Seismic Reliability and Risk Assessment of

Structures Based on Fragility Analysis A Review

M. G. Sfahani*, Hong Guan and Yew-Chaye Loo

Griffith School of Engineering, Griffith University Gold Coast Campus, Queensland 4222, Australia

(Received: 16 January 2014; Received revised form: 27 May 2015; Accepted: 10 June 2015)

Abstract: A practical approach to explain the consequences of seismic hazards for

society and decision making organisations is predicting the seismic risk. In this regard,
the Probabilistic Seismic Hazard Analysis (PSHA) is used extensively to investigate
the probability of different seismic hazard levels at a geographical location. In
addition, a further advancement is the introduction of Probabilistic Seismic Demand
Analysis (PSDA) method because it provides a new insight into the Performance-
Based Earthquake Engineering (PBEE) by evaluating the seismic risk specifically for
a structure. To evaluate this seismic risk, the probability that the structural seismic
demands may exceed a specific value is calculated under different ground motion
intensities through a probabilistic approach. This approach is called the fragility
analysis. This paper provides a review of recent research advancements in seismic
fragility analysis. Different methods and related solutions which can be used for the
fragility analysis are discussed. In addition, uncertainty quantification, as a significant
feature in fragility analysis, is described and the important parameters which may
influence the seismic fragility of a structure are explained. Finally, the authors offer
their recommendations for improving the fragility analysis for further studies in the
future.

Key words: fragility analysis, seismic hazard, seismic demand, uncertainty quantification, performance limit-states.

1. INTRODUCTION unable to satisfy the requirements of the well-accepted

1.1. General Remarks seismic standards. Consequently, while the new
Natural hazards have always had undesirable influences earthquake-resistant structures are designed to have an
on infrastructure, human life, economy and communities. appropriate balance of stiffness and ductility, the existing
Recent advances in computational science and aged structures are more susceptible to earthquakes.
technology have resulted in higher safety levels against Therefore, seismic performance assessment of the
natural hazards for the benefit of human societies. existing structures is a key element in the process of
However, earthquakes remain one of the most achieving the safety against seismic hazard.
unpredictable natural hazards due to the randomness of
the return period and the shaking intensity. Initial 1.2. Code-Based Seismic Performance
academic and scientific attempts for mitigating fatalities Assessment
and financial losses of earthquakes started in the early Primary code-based guidelines for seismic performance
1900s and have resulted in several structural seismic assessment and rehabilitation of structures were
provisions and building standards. Since this presented by FEMA 273 (1997) and FEMA 274 (1997),
achievement has been accomplished only in the past few and later by FEMA 356 (2000). These guidelines were
decades, a large number of structures remained which are the culmination of a long-term project, sponsored by the

*Corresponding author. E-mail address: [email protected]; Fax: + 617-55528065; Tel: +617-55527608.

Advances in Structural Engineering Vol. 18 No. 10 2015 1653

 Seismic Reliability and Risk Assessment of Structures Based on Fragility Analysis A Review

Federal Emergency Management Agency (FEMA) of are then encouraged to carry out performance-based
the United States, to determine the seismic performance seismic evaluation and design by using ASCE 41-06
assessment of structures. They were developed by a which results in inappropriate engineering analysis,
group of specialists in earthquake engineering and overestimation, dispensable strengthening of structures,
seismic rehabilitation, from the Applied Technology and additional costs (Paret et al. 2011).
Council (ATC), the National Earthquake Hazards
Reduction Program (NEHRP), and a consortium of 1.3. Current Research
universities for research in earthquake engineering A probabilistic earthquake engineering framework is an
(CUREE). Unlike the prescriptive procedure of the alternative for seismic performance assessment of the
traditional building codes for design of new structures, structures that incorporates the details of seismic
four different performance levels are proposed in these analysis or design assumptions. Using such a
guidelines (operational, immediate occupancy, life framework, the seismic damage vulnerability of a
safety, and collapse prevention) in which each structure can be represented in the form of a series of
represents a specific seismic damage level (slight, fragility curves which allows a decision maker to adopt
moderate, extensive, and complete, respectively). a rehabilitation level according to his/her engineering
Then, the combination of the seismic damage level and judgment. Recently, a performance-based earthquake
seismic hazard level indicates the likelihood of engineering (PBEE) framework has been developed by
exceeding a performance level, which is a beneficial the Pacific Earthquake Engineering Research (PEER)
tool for determining the rehabilitation strategy. Centre and other research organisations (Cornell and
Nevertheless, in 2007, ASCE 41-06 (2007) was Krawinkler 2000; Gnay and Mosalam 2013). This
introduced as the only standard for seismic performance framework estimates the probabilistic seismic
assessment of structures and it has since superseded performance of structures in terms of system-level
other guidelines. ASCE 41-06 provides two decision variables (repair cost, casualties, and loss of
rehabilitation methods: a simplified method which is functionality). The framework consists of four stages:
aimed for small, regular, and simple structures and a (a) hazard analysis (intensity of the ground motions
systematic method for predicting performance of threatening the structure); (b) structural analysis (input
complex structures and minimising the cost of seismic loads and output structural responses); (c)
rehabilitation. Moreover, four different structural damage analysis (current condition and after-event
analysis methods (linear static, nonlinear static, linear performance of the structure); and (d) loss estimation
dynamic, and nonlinear dynamic), with related (dollars, deaths, and downtime). In terms of the first
acceptance criteria, were defined by ASCE 41-06 in stage of this framework, hazard analysis, the mean
which, when the complexity of a structure increases, a annual frequency (MAF) of seismic hazard is estimated
nonlinear and/or dynamic analysis is required. for a range of ground motion intensities with different
ASCE 41-06 represents the current standard of magnitudes and epicentral distances. In fact, this stage is
practice to achieve an objective structural performance established on the Probabilistic Seismic Hazard
level against seismic loads, however, compliance to this Analysis (PSHA) as proposed by Cornell (1968). The
standard does guarantee the achievement of the desired hazard maps (or curves), which are now provided by
performance. Earthquakes are accompanied by different seismic codes, in the form of Peak Ground Motion
sources of uncertainty, such as ambiguities in location (PGA) or Spectral acceleration (Sa), rely heavily on
(epicentre), and magnitude of the ground motions. The PSHA which, itself, is composed of four steps: (a)
uncertainty increases in the structural engineering phase identifying earthquake sources and magnitudes; (b)
by encompassing the structural modelling uncertainties identifying earthquake distances; (c) predicting ground
as well as the randomness of structural seismic motion intensities; and (d) combining information of the
responses. However, although compliance to ASCE 41- first three steps for achieving the hazard curves. Earlier
06 suggests the use of seismic hazard maps to find the development of PSHA was reviewed and summarised
maximum considered earthquake for structural by McGuire (2008).
analyses, there is no provision to investigate the In regard to the second stage of the abovementioned
variation of seismic responses when multiple seismic framework, structural analysis, advances in computer
events are considered. Under these circumstances, sciences and numerical modelling during past decades
ASCE 41-06 seems to discourage, or even prohibit, the have provided the possibility of performing complex
use of engineering judgment and decision making by its structural analysis. This scientific advancement has
prescriptive procedure (Searer et al. 2008). As a been particularly accelerated by the introduction of
consequence, even inexperienced structural engineers Incremental Dynamic Analysis (IDA) for investigating

1654 Advances in Structural Engineering Vol. 18 No. 10 2015

M. G. Sfahani, Hong Guan and Yew-Chaye Loo

the nonlinear seismic responses of structures under choosing a proper ground motion intensity in fragility
different shaking levels of a same seismic event analysis is highlighted and the defined characteristics
(Vamvatsikos and Cornell 2002). Nevertheless, ongoing for an optimal ground motion intensity are studied.
earthquake risk studies mainly contribute to the third Factors required to be considered for seismic record
stage of the framework, which attempts to determine the selection are discussed in the same section. Similar to
risk of seismic structural damage levels instead of the Section 4, in Section 5 the significance of using an
risk of seismic hazard levels (first stage), in order to appropriate damage measure in earthquake engineering
present an overview of the seismic performance of the is emphasized. In this section, the current definition of
structure (Bazzurro 1998; Shome 1999). In these structural performance during an earthquake is
studies, a structural performance level is demonstrated described and the importance of considering cumulative
by specifying a seismic damage level calling the limit- damage indices in seismic damage analysis is
state (Bazzurro 1998). As a consequence, it is possible underlined. In Section 6 the flexibility of fragility
to evaluate the MAF of exceeding the limit-states apart analysis to include the joint impact of additional hazard
from the MAF of the seismic hazard levels. This loads (e.g. corrosion, soil liquefaction, snow load,
procedure is referred to as the Probabilistic Seismic hurricane, and flood) is considered.
Demand Analysis (PSDA) (Bazzurro 1998; Shome
1999; Carballo and Cornell 2000; Luco 2002). PSDA 2. FRAGILITY ANALYSIS
requires structure-specific nonlinear analysis with a Seismic fragility analysis was initially introduced and
relatively small set of recorded accelerograms for used for the purpose of risk and safety assessment of
seismic hazard analysis, and it provides the next stage of structural components and mechanical assemblies in
development for seismic assessment that considers the nuclear power plants (Kennedy et al. 1980; Kaplan et al.
uncertainties in nonlinear response and capacity (Shome 1983). The outcome of analysis depicts the seismic risk
1999). Indeed, PSDA has become the basis of extensive exposure in the form of a curve, namely fragility curve,
research in earthquake engineering for structural which represents the probability of exceeding a
damage analysis. The most significant result of PSDA is structural performance level in response to increasing
a set of curves which illustrate the likelihood of earthquake intensity. Figure 1 shows schematic fragility
exceeding a structural performance level, caused by curves of building structures where fragility curves were
different ground shaking intensities, namely the fragility plotted for different structural performance levels. In
curve. Recently, during the SAC/FEMA* project fragility analysis, the structural performance levels are
(Mahin et al. 2002), the procedure for developing represented in the form of seismic damage measures
seismic fragility curves of structures has been revised, (DM), namely limit-states (LS). Consequently, each
and hence, a new insight was provided into the seismic fragility curve demonstrates the probability of
risk analysis (Shome et al. 1998; Luco and Cornell exceeding a LS in response to increasing earthquake
2000). intensity.

