See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/334587900

Probabilistic Seismic Resilience Analysis for Bridges

Conference Paper · December 2015

CITATIONS READS

0 56

5 authors, including:

Yang Liu Da-Gang Lu

Sichuan University Harbin Institute of Technology
17 PUBLICATIONS   11 CITATIONS    183 PUBLICATIONS   695 CITATIONS   

SEE PROFILE SEE PROFILE

Fabrizio Paolacci
Università Degli Studi Roma Tre
156 PUBLICATIONS   988 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

SPECIAL ISSUE - Frontiers in built environment journal (Seismic Engineering): Vulnerability Assessment Methods for Storage Tanks in Oil and Gas Industries under
Natech Events View project

Soft Design and Soft Computing for Engineering Structures View project

All content following this page was uploaded by Da-Gang Lu on 20 July 2019.

The user has requested enhancement of the downloaded file.

Probabilistic Seismic Resilience Analysis for Bridges Shocked by Near-
fault Pulse-like Ground Motions
Y.Liu & D.G. Lu
School of Civil Engineering, Harbin Institute of Technology, Harbin, P.R. China
F. Paolacci
Department of Engineering, University of Roma Tre, Rome, Italy

ABSTRACT: This paper developed a probabilistic approach to estimate the seismic resilience of the short-
medium span bridges shocked by near-fault pulse-like ground motions. The seismic resilience demand of
bridges is quantified. A two-span steel-concrete composite bridge is taken as a case study, whose finite element
model is established by OpenSees. The double-side pulse-like ground motion records and non-pulse ground
motions records are selected from 5 earthquakes based on the PEER NGA strong ground motion database. The
spectral acceleration (Sa) is chosen as the intensity measure (IM), while the displacement of the middle pier is
selected as the engineering demand parameters (EDP). A series of nonlinear dynamic time-history analysis of
the case-study and regression analysis are aimed at simplifying the model of the engineering demand. Then,
several techniques are available to calculate fragility curves for this type of bridge. Finally, the resilience anal-
ysis based on numerical approach is conducted with the aim to evaluate the capability of this type bridges to
reduce the damage induced by earthquakes and minimize subsequent losses with an effective recovery. The
results show that this type bridges are good of seismic resilience and the near-fault earthquakes have significant
effect on that performance.

Nowadays, several probabilistic seismic analysis

progress have been achieved in performance-based
1 INTRODUCTION earthquake engineering framework, many of them fo-
cusing on seismic demand analysis and fragility as-
The concept of seismic resilience is fast developing sessment works, and some of them promoting the
in recent years, which plays the important role for a probabilistic treatment of resilience. A seminal work
performance assessment especially with the infra- of probabilistic seismic demand analysis (PSDA) of
structures. The condition of bridges plays a key role research is the study carried out by Shome (Shome,
with respect to the functionality of transportation net- 1999) and Cornell et al. (Cornell et al., 2000), in
works, which must be good of seismic resilience to which a probabilistic approach was utilized to com-
guarantee the efficient flow of people and services. In bine the uncertainty involved in demand and capacity
many studies, the performance-based techniques are in order to derive a closed-form formulation of the
adopted to predefined standards of performance dur- damage given the ground motion intensity meas-
ing earthquakes, the resilience can be considered as a ure(IM). Some recent researches about probabilistic
performance indicator that quantifies the residual resilience analysis promoted by Frangopol (Deco et
functionality along with the effort of the society in re- al., 2013) and Bocchini et al., (Karamlou et al., 2015)
sponding to a seismic event. (Deco et al., 2013) in which the resilience and costs can be calculated.
A large number of earthquake damage show that This paper focuses on the probabilistic seismic re-
near-fault pulse-like ground motions are different silience estimate method for bridges possible shocked
from far field ground motions. The former contain in extreme earthquakes. Probabilistic seismic demand
large velocity pulse, thus it can produce great destruc- and fragility analyses are two key elements of the
tive power to structures. The complexity responses of seismic resilience assessment to be used. Based on the
structures shocked by near-fault ground motions can total probability theorem, the expected functionality
be realized in other studies. To take into account the of the bridge after an extreme event can be estimated.
inherent uncertainties in the seismic hazard and re- With respect to the widely using of composite
sponse of the structures, the probabilistic approach brides, a two-span steel-concrete composite (SCC)
should be adopted. bridge is taken as a case study, whose 3D finite ele-
ment model is established by OpenSees software. The
results show that this type bridges are good of seismic Hence, the logarithmic standard deviation could be
resilience and the near-fault earthquakes have sig-in- obtained by the following expression:
dicant effect on their performance.
N
2
ln( d i ) ln( mD )
i 1
2 ASSESSMENT METHODOLOGY EDP | IM (3)
N 2

