Performance of Two Real-Life California Bridges under

Regional Natural Hazards

Taner Yilmaz, S.M.ASCE1; Swagata Banerjee, A.M.ASCE2; and Peggy A. Johnson, A.M.ASCE3

Abstract: The performance of two real-life California bridges is assessed under a possible regional multihazard condition involving floods and
earthquakes. For flood events with varied frequency, expected scour depths at bridge piers are calculated and incorporated in finite-
Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

element analyses (FEAs) of the bridges under earthquakes that represent regional seismic hazards. Based on FEA results, fragility curves of
bridges are developed at component and system levels. Fragility surfaces are generated to acquire comprehensive knowledge on bridge failure
probability for the combined effect of earthquake and flood events of varying frequency. Quantified bridge vulnerability is applied to a risk evalu-
ation framework that combines hazard probability with bridge failure consequences. Obtained results depict that the occurrence of flood events
can increase the seismic vulnerability and risk of bridges, although the amount of increase depends on bridge attributes. Bigger (large-diameter)
foundations tend to reduce the impact of flood hazard on bridge seismic performance. Moreover, the multihazard response of the bridges indi-
cates that the seismic design philosophy used for bridges can play a role in attaining the seismic safety of bridges with flood-induced scour at the
foundations. DOI: 10.1061/(ASCE)BE.1943-5592.0000827. © 2015 American Society of Civil Engineers.
Author keywords: Multihazard; Earthquake; Flood; Fragility curve; Fragility surface; Risk curve.

Introduction concern. The present study elaborates on this issue by analyzing

two highway bridges in California for the combined effect of re-
The present worldwide emphasis on building and maintaining sus- gional earthquake and flood hazards.
tainable civil infrastructure systems requires all infrastructure com- Earthquakes and floods are low-probability, high-consequence
ponents to remain functional during natural disasters. Bridges are events. Flood events can cause scour at bridge foundations by
the most vulnerable components of a highway transportation sys- removing bed materials from around bridge piers (Arneson et al.
tem. The overarching need for the safety and serviceability of the 2012). To ensure bridge stability against scour, AASHTO (2010)
critical transportation infrastructure system under extreme natural mandated foundation stability checks against flood-induced scour
hazards requires highway bridges to be analyzed and designed not for a design-level flood event. However, no guideline is currently
only for discrete hazard events, but also for possible multiple- available in design codes to ensure the seismic safety of bridges
hazard (or multihazard) conditions. For a highway network span- with scoured foundations. Scour causes loss in the load-carrying
ning a seismically active flood-prone region, the occurrence of capacity of bridge piers and makes bridges more flexible under lat-
earthquakes in the presence of flood-induced scour is a possible eral loading. Thus, scour may eventually increase the risk of bridge
multihazard scenario for bridges in the network. The highway failure under lateral loading, such as that due to earthquakes.
transportation network in California represents such a network. A Conversely, the increased flexibility of bridges due to scour may
recent report (Marshall 2014) revealed that earthquakes and help in reducing seismic inertial forces. Nonetheless, the seismic
floods are the biggest natural threats to highway bridges in design philosophy used for a bridge may play a role in attaining its
California. Hence, the aforementioned multihazard scenario has overall seismic safety in the case of scour resulting from flood
the potential to influence the seismic design and retrofit of events. Hence, to ensure the seismic safety of bridges under possi-
California bridges when network safety and functionality are of ble flood conditions, it is important to check bridge seismic per-
formance in the presence and absence of scour resulting from flood
events of varying frequency.
1 There are only a few past studies that discussed the combined
Assistant Professor, Dept. of Civil Engineering, Ozyegin Univ.,
34794 Istanbul, Turkey; formerly, Graduate Research Assistant, Dept. of effect of earthquake and flood-induced scour on the performance of
Civil and Environmental Engineering, Pennsylvania State Univ., bridges (Ghosn et al. 2003; Wang et al. 2012; Prasad and Banerjee
University Park, PA 16802. E-mail: [Link]@[Link] 2013; Banerjee and Prasad 2013; Alipour et al. 2013; Wang et al.
2 2014a, 2014b). In these studies, example bridges with different
Assistant Professor, Dept. of Civil Engineering, Indian Institute of
Technology Bombay, Powai, Mumbai 400076, India; formerly Assistant structural attributes were analyzed for various combinations of
Professor, Dept. of Civil and Environmental Engineering, Pennsylvania scour depths and seismic events, and bridge failure probabilities for
State Univ., University Park, PA 16802 (corresponding author). E-mail: each of these combinations were determined. It is generally
swagata@[Link] observed that bridge seismic fragility characteristics change with
3
Professor, Dept. of Civil and Environmental Engineering, Penn- scour depth depending on the type of bridge foundation. Bigger
sylvania State Univ., University Park, PA 16802. E-mail: padj6@engr
(large-diameter) foundations tend to minimize the impact of scour
.[Link]
Note. This manuscript was submitted on November 4, 2014; approved on bridge dynamic response (Banerjee and Prasad 2013). Other
on June 25, 2015; published online on December 14, 2015. Discussion pe- than analyzing bridges as structural systems, bridge components,
riod open until May 14, 2016; separate discussions must be submitted for such as piers, are also analyzed to evaluate their failure probabilities
individual papers. This paper is part of the Journal of Bridge for a combined load effect of scour, truck, and earthquake (Liang
Engineering, © ASCE, ISSN 1084-0702. and Lee 2013a, 2013b).

© ASCE 04015063-1 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063

 All of these previous studies provided valuable information and length of 231.2 m and is founded on a soil medium comprised of
insight to the multihazard problem of earthquakes and floods primarily gravels with a sand matrix. The entire structure contains
involving bridges; however, region-specific hazard information is two individual bridges that provide simultaneous service to both
lost in these studies because of their use of example bridges or directions of traffic. Bridge superstructures are composed of pre-
bridge components. In addition, most of these studies considered stressed concrete box girders and connected with a monolithic con-
scour to be a source of hazard; hence, possible bridge scour depths crete closure pour at the middle. The bridge has seat-type abut-
are considered without evaluating flood potential at bridge sites ments. Each column bent on either side of the bridge superstructure
(Ghosn et al. 2003; Wang et al. 2012; Alipour et al. 2013; Liang and is composed of two circular reinforced concrete piers with a 1.83-m
Lee 2013a, 2013b; Wang et al. 2014a, 2014b). Note that bridge diameter monolithically connected with bridge girders. All piers
scour is a consequence of flood hazard, and hence it does not neces- have equal cross-sectional and material properties. At the founda-
sarily represent any load effect to a bridge. A regional flood event of tions, the piers are extended below ground level; these extended
the same intensity may result in different scour depths for different parts are referred to as pile shafts. The pile shafts have the same
bridges. Moreover, because of irregular bed elevation along the cross-sectional and material properties as those of the piers until a
Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

