting, updating and activating the par-
Wang, M. L., & Shah, S. P. (1987). Reinforced concrete hysteresis ameter. These functions are now defined in the header files and the corre-
model based on the damage concept. Earthquake Engineering & sponding definitions are added to the .cpp files of the newly added
Structural Dynamics, 15(8), 993–1003. materials. An OPS function is now defined in the model builder files for
Williams, M. S., & Sexsmith, R. G. (1997). Seismic assessment of con- the user to define material inputs. In the header files of OPS function (i.e,
crete bridges using inelastic damage analysis. Engineering Structures, “OPS.Globals.h”), the defined input functions are added to the new
19(3), 208–216. doi:10.1016/S0141-0296(96)00104-6 material files respectively and all the parameters are initialized in the con-
Yanweerasak, T., Pansuk, W., Akiyama, M., & Frangopol, D. M. structor. Finally, the new material properties are defined in the software
(2018). Life-cycle reliability assessment of reinforced concrete development files namely material.vcxproj and material.vcxproj.filters.
bridges under multiple hazards. Structure and Infrastructure Incorporating these changes, OpenSees source code should now be suc-
Engineering, 14(7), 1011–1024. doi:10.1080/15732479.2018.1437640 cessfully compiled to a new OpenSees executable.

Figure A1: Flowchart elaborating the addition of new material definitions to the OpenSees source code.

You might also like