2.1 Quantitative identification of near-fault pulse- 2.3 Fragility curves

like ground motions Fragility is traditionally defined as the conditional
In this paper the method based on energy to quantita- probability of demand (D) exceeding the capacity (C)
tive identify the NF pulse-like ground motions is of a component (or system), given a certain value of
used. First, the real ground motion time-histories are a representative IM. Analytical fragility curves are
matched to the simplified mathematical model by the generated generally through numerical analysis. As-
Least square method. Then, the simplified time-histo- suming that the capacity follows a lognormal distri-
ries with characteristics of main pulse are got. Sec- bution, the fragility can be defined by the convolution
ond, the perk-point method (PPM) is used to identify formula (Lu et al., 2014):
the parameters (e.g.: period of pulse, perk velocity) of ln mD mC
the main pulse of real ground motions. Finally, the P D C IM
2 2
energy index Ep is defined to identify the pulse-like D IM C
ground motion (Baker, 2007). When the Ep is greater (4)
than 0.3, the NF ground motion records include obvi- b ln IM ln mC ln a
ous velocity pulses (Zhai et al., 2012).
2 2
Through the least square method, the velocity pulse D IM C
in real ground motion time-histories are matched the where, the mC and C are the median and dispersion
simplified mathematical model. Based on the velocity of the lognormally distributed capacity.
pulse model from the real ground motion, the PPM
(Osaki, 2008) is used to define the time interval of
valley and peak in time-history curves as the pulse 2.4 Probabilistic resilience
period.
2.2 Probability seismic demand models The definition of seismic resilience is ‘the ability of
social units (e.g., organizations and communities) to
The probabilistic seismic demand analysis (PSDA) mitigate hazards, contain the effects of disasters when
provides the probabilistic relationship between the in- they occur, and carry out recovery activities in ways
tensity measures of ground motions and structural re- that minimize social disruption and mitigate the ef-
sponse parameters. Assuming that the demand fol- fects of future earthquakes’. (Bruneau M et al., 2003)
lows a lognormal distribution (Cornell et al., 2000), However, resilience is a multi-faceted concept, which
the closed form formulation of the conditional proba- covers too broadly social and technical aspects to be
bility of exceedance of the demand D with given IM defined as a single analytical definition. In this paper,
can be derived as: the following definition is adopted (Deco et al.,
2013):
ln(d / mD )
PD d IM 1 (1) t0 th