1.4. Aims and Scope

Although seismic fragility analysis has undergone 1.0
Slight
continued development, efficient, practical and more damage
Moderate
accurate fragility curves are just at the initial stage of damage
development. This paper presents a state-of-the-art Extensive
damage
MAF

review of seismic fragility analysis. In addition to the 0.5

introduction (i.e. Section 1), this paper consists of six Complete
damage
sections. In Section 2, conventional methods for
developing fragility curves, as well as novel closed-
form solutions for fragility function, are described. In
Section 3, it is explained how the structural uncertainties 0.0
Weak Medium Strong
are reflected in the fragility analysis. Also, the latest shaking shaking shaking
IM
studies for performing hybrid fragility analysis, as well
as the techniques for structural system reliability Figure 1. Example of building fragility curves for slight, moderate,
analysis, are reviewed. In Section 4 the importance of extensive and complete damage

* The SAC project for reduction of earthquake hazards in steel moment resisting frames (SMRF) in US involved three joint venture partners: the SEAOC
(Structural Engineers Association of California), ATC, and CUREE.

Advances in Structural Engineering Vol. 18 No. 10 2015 1655

 Seismic Reliability and Risk Assessment of Structures Based on Fragility Analysis A Review

One advantage of calculating the fragility curves is testing is not required. However, being purely
that it permits the accomplishment of the last stage of established on expert opinion causes the uncertainty
PEERs PBEE framework, where the fragility curves level of judgmental fragility curves to increase.
are employed as the inputs for loss estimation. The most Analytical fragility curves are developed to avoid the
facilitating option for this purpose is the computer costly experimental testing involved in empirical
software Hazus-MH which contains models for fragility curves and/or the uncertain data gathering on
estimating potential losses from earthquakes, floods and which the judgemental fragility curves are based. For
hurricanes (Hazus 2014). Hazus uses the Geographic this purpose, structures are modelled by computer
Information Systems (GIS) based PSHA data software and analysed under simulated earthquake loads
standardized classified inventory of infrastructures and (Porter et al. 2001). The output seismic responses are
the calculated fragility curves to estimate the direct and then imported into the fragility function to derive the
indirect loss (Kircher et al. 2006). The outputs fragility curve. To formulate the analytical fragility
demonstrate the limits of identified high-risk locations function and obtain a solution for it, the probability of
due to a natural hazard (e.g. earthquake) in the form of seismic demand (DM) exceeding a limit-state (LS)
graphical exposures which can be used for mitigation conditional on seismic intensity measure (IM) is
and recovery as well as preparedness and response expressed through the Total Probability Theorem
(Kircher et al. 2006). (Benjamin and Cornell 1970), as

2.1. Calculation of Fragility Curves

Based on the sources of seismic data, fragility curves are
P[ DM LS ] = P[ LS[ DM ] . dP[ DM | IM] . dP[IM ]
generally divided into three main categories: empirical, = P[ DM LS | IM = xi ]P[ IM = xi ] (1)
all x i
judgmental and analytical (Rossetto and Elnashai 2003).
Empirical fragility curves are developed based on the
In fact, the above equation denotes the two essential
inspection and observation of the recorded structural
elements in seismic risk analysis: local seismicity
damage from the post-earthquake surveys, and therefore
P[IM = xi] and the probability of seismic damage and
are considered to be highly reliable. A procedure for
structural fragility P[DM LS|IM = xi]. The former is
developing fragility curves from empirical data has been
illustrated by seismic hazard curves (or maps) while the
proposed by Shinozuka et al. (2000). Typical empirical
latter is demonstrated using fragility curves. To find a
fragility curves can be found in the literature based on
closed-form solution for the fragility function, it is
previous earthquakes, such as 1994 California (Basoz et
required to establish a relation between the ground
al. 1999), 1995 Kobe (Karim and Yamazaki 2001),
motion intensity (input) and the seismic demand
2006 Indonesia (Khalfan 2013) and 2009 Italy (DAyala
(output). A power model, as follows, was proposed for
and Paganoni 2011), for bridges and masonry structures.
this purpose (Cornell et al. 2002; Jalayer 2003),
Also, in the absence of real earthquake data, large scale
shaking table tests have been used to generate empirical
(experimental) fragility curves (Bothara et al. 2010; DM = a ( IM )b (2)
Banerjee and Shinozuka 2008a).
The most practical way to estimate the parameters, a
However, because of insufficient real earthquake data
and b, is to conduct nonlinear dynamic analyses and plot
and costly experimental testing, the use of empirical
DM versus IM followed by a regression analysis
fragility curves is infrequent. An alternative is to
(Vamvatsikos and Cornell 2002; Jalayer 2003;
develop judgemental fragility curves in which the
Fragiadakis and Vamvatsikos 2010; Fajfar and Dolsek
experts opinions are requested and collected on the
2011). It is assumed that both the DM and IM are log-
vulnerability of structural components to future
normally distributed with a dispersion value, in which
earthquakes. Mosleh and Apostolakis (1986) have
DM|IM represents the respective dispersion of median
developed a probabilistic approach for approximating
seismic demand over the range of ground motion
the probability distribution using experts estimates.
intensities (Jalayer 2003). Consequently, the seismic
Porter et al. (2007) also suggested and produced a
fragility curve for a specific limit-state (LS) (the
questionnaire for soliciting expert judgment on
structural capacity (C) in this case) is obtained as
structural components fragility. Typical judgmental
fragility curves for bridge components can be found in
ATC-40 (1996). Employing the judgmental fragility
In( DM ) In(C )
curves is more beneficial because sufficient data can be P[ DM C | IM = xi ] = (3)
2 DM | IM + 2
collected, and it is less costly since expensive laboratory c

1656 Advances in Structural Engineering Vol. 18 No. 10 2015

M. G. Sfahani, Hong Guan and Yew-Chaye Loo

In other word, the above equation denotes the simplifying assumptions which result in some errors.
probability of failure where c is the dispersion value of The dispersion value of the median seismic demand
median seismic capacity and is the standard normal DM|IM, which is considered to be constant for the whole
distribution. Some analytical fragility studies have been range of IM, is highly uncertain in the presence of an
carried out, using the above closed-form solution, outlier. An outlier is an observation which deviates
during the 1990s (Hwang and Hou 1994; Singhal and significantly from the rest of the distribution and arouses
Kiremidjian 1996; Hwang and Jaw 1990). Nevertheless, the suspicion that it follows a different distribution
as a part of the SAC/FEMA project a decision was made (Aslani and Miranda 2005). When a structure
to also include probability of seismic hazard approaches the collapse stage, a small increment in the
(seismicity) in the fragility function (Cornell et al. 2002; IM produces very large DM in which the result
Mahin et al. 2002; Jalayer 2003). In regard to this, the disregards the distribution of the remaining samples. To
seismic hazard is also approximated by a power model capture this variation, it was suggested to estimate DM
(Cornell 1968) at three different levels of ground motion intensity
through a modified power model (Aslani and Miranda
H ( IM ) = P[ IM xi ] k0 IM
k1
= k0 exp( k1InIM
M) (4) 2005), as follows

where k0 and k1 are positive coefficients representing the DM = 1 2IM ( IM ) 3

(7)
intercept and slope of linear regression of seismic
hazard in the logarithmic space, respectively.
Subsequently, a fragility curve is obtained to
demonstrate the annual probability of exceeding a limit- DM |IM = 1 + 2 ( IM ) + 3 ( IM ) 2
state as follows (8)
where i and i are constants and are computed for the
1 k12 three pairs of IM-DM and IM DM|IM at the
P[ DM LS ] = H ( IM LS )exp DM
2
|IM 2 (5) corresponding IM level. Moreover, the approximation
2 b of the seismic hazard by a power-model (Eqn 4)
indicates a linear distribution in log-log scale. As shown
and the probability of failure will take the following in Figure 2, this assumption is only suitable for small
form changes of IM while the closed-form solution is
calculated by integration over the whole range of
1 k12 considered IM. This error can be reflected as a function
PoF = P[ DM C ] = H ( IM C ) exp 2
( DM 2
| IM + c ) (6) of the curvature (second derivative) of seismic hazard
2
2 b
curve in log-log space (Bradley and Dhakal 2008).
In order to reduce this error, a second-order
where IMLS and IMC are the corresponding ground
approximation was proposed (Vamvatsikos 2013), as
motion intensities to the limit-state (LS) and seismic
follows
capacity (C), based on the achieved power model (Eqn
2). If there was no dispersion ( = 0) then the
probability of exceeding a limit state (or failure 1
probability) could be determined by substituting these
values into the seismic hazard approximation (Eqn 4).
However, the dispersion values increase the probability
MAF

since the value of the exponential phrase in Eqn 6, 0.1

namely the correction factor, was reported to be
typically between 1.5 and 3 (Cornell et al. 2002; Jalayer Seismic hazard
2003). First-order approximation
Second-order approximation
0.01
0.01 0.1 1
2.2. Error of Closed-Form Solution
IM (g)
The above closed-form solution (Eqns 5 and 6) are the
basis of the demand and capacity factor design (DCFD) Figure 2. 5% damp spectral acceleration seismic hazard curve for
method for performance-based seismic design and T = 1s at Central Los Angeles, California, and the first and second
assessment in the SAC/FEMA project (Jalayer 2003). order approximations using Eqns 4 and 9 in the vicinity of
However, it is worth noting this solution employs IM = 0.18g