cross section of a flood channel, it is less likely that a bridge with certain depth, below which steel casings are used around the pile
multiple piers in the water will have the same scour depth at all pier shafts. The diameter of the pile shafts with steel casing is 2.44 m. At
locations. Hence, equal scour depth at all bridge piers, as has been greater depths, concrete shafts are socketed into rock layers.
considered in past studies (Wang et al. 2014a, 2014b; Prasad and The second bridge, henceforth referred to as Bridge 2, is located
Banerjee 2013; Banerjee and Prasad 2013), may not be a realistic in San Joaquin County. Built in 1972, this bridge is on Interstate 5
assumption for bridges with multiple piers in the water. Therefore, and crosses the San Joaquin River. Fig. 2 provides schematic draw-
region-specific analyses with real-life bridges are required to accu- ings and general geometric details of the bridge. The bridge has six
rately predict the impact of regional flood events on bridge seismic spans with a total length of 235.3 m. It has a reinforced concrete box
behavior. Besides, bridges should be selected from different regions girder that is monolithically connected to wall-type piers. Each pier
having moderate to high seismic and flood hazards. This is indeed is founded on a group of precast prestressed concrete piles with a re-
important, as the characteristics of a flood event with a specified fre- inforced concrete pile cap. Details of these pile foundations are
quency (such as a 100-year flood) may change from one region to given in Fig. 2. The soil medium underlying the bridge foundation
the other depending on various factors, such as topology and annual mainly consists of silty sands. The bridge has integral abutments on
rainfall. The same is equally true for seismic hazard. Bridge both sides and an in-span hinge between Bent 3 and Bent 4. At the
response obtained from such a study can be utilized further for risk hinge, the bridge girder is separated by a 25.4-mm (1-in.) gap such
and uncertainty analyses under the same multihazard condition. that one side of the girder sits on elastomeric bearings placed on the
Obtained results will also facilitate the evaluation of load factors other side of the girder.
combining earthquakes and floods when these two are considered in Consideration of these real-life bridges is a distinct feature of the
a multihazard framework for the design of new highway bridges. current study. Such a consideration made it possible to use region-
With the objective of analyzing real-life bridges under the re- specific hazard information. It also paved the path to assess the risk
gional multihazard scenario, the study selected two bridges in of the bridges under the multihazard condition of flood-induced
California that were constructed at different times and are located in scour and earthquake. With the exception of Banerjee and Prasad
different regions. The multihazard performance of these two (2013), no previous study on this multihazard issue has extended its
bridges is assessed by considering a flood event followed by a seis- scope to the multihazard risk assessment of bridges. In comparison
mic event. The impact of flood hazard on these bridges is consid- with Banerjee and Prasad (2013), the current study adopted more
ered through scour depths at bridge foundations. The potential of sophisticated modeling approaches for several nonlinear compo-
occurring liquefaction at both bridge sites is not considered in this nents (such as piers, abutments, foundation–structure interaction,
study to isolate the multihazard effect of earthquake and flood- and bearings) of both bridges, as detailed further in following sec-
induced scour on bridge performance from that caused by any other tions. The overall performance of the bridges is evaluated based on
seismic-induced hazard, such as liquefaction. To capture wide the response of all of these nonlinear components for the multiha-
ranges of the two hazards, flood events of varying frequency and zard condition. Moreover, fragility surfaces are developed for both
various levels of seismic hazard are considered based on regional bridges that provide the probability of exceeding a specified damage
hazard information acquired for the bridge sites. The multihazard level under joint intensities of seismic and flood hazards. Hence, in
performance of these bridges is expressed in the form of fragility addition to generating new knowledge (which is discussed at the
curves and surfaces. Whereas the fragility curves represent bridge end of this article), the current study provides a realistic perspective
vulnerability for specific combinations of flood and seismic haz- on the multihazard issue considered herein.
ards, the same is expressed in fragility surfaces for all possible com-
binations of these two natural hazards. The vulnerability informa-
tion expressed in fragility curves and surfaces is used to generate Regional Seismic and Flood Hazards
risk curves of these bridges. These risk curves provide annual
exceedance probabilities of various levels of bridge restoration cost Regional Seismic Hazard
for the specified multihazard condition.
Identification of region-specific seismic and flood hazard levels
and selection of their critical combinations are important because
California Bridges under Analysis structural safety depends on maximum demands from multiple
hazards. To pursue this study, regional seismic hazard is consid-
The first bridge, henceforth referred to as Bridge 1, is located in ered through (1) seismic hazard curves that provide information
Shasta County. This bridge is on State Highway 44 and crosses the on regional seismicity and (2) historic ground motion data sets
Sacramento River. Built in 2010, the bridge replaces an old bridge that can be used for time-history analyses of bridges. Seismic haz-
at this site. Schematic drawings and general geometric details of the ard curves provide annual exceedance probabilities of seismic
bridge are presented in Fig. 1. The bridge has four spans with a total events having various intensity levels. These curves are utilized

© ASCE 04015063-2 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063

 D = 2.440
231.2
50.68 64.92 64.92 50.68 D = 1.830
#36 (D=36 mm)
25 bundles
total of 50
Abut. 1 Abut. 5
#25 (D=25 mm) hoops
Bent 2 Bent 3 Bent 4 Clear cover = 75 mm 25 mm steel casing

(a) (b) CIDH Piles / Pier columns Pile shaft with steel casing
Closure Pour Center Line
16.86~20.50 22.58

1.40 3.44~4.35 3.44~4.35 3.44~4.35 3.44~4.35 1.40 1.40 3.43 1.451 4.875 4.875 4.88 1.40
Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

0.15 0.15 0.15 0.15

0.265
2.60
0.25

0.30 0.30
2.60

0.265
1.00

9.75

B2 B3 B4 B2 B3 B4 B2 B3 B4 B2 B3 B4
H = 6.212 6.387 5.999 H= 6.397 6.598 6.238 H= 6.589 6.789 6.238 H= 6.335 6.536 6.176

D = 1.83
B2 B3 B4
142.050 142.100 142.710
River bed

LEFT BRIDGE 139.000 RIGHT BRIDGE

Steel casing cutoff elev.
2.44
B2 B3 B4
10.51 13.00
131.120 130.870 128.870
(c) Pile socket elev. *All units are in meter unless otherwise stated

Fig. 1. (a) Schematic view of Bridge 1; (b) cross sections of bridge pier and pile shaft with and without steel casings; (c) general elevation view of a
typical bent

for risk evaluation of bridges, as discussed later in this article. earthquake records is generated for Bridge 1, among which the
Generation of site-specific seismic hazard curves is beyond the number of records having PGA values of 0.1–0.2 g, 0.2–0.3 g, 0.3–
scope of the present study. Hence, seismic hazard curves devel- 0.4 g and > 0.4 g are 62, 19, 15, and 8, respectively. For Bridge 2, 160
oped by the USGS (2013a) are considered here. Fig. 3 shows the ground motion records are utilized to form the data set, in which 64,
seismic hazard curves at locations of Bridge 1 and Bridge 2 for 63, 24, and 9 recordings, respectively, fall into the aforementioned
site-specific soil conditions. Peak ground acceleration (PGA) is PGA ranges. Ground motions with PGAs less than 0.1 g are ignored
used here as the measure of seismic intensity, which was selected from both data sets.
because it facilitates easy interpretation between seismic intensity
and hazard level. Regional Flood Hazard
A large set of ground motions with varying hazard levels is de-
sirable for seismic vulnerability analysis of bridges. Selected Flood hazard curves are generated to express regional flood hazards
ground motions should reflect the seismic characteristics of the at bridge sites. These curves provide peak-flow discharges corre-
region of interest and be compatible with local soil conditions. For sponding to flood events having various annual exceedance proba-
this purpose, corrected and filtered real ground motion time histor- bilities. Flood hazard curves at bridge sites are developed through
ies recorded at the bridge sites are obtained from the Next flood-frequency analysis (Interagency Advisory Committee on
Generation Attenuation (NGA) database of the Pacific Earthquake Water Data 1982) by processing annual peak-discharge data that
Engineering Research Center (PEER 2013). Two separate sets of are recorded at the stream gage stations nearest to the bridge sites
ground motions are developed for the two bridges. These ground [obtainable from USGS (2013b) National Water Information
motion data sets constitute earthquakes recorded within a 170-km System]. It is realized that because of the complexity of river basins,
radius from the respective bridge sites. The horizontal orthogonal flow discharges at bridge sites may not always be the same as those
components of these earthquake records are considered for time- that are recorded at the nearest stream gage stations. However, hav-
history analysis. Note that relatively low seismic activity, as com- ing a stream gage installed exactly at a bridge site is less probable in
pared with other parts of California, has been recorded historically reality. For this reason, the USGS has developed the National
at both bridge sites. However, the possibility of future strong earth- Streamflow Statistics (NSS) software program (Ries 2007), which
quakes occurring in these locations cannot be ignored. Therefore, compiles regional regression equations for estimating streamflow
some of the ground motion records are scaled with a factor of 2 statistics at ungaged sites. These regional regression equations were
such that performance of these bridges under strong ground derived such that streamflow statistics can be transferred from
motions can be observed. Thus, a data set containing 104 gaged to ungaged sites through the use of watershed and climatic