D IM
Q t dt
t0
R (5)
in which mD is the median of the demand parameter, th
D IM is the dispersion of the demand about its median where R is the resilience index, t0 is the time of occur-
conditioned on IM, and is the standard normal rence of the extreme event, th is the investigated time
cumulative distribution function. Additionally, mD horizon, and Q t is the time-variant functionality of
can be assumed linear in the log-log space, whereas the structure or system. Based on the total probability
the conditional dispersion is constant (Cornell et al., theorem, the expected functionality of the structure at
2000). Thus, the probabilistic demand model (PSDM) each time instant t after an earthquake can be esti-
can be expressed as: mated. If a collection of events with a limited range
of IMs is of interest, the following equation can be
mD aIM b (2) used to compute the expected functionality Q t
(Karamlou et al., 2015):
in which, a and b are regression parameters, which nIM nDS
could be assessed by regression analysis.
Q t P IM i P( DS j IM i )Q j t (6)
i j
in which Q j t is the functionality recovery func- the refined model, the nonlinearities of material are con-
tion of the structure at time t with an initial damage sidered. The shear force connections models between steel
level j. In the case of highway bridges, functionality girders and concrete slab, also between crossbeam and
is usually presented in the form of the percentage of composite beam, are built. The plane models for each
traffic-carrying capacity in normal conditions (ATC- girder 2.65m wide at transverse direction are firstly built
and then the 3D model is achieved by adding transverse
13, 1985). Pj DS j IM i is the conditional prob-
connections to the plane models of girders. In this 3D
ability of being in the damage state j given the event model, steel girder and concrete slab were modeled using
scenario with IM equal to i, could be computed by nonlinear beam elements with fiber sections. The rein-
fragility analysis. Finally, Pi IM i is the proba- forced concrete pier was also modeled by beam elements.
bility of occurrence of extreme events generating in- Vertical shear studs connecting the steel girders to slab as
tensity i, computed by a standard probabilistic seis- well as the shear studs within the CCB are modeled using
mic hazard analysis performed on the region where nonlinear links with elastic-plastic behavior. For materials,
the bridge is located (FEMA445, 2006). the concrete01 type was selected as the mechanical model
of concrete in slab according to Kent and Park model
where the tensile strength of concrete is neglected. Mean-
3 A CASE STUDY while, the steel01 type, an elastic-perfectly plastic model,
has been selected as the mechanical model of steel in gird-
3.1 Finite element model ers.
The short-to-medium span steel-concrete compo-
site bridges have been used more and more world-
wide, owing to their several advantages. (Liu et al.,
2015). The case study is illustrated by a typical 2-span
overpasses, currently under investigation within the
European Project SEQBRI (www.seqbri.it). The
straight bridge is 40.0 m long and consists of two
spans of 20.0 m (Figure1). The total width of the road
cross-section is 10.6 m, with carriageway 6.5 m wide
and 2 sidewalks 2.05m wide. This is a typical over-
pass cross-section with 2 traffic lanes. Four main
girders HE600B made of S460 steel girders are de-
signed, with center distance of 2.65 m. The thickness
Figure2. 3-D FE model
of the concrete slab is equal to 25cm. The steel gird-
ers are fixed to reinforced concrete crossbeams at
The joint of composite girder, concrete crossbeam and
abutments. By this diaphragm, the deck is simply sup-
pier is difficult to reproduce. The refined crossbeam model
ported on normal (low-damping) damping rubber
is built to simulate the complicated mechanical behavior.
bearings. At the intermediate pier, the steel girders are
The vertical head plates welded on the steel girders and
fixed to a reinforced concrete crossbeam 0.9 m wide
directly in contact with the transverse concrete beam are
monolithically connected to the pier. The wall type
modeled by rigid links. Links standing for shear headed
pier is 7.0 m height, 0.6 m thick and 7.0 m wide. It is
studs within the joint are connected to these rigid links ac-
assumed that the foundation soil is categorized as
cording to concrete crossbeam configuration. In order to
type B according to EN1998 and the SSI effect is ne-
reproduce the constraint in compression due to the pres-
glected.
ence of the concrete crossbeam, four compression gap el-
40.00 ements have been modeled both at left and right sides of
20.00 20.00
the CCB joints, as shown in Figure 3. The compression
gaps of the numerical model link the vertical end-plate to
the axis of CCB beam at mid-support, both in the left and
in the right side of the joint. The two levels of pretension
bars in the bottom area of the CCB have been modeled by
10.60
2.05 3.25 3.25 2.05 two elastic truss elements.