Advances in Structural Engineering Vol. 18 No. 10 2015 1657

 Seismic Reliability and Risk Assessment of Structures Based on Fragility Analysis A Review

shortcomings of this approach were the high sensitivity of

H ( IM ) k0 exp( k2ln 2 IM k1ln IM ) (9)
the predicted results to the level of confidence chosen and
where k2 0 denotes the curvature of the seismic the need for distinguishing between aleatory and epistemic
hazard model in log-log space and k0, k1 > 0. uncertainties, which is a subjective and debatable
Subsequently, Eqn 5 is re-written (Vamvatsikos 2013), distinction (Liel et al. 2009). Cornell et al. (2002)
as follows challenged the idea that using a mean estimation of PoF,
rather than the median estimate (Eqn 6), can capture the
effect of uncertainty in the same fragility function:
P[ DM LS ] = p k01 p
1 k12 2 (10) 1 k12
[ H ( IM LS
)] exp p 2 DM |IM
P 2
PoF = H (saC )exp ( DR
2
+ DU
2
+ CR
2
+ CU ) (12)
2 b 2 b
2

where
In this regard, the dispersion measures DR and CR,
which were denoted as D|Sa and C in the previous
1 section, highlight the effect of randomness while, DU
p=
1 + 2 k2 DM
2 (11) and CU reflect the dispersion measures according to an
| IM
epistemic uncertainty source. Albeit, Yun et al. (2002)
Furthermore, similar to the approximation of seismic clarified that this generic development of uncertainty
hazard, alternative distribution models have been can be expanded to involve a larger set of dispersions
suggested for the DM-IM relation such as a linear model such as uncertainty associated with damping estimation
in semi-log space (Romo et al. 2013). (damping), live load (live load), and material properties
(material properties).
3. UNCERTAINTY QUANTIFICATION AND There are two major approaches in uncertainty
HYBRID SIMULATION quantification. In the forward approach (uncertainty
In the presence of uncertainties, the seismic risk to civil propagation), the output results of numerical modelling
structures cannot be eliminated. Nevertheless, it must be and dynamic analysis are evaluated to reflect the
managed by engineers, code-writers and other regulatory uncertain input data usually by a mean and variance
authorities (Ellingwood and Kinali 2009). In this regard, it measure (Lee and Chen 2009). Majority of uncertainty
is essential to primarily distinguish between the types of analysis techniques was developed for uncertainty
uncertainties in order to identify their sources (Kiureghian propagation problems (Lee and Chen 2009). In this
and Ditlevsen 2009). The uncertainty which is assumed to regard, different sources of uncertainty were investigated,
be the intrinsic randomness of a phenomenon is referred to in relation to record selection (Padgett and DesRoches
as aleatoric and the epistemic uncertainty is the one which 2007), material strength and structural component
is caused by lack of knowledge or seismic data, imperfect stiffness (Somja et al. 2013; Celik and Ellingwood 2010;
modelling, and simplification (Celik and Ellingwood Vamvatsikos and Fragiadakis 2010; Liel et al. 2009;
2010). For example, it is generally understood that the Padgett and DesRoches 2007; Montiel and Ruiz 2007),
seismic hazard curves (maps), which are provided by the structural damping ratio (Celik and Ellingwood 2010),
seismic codes, represent the mean seismic hazard. The gross geometry (Padgett and DesRoches 2007), structural
assumption of hazard levels according to these curves capacity and system loads (Liel et al. 2009; Tesfamariam
represents the aleatoric uncertainty while the epistemic and Sadiq 2006; Montiel and Ruiz 2007), support or
uncertainty is associated with the plausible alternative connection conditions (Celik and Ellingwood 2010),
ground motion models, which are averaged to obtain the influences of partition walls and other non-structural
mean hazard (Abrahamson and Bommer 2005). In seismic elements, and two-dimensional modelling of three-
risk analysis, the sources of uncertainty which arise from dimensional structures (Ellingwood 2001).
the ambiguities in structural and modelling parameters are
mainly attributed to the epistemic uncertainty (Pat- 3.2. Hybrid Simulation
Cornell 1996). The other approach in uncertainty analysis, namely the
hybrid simulation, is an inverse uncertainty
3.1. Uncertainty Modelling quantification in which the discrepancies between the
The conventional approach for uncertainty quantification experimental measurements of a system and the
in fragility analysis was the application of confidence numerical simulation results, and subsequently the
interval method (Ellingwood and Kinali 2009). The uncertainties in the model, are estimated (model

1658 Advances in Structural Engineering Vol. 18 No. 10 2015

M. G. Sfahani, Hong Guan and Yew-Chaye Loo

calibrating) (Matthies 2007). However this approach is Experimental studies can also be a source data for
more difficult to adopt than the forward uncertainty calibrating analytical fragility curves. In this regard,
propagation, nevertheless, hybrid simulation has the Kwon and Elnashai (2006) used the results of shaking
advantage that the numerical modelling results can be table test of a one-third scale ordinary three-story
updated by the measured data (Mottershead and moment resisting RC frame to calibrate the analytical
Friswell 1993). The resulting fragility curves, in this fragility curves considering material and ground motion
case, are called the hybrid fragility curves. uncertainties. The use of large scale experimental tests,
In this regard, Singhal and Kiremidjian (1998) for calibrating and updating the analytical fragility
presented a Bayesian approach to update analytical curves, has been considered by other researchers as well
fragility functions of RC frame buildings using the (Banerjee and Shinozuka 2008a; Kramar et al. 2010;
observational damage data from the 1994 Northridge Banerjee and Chi 2013). In comparison to the use of
earthquake. Nevertheless, because of the limited data empirically collected data for updating the numerical
covering only six levels of ground motion intensities, model of the structure, using a large scale experimental
the fragility functions were updated at discrete points. test results in a more reliable updated model because
Similarly, Kappos et al. (1998) used the empirical 1978 sufficient detailed data can be obtained. Nevertheless,
Thessaloniki (Greece) earthquake data to validate the the cost of large scale testing is worth noting. To
numerically calculated damage fragility matrices for overcome this shortcoming, a Bayesian approach was
seismic loss evaluation of reinforced concrete buildings. proposed, by Koutsourelakis (2010), which has the
In addition, Karim and Yamazaki (2001) verified the advantage of estimating the empirical seismic data
numerically constructed fragility curves for bridge piers samples for the whole range of values of ground motion
against the empirical fragility curves, which were intensity. Subsequently, unlike the approach proposed
developed based on damage data from the 1995 Kobe by Singhal and Kiremidjian (1998), which requires at
earthquake. Moreover, Banerjee and Shinozuka (2008b) least one measured seismic data sample for each level of
developed a mechanistic model for investigating the ground shaking, the entire fragility function can be
seismic vulnerability of concrete bridges in the form of updated with limited measured data. Furthermore, Lin et
fragility curves. The optimal choices of the threshold al. (2012) recently proposed a procedure for performing
limit-state values, in this model, were calibrated by the hybrid fragility analysis, using multiple data sources, for
empirical fragility curves which were derived based on an integrated seismic risk assessment (see Figure 3). A
the 1994 Northridge earthquake damage data. typical procedure like Figure 4 may have been

Model
improvement

Structural
Real structure
real EQ event

response
measurement Finite element
Seosor
& model
instrumentation
system updating, I
identification

Experimental
improvement

testing
Source

Ground motion Seismic hazard &

measurement analysis updated
hybrid
modelling
Reliable impact
assessment

Seismic Finite element

Impact fragility model
assessment analysis updating, II

Figure 3. The proposed integrated framework for seismic risk assessment by Lin et al. (2012)

Advances in Structural Engineering Vol. 18 No. 10 2015 1659

 Seismic Reliability and Risk Assessment of Structures Based on Fragility Analysis A Review

in which the upper and lower bounds of the system

Limit-state j 1

Limit-state j
Component 2 are as follows:

Interval space Ptypical (j) n

limit-state j max P ( Failcomponent i ) P ( Failsystem )
i =1
(14)
limit-state j 1

Pinterval (j & j 1)
(
1 in=1 1 P Failcomponent i )
Interval space An important issue which should be considered here
is the effect of the interval space between the two
typical consecutive seismic demand limit-states j and j
Componet 1
1, as shown in Figure 4, on the probability of the
Figure 4. The borders of consecutive limit-states and the domains system-level failure event. To investigate this issue,
of typical and augmented failure probability of the whole system Duen~as-Osorio and Padgett (2011) described that the
probability of the overall performance of the system
exceeding the seismic demand limit-state j, is coupled
developed to encourage the researchers, in the various with the simultaneous failure of all important
areas of earthquake risk analysis, to investigate the components in the previous limit-state j 1. Therefore,
seismic risk problems across different engineering a wider confidence bound was proposed in order to
disciplines, however, raising the interests is not so assess the system-level reliability at the limit-state j,
simple when taking the costs of the whole process into namely the augmented system-level failure (Dueas-
consideration. Osorio and Padgett 2011):

3.3. System-Level Reliability Paugmented j = Ptypical j + Pint erval j& j 1 (15)

An important issue in fragility analysis is to assess the
reliability and safety of structures for the whole where Ptypical and Pinterval are illustrated in Figure 5. The
structural system (Ditlevsen and Madsen 1996). A effect of an augmented failure event on a typical system
major difficulty in this regard is defining and computing was quantified to be a 4%20% change in the median
the system-level failure event probability, under seismic moderate limit-state in fragility analysis (Dueas-Osorio
hazards, as a function of component-level failure events. and Padgett 2011).
Initial approaches to establish system-level failure To improve the reliability bounds, Monte Carlo
events relied on the expert opinion (judgmental fragility Simulation (MCS) had been used (Nielson and
analysis) and reflecting the overall performance of the DesRoches 2007). Nevertheless, system-level failure
system, in terms of residual capacity as a function of
structural damage (ATC-40 1996). In this regard,
seismic system failure events are defined based on In (DM) = ln (a) + b ln (IM)
empirical data from postearthquake inspection to
correlate structural damage and overall performance of N (In (DM ))
the system. However, as discussed before, using

judgmental fragility analysis has major shortcomings.