© ASCE 04015063-3 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063

 235.3
7.32
39.90 44.20 36.88 44.20 44.20 18.57

In-span hinge

Abut. 1 Bent 5 Abut. 7

Bent 2 Bent 3 Bent 4 Bent 6

(a)
#8 @ 0.30 m #4 @ 0.46 m

16.154
Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

1.067 2.337 2.337 2.337 2.337 2.337 2.337 1.067

m
#8 @ 0.30 m

7
0.162

45
0.178

0.
At the top of columns:

R
Additional #8 @ 0.60 m for Pier 2 & Pier 3

0.305
2.296

0.203 Additional #8 @ 0.30 m for Pier 6

0.14
*All units are in meter unless otherwise stated

Abut 1 Bent 2 Bent 3 Bent 4 Bent 5 Bent 6 Abut 7

4.57
H pier 14.041 14.660 15.889 9.803 15.757

H pier Hpile 22.016 8.814 14.935 8.534 12.116 6.172 23.815

H river bed 3.658 2.561 3.048 5.639 3.810

10.262 Number of piles

(row X column) 1X9 3 X 10 3 X 10 3 X 9 3X8 3X9 1X5
H river bed
S trans
1.930 1.118 1.118 1.27 1.448 1.27 3.861
0.914

Transverse spacing of piles

S long
Hpile 1.219 1.219 1.219 1.219 1.219
Longitudinal spacing of piles

S trans D=0.381
(b) (c)

Fig. 2. (a) Schematic view of Bridge 2; (b) general elevation of a pier; (c) pier cross section

0
10 10000
Annual Probability of Exceedance

Bridge-1
Annual Peak Discharge (m3/s)

-1
10
Recorded Data
10-2
1000
-3
10

10-4
100
-5
10 Bridge-1 Bridge-2
10 -6 Bridge-2 Recorded Data

-7 10
10
99.8 99 95 80 60 40 20 5 1
0.001 0.01 0.1 1 10
Probability of Exceedance (%)
PGA (g)
Fig. 4. Mean flood hazard curves at Bridge 1 and Bridge 2 sites
Fig. 3. Seismic hazard curves at bridge sites

characteristics as explanatory or predictor variables (Lagasse et al. County (FEMA 2009a) mentioned that the Shasta Dam located
2013). For the present study, comprehensive river basin information upstream from Bridge 1 (where data are recorded) controls up
for the Sacramento and San Joaquin Rivers is obtained from flood in- to 1% of the annual chance flood (i.e., 100-year flood) with a
surance studies performed by the Federal Emergency Management maximum discharge of 2,237 m3/s at the bridge site. Therefore,
Agency (FEMA 2009a, 2009b) at the bridge locations. The obtained the flood hazard curve of Bridge 1 has a maximum flow dis-
information is utilized to corroborate flood hazard curves developed charge of 2,237 m3/s, as shown in Fig. 4. The FEMA flood in-
for the Bridge 1 and Bridge 2 sites. surance study for San Joaquin County (FEMA 2009b) did not
Fig. 4 shows mean (i.e., at 50% confidence level) flood haz- mention any such limitation of flow discharge at the Bridge 2
ard curves developed at the Bridge 1 and Bridge 2 locations. site. Therefore, the results obtained from flood-frequency analy-
Flow data from the nearest stream gauge stations are also plot- sis were directly used for the mean flood hazard curve at the
ted in the figure. The FEMA flood insurance study of Shasta Bridge 2 site.

© ASCE 04015063-4 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063

 Flood-Induced Scour at Bridge Foundations the bridges include some realistic assumptions and idealizations,
as described in the following discussion.
In the present study, flood events with annual exceedance probabil-
ities of 90%, 50%, 10%, 5%, 2%, and 1% (corresponding to 1.1- Bridge Girders
year, 2-year, 10-year, 20-year, 50-year, and 100-year floods) are
For performance-based seismic analysis of bridges, the use of a
studied for the multihazard risk evaluation of Bridge 1 and Bridge
single-beam stick model is a reasonable assumption because
2. As can be observed from Fig. 4, flood events with return periods
bridge girders are generally expected to remain elastic during
of more than 20 years have the same peak discharge as that of the
seismic excitations. Likewise, the present study uses linear elastic
20-year flood event at the site of Bridge 1. Therefore, the multiha-
beam elements to represent bridge girders. These elements are
zard analyses with 50-year and 100-year flood events are redundant
assumed to run along the center lines of the respective girders
for this bridge.
such that equivalent stiffness and mass properties calculated at
For each of the investigated flood events, corresponding peak-
discharge values are obtained from the mean flood hazard curves the centroid of the girders can be lumped on them. As per
shown in Fig. 4, and resulting scour depths ys at bridge foundations Caltrans (2010) recommendations, an effective flexural rigidity
Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

are estimated using the following equation suggested by Arneson equal to the average of 0.5–0.75 times the gross stiffness is used
et al. (2012): for Bridge 2’s reinforced concrete box girder sections. For Bridge
1’s prestressed concrete box girder sections, no stiffness reduction

0:65 is applied.
a
ys ¼ 2:0y1 K1 K2 K3 Fr10:43 (1)
y1
Bridge Piers
Displacement-based fiber elements in OpenSees are used to model
where y1 is the flow depth directly upstream from the bridge pier; a the circular reinforced concrete extended shafts of Bridge 1 and the
is the pier width; and K1 , K2 , and K3 are correction factors for pier reinforced concrete wall-type piers of Bridge 2. As recommended
nose shape, angle of attack of flow, and bed condition, respectively.
pffiffiffiffiffiffiffi by Caltrans (2010), the material model for concrete proposed by
Fr1 is the Froude number as defined by V= gy1 , where V and g, Mander et al. (1988) is considered for the stress-strain relations of
respectively, represent the mean velocity of the upstream flow and both unconfined and confined concrete sections. In OpenSees,
the gravitational acceleration. Taking close approximations of river Concrete07 and Steel02 materials are assigned to define material
cross sections at the bridge sites, Manning’s equation is used to
models of concrete and reinforcing steel, respectively. The wall-
determine y1 and V simultaneously for reasonable values of stream
type piers in Bridge 2 were not confined enough (as can be seen
slope and roughness coefficients. For the studied bridges, determin-
from Fig. 2), which made the confined concrete properties in these
istic scour depth estimates at all bent locations are listed in Table 1.
piers nearly equal to the unconfined concrete properties.
As can be observed from this table, scour depths for Bridge 2 do not
To validate the modeling of bridge piers in OpenSees, the same
reach the pile cap at Bent 1 and Bent 3, and marginally go below the
modeling approach is used for numerous reinforced concrete col-
top elevation of the pile cap at Bent 2. It is important to note here
umns that were experimentally tested in the past under cycling load-
that the values presented in Table 1 are reasonable estimates of
ing. The lateral load-displacement responses of these columns
scour. In cases where an exact magnitude of scour is needed, a com-
obtained from OpenSees are observed to be well in accordance with
prehensive calculation of scour through detailed hydraulic analysis
that acquired from past experiments. Hence, the element formula-
is warranted. Such a detailed hydraulic analysis is beyond the scope
of the present study. tion employed for piers is sufficient for the use in nonlinear model-
ing of the bridges.