2.5% 2.5% 3.2 Selection of input ground motions

Unit:m
A subset ground motion records based on PEER
strong ground motion database of 60 near-fault (NF)
Figure1. Longitudinal section and cross section at span pulse-like and 60 far-field (FF) have been selected.
The group of NF records are formed with closest site-
For seismic demand analysis, the 3D finite element (FE) to-source distances (R) less than 20 km and moment
model is developed by OpenSees software (Figure2). In magnitudes (MW) greater than 6.0. The ratio of PGV
and PGA is greater than 0.2 for making sure the 3.4 Fragility analysis
ground motions including obvious velocity pulse. Berry and Eberhard (2003) developed empirical
Through directly observing, the ground motions rec- equations to estimate deformations at bar buckling
ords in NF group should have double sides velocity and fracture and concrete cover spalling based on the-
pulse. On the opposite rules, the FF ground motion oretically expected trends in drift ratios. The damage
records are selected with matching the PGA and MW statement of the pier has been listed in Table 2. Fi-
of the NF records.
nally, the demand fragility curves of piers are devel-
The selected NF ground motions records come oped, which are shown in Figure 4 (a-d).
from 5 strong earthquakes. These strong shock earth-
quakes are N. Palm Springs Earthquake (America,
Table 2. Damage index for pier drift ratio
1986), Imperial Valley Earthquake (America, 1979),
ChiChi Earthquake (Taiwan, 1999), Kocaeli Earth- Slight Moderate Extensive Complete
quake (Turkey, 1999) and Northridge Earthquake damage damage damage damage
(America, 1994); 80 FF ground motions records from Damage index
1.07% 1.31% 4.31% 8.96%
7 strong shock earthquakes are selected, which are the (Drift ratio)
same 5 strong earthquakes as NF and Kobe Earth-
quake (Japan, 1995), Whittier Narrows Earthquake
(America, 1987).
Based on the former research, all the selection NF
ground motions are up to standards of near-fault
pulse-like ground motions (Liu et al., 2015). Due to
limited space, only the energy index Ep of selected NF
ground motions are shown in Figure 3.
20
Frequence of Ep
18
16
14
12
Count

10
8
6
4 (a) fragility curves for slight damage
2
0
0.3 0.4 0.5 0.6 0.7 0.8 0.9
Ep

Figure3. Energy index Ep of selected NF ground motions

3.3 Probabilistic seismic demand analysis

The total of 120 nonlinear analyses data provides the
parameters of the conditional probability distribution
of the D given IM. Through significance testing of re-
gression coefficients, the logarithmic standard devia-
tion and coefficients of determination R2 can be
D IM

used to test the goodness of fit.

The results of regression are shown in Table1. Accord-
ing to the regression results, the goodness of fit for FF
ground motions is better than NF ground motions with the (b) fragility curves for moderate damage
IM being Sa. This may be due to the more uncertainty of
NF ground motions.