Early attempts for evaluating analytical system-level
In (DM)

reliability (fragility curve) were based on computation

of confidence bounds (from statistically-independent to

fully-coupled) to investigate the correlated performance
of the system components (Song and Kiureghian 2003;
Choi et al. 2004). These studies revealed that the total
system reliability at a particular limit-state is the union
of the reliabilities of the system components in the same
limit-state: In (IM)

Figure 5. Distribution of the seismic demand in response to

P[ Failsystem ] = U in=1P[ Failcomponent i ] (13) shaking intensity in the log-log space

1660 Advances in Structural Engineering Vol. 18 No. 10 2015

M. G. Sfahani, Hong Guan and Yew-Chaye Loo

definitions range from considering a single structural 4.1. Optimal Intensity Measure
component failure to simultaneous failure of all The list of IMs which was proposed for earthquake
components. Thus, the use of traditional MCS may engineering is already large and seems to be growing
become computationally demanding given that each (Ye et al. 2013; Giovenale et al. 2004; Baker and
definition of system-level failure events requires a full Cornell 2005; Cimellaro et al. 2011; Luco and Cornell
numerical simulation. An alternative to reduce the 2007; Padgett et al. 2008). Nevertheless, it is important
computational cost is to use MCS in combination with to choose an appropriate and comprehensive IM which
Response Surface Methodology (RSM), because the is correlated with the structural performance. In this
simulation is performed on a polynomial response regard, several features are defined for an optimal IM.
surface function rather than a complex dynamic model Among these, the most significant and recognised factor
(Towashiraporn 2004). Seo and Linzell (2013) have is known as efficiency, which describes the
recently applied RSM for system-level fragility corresponding damage measure (DM) variation
analysis. However, in using RSM it is required to according to the given IM in which, a more efficient IM
choose the order of the RSM polynomial. In this regard, results in relatively less variability (Shome 1999). In
Stander and Craig (2002) criticised that while a two- other words, applying an efficient IM produces a
polynomial function (linear approximations) results in a comparatively smaller dispersion measure of the
rapid solution, it may only be useful within a certain calculated seismic demand in the non-linear dynamic
subregion of the design space. On the other hand, in analyses of the structure which was previously defined
addition to the higher possibility of bias error as DM|IM in this study. The benefit in this case is the
for a nonlinear estimation, a higher order approxi- improvement of the reliability of analysis by reaching a
mation (quadratic polynomials) is time consuming narrower confidence interval band () of the conditional
computationally (Stander and Craig 2002). Also, a median DM value for a given IM level as shown in
weighting scheme of component failure can be used for Figure 5.
establishing system-level failure based on the To evaluate the efficiency of an IM the most direct
contribution of load-carrying capacity, repair cost, or and rigorous approach is performing the correlation
impact on functionality (Tesfamariam and Sadiq 2008; analysis of the DM-IM, for different IMs, through
Zhang and Huo 2009). This approach indicates that all nonlinear time-history analysis (Ye et al. 2013).
the structural components do not influence the system- Nevertheless, the main difficulty of this exercise is that
level failure equally. Nevertheless, a difficulty in a certain number of nonlinear time-history analyses
application of this approach is establishment of weights must be re-scaled and re-run every time a new candidate
for different components because it may bring about IM is proposed. An alternative to reduce the amount of
additional uncertainties. Moreover, a matrix-based computation is to use Incremental Dynamic Analysis
system reliability (MSR) method has been proposed and (IDA). An IDA curve is a continuous plot of DM versus
applied by Song and Kang (2009). Nevertheless, using IM which is derived by scaling a record up or down to
this method the total number of system failure events, different IM levels (Vamvatsikos and Cornell 2002). In
from no component failure to simultaneous failure of all addition, a stripe analysis technique, as described by
components, is 2N where N denotes the number of Jalayer (2003), can be employed to summarise the IDA
components. This means that considering an additional curves properly in which stripes of DM values can be
structural component, for the system-level fragility generated at any IM values containing as many DM
analysis, the total probable system-level failure events values as the number of records. Subsequently, it is
are doubled and the computation costs rise possible to change the IM in terms of the representative
exponentially. DM-IM curves without re-running the analyses.
Applying the stripe analysis is, also, beneficial in the
4. GROUND MOTION INTENSITY MEASURE way that it does not introduce any bias in the results of
The ground motion intensity, or simply the intensity DM values using different IM (Jalayer 2003). Also, an
measure (IM), is the most significant index to describe approximate method has been introduced by Giovenale
the earthquake characteristics from the structural et al. (2004) which estimates the DM-IM dispersion to
engineering point of view. Considering the fragility decrease the efforts of re-running the analysis and
analysis formulation (Eqns 1 to 6), it is understood that computational costs.
the seismic fragility curve is a function of IM. Thus, a Padgett et al. (2008) challenged the idea that the
more appropriate IM, which represents more precisely direct correlation between the IM and DM should also
the ground shaking forces and characteristics, results in be considered for assessing the efficiency of an IM. As
more accurate fragility curves. such, a new feature, namely practicality, was defined

Advances in Structural Engineering Vol. 18 No. 10 2015 1661

 Seismic Reliability and Risk Assessment of Structures Based on Fragility Analysis A Review

in which an IM is considered practical when the that the number of required records to obtain a stable
resulting DM is directly dependent upon it. It was estimate of the response is related to the type of DM
explained that the regression parameter b (see Eqn 2 used in the analysis. Reyes and Kalkan (2012) also
and Figure 5) is the index for practicality wherein, a demonstrated that the procedure of ASCE/SEI 7-10 is
steeper linear IM-DM distribution implies a too conservative in the case of using less than seven
more practical IM. Subsequently, by combining records (choosing the maximum DM value). In addition,
practicality with efficiency, Padgett et al. (2008) it was reported that increasing the number of records
proposed the proficiency for assessing optimal IMs. from 7 to 10 has little effect on the accuracy of the
In this regard, the modified dispersion measure nonlinear analysis results. Moreover, the study by
shows the practicality which is defined Cimellaro et al. (2011) indicates that the minimum
DM|IM number of records for an accurate estimation of seismic
by = . demand can be reduced when Sa is used instead of PGA.
b Also, in addition to the optimum number of required
Furthermore, an optimal IM should yield DMs records for an accurate nonlinear dynamic analysis,
conditionally independent of earthquake characteristics some qualitative factors have been addressed which
(e.g. magnitude (M) and epicentral distance (R)) and site influence the record selection such as soil profile (Lee et
characteristics (e.g. soil and fault type). For example, al. 2000), motion duration (Hancock and Bommer 2007;
considering Eqn 1, P [DM LS|IM] is the MAF of Iervolino et al. 2006), earthquake magnitude (M) and
exceeding each IM value. There is no need to condition distance (R) (Elefante et al. 2010), acceleration to
Eqn 1 upon additional parameters for a sufficient IM velocity ratio (a/v) (Tso et al. 1992), seismotectonic and
while an insufficient IM requires re-writing the geophysical parameters (Kawaga et al. 2004).
conditional term in the form of P [DM LS|IM, M, R,
ect.]. Therefore, a factor namely sufficiency was 5. SEISMIC DAMAGE MEASURE
defined which is evaluated by performing a regression Similar to the significance of IM discussed above,
analysis on the PSDM residuals relative to the ground seismic damage measure (DM) largely influences the
motion characteristics, M and R (Luco and Cornell fragility function. This is because the prescribed limit-
2007). On the other hand, it is noteworthy that hazard states (LS) by the seismic codes, which impact the level
maps are normally available for peak ground of calculated failure probability, are highly uncertain
acceleration (PGA) or alternatively for a discrete and the amount of such an uncertainty varies from one
number of values of spectral acceleration (Sa). Although DM to another (Jalayer 2003). Excessive fragility
novel hazard curves can be calculated for a more studies have been accomplished using the peak transient
efficient new IM through PSHA, the calculation of the displacement of a structure, or simply its lateral story
corresponding hazard curves might be excessively drift, as the DM under seismic loads (Karavasilis et al.
demanding. Therefore, hazard computability was 2006; Romo et al. 2013; Gnay and Mosalam 2013).
defined by Giovenale et al. (2004) as the effort required However, lateral drift is unable to signal the location
in order to determine the hazard curve in PSHA. It was and intensity of seismic damage as well as the residual
explained that, however, a particular IM may be more capacity of the structure. On the other hand, inelastic
efficient or proficient; it may therefore be less desirable seismic displacements, i.e. the residual deformation,
because of its hazard computability. play an important role in defining the seismic
performance level by providing an indicator for the
4.2. Ground Motion Record Selection effort required to straighten and repair structures.
In addition to the defined metrics, as the characteristics Consequently, the use of residual drift demand, as an
of an optimal IM, the number of ground motion records alternative to DM, has been recommended for seismic
which should be used for scaling and nonlinear dynamic reliability assessment of structures (Ruiz-Garca and
analysis purposes is a factor that influences both the Miranda 2010; Rveillre et al. 2012; Bojrquez and
result accuracy and the computational time. In this Ruiz-Garca 2013). The structural ductility ratio has also
regard, according to the ASCE/SEI 7-10 (2010), a been utilised for seismic risk analyses to identify the
minimum of three ground motion records is required, location of possible damage for strengthening purposes
and in the case of using less than seven ground motion (Hong and Hong 2007; Mehani et al. 2013; Freddi et al.
records, the design value of DM should be taken as the 2013). However, all the above mentioned DMs still
maximum of the resulted DMs. For a higher number of cannot capture the cyclic effects of seismic loads.
ground motion records the average of the DMs was The dynamic nature of seismic loads asserts the
proposed. Nevertheless, Hancock et al. (2008) discussed necessity of provisions to include the impacts of