Finite-Element Analyses of Bridges Foundations

At the foundations, interaction between the foundations and the
surrounding soil is represented through a series of soil springs
Bridge Modeling placed at various depths along the length of the foundation ele-
The generation of fragility curves for the bridges requires a large ments. These soil springs are basically zero-length elements that
number of numerical simulations of the bridges subjected to various characterize the nonlinear soil resistances developed due to the
combined seismic and flood hazard levels. The simulations of the movement of bridge foundations in three translational directions.
bridges are performed with the finite-element analysis (FEA) plat- For both bridges, nonlinear soil resistances in two horizontal direc-
form OpenSees. The three-dimensional finite-element models of tions are modeled with conventional p-y springs. For Bridge 2, shaft

Table 1. Estimated Maximum Scour Depths (ys in m) at Foundations of Bridge 1 and Bridge 2

Bent/Pier 1.1-year flood 2-year flood 10-year flood 20-year flooda 50-year flooda 100-year flooda
Bridge 1 Q = 318.5 m3/s Q = 804.6 m3/s Q = 1,963.3 m3/s Q = 2,237.0 m3/s Q = 2,237.0 m3/s Q = 2,237.0 m3/s
Bent 2 1.85 2.47 3.16 3.28 3.28 3.28
Bent 3 1.83 2.46 3.16 3.27 3.27 3.27
Bent 4 1.28 2.33 3.08 3.20 3.20 3.20
Bridge 2 Q = 90.2 m3/s Q = 322.0 m3/s Q = 1,121.4 m3/s Q = 1,589.9 m3/s Q = 2,349.6 m3/s Q = 3,044.1 m3/s
Pier 2b 0.00 0.00 1.92 2.13 2.37 2.53
Pier 3 1.32 1.74 2.21 2.38 2.58 2.73
Pier 4 1.37 1.78 2.24 2.40 2.61 2.75
a
Flood events with return period equal to or more than 20 years have the same peak discharge.
b
The pier is not under water for the no-flood, 1.1-year-flood, and 2-year-flood conditions.

© ASCE 04015063-5 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063

 resistance (along the pile length) is modeled with t-z springs. For the Bearings in Bridge 1
modeling of soil springs, pySimple1 and tzSimple1 materials from Polytetrafluoroethylene (PTFE)/elastomeric bearings are used at
OpenSees library are utilized. At various depths, p-y relations are both abutments of Bridge 1. These bearings are generally designed
calculated following the recommendations of the American for thermal expansion because they can accommodate horizontal
Petroleum Institute (API 2003), and Mosher’s (1984) relation is translations (because of the low friction provided at the interface of
utilized to calculate t-z relations. Full fixity conditions are consid- PTFE disks and the stainless steel surface) and rotations (because of
ered at the pile bottoms for both bridges. The pile shafts with steel the elastomeric bearing pad at the bottom of the PTFE disks)
casings of Bridge 1 and the prestressed concrete piles of Bridge 2 (Konstantinidis et al. 2008). To capture the complex nature of this
are modeled with linear elastic beam-column elements because bearing, two linear and two nonlinear elements are introduced.
they are expected to stay elastic during seismic excitations. Bridge Linear elements are employed to account for the vertical deforma-
foundations in the presence of scour are modeled by removing tion and rotation of the bridge girder at the global transverse axis of
soil springs from the top to accommodate scour depths. In so the bridge. In two horizontal directions, nonlinear elements with
doing, the properties of the remaining springs are recalculated ev- elastic–perfectly plastic hysteretic backbone curves are assigned to
Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

ery time with respect to the new river bed elevation after scour model lateral seismic responses. These backbone curves are charac-
occurs. terized with initial stiffness and yield force that are equal to, respec-
tively, the shear stiffness of the elastomeric bearing pad and the fric-
Abutments tion force developing at the PTFE–stainless steel interface. Bearing
The modeling of bridge abutments differs from Bridge 1 to Bridge properties in each horizontal direction of the bridge are considered
2. The key components for the modeling of the seat-type abutments to be the same. Typical design values of the shear modulus of elas-
of Bridge 1 are bearings, shear keys, and abutment response in the tomer and the friction coefficient at the PTFE–stainless steel inter-
longitudinal and transverse directions, whereas the same for the in- face are taken as G ¼ 107:5 psi (Caltrans 2000) and m ¼ 0:06
tegral abutments of Bridge 2 are abutment piles and abutment (Caltrans 1994), respectively. Such modeling of PTFE/elastomeric
response in the longitudinal and transverse directions. Fig. 5 shows bearings is validated by comparing numerical load-deformation
schematic abutment models of Bridge 1 and Bridge 2. The current response of a PTFE/elastomeric bearing with that obtained from an
subsection mostly discusses abutment response in the longitudinal experimental study performed by Konstantinidis et al. (2008). This
and transverse directions, leaving bearing and shear keys of Bridge comparison confirmed the use of the aforementioned backbone
1 for the following two subsections. curves for a realistic modeling of PTFE/elastomeric bearings in
In the longitudinal direction of Bridge 1, the assembly of large structures (such as Bridge 1) with several nonlinear
backwall–backfill interaction and the gap between bridge girder components.
and abutment backwall is modeled in OpenSees with elastic–
perfectly plastic gap elements. During seismic excitations, passive Shear Keys in Bridge 1
resistance arises from the backwall–backfill interaction when the Exterior shear keys are provided only in the transverse direction of
bridge girder pushes on the backwall after the gap is completely the Bridge 1 abutments. According to the geometric and material
closed. This resistance is represented with an elastic–perfectly properties of these shear keys, they are expected to fail in a com-
plastic backbone curve, as suggested by Caltrans (2010), and the bined shear–flexure mode instead of a pure shear mode. This fail-
recommended passive resistance capacity is increased by 50% in ure mode follows the observations made from an experimental
order to account for dynamic loading conditions. The same mod- study on similar exterior shear keys (Bozorgzadeh et al. 2007). In
eling technique, except for a gap in element definition, is used for OpenSees, these elements are characterized with a nonlinear force–
the backwall–backfill interaction at the abutment in the longitudi- deformation relation on the basis of the hysteretic model proposed
nal direction of Bridge 2. For this bridge, abutment piles take an by Megally et al. (2002).
active role with the movement of the integrated bridge girder–
abutment system (due to integral abutments) during seismic exci- In-Span Hinge in Bridge 2
tations. These abutment piles are modeled in a way similar to the The modeling of the in-span hinge in Bridge 2 includes the
modeling of the foundation piles. In the transverse direction, the longitudinal response of the elastomeric bearings and the
abutments of both bridges are modeled as recommended by pounding of adjacent bridge decks. Linear elastic–perfectly
Aviram et al. (2008). plastic elements are used to model the longitudinal response of

TOP VIEW 3D VIEW 1. Abutment spring in

1. Bearing element in
3 transverse direction transverse direction
2. Shear key element 1 Bridge girder 2. Abutment back wall
2 1 3. Abutment spring in element in longitudinal
transverse direction direction
4. Abutment back wall 3. Abutment pile (soil
Bridge springs not shown)
4 element with a gap
i j girder 1
property
6 5. Bearing element in 2
longitudinal direction
5 6. Full restraint except free
Abutment rigid Girder rigid 3
link movement in transverse
link direction
(a) (b)