Table 1. The regression parameters and determination coeffi-

cients
lna b D IM
R2
NF 1.07 1.13 0.247 0.880
FF 0.61 0.88 0.226 0.901
 has been selected to be in Los Angeles, California,
USA.
The functionality recovery functions from ATC-13
(ATC-13, 1985) have been adapted to compute resil-
ience. These recovery profiles have been computed
on the basis of experts surveys, and are presented in
the form of the mean time required to reach 30%,
60%, and 100% of the normal functionality of the
structure. The selected functions are associated with
the recovery profiles of social function class 25c,
which corresponds to conventional highway bridges.
The damage states 2 through 5 presented in ATC-13
are assumed to be the corresponding damage states
associated with slight through complete damage limit
states. The functionality recovery function for each
(c) fragility curves for extensive damage limit state used in this study is shown in Figure 5. It
is worth mentioning that there is significant uncer-
tainty on the real restoration process of a bridge,
while there is only a limited number of studies avail-
able in the literature investigating this uncertainty.
slight extensive
moderate complete
1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1
(d) fragility curves for complete damage
0.0
0.1 1 10 100
Figure4. Comparison of fragility curves of piers with the given t(Days)
IMs on longitude
Figure5. Functionality recovery profiles from ATC-13
The Figure 4 shows that the exceeding probability
of damage for bridge under NF pulse-like ground mo- The probabilities of exceeding the damage states
tions are much larger than those under FF ground mo- at each level of IM, computed by means of the system
tions as the demand analysis results. The probabilities fragilities (Fig. 4), have been used to calculate the
of different damage statements, while the inputted corresponding limit state probabilities of occurrence.
ground motions are with obvious velocity pulse, are A probabilistic seismic hazard analysis has been per-
about 30% larger. In the considered range of IMs, the formed on the site of the location of the bridge based
max probability of bars yielding of bridge constructed on the United States Geological Survey (USGS) haz-
in NF region is about 90% while the probability is ard curves (USGS, 2015). The probabilities of occur-
about 60% with the bridge in FF region. At the ex- rence of each investigated IM have been assessed
treme statement, the probability of bars fracture of from hazard curves assuming a life of 100 years for
bridge is about 30% under the NF pulse-like ground the bridge. Eventually, the expected functionality of
motions. However, the probability of bars fracture the bridge has been computed using Equation (6).
with the bridge shocking by FF ground motions is Figures 6 (a-d) illustrate the expected functionality re-
only about 5% at the same extreme statement. covery profile of the bridge. Two profiles presented
in the figures account for the different damage caused
3.5 Probabilistic resilience analysis by near-fault and far-field earthquakes. And the ex-
pected functionality recovery of bridge is showed in
The system fragility curves computed in the previous figure 6 (e). Considering two assumptions that the to-
section have been used to calculate the resilience of
tal recovery time is a constant and the initial damage
SCC bridges following the methodology presented in
Section 2.3. For this analysis, the geographic location for different statements are same with being 0, the re-
sults show that the functionality paths obtained are so
different. It can be easily seen that the damage caused 1.0

by near-fault earthquakes is more difficult to be re-

covered. In other words, the seismic resilience de-

Q(t) for complete damage

0.8

mand of bridge shocked by near-fault earthquakes is

larger than resilience demand of far-field earth- 0.6

quakes.
The resilience of bridge can be determined using 0.4
Near-fault
equation (5), and the resilience indexes of bridge are Far-field
shown in table 2. It can be found that the resilience of 0.2

bridge shocked by far-field earthquakes with 0.98 is

larger than near-fault earthquakes with 0.89. 0.0
50 100 150 200 250 300
t(Days)
Table 2. The expected resilience index of bridge (d) Expected functionality recovery for complete damage
Near-fault Far-field 1.0

Expected resilience index 0.89 0.98

1.0 0.8

0.8 0.6
Q(t) for slight damage

Q(t)
0.6 0.4
Near-fault
Far-field
0.4 0.2
Near-fault
Far-field
0.2 0.0
0.1 1 10 100
t(Days)
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 (e) Expected functionality recovery of bridge
t(Days)
Figure6. Expected functionality recovery of bridge at different
(a) Expected functionality recovery for slight damage damage levels
1.0

4 CONCLUSIONS
Q(t) for moderate damage

0.8

0.6 The probabilistic seismic resilience of bridges is esti-

mated by a simulation approach, and the resilience
0.4 demand after the near-fault earthquake events is
Near-fault quantified. The procedure of calculation is clear and
Far-field
0.2
easy to be conducted with some assumptions. The re-
sults show that the near-fault earthquakes have signif-
0.0
icant effect on resilience, the performance for which
0 1 2 3 4 5 6 7 8 9 is about 10% larger than far-field earthquakes.
t(Days)