1662 Advances in Structural Engineering Vol. 18 No. 10 2015

M. G. Sfahani, Hong Guan and Yew-Chaye Loo

low-cycle fatigue and energy dissipation on structural effects of airborne chloride in marine environments with
capacity. A number of DMs have been suggested in the seismic reliability estimation of RC bridge piers. In
literature to reflect such an impact, namely the addition to the environmental deteriorating condition, as
cumulative damage indices (Krawinkler and Zohrei a long-term effect of ageing of structures, it was
1983; Fajfar 1992; Park and Ang 1985; Cosenza et al. suggested that the influences of accelerating
2009). These indices are particularly most appropriate to deteriorating factors (e.g. de-icing salt) should be taken
be used when investigating the sustainability of into consideration. To reflect the ageing effects,
structures in aftershocks (Luco et al. 2004), or when a Vamvatsikos and Dolsek (2011) proposed a closed-form
previously damaged structure affected by a sequence of solution, based on the equivalent constant rate (ECR)
earthquakes is considered (Mander and Rodgers 2013). which was introduced by Yeo and Cornell (2009) for
Moreover, the influence of ground motion duration, post-quake environment decision-making. In addition,
which is assumed to affect the total energy induced by Ghosh and Padgett (2010) developed time-dependent
earthquake loads (Hancock and Bommer 2007), can be seismic fragility curves in which instead of decreasing
investigated through the cumulative damage theory the structural capacity over time, the seismic demand
(Raghunandan and Liel 2013; Iervolino et al. 2006; was considered to be increasing as a function of time.
Teran-Gilmore and Jirsa 2007; Bojrquez et al. 2008). The impact of aging was then illustrated not only for the
Subsequently, fragility curves have been developed fragility of structural components but also for the
using the cumulative damage indices as the DM in a reliability of the whole system.
fragility function (Hwang and Huo 1994; Singhal and
Kiremidjian 1996; Karim and Yamazaki 2001; Khaloo 6.2. Soil-Structure Interaction
and Tonekaboni 2013; Bojrquez et al. 2008). Bedrock configuration and soil characteristics play a
significant role in preserving the general stability and
6. JOINT IMPACT IN FRAGILITY ANALYSIS system reliability of a structure. Seismic fragility
In infrastructure management, in order for the structure analysis of structures is often performed without
to preserve its sustainability during its life-cycle, it is consideration of soil-structure interaction (SSI) in that
particularly important to analyse the failure risk not only the base of the structure is usually assumed to be fixed.
for the seismic events but also for other hazards. Recent The most common approach to consider the flexibility
advances in computational methods have provided the of the foundation is using a lumped spring model where
possibility of performing compound risk analysis of the properties are accounted for based on the field tests.
structures in highly critical conditions. In this regard, A seismic fragility analysis of bridges, carried out by
some researchers attempted to combine the seismic Kwon and Elnashai (2007), shows that the embankment
fragility analysis of structures with additional hazards and abutment behave highly nonlinear with soil
probability. material model which is incomparable to the simplified
lumped spring model. On the other hand, soil
6.1. Environmental Deterioration liquefaction is a serious problem during an earthquake
Using the current guidelines for seismic performance which may result in ground subsidence, loss of bearing
assessment of structures (FEMA 273 1997; FEMA 274 capacity, shifting/tilting of foundation, pile failure,
1997; FEMA 356 2000; ASCE 41-06 2007), the slope/embankment failure, abutment failure, lateral
structural capacity is calculated as time-independent. spreading, and subsequently the instability of the
This issue is at odds with the accurate condition of structure (Bignell and LaFave 2009).
ageing structures in deteriorating environmental Koutsourelakis et al. (2002) established seismic
situation which influences the structural capacity. Choe fragility curves for soil-structure systems due to soil
et al. (2008) primarily developed a seismic drift and liquefaction by considering the variability of soil
shear force capacity model for corroded RC structures. parameters. Zhang et al. (2008) evaluated the seismic
In their study, the combination of time-dependent fragility curves of structures subjected to both dynamic
corrosion rate and seismic fragility was presented by loads and liquefaction-induced lateral spreading, based
calculating the chloride-induced corroded structural on nonlinear dynamic and equivalent static procedures.
capacity and related fragility curve in various time It was reported that structures demonstrate different
intervals. Kumar et al. (2009) considered the effects of seismic capacities under seismic loads and lateral
chloride-induced corrosion of steel reinforcement in RC spreading. Other researchers discussed that the seismic
bridges for seismic fragility analysis with special fragility due to soil liquefaction should be taken into
attention to the seismic damage accumulation. Akiyama consideration the same way as the structural
et al. (2011) investigated a procedure for integrating the component fragility for the system-level failure

Advances in Structural Engineering Vol. 18 No. 10 2015 1663

 Seismic Reliability and Risk Assessment of Structures Based on Fragility Analysis A Review

fragility analysis (Bignell and LaFave 2010; Aygn et quantification and the methods for system-level
al. 2011). A case study by Padgett et al. (2010) shows reliability assessments has revealed that the current
that the seismic fragility, in this regard, is influenced methods either demand large time/memory or use
by the variation in some of the soil modelling simplifying assumptions.
parameters such as undrained shear strength of soil, Moreover, the studies on fragility analysis using
soil shear modulus, gap between deck and abutment, alternative IM and DM have been reviewed. To present
ultimate capacity of soil, and fixed and expansion the most optimal IMs, the list of those available is
bearing coefficients of friction. Moreover, Wang et al. reduced to the Peak Ground Acceleration (PGA) and the
(2013) reported that inclusion of vertical ground spectral acceleration at the fundamental period of a
motions (VGM) in seismic fragility analysis of structure Sa(T1). The latter, being a structure-specific
coupled structure-soil-foundation systems, with IM, is proven to be more efficient than the former.
liquefaction potential, influences the failure Albeit, using Sa(T1), the contributions of higher modes
probability that neglecting VGM results in to the overall dynamic response and the increase of the
underestimating the probability of failure. fundamental period of structure associated with non-
linear behaviour are ignored.
6.3. Multi-Hazard Analysis The principal uncertainty in fragility analysis,
Although in the most of geographical locations one however, arises from the prescriptive limit-states (LS)
natural hazard is dominant, in some certain areas and structural capacities. More seismic data are required
multiple hazards threaten the safety of structures. The to be collected, from real seismic events and/or full-size
multi-hazard analysis is the state-of-the-art in structural experimental tests, to confidently express the structural-
performance-based design since it considers more than a specific LS. Meanwhile, the use of confidence bound
single hazard, during the life-cycle of a structure, to methods seems to be an appropriate approach for
increase the structural safety and reduce subsequent reflecting the existing uncertainty of LS in fragility
costs of rehabilitation. In addition, a multi-hazard curves. Also, using threshold drift values for evaluating
analysis can support the improvements in design and cumulative LS in some studies is in contrast with the
construction practices, insurance underwriting, and theories of low-cycle fatigue. Therefore, energy
planning for the communitys response to disasters. For dissipation and cumulative LS must be re-evaluated by
this reason, additional hazard sources such as snow further research.
loads (Lee and Rosowsky 2006), hurricanes (Li and With regard to the DM selection, gradual changes of
Ellingwood 2009), and floods (Banerjee and Prasad the modal parameters will be of great significance in
2011; Dong et al. 2013) were considered by some future investigations. Such a DM also allows the study
researchers in addition to the seismic fragility analysis. of the effective contribution of ground motion duration
Nevertheless, the main challenge in multi-hazard on seismic responses by plotting the changes of modal
analysis is optimisation of the allocation of hazard parameters against the ground motion record time,
mitigation resources. giving consideration for varying time interval.
Furthermore, in depth studies of multi-hazard analysis
7. SUMMARY, RECOMMENDATIONS AND through fragility functions are highly acknowledged,
PROSPECTIVES however the circumstances of occurrence of coupled
Extensive research on seismic risk and reliability has hazards need to be justified more clearly.
brought about a significant advancement for
mitigation and recovery, as well as preparedness and 8. CONCLUSION
response against seismic hazard. However, there still A review of literature is presented herein covering the
remain some limitations with regard to the calculation current methods of practice for seismic reliability and
process of seismic fragility and structural reliability. risk assessment of structures based on fragility analysis,
Although the use of power model approximations for and recent improvements of these methods of analysis.
estimating seismic hazard and DM-IM distribution Major achievements and inherent limitations have also
make the closed-form solution quite rapid and been discussed. Research on seismic fragility analysis is
appealing, the inherent errors of such a solution are still very active because the fragility curves can be used
non-negligible. Nevertheless, these errors can be in risk assessment and mitigation planning for a
reduced by employing a bilinear or trilinear DM-IM communitys long-term strategy to reduce seismic
distribution or higher-order (e.g. second-order or damage and losses. The research gaps identified have
fractional-order) approximations of seismic hazard. In motivated the proposed area of study and assisted in
addition, the discussion on the uncertainty formulating the future research directions.