Fig. 5. Abutment modeling: (a) Bridge 1; (b) Bridge 2

© ASCE 04015063-6 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063

 the elastomeric bearings. The initial stiffness of this element is pounding of the two adjacent bridge decks during seismic
represented by the shear stiffness of the elastomer, and its excitations.
force capacity is taken to be equal to the frictional resistance
developed between the concrete surface and elastomer. The
Modal Analyses of Bridges
interface friction coefficient is taken as 0.40 (Caltrans 2010).
The pounding between adjacent bridge decks is modeled with Modal analyses of bridges in the presence and absence of flood-
zero-length elements having bilinear hysteretic backbone induced scour are performed prior to nonlinear time-history analy-
curves and a gap representing the physical gap between adja- ses. Fig. 6 and Fig. 7 depict the first four fundamental modes (and
cent bridge decks at the hinge location. The properties of this related modal periods) of Bridge 1 and Bridge 2, respectively, at
backbone curve are obtained from Muthukumar (2003). This their original states (i.e., without scour). Table 2 lists the fundamen-
backbone curve approximates the total energy lost due to tal modal periods of these bridges at all flood levels. As can be seen
Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

Longitudinal: T1=0.66 sec. Transverse: T2=0.54 sec.

Vertical: T3=0.54 sec. Torsional: T4=0.46 sec.

Fig. 6. Fundamental mode shapes of Bridge 1

Longitudinal-1: T1=0.72 sec.

Longitudinal-2: T2=0.67 sec.

Transverse: T3=0.44 sec. Vertical: T4=0.42 sec.

Fig. 7. Fundamental mode shapes of Bridge 2

Table 2. Modal Periods (in sec) of Bridge 1 and Bridge 2

No 1.1-year 2-year 10-year 20-year 50-year 100-year

Bridge Mode flood flood flood flood flooda flooda flooda
Bridge 1 — ys ðmaxÞ ¼ 1:85m ys ðmaxÞ ¼ 2:47m ys ðmaxÞ ¼ 3:16m ys ðmaxÞ ¼ 3:28m ys ðmaxÞ ¼ 3:28m ys ðmaxÞ ¼ 3:28m
Longitudinal 0.66 0.72 0.74 0.77 0.77 0.77 0.77
Transverse 0.54 0.58 0.60 0.62 0.62 0.62 0.62
Vertical 0.54 0.54 0.54 0.55 0.55 0.55 0.55
Torsional 0.46 0.48 0.48 0.49 0.49 0.49 0.49
Bridge 2 — ys ðmaxÞ ¼ 1:37m ys ðmaxÞ ¼ 1:78m ys ðmaxÞ ¼ 2:24m ys ðmaxÞ ¼ 2:40m ys ðmaxÞ ¼ 2:61m ys ðmaxÞ ¼ 2:75m
1st longitudinal 0.72 0.72 0.72 0.72 0.73 0.73 0.73
2nd longitudinal 0.67 0.67 0.67 0.68 0.68 0.68 0.68
Transverse 0.44 0.45 0.46 0.46 0.47 0.47 0.47
Vertical 0.42 0.42 0.42 0.42 0.42 0.42 0.42
a
Flood events with return period equal to or more than 20 years have the same peak discharge.

© ASCE 04015063-7 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063

 from this table, the modal periods of Bridge 1 in the longitudinal Bridge 2 under the ground motion NGA0790 (1989 Loma Prieta
and transverse directions increase with the increase in scour depth Earthquake, scaled with 2.0). Only a portion of the response time
up to that results from a 10-year flood event. For Bridge 2, the histories is displayed in this figure to allow for a close observation
change in the fundamental modal period is small as compared with of bridge seismic responses at different flood conditions. As
that for Bridge 1, and no change is observed beyond the 20-year shown in Fig. 8, a slight change in pier-top displacement is
flood event. observed for Bridge 1 with an increasing flood hazard level,
whereas no such change is found for Bridge 2. Note that the
responses plotted here are for two specific earthquakes and at two
Fragility Curves and Surfaces specific locations on the bridges. For both bridges, pier-top dis-
placement is not the only factor controlling the resulting seismic
Time-History Analyses of Bridges damage. The fragility analysis presented later in the article pro-
vides a comprehensive overview of bridge responses at different
Nonlinear time-history analyses of Bridge 1 and Bridge 2 are per- bridge components and for all ground motions considered for
Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

formed under the ground motions selected for the respective bridge analysis of the two bridges.
site. The responses of potentially critical bridge components (such
as piers, abutments, bearings, and shear keys) are recorded for each
Component-Level Damage States
analysis. To exemplify, the response time histories of Bridge 1 and
Bride 2 at the pier tops are shown in Fig. 8. For Bridge 1, the trans- The obtained results from the time-history analyses are processed to
verse displacement at the top of a pier (left bridge, left column) in generate fragility curves of Bridge 1 and Bridge 2 at the component
Bent 3 under the ground motion NGA0008 (1941 Northern and system levels. Component-level fragility curves characterize
California Earthquake, scaled with 2.0) is plotted in Fig. 8(a). Fig. 8 performance of various bridge components (such as piers) at differ-
(b) shows the longitudinal displacement at the top of Pier 2 of ent damage states, whereas the same at the system level indicate the
overall performance of the bridge. These damage states essentially
signify intermediate and ultimate limit states [such as minor dam-
60 age, moderate damage, major damage, and collapse; FEMA (2013)]
No flood
of various bridge components. For each component, damage states
1.1-yr flood
40 are quantitatively represented with threshold limits (as obtained
Pier Top Displacement (mm)

2-yr flood from literature and detailed later). Following a time-history analy-
sis, bridge response at a component is compared with its predefined
20 threshold limits; if the response is lower than the threshold limit of a
damage state, say, moderate damage, and higher than that for its
0 previous damage state (i.e., minor), the component is said to have
minor damage. When developing system-level fragility curves of
bridges, it is desirable that the damage states of different bridge
-20 components be allied to the global performance of bridges [such as
fully operational, operational, life safety, and collapse; FHWA
-40 (2006)] such that the component- and system-level damage states
10-yr flood
complement each other.
20-yr flood
In this study, component-level fragility curves are generated
-60 for bridge components that have relatively high damage potential
5 6 7 8 9 10 11 12 13 14 15 and may lead to moderate to major damage of bridges (if not col-
(a) Time (sec) lapse) under the multihazard condition. Such components are
identified as piers, abutments, shear keys, and abutment bearings
80 for Bridge 1, and as piers, abutments, and bearings at the in-span
No flood
1.1-yr flood
hinges for Bridge 2. For the components having a direct effect
60 on the vertical stability and load-carrying capacity of a bridge, a
Pier Top Displacement (mm)

2-yr flood
full range of damage states is addressed, including major damage
10-yr flood
and complete collapse that may result in closure of the bridge
40 20-yr flood
from traffic. On the other hand, for the components that do not
directly affect the vertical stability of the bridge, only their lower
20 damage states (such as minor and moderate damage) are consid-
ered. The threshold limits of various component damage states
are obtained from previous studies, as detailed in the following
0 paragraphs and summarized in Tables 3 and 4 for Bridge 1 and
Bridge 2, respectively.
-20 Bridge piers may have two major seismic failure modes, flexural
50-yr flood and shear. For both bridges, the shear capacity of each bridge pier is
100-yr flood compared with the developed shear force. In all ground motion
-40 analyses, shear demand is observed to be lower than shear capacity;
10 11 12 13 14 15 16 17 18 19 20
this confirmed the absence of shear failure in the bridge piers.
(b) Time (sec)
Hence, damage of bridge piers under the multihazard scenario is
Fig 8. Sample pier-top horizontal displacement histories: (a) Bridge 1; evaluated based on the flexural response of the piers. Curvature duc-
(b) Bridge 2 tility mw is used as the engineering demand parameter (EDP) for the
assessment of flexural damage of bridge piers. The threshold limits