(b) Expected functionality recovery for moderate damage

1.0 ACKNOWLEDGEMENT

The financial supports received from the National Sci-

Q(t) for extensive damage

0.8
ence Foundation of China (Grant Nos. 91315301,
51378162, 51178150), the Research Fund from Ministry
0.6
of Science and Technology of China (2013BAJ08B01),
and the Open Research Fund of State Key Laboratory for
0.4 Disaster Reduction in Civil Engineering (SLDRCE12-
Near-fault MB-04) are gratefully appreciated. The Present work is
Far-field
0.2
also supported by the funds of the European Project SE-
QBRI: “Performance-Based Earthquake Engineering
Analysis Of Short-Medium Span Steel-Concrete Compo-
0.0
20 30 40 50 60 70 80 90 site Bridges”, Grant N° RFSR-CT-2012-00032. Any opin-
t(Days)

(c) Expected functionality recovery for extensive damage

ions, findings and conclusions or recommendations ex-
pressed in this paper are those of the authors and do not
necessarily reflect those of SEQBRI sponsors.

REFERENCES
DecòA, Bocchini P, Frangopol D M. 2013. A probabilistic ap-
proach for the prediction of seismic resilience of bridges.
Earthquake Engineering & Structural Dynamics, 42(10):
1469-1487.
Shome N. 1999. Probabilistic seismic demand analysis of non-
linear structures. Ph.D. Thesis, Stanford University.
Cornell CA, Jalayer F, Hamburger RO. Foutch DA. 2000. Prob-
abilistic basis for 2000 SAC federal emergency manage-
ment.
Karamlou A, Bocchini P. 2015. Computation of bridge seismic
fragility by large‐scale simulation for probabilistic resilience
analysis. Earthquake Engineering & Structural Dynamics.
Baker, J.W. 2007. Quantitative classification of near-fault
ground motions with wavelet analysis. Bulletin of the Seis-
mological Society of America, 97(5): 1486-1501.
Zhai, C., Chang, Z., Li, S., Chen, Z.Q., Xie, L. 2013. Quantita-
tive Identification of Near-Fault Pulse-Like Ground Mo-
tions Based on Energy. Bulletin of the Seismological Soci-
ety of America, 103(5): 2591-2603.
Osaki, Y. 2008. Spectral Analysis in Earthquake Engineering,
Beijing: Seismological Press.
Paolacci, F. & Giannini. R. 2014. Toward a performance-based
earthquake engineering analysis of short-medium span steel-
concrete composite bridges. Second European Conference
on Earthquake Engineering and Seismology. Istanbul.
Lu D, Yu X, Jia M, et al. 2014. Seismic risk assessment for a
reinforced concrete frame designed according to Chinese
codes[J]. Structure and Infrastructure Engineering, 10(10):
1295-1310.
Bruneau M, Chang SE, Eguchi RT, Lee GC, O’Rourke TD,
Reinhorn AM, Shinozuka M, Tierney K, Wallace AW, von
Winterfeldt D. 2003. A framework to quantitatively assess
and enhance the seismic resilience of communities. Earth-
quake Spectra, 19(4):733–752.
Rojahn, C., & Sharpe, R. L. 1985. Earthquake damage evalua-
tion data for California. Applied technology council.
Federal Emergency Management Agency (FEMA). 2006. Next-
generation performance-based seismic design guidelines.
Technical Report FEMA-445, Washington, D.C.
Liu Y, Lu D G, Paolacci F. 2015. Multivariate Probabilistic Seis-
mic Demand Analysis of Steel-concrete Composite Bridges
under Near-fault Pulse-like Ground Motions. Multi-span
Large Beidges, Porto.
Berry, M.P. & Eberhard, M. 2003. Performance Models for
Flexural Damage in Reinforced Concrete Columns. PEER
Report, 2003/18.
USGS. U.S. Geological Hazard Science Center. 2015. Unite
States Geological Service, (Available from: http:// earth-
quake.usgs.gov/) [accessed on October.2015].

View publication stats