1664 Advances in Structural Engineering Vol. 18 No. 10 2015

M. G. Sfahani, Hong Guan and Yew-Chaye Loo

REFERENCES southern illinois wall pier supported highway bridges,

Abrahamson, N.A. and Bommer, J.J. (2005). Probability and Earthquake Engineering & Structural Dynamics, Vol. 39, No. 7,
uncertainty in seismic hazard analysis, Earthquake Spectra, Vol. pp. 709729.
21, No. 2, pp. 603607. Bojorquez, E., Ruiz, S.E. and Teran-Gilmore, A. (2008).
Akiyama, M., Frangopol, D.M. and Matsuzaki, H. (2011). Life- Reliability-based evaluation of steel structures using energy
cycle reliability of RC bridge piers under seismic and airborne concepts, Engineering Structures, Vol. 30, No. 6, pp.
chloride hazards, Earthquake Engineering & Structural 17451759.
Dynamics, Vol. 40, No. 15, pp. 16711687. Bojrquez, E. and Ruiz-Garca, J. (2013). Residual drift demands in
ASCE/SEI 41-06 (2007). Seismic Rehabilitation of Existing moment-resisting steel frames subjected to narrow-band
Buildings, American Society of Civil Engineers, Reston, USA. earthquake ground motions, Earthquake Engineering &
ASCE/SEI SEI 7-10 (2010). Minimum Design Loads for Buildings Structural Dynamics, Vol. 42, No. 11, pp. 15831598.
and Other Structures, American Society of Civil Engineers, Bothara, J.K., Dhakal, R.P. and Mander, J.B. (2010). Seismic
Reston, USA. performance of an unreinforced masonry building: An
Aslani, H. and Miranda, E. (2005). Probability-based seismic experimental investigation, Earthquake Engineering &
response analysis, Engineering Structures, Vol. 27, No. 8, Structural Dynamics, Vol. 39, No. 1, pp. 4568.
pp. 11511163. Bradley, B.A. and Dhakal, R.P. (2008). Error estimation of closed-
ATC-40 (1996). Seismic Evaluation and Retrofit of Concrete form solution for annual rate of structural collapse, Earthquake
Buildings Volume 1, Applied Technology Council, Redwood Engineering & Structural Dynamics, Vol. 37, No. 15, pp.
City, USA. 17211737.
Aygn, B., Dueas-Osorio, L., Padgett, J.E. and Desroches, R. (2011). Carballo, J.E. and Cornell, C.A. (2000). Probabilistic Seismic
Efficient longitudinal seismic fragility assessment of a Demand Analysis: Spectrum Matching and Design, Report No.
multispan continuous steel bridge on liquefiable soils, Journal of RMS-41, 313, Stanford University, USA.
Bridge Engineering, ASCE, Vol. 16, No. 1, pp. 93107. Celik, O.C. and Ellingwood, B.R. (2010). Seismic fragilities for
Baker, J.W. and Cornell, C.A. (2005). A vector-valued ground non-ductile reinforced concrete frames role of aleatoric and
motion intensity measure consisting of spectral acceleration and epistemic uncertainties, Structural Safety, Vol. 32, No. 1,
epsilon, Earthquake Engineering & Structural Dynamics, Vol. pp. 112.
34, No. 10, pp. 11931217. Choe, D.E., Gardoni, P., Rosowsky, D.V. and Haukaas, T. (2008).
Banerjee, S. and Chi, C. (2013). State-dependent fragility curves of Probabilistic capacity models and seismic fragility estimates for
bridges based on vibration measurements, Probabilistic RC columns subject to corrosion, Reliability Engineering &
Engineering Mechanics, Vol. 33, pp. 116125. System Safety, Vol. 93, No. 3, pp. 383393.
Banerjee, S. and Prasad, G.G. (2011) Analysis of bridge Choi, E.S., DesRoches, R. and Nielson, B. (2004). Seismic fragility
performance under the combined effect of earthquake and flood- of typical bridges in moderate seismic zones, Engineering
induced scour, Vulnerability, Uncertainty, and Risk: Analysis, Structures, Vol. 26, No. 2, pp. 187199.
Modelling, and Management, Hyattsville, Maryland, USA, pp. Cimellaro, G.P., Reinhorn, A.M., DAmbrisi, A. and De Stefano,
889896. M. (2011). Fragility analysis and seismic record selection,
Banerjee, S. and Shinozuka, M. (2008a). Experimental verification Journal of Structural Engineering, ASCE, Vol. 137, No. 3,
of bridge seismic damage states quantified by calibrating pp. 379390.
analytical models with empirical field data, Earthquake Cornell, C.A. (1968). Engineering seismic risk analysis, Bulletin
Engineering & Engineering Vibration, Vol. 7, No. 4, of the Seismological Society of America, Vol. 58, No. 5, pp. 1583.
pp. 383393. Cornell, C.A., Jalayer, F., Hamburger, R.O. and Foutch, D.A.
Banerjee, S. and Shinozuka, M. (2008b). Mechanistic (2002). Probabilistic basis for 2000 sac federal emergency
quantification of RC bridge damage states under earthquake management agency steel moment frame guidelines, Journal of
through fragility analysis, Probabilistic Engineering Mechanics, Structural Engineering, ASCE, Vol. 128, No. 4, pp. 526533.
Vol. 23, No. 1, pp. 1222. Cornell, C.A. and Krawinkler, H. (2000). Progress and challenges
Basoz, N.I., Kiremidjian, A.S., King, S.A. and Law, K.H. (1999). in seismic performance assessment, PEER Center News,
Statistical analysis of bridge damage data from the 1994 pp. 13.
Northridge, Ca, Earthquake, Earthquake Spectra, Vol. 15, No. 1, Cosenza, E., Manfredi, G. and Polese, M. (2009). Simplified
pp. 2554. method to include cumulative damage in the seismic response of
Bazzurro, P. (1998). Probabilistic Seismic Demand Analysis, PhD single-degree-of-freedom Ssystems, Journal of Engineering
Thesis, Stanford University, USA. Mechanics, ASCE, Vol. 135, No. 10, pp. 10811088.
Benjamin, J.R. and Cornell, C.A. (1975). Probability, Statistics and DAyala, D.F. and Paganoni, S. (2011). Assessment and analysis of
Decision for Civil Engineers, McGraw-Hill, New York, USA. damage in Laquila historic city centre after 6th April 2009,
Bignell, J. and LaFave, J. (2009). Analytical fragility analysis of Bulletin of Earthquake Engineering, Vol. 9, No. 1, pp. 81104.

Advances in Structural Engineering Vol. 18 No. 10 2015 1665

 Seismic Reliability and Risk Assessment of Structures Based on Fragility Analysis A Review