© ASCE 04015063-8 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063

 Table 3. Damage Threshold Limits for Bridge 1

Component EDP Minor Moderate Major Collapse

Piers Curvature ductility 1:0  m f < 4:0 4:0  m f < 8:0 8:0  m f < 12:0 12:0  m f
Abutment Longitudinal deformation in passive direction (mm) 32  Dlong;p < 96 96  Dlong;p — —
Shear key Transverse deformation (mm) 17  Dsk < 201 201  Dsk < 422 422  Dsk —
PTFE/elastomeric Longitudinal deformation (mm) 40  Db;long < 213 213  Db;long < 387 387  Db;long < 560 560  Db;long
bearing Transverse deformation (mm) 40  Db;trans < 213 213  Db;trans — —

Table 4. Damage Threshold Limits for Bridge 2

Component EDP Minor Moderate Major Collapse

Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

Piers Curvature ductility 1:0  m f < 2:0 2:0  m f < 3:5 3:5  m f < 5:0 5:0  m f
Abutment Longitudinal deformation in passive direction (mm) 57  Dlong;p < 170 170  Dlong;p — —
Longitudinal deformation in active direction (mm) 38  Dlong;a < 102 102  Dlong;a — —
Elastomeric Longitudinal deformation (mm) 30  Db;long < 88 88  Db;long < 282 282  Db;long < 335 335  Db;long
bearing

of curvature ductility are taken from Ramanathan (2012). This liter- to slide. On the other extreme, the collapse state of this bearing is
ature defined these threshold limits on the basis of a comprehen- defined with the deformation for which the bridge deck falls off
sive review of lateral load tests of reinforced concrete bridge piers from the bearing. In between minor damage and collapse, two inter-
and by categorizing test results according to pre-1971 brittle piers, mediate damage states (moderate and major) are assumed to be
1971–1990 strength-degrading piers, and post-1990 ductile piers. evenly distributed. Threshold limits of Db for moderate and major
Accordingly, the threshold limits of the curvature ductility of the damage are decided based on the experimental results obtained by
piers of Bridge 1 and Bridge 2 are obtained for the minor-damage, Konstantinidis et al. (2008) on PTFE/elastomeric bearings under
moderate-damage, major-damage, and collapse states, as presented seismic loading, in which gradual deformation of PTFE bearing
in Tables 3 and 4. pads was observed until excessive shedding. These threshold values
Seismic damage at bridge abutments can be initiated from three are listed in Table 3. For bearing deformation in the transverse
different abutment deformations: longitudinal deformation in the direction, only minor- and moderate-damage threshold limits are
passive direction (Dlong;p ) and in the active direction (Dlong;a ) and specified, with the same values adopted for bearing deformation in
deformation in the transverse direction (Dtrans ). For Bridge 1, abut- the longitudinal direction.
ment damage in the longitudinal direction is identified only due to Seismic damage of elastomeric bearings at the in-span hinge of
its passive deformation (Dlong;p ); deformation in the active direction Bridge 2 is evaluated based on their horizontal longitudinal deforma-
is ignored in the model definition because of the nature of this abut- tion (Db;long ). The threshold limits at the moderate-damage and col-
ment. The threshold limit of Dlong;p at the minor-damage state is lapse states correspond to the longitudinal deformation when sliding
taken as the yield displacement of the backwall–backfill interaction starts at the concrete–elastomer interface and the bridge deck falls off
curve. At moderate damage, the threshold limit of Dlong;p is taken to
from the bearings, respectively. Threshold limit at major damage is
be the maximum displacement, which is 5% of the backwall height
taken as the average of that for the moderate-damage and collapse
for granular backfills (Shamsabadi et al. 2007). Similar to Dlong;p ,
states.
threshold limit of Dtrans at the minor-damage state is taken as the
yield displacement of the backbone curve in the transverse direc-
tion. However, abutment deformations in the transverse direction Component-Level Fragility Curves
are found to be within elastic range (i.e., lower than the minor- A seismic fragility curve can be defined with a two-parameter log-
damage threshold limit) for both bridges under all ground motions. normal distribution (Shinozuka et al. 2000), as follows:
Hence, no further damage state involving Dtrans is defined for any of
the bridges. In addition to Dlong;p , the abutment damage of Bridge 2  
lnðxj =ck Þ
is identified in terms of Dlong;a . In this case, the threshold limits for Fðxj ; ck ; z k Þ ¼ U (2)
zk
the minor- and moderate-damage states are directly taken from
Ramanathan (2012).
The seismic damage of shear keys in Bridge 1 is assessed with where the fragility function FðÞ represents the failure probability
respect to the transverse deformation of the shear key elements of a bridge component at damage state k (such as minor, moderate,
(Dsk ). Threshold limits of Dsk at the minor-, moderate-, and major- major, and collapse state) under a ground motion j with PGA xj .
damage states are determined from the element load-deformation Fragility parameters ck and z k refer to the median and lognormal
curve. Performance levels, such as the onset of yielding of shear standard deviation at damage state k, respectively. In the current
key reinforcement and yielding of all rebars crossing the crack study, fragility parameters are estimated by using the method of
zone, are identified in order to define these damage states. maximum likelihood, in which the likelihood function L is
The seismic damage of PTFE/elastomeric bearings in Bridge 1 expressed as
is assessed based on the horizontal deformation of bearing elements
(Db ). Beyond yielding, the threshold limit of Db at the minor-dam- Y
N
age state is set to 40 mm, considering that damage in this component L¼ ½Fðxj ; ck ; z k Þrj ½1  Fðxj ; ck ; z k Þ1rj (3)
does not necessarily begin to occur right after the bridge deck starts j¼1

© ASCE 04015063-9 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063

 in which rj expresses the damage condition of the bridge component between the two ends do not change enough to cause notable varia-
at damage state k under PGA xj . It takes a value equal to 1 or 0 tion in pier flexural damage and its seismic vulnerability with
depending on whether or not the component damage state k is increasing scour depth. Among other bridge components, the change
exceeded for xj . The log-standard deviation indicates the shape of in seismic vulnerability with scour depth is observed only for bridge
the fragility curves. A single dispersion value of z k ¼ 0:6 is bearings (in the longitudinal direction) at the minor-damage state.
adopted from FEMA (2013) for all damage states in order to prevent This is due to increased longitudinal displacement at the superstruc-
the intersection of any two fragility curves across damage states and ture level. Overall, the result presented in Fig. 9 indicates that re-
multihazard cases. gional flood hazard does not impose any significant threat to the seis-
Fig. 9 shows the seismic fragility curves of key nonlinear compo- mic vulnerability of bridge components, particularly at higher
nents of Bridge 1 at various flood hazard levels. Median values of all damage levels. This is a reasonable outcome for a newly constructed
fragility curves are given in these figures. Because the same log- bridge with ductile piers and large-diameter pile shafts socketed into
standard deviation value is used for all fragility curves, these curves rock layers.
can be compared in terms of their median values. A lower median The seismic fragility curves developed for critical compo-
Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

value signifies higher seismic vulnerability. As Fig. 9 indicates, nents of Bridge 2 are shown in Fig. 10. This figure suggests
bridge piers are the only component that can lead to major damage that piers are the only component of Bridge 2 that may lead to
and complete collapse of the bridge under the multihazard scenario. major damage and collapse of the bridge under seismic excita-
However, the seismic vulnerability of this component is found to be tions in the presence and absence of scour. Significant changes
least sensitive to the flood hazard, except for that at the minor-dam- in the seismic fragility of this component are observed at the
age state. This is because displacement and rotation at the top and at major-damage and collapse states with the increase in flood haz-
the foundation level of the piers increase simultaneously in the pres- ard level. This is because the exposed height of the bridge piers
ence of scour at the foundations. Because of this simultaneous increases with scour depth (“H river bed” decreases; Fig. 2),
increase, resultant displacement and rotation of the bridge piers making the bridge piers more vulnerable under seismic ground

1.0 1.0
Probability of Exceeding a Damage State

cmin. of Exceeding a Damage State

Minor Damage
No flood
No fl. 0.409 1.1-yr flood Minor Damage
0.8 1.1-yr 0.364 0.8 2-yr flood cmin. = 0.741
2-yr 0.364 10-yr flood
Moderate Damage
10-yr 0.364
0.6 cmod. = 0.688 0.6
20-yr flood
20-yr 0.364

0.4 Major Damage & 0.4

Collapse State Moderate Damage
cmaj. = ccol. = 0.815 cmod. = 0.894
Probability

0.2 0.2
r

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
(a) PGA (g) (b) PGA (g)

1.0 1.0
Probability of Exceeding a Damage State
Probability of Exceeding a Damage State

No flood
Minor Damage 1.1-yr flood
0.8 cmin. 0.8
Minor damage due to
shear key deformation
2-yr flood
No flood 0.451
10-yr flood
cmin. = 0.815
1.1-yr flood 0.423
0.6 0.6 20-yr flood
2-yr flood 0.395
10-yr flood 0.384
20-yr flood 0.384
0.4 0.4

Moderate Damage 0.2

0.2 Minor damage due to
cmod. = 1.935 bearing transverse deformation
cmin. = 1.035
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
(c) PGA (g) (d) PGA (g)

Fig. 9. Component-level fragility curves of Bridge 1: (a) pier flexural damage; (b) abutment passive deformation; (c) bearing longitudinal de-
formation; (d) shear key and bearing transverse deformations [Note: when the fragility curves for different flood conditions have the same me-
dian (c) value at a certain damage state, these curves and their legends are clustered on the figure, and the common median value is reported next
to the clustered curve]

© ASCE 04015063-10 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063

 1.0 1.0

Probability of Exceeding a Damage State

Probability of Exceeding a Damage State
ccol.
No flood 0.842
Minor 0.8
0.8 Damage
Moderate 1.1-yr flood 0.748
Damage
2-yr flood 0.720
10-yr flood 0.689
0.6 0.6
cmin. cmod. cmaj. 20-yr flood 0.689
No fl. 0.230 0.407 0.684 50-yr flood 0.634

0.4 1.1-yr 0.224 0.407 0.659 0.4 100-yr flood 0.634

2-yr 0.224 0.407 0.659
10-yr 0.222 0.359 0.599
0.2 Major 20-yr 0.222 0.359 0.566 0.2
Damage
50-yr 0.222 0.359 0.566
Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

100-yr 0.222 0.354 0.525

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
(a) PGA (g) (b) PGA (g)

1.0 1.0
Probability of Exceeding a Damage State

Probability of Exceeding a Damage State

cmin. cmod. No flood
Minor Damage No fl. 0.204 1.081 1.1-yr flood
0.8 1.1-yr 0.204 1.081 0.8 2-yr flood
2-yr 0.202 1.081 10-yr flood
10-yr 0.197 1.081 20-yr flood Minor damage due to
0.6 0.6
20-yr 0.192 0.980 50-yr flood bearing deformation
50-yr 0.192 0.980 100-yr flood cmin.= 1.087
0.4 100-yr 0.192 0.980 0.4

0.2 Moderate Damage 0.2 Minor damage due to

abutment passive deformation
cmin.= 1.245
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
(c) PGA (g) (d) PGA (g)

Fig. 10. Component-level fragility curves of Bridge 2: (a) pier flexural damage at the minor-, moderate-, and major-damage states; (b) pier flexural
damage at the collapse state; (c) abutment active deformation; (d) abutment passive deformation and bearing longitudinal deformation [Note: when
the fragility curves for different flood conditions have the same median (c) value at a certain damage state, these curves and their legends are clustered
on the figure, and the common median value is reported next to the clustered curve]

motions. For all other cases including minor and moderate dam- Fragility Surfaces
age of bridge piers, slight to no change in seismic vulnerability
Fragility surfaces provide a comprehensive visualization of the
is observed with increasing flood hazard level.
combined effect of earthquake and flood hazards on bridge fail-
ure probabilities at various damage levels. In these, hazard
intensities are generally plotted along two horizontal axes, and
System-Level Fragility Curves
the surface denotes the exceedance probability of a bridge dam-
The system-level fragility curves of Bridge 1 and Bridge 2 are pre- age state. Some studies have been performed in the past to de-
sented in Fig. 11. As previously mentioned, the damage state crite- velop fragility surfaces by considering two intensity measures
ria of critical bridge components are described such that global from the same natural hazard (such as earthquake), although
damage states for the bridges can be defined in a consistent manner. very little effort is made to generate fragility surfaces demon-
For both of the bridges, the moderate-damage, major-damage, and strating the combined effect of more than one natural hazard
collapse states are controlled by the damage in bridge piers. (Lee and Rosowsky 2006; Wang et al. 2014a). To develop fra-
Therefore, no change in bridge seismic fragility with increasing gility surfaces, the present study considers peak annual flow
flood hazard level is observed for Bridge 1 at these damage states, discharge as the flood hazard intensity measure and PGA as the
whereas the change is significantly large at the major-damage and earthquake hazard intensity measure. These intensity measures
collapse states for Bridge 2. At various flood hazard levels, the are considered to be two statistically independent random varia-
minor-damage state of Bridge 1 is jointly governed by bearing bles, and their joint cumulative probability distribution provides
response in the longitudinal direction and the response of bridge the failure probability of the bridge under the multihazard sce-
piers. For Bridge 2, abutment active deformation primarily governs nario. A bivariate lognormal distribution is used to define the
this damage state. Accordingly, variations in fragility curves are joint probability density. Hence, the fragility surface is defined
observed with increasing flood hazard level. here as

© ASCE 04015063-11 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063

 1.0 Similar to fragility curves, distribution parameters for fragility
Probability of Exceeding a Damage State cmin. Minor Damage surfaces can be calculated through the maximum likelihood
No fl. 0.409 method. The log-standard deviation for seismic hazard z 1;k is taken
0.8 1.1-yr 0.364 to be equal to 0.6 in order to keep consistency between fragility
2-yr 0.354 Moderate Damage
curves and surfaces. The likelihood function is given as
10-yr 0.345
0.6 cmod. = 0.688 Y
N
20-yr 0.345
L¼ ½Fðx1;j ; x2;j ; c1;k ; c2;k ; z 1;k ; z 2;k Þrj
j¼1
0.4 Major Damage &
Collapse State ½1  Fðx1;j ; x2;j ; c1;k ; c2;k ; z 1;k ; z 2;k Þ1rj (7)
cmaj. = ccol. = 0.815
0.2 where j is a multihazard case involving earthquake and flood haz-
ards, rj ¼ 1 or 0 depending on whether or not a damage state k is
Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

exceeded for a ground motion with PGA ¼ x1;j and a flood event
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 with peak-flow discharge ¼ x2; j , and N represents the total number
(a) PGA (g) of multihazard cases.
Fig. 12 shows the fragility surfaces of Bridge 1 and Bridge 2
1.0 developed based on system-level bridge damage data, and corre-
Probability of Exceeding a Damage State