Ditlevsen, O. and Madsen, H.O. (1996). Structural Reliability Gnay, S. and Mosalam, K.M. (2013). Peer performance-based
Methods, John Wiley & Sons, Chichester, UK. earthquake engineering methodology, revisited, Journal of
Dong, Y., Frangopol, D.M. and Saydam, D. (2013). Time-variant Earthquake Engineering, Vol. 17, No. 6, pp. 829858.
sustainability assessment of seismically vulnerable bridges Hancock, J. and Bommer, J.J. (2007). Using spectral matched
subjected to multiple hazards, Earthquake Engineering & records to explore the influence of strong-motion duration on
Structural Dynamics, Vol. 42, No. 10, pp. 14511467. inelastic structural response, Soil Dynamics and Earthquake
Dueas-Osorio, L. and Padgett, J.E. (2011). Seismic reliability Engineering, Vol. 27, No. 4, pp. 291299.
assessment of bridges with user-defined system failure events, Hancock, J., Bommer, J.J. and Stafford, P.J. (2008). Numbers of
Journal of Engineering Mechanics, ASCE, Vol. 137, No. 10, pp. scaled and matched accelerograms required for inelastic dynamic
680690. analyses, Earthquake Engineering & Structural Dynamics, Vol.
Elefante, L., Jalayer, F., Iervolino, I. and Manfredi, G. (2010). 37, No. 14, pp. 15851607.
Disaggregation-based response weighting scheme for seismic Hazus. (2014). Hazus: The Federal Emergency Management
risk assessment of structures, Soil Dynamics and Earthquake Agencys (FEMAs) Methodology for Estimating Potential Losses
Engineering, Vol. 30, No. 12, pp. 15131527. from Disasters. http://www.fema.gov/hazus.
Ellingwood, B.R. (2001). Earthquake risk assessment of building Hong, H.P. and Hong, P. (2007). Assessment of ductility demand
structures, Reliability Engineering & System Safety, Vol. 74, No. and reliability of bilinear single-degree-of-freedom systems
3, pp. 251262. under earthquake loading, Canadian Journal of Civil
Ellingwood, B.R. and Kinali, K. (2009). Quantifying and Engineering, Vol. 34, No. 12, pp. 16061615.
communicating uncertainty in seismic risk assessment, Hwang, H.H.M. and Huo, J.R. (1994). Generation of hazard-
Structural Safety, Vol. 31, No. 2, pp. 179187. consistent fragility curves, Soil Dynamics and Earthquake
Fajfar, P. (1992). Equivalent ductility factors, taking into account Engineering, Vol. 13, No. 5, pp. 345354.
low-cycle fatigue, Earthquake Engineering & Structural Hwang, H.H.M. and Jaw, J.W. (1990). Probabilistic damage
Dynamics, Vol. 21, No. 10, pp. 837848. analysis of structures, Journal of Structural Engineering, ASCE,
Fajfar, P. and Dolsek, M. (2012). A practice-oriented estimation of Vol. 116, No. 7, pp. 19922007.
the failure probability of building structures, Earthquake Iervolino, I., Manfredi, G. and Cosenza, E. (2006). Ground motion
Engineering & Structural Dynamics, Vol. 41, No. 3, pp. duration effects on nonlinear seismic response, Earthquake
531547. Engineering & Structural Dynamics, Vol. 35, No. 1, pp. 2138.
FEMA 274 (1997a). NEHRP Commentary on the Guidelines for the Jalayer, F. (2003). Direct probabilistic seismic analysis:
Seismic Rehabilitation of Buildings, Federal Emergency implementing non-linear dynamic assessments, PhD Thesis,
Management Agency, Washington, D.C., USA. Stanford University, CA, USA.
FEMA 273 (1997b). NEHRP Guidelines for the Seismic Kagawa, T., Irikura, K. and Somerville, P.G. (2004). Differences in
Rehabilitation of Buildings, Federal Emergency Management ground motion and fault rupture process between the surface and
Agency, Washington, D.C., USA. buried rupture earthquakes, Earth Planets and Space, Vol. 56,
FEMA 356 (2000). Pre-Standard and Commentary for the Seismic No. 1, pp. 314.
Rehabilitation of Buildings, Federal Emergency Management Kaplan, S., Perla, H.F. and Bley, D.C. (1983). A methodology for
Agency, Washington, D.C., USA seismic risk analysis of nuclear power plants, Risk Analysis, Vol.
Fragiadakis, M. and Vamvatsikos, D. (2010). Fast performance 3, No. 3, pp. 169180.
uncertainty estimation via pushover and approximate ida, Kappos, A.J., Stylianidis, K.C. and Pitilakis, K. (1998).
Earthquake Engineering & Structural Dynamics, Vol. 39, No. 6, Development of seismic risk scenarios bsed on a hybrid method
pp. 683703. of vulnerability assessment, Natural Hazards, Vol. 17, No. 2,
Freddi, F., Tubaldi, E., Ragni, L. and DallAsta, A. (2013). pp. 177192.
Probabilistic performance assessment of low-ductility Karavasilis, T.L., Bazeos, N. and Beskos, D.E. (2006). Maximum
reinforced concrete frames retrofitted with dissipative braces, displacement profiles for the performance based seismic design
Earthquake Engineering & Structural Dynamics, Vol. 42, No. 7, of plane steel moment resisting frames, Engineering Structures,
pp. 9931011. Vol. 28, No. 1, pp. 922.
Ghosh, J. and Padgett, J.E. (2010). Aging considerations in the Karim, K.R. and Yamazaki, F. (2001). Effect of earthquake ground
development of time-dependent seismic fragility curves, Journal motions on fragility curves of highway bridge piers based on
of Structural Engineering, ASCE, Vol. 136, No. 12, pp. numerical simulation, Earthquake Engineering & Structural
14971511. Dynamics, Vol. 30, No. 12, pp. 18391856.
Giovenale, P., Cornell, C.A. and Esteva, L. (2004). Comparing the Kennedy, R.P., Cornell, C.A., Campbell, R.D., Kaplan, S. and Perla,
adequacy of alternative ground motion intensity measures for the H.F. (1980). Probabilistic seismic safety study of an existing
estimation of structural responses, Earthquake Engineering & nuclear power plant, Nuclear Engineering and Design, Vol. 59,
Structural Dynamics, Vol. 33, No. 8, pp. 951979. No. 2, pp. 315338.

1666 Advances in Structural Engineering Vol. 18 No. 10 2015

M. G. Sfahani, Hong Guan and Yew-Chaye Loo

Khalfan, M. (2013). Fragility Curves for Residential Buildings in Liel, A.B., Haselton, C.B., Deierlein, G.G. and Baker, J.W. (2009).
Developing Countries: A Case Study on Non-Engineered Incorporating modeling uncertainties in the assessment of
Unreinforced Masonry Homes in Bantul, Indonesia, MSc Thesis, seismic collapse risk of buildings, Structural Safety, Vol. 31,
McMaster University, Canada. No. 2, pp. 197211.
Khaloo, A. and Tonekaboni, M. (2013). Risk based seismic Lin, S.L., Li, J., Elnashai, A.S. and Spencer Jr, B.F. (2012).
assessment of structures, Advances in Structural Engineering, Nees integrated seismic risk assessment framework (Nisraf),
Vol. 16, No. 2, pp. 307314. Soil Dynamics and Earthquake Engineering, Vol. 42,
Kircher, C.A., Whitman, R.V. and Holmes, W.T. (2006). Hazus pp. 219228.
earthquake loss estimation methods, Natural Hazards Review, Luco, N. (2002). Probabilistic Seismic Demand Analysis, Smrf
Vol. 7, No. 2, pp. 4559. Connection Fractures, and near-Source Effects, PhD Thesis,
Kiureghian, A.D. and Ditlevsen, O. (2009). Aleatory or Stanford University, USA.
epistemic? Does it matter?, Structural Safety, Vol. 31, No. 2, Luco, N., Bazzurro, P. and Cornell, C.A. (2004). Dynamic versus
pp. 105112. static computation of the residual capacity of a mainshock-
Koutsourelakis, P.S. (2010). Assessing structural vulnerability damaged building to withstand an aftershock, Proceedings of the
against earthquakes using multi-dimensional fragility surfaces: A 13th World Conference on Earthquake Engineering, Vancouver,
bayesian framework, Probabilistic Engineering Mechanics, Vol. British Columbia, Canada.
25, No. 1, pp. 4960. Luco, N. and Cornell, C.A. (2000). Effects of connection fractures
Koutsourelakis, S., Prvost, J.H. and Deodatis, G. (2002). Risk on smrf seismic drift demands, Journal of Structural
assessment of an interacting structure-soil system due to Engineering, ASCE, Vol. 126, No. 1, pp. 127136.
liquefaction, Earthquake Engineering & Structural Dynamics, Luco, N. and Cornell, C.A. (2007). Structure-specific scalar
Vol. 31, No. 4, pp. 851879. intensity measures for near-source and ordinary earthquake
Kramar, M., Isakovic, T. and Fischinger, M. (2009). Seismic ground motions, Earthquake Spectra, Vol. 23, No. 2,
collapse risk of precast industrial buildings with strong pp. 357392.
connections, Earthquake Engineering & Structural Dynamics, Mahin, S., Malley, J. and Hamburger, R. (2002). Overview of the
Vol. 39, No. 8, pp. 847868. FEMA/SAC program for reduction of earthquake hazards in steel
Krawinkler, H. and Zohrei, M. (1983). Cumulative damage in steel moment frame structures, Journal of Constructional Steel
structures subjected to earthquake ground motions, Computers Research, Vol. 58, No. 58, pp. 511528.
& Structures, Vol. 16, No. 14, pp. 531541. Mander, J.B. and Rodgers, G.W. (2013) Cyclic fatigue demands on
Kumar, R., Gardoni, P. and Sanchez-Silva, M. (2009). Effect of structures subjected to the 2010-2011canterbury earthquake
cumulative seismic damage and corrosion on the life-cycle cost of sequence, NZSEE Technical Conference and AGM, Wellington,
reinforced concrete bridges, Earthquake Engineering & New Zealand.
Structural Dynamics, Vol. 38, No. 7, pp. 887905. Matthies, H.G. (2007). Quantifying uncertainty: Modern
Kwon, O.S. and Elnashai, A. (2006). The effect of material and computational representation of probability and applications, In
ground motion uncertainty on the seismic vulnerability curves of Extreme Man-Made and Natural Hazards in Dynamics of
RC structure, Engineering Structures, Vol. 28, No. 2, pp. Structures, A. Ibrahimbegovic and I. Kozar, ed., pp. 105135.
289303. McGuire, R.K. (2008). Probabilistic seismic hazard analysis: Early
Kwon, O.S. and Elnashai, A. (2007) Fragility analysis of a bridge history, Earthquake Engineering & Structural Dynamics, Vol.
with consideration of soil-structure-interaction using multi- 37, No. 3, pp. 329338.
platform analysis, Structural Engineering Research Frontiers, Mehani, Y., Bechtoula, H., Kibboua, A. and Naili, M. (2013).
Long Beach, California, USA, pp. 114. Damage quantification of steel moment resisting frames using
Lee, K.H. and Rosowsky, D.V. (2006). Fragility analysis of ductility parameters, KSCE Journal of Civil Engineering, Vol.
woodframe buildings considering combined snow and earthquake 17, No. 6, pp. 13941402.
loading, Structural Safety, Vol. 28, No. 3, pp. 289303. Montiel, M.A. and Ruiz, S.E. (2007). Influence of structural
Lee, L.H., Lee, H.H. and Han, S.W. (2000). Method of selecting capacity uncertainty on seismic reliability of buildings under
design earthquake ground motions for tall buildings, The narrow-band motions, Earthquake Engineering & Structural
Structural Design of Tall Buildings, Vol. 9, No. 3, pp. 201213. Dynamics, Vol. 36, No. 13, pp. 19151934.
Lee, S.H. and Chen, W. (2009). A comparative study of uncertainty Mosleh, A. and Apostolakis, G. (1986). The assessment of
propagation methods for black-box-type problems, Structural probability distributions from expert opinions with an application
and Multidisciplinary Optimization, Vol. 37, No. 3, pp. 239253. to seismic fragility curves, Risk Analysis, Vol. 6, No. 4,
Li, Y. and Ellingwood, B.R. (2009). Framework for multihazard pp. 447461.
risk assessment and mitigation for wood-frame residential Mottershead, J.E. and Friswell, M.I. (1993). Model updating in
construction, Journal of Structural Engineering, ASCE, Vol. structural dynamics: A survey, Journal of Sound and Vibration,
135, No. 2, pp. 159168. Vol. 167, No. 2, pp. 347375.

Advances in Structural Engineering Vol. 18 No. 10 2015 1667

 Seismic Reliability and Risk Assessment of Structures Based on Fragility Analysis A Review

Nielson, B.G. and DesRoches, R. (2007). Seismic fragility Searer, G.R., Paret, T.F. and Freeman, S.A. (2008). ASCE-31 and
methodology for highway bridges using a component level ASCE -41: What good are they?, ASCE Structures Congress,
approach, Earthquake Engineering & Structural Dynamics, Vol. Vancouver, Canada, Vancouver, British Columbia, Canada, pp.
36, No. 6, pp. 823839. 18.
Padgett, J.E. and DesRoches, R. (2007). Sensitivity of seismic Seo, J. and Linzell, D.G. (2013). Use of response surface
response and fragility to parameter uncertainty, Journal of metamodels to generate system level fragilities for existing
Structural Engineering, ASCE, Vol. 133, No. 12, pp. 17101718. curved steel bridges, Engineering Structures, Vol. 52, No. 1, pp.
Padgett, J.E., Ghosh, J. and Dueas-Osorio, L. (2010). Effects of 642653.
liquefiable soil and bridge modelling parameters on the seismic Shinozuka, M., Feng, M.Q., Lee, J. and Naganuma, T. (2000).
reliability of critical structural components, Structure and Statistical analysis of fragility curves, Journal of Engineering
Infrastructure Engineering, pp. 119. Mechanics, ASCE, Vol. 126, No. 12, pp. 12241231.
Padgett, J.E., Nielson, B.G. and DesRoches, R. (2008). Selection of Shome, N. (1999). Probabilistic Seismic Demand Analysis of
optimal intensity measures in probabilistic seismic demand Nonlinear Structures, PhD Thesis, Stanford University, USA.
models of highway bridge portfolios, Earthquake Engineering Shome, N., Cornell, C.A., Bazzurro, P. and Carballo, J.E. (1998).
& Structural Dynamics, Vol. 37, No. 5, pp. 711725. Earthquakes, records, and nonlinear responses, Earthquake
Paret, T.F., Searer, G.R. and Freeman, S.A. (2011). ASCE 31 and Spectra, Vol. 14, No. 3, pp. 469500.
41: Apocalypse now, Proceedings of the 2011 Structures Singhal, A. and Kiremidjian, A.S. (1998). Bayesian updating of
Congress, Las Vegas, Nevada, USA, pp. 27412755. fragilities with application to RC frames, Journal of Structural
Park, Y.J. and Ang, A.H.S. (1985). Mechanistic seismic damage Engineering, ASCE, Vol. 124, No. 8, pp. 922929.
model for reinforced-concrete, Journal of Structural Singhal, A. and Kiremidjian, A.S. (1996). Method for probabilistic
Engineering, ASCE, Vol. 111, No. 4, pp. 722739. evaluation of seismic structural damage, Journal of Structural
Pat-Cornell, M.E. (1996). Uncertainties in risk analysis: Six levels Engineering, ASCE, Vol. 122, No. 12, pp. 14591467.
of treatment, Reliability Engineering & System Safety, Vol. 54, Somja, H., Nofal, S., Hjiaj, M. and Degee, H. (2013). Effect of the
No. 23, pp. 95111. steel material variability on the seismic capacity design of steel-
Porter, K., Kiremidjian, A.S. and LeGrue, J.S. (2001). concrete composite structures: A parametric study, Bulletin of
Assembly-based vulnerability of buildings and its use in Earthquake Engineering, Vol. 11, No. 4, pp. 10991127.
performance evaluation, Earthquake Spectra, Vol. 17, No. 2, pp. Song, J. and Kang, W.H. (2009). System reliability and sensitivity
291312. under statistical dependence by matrix-based system reliability
Porter, K., Kennedy, R.P. and Bachman, R. (2007). Creating method, Structural Safety, Vol. 31, No. 2, pp. 148156.
fragility functions for performance-based earthquake Song, J. and Kiureghian, A.D. (2003). Bounds on system reliability
engineering, Earthquake Spectra, Vol. 23, No. 2, pp. 471489. by linear programming, Journal of Engineering Mechanics,
Raghunandan, M. and Liel, A.B. (2013). Effect of ground motion ASCE, Vol. 129, No. 6, pp. 627636.
duration on earthquake-induced structural collapse, Structural Stander, N. and Craig, K.J. (2002). On the robustness of a simple
Safety, Vol. 41, pp. 119133. domain reduction scheme for simulation-based optimization,
Rveillre, A., Gehl, P., Seyedi, D. and Modaressi, H. (2012) Engineering Computations, Vol. 19, No. 34, pp. 431450.
Development of seismic fragility curves for mainshock- Teran-Gilmore, A. and Jirsa, J.O. (2007). Energy demands for
damaged reinforced-concrete structures, Proceedings of the 15th seismic design against low-cycle fatigue, Earthquake
World Conference on Earthquake Engineering, Lisbon, Portugal. Engineering & Structural Dynamics, Vol. 36, No. 3, pp.
Reyes, J.C. and Kalkan, E. (2012). How many records should be 383404.
used in an ASCE/SEI-7 ground motion scaling procedure ?, Tesfamariam, S. and Sadiq, R. (2006). Probabilistic risk analysis
Earthquake Spectra, Vol. 28, No. 3, pp. 12231242. using ordered weighted averaging (Owa) operators, Stochastic
Romo, X., Delgado, R. and Costa, A. (2013). Alternative closed- Environmental Research and Risk Assessment, Vol. 22, No. 1, pp.
form solutions for the mean rate of exceedance of structural limit 115.
states, Earthquake Engineering & Structural Dynamics, Vol. 42, Towashiraporn, P. (2004). Building Seismic Fragilities Using
No. 12, pp. 18271845. Response Surface Metamodels, PhD Thesis, Georgia Institute of
Rossetto, T. and Elnashai, A. (2003). Derivation of vulnerability Technology, USA.
functions for European-type RC structures based on Tso, W.K., Zhu, T.J. and Heidebrecht, A.C. (1992). Engineering
observational data, Engineering Structures, Vol. 25, No. 10, pp. implication of ground motion a/V ratio, Soil Dynamics and
12411263. Earthquake Engineering, Vol. 11, No. 3, pp. 133144.
Ruiz-Garca, J. and Miranda, E. (2010). Probabilistic estimation of Vamvatsikos, D. (2013). Derivation of new SAC/FEMA
residual drift demands for seismic assessment of multi-story performance evaluation solutions with second-order hazard
framed buildings, Engineering Structures, Vol. 32, No. 1, pp. approximation, Earthquake Engineering & Structural
1120. Dynamics, Vol. 42, No. 8, pp. 11711188.

1668 Advances in Structural Engineering Vol. 18 No. 10 2015

M. G. Sfahani, Hong Guan and Yew-Chaye Loo

Vamvatsikos, D. and Cornell, C.A. (2002). Incremental dynamic Yeo, G.L. and Cornell, C.A. (2009). Equivalent constant rates for
analysis, Earthquake Engineering & Structural Dynamics, Vol. post-quake seismic decision making, Structural Safety, Vol. 31,
31, No. 3, pp. 491514. No. 5, pp. 443447.
Vamvatsikos, D. and Dolsek, M. (2011). Equivalent constant rates Yun, S.Y., Hamburger, R.O., Cornell, C.A. and Foutch, D.A. (2002).
for performance-based seismic assessment of ageing structures, Seismic performance evaluation for steel moment frames,
Structural Safety, Vol. 33, No. 1, pp. 818. Journal of Structural Engineering, ASCE, Vol. 128, No. 4, pp.
Vamvatsikos, D. and Fragiadakis, M. (2010). Incremental dynamic 534545.
analysis for estimating seismic performance sensitivity and Zareian, F. and Krawinkler, H. (2010). Structural system parameter
uncertainty, Earthquake Engineering & Structural Dynamics, selection based on collapse potential of buildings in earthquakes,
Vol. 39, No. 2, pp. 141163. Journal of Structural Engineering, ASCE, Vol. 136, No. 8, pp.
Wang, Z., Dueas-Osorio, L. and Padgett, J.E. (2013). Seismic 933943.
response of a bridge-soil-foundation system under the combined Zhang, J. and Huo, Y. (2009). Evaluating effectiveness and
effect of vertical and horizontal ground motions, Earthquake optimum design of isolation devices for highway bridges using
Engineering & Structural Dynamics, Vol. 42, No. 4, pp. the fragility function method, Engineering Structures, Vol. 31,
545564. No. 8, pp. 16481660.
Ye, L., Ma, Q., Miao, Z., Guan, H. and Zhuge, Y. Zhang, J., Huo, Y., Brandenberg, S.J. and Kashighandi, P. (2008).
(2013). Numerical and comparative study of Effects of structural characterizations on fragility functions of
earthquake intensity indices in seismic analysis, The Structural bridges subject to seismic shaking and lateral spreading,
Design of Tall and Special Buildings, Vol. 22, No. 4, pp. Earthquake Engineering and Engineering Vibration, Vol. 7, No. 4,
362381. pp. 369382.

Advances in Structural Engineering Vol. 18 No. 10 2015 1669

Copyright of Advances in Structural Engineering is the property of Multi-Science Publishing
Co Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv
without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.