sponding fragility parameters are listed in the attached table. As the

figure shows, the fragility surface for Bridge 1 is developed only at
0.8 Moderate the minor-damage state [Fig. 12(a)]. For the other three damage
Damage
states of the bridge, the median fragility parameters showed no vari-
Minor ation with increasing flood hazard level (Fig. 11). Hence, the fragil-
0.6 Damage
ity surfaces of the bridge at these three damage states would be lin-
No fl. ear extensions of the respective fragility curves along the axis
0.4 1.1-yr fl. presenting flood hazard intensity. For Bridge 2, however, the fragil-
2-yr fl. ity surfaces are developed for all four damage states. At minor dam-
Collapse 10-yr fl. age of both bridges and moderate damage of Bridge 2, the median
State
0.2 20-yr fl. values of peak-flow discharge are calculated to be very small, indi-
50-yr fl. cating insignificant impact of flood hazard on the multihazard per-
Major Damage 100-yr fl. formance of these bridges at these damage states. The same is also
0.0 evident from the fragility curves of these bridges developed at the
0.0 0.2 0.4 0.6 0.8 1.0 1.2 same damage states, which show slight changes in bridge fragility
(b) PGA (g) characteristics with increasing flood hazard level (Fig. 11). The fra-
gility surfaces at the major-damage and collapse state of Bridge 2
Fig. 11. System-level fragility curves: (a) Bridge 1; (b) Bridge 2
demonstrate the significance of flood hazard on bridge multihazard
[Note: when the fragility curves for different flood conditions have the
performance, which is also recognized from the fragility curves of
same median (c) value at a certain damage state, these curves and their
the bridge at these higher damage states.
legends are clustered on the figure, and the common median value is
reported next to the clustered curve]
Risk Evaluation

Fðx1 ; x2 ; c1;k ; c2;k ; z 1;k ; z 2;k Þ A risk-based framework is utilized to predict the negative conse-
quences from future occurrences of the aforementioned multihazard
xð2 xð1 event in California. For the two California bridges, expected risk
due to regional hazards is expressed in the form of risk curves. In
¼ f ðx1 ; x2 ; c1;k ; c2;k ; z 1;k ; z 2;k Þdx1 dx2 (4)
general, postevent consequences of bridge damage are represented
0 0 by socioeconomic losses arising from various sources, such as post-
event bridge restoration, traffic delay, network downtime, and loss
where x1 and x2 represent samples from random variables represent- of opportunity. Among these, the postevent bridge restoration cost
ing PGA and peak-flow discharge, respectively; c1;k and c2;k are is considered in the present risk assessment framework because it is
corresponding median values at damage state k, and z 1;k z 2;k are the direct consequence from bridge damage under regional multiha-
log-standard deviations. The probability density function f ð::Þ of zard events. Bridge owners are most concerned with this loss.
the bivariate lognormal distribution can be defined in the following Following a multihazard scenario m, bridge restoration cost CRm
equations: can be estimated as (Zhou et al. 2010)


1 q X
4
f ðx1 ; x2 ; c1;k ; c2;k ; z 1;k ; z 2;k Þ ¼ exp  (5) CRm ¼ pm ðDS ¼ k j am ; dm Þ Cn rk (8)
2p x1 x2 z 1;k z 2;k 2
k¼1

!2 !2 where pm ðDS ¼ kjam ; dm Þ is the probability that the bridge can sus-
lnðx1 =c1;k Þ lnðx2 =c2;k Þ tain the damage state k under a ground motion with PGA ¼ am that
q¼ þ (6) occurs in the presence of scour resulting from a flood event with
z 1;k z 2;k
peak-flow discharge ¼ dm . For the no-flood condition, pm ðDS ¼

© ASCE 04015063-12 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063

 Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

© ASCE
04015063-13

J. Bridge Eng., 2016, 21(3): 04015063

Fig. 12. Fragility surfaces of Bridge 1 and Bridge 2

J. Bridge Eng.
 k j am ; dm Þ is calculated for seismic hazard only. Cn represents increasing flood hazard levels. For an example, there is 0.03% an-
bridge replacement cost and rk is the damage ratio corresponding to nual chance that the expected bridge restoration cost will exceed
the damage state k. Values of rk as recommended in FEMA (2013) $2,000,000 because of regional seismic hazard only, whereas the
are considered in this study. Bridge replacement costs are calculated same figure is $2,740,000 (37% increase) for regional seismic haz-
by multiplying bridge deck areas with the average unit replacement ard in the presence of scour resulting from a 100-year flood event.
costs. According to construction statistics of Caltrans (2012), aver- Such an increase in the risk of Bridge 2 is an obvious outcome of
age bridge replacement cost per unit area of prestressed concrete enhanced bridge seismic vulnerability, particularly at higher dam-
box girder and reinforced concrete box girder bridges are taken as age levels, in the presence of flood-induced scour.
$1,805.6/m2 and $1,833.3/m2, respectively. A comparison between the response of Bridge 1 and Bridge 2
Fig. 13 shows the risk curves of Bridge 1 and Bridge 2 under re- under the multihazard effect of earthquakes and floods signifies the
gional multihazard scenarios. These curves represent annual influence of bridge design philosophy on bridge performance under
exceedance probabilities of different levels of bridge restoration the multihazard condition. Bridge 1 is a newly constructed bridge;
costs (i.e., risks) due to various intensities of earthquake and flood specific features of the bridge, such as ductile piers and pile shafts,
Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

hazards. To generate these curves, information presented in regional and large-diameter pile shafts, helped in minimizing the impact of
seismic hazard curves is utilized. For a combination of earthquake flood events on the seismic vulnerability of the bridge. Such features
and flood hazards, values of pm ðDS ¼ k j am ; dm Þ) for the two are not present in Bridge 2. Moreover, the increase in the exposed
bridges are obtained from their respective fragility curves (Fig. 11). height of the bridge piers with increasing scour depth makes Bridge 2
For Bridge 1, no change is observed in the bridge risk curve with more seismically vulnerable as the flood hazard level increases. In a
increasing flood hazard level. This observation is obvious because way, the results obtained for Bridge 1 are in accordance with those
the seismic fragility characteristics of the bridge are mostly insensi- observed in Banerjee and Prasad (2013); bigger (large-diameter) foun-
tive to flood hazard. For varying flood hazard levels, the slight varia- dations tend to reduce the impact of flood hazard on bridge seismic
tion observed in the fragility curves of the bridge at the minor-dam- performance. However, the present study showed that the previous
age state was not enough to produce any notable variation in the statement is particularly true when extended pile shafts are used as
bridge risk curve. For Bridge 2, however, seismic risk increases with foundation elements. In cases where pile groups are used for bridge
foundations (as considered in Banerjee and Prasad 2013), outcomes
100 from Bridge 2 indicate that the multihazard performance of bridges
No flood greatly depends on bridge design features.
1.1-yr flood
Annual Probability of Exceedance

-1
10 2-yr flood
10-yr flood
Conclusions
20-yr flood
Two highway river-crossing bridges in California, constructed at dif-
10-2
ferent times and located in different regions, are analyzed for a re-
gional multihazard scenario involving earthquakes and floods. For a
10-3 number of flood cases with varied intensity level, scour depths are
calculated at bridge piers. These scour depths are observed to vary
from one pier to another because of irregular river bed elevations
10-4 across cross sections. The multihazard effect of flood and seismic
events on bridge performance is observed through fragility curves
generated at the component and system levels of both bridges. It is
10-5 found that the component- and system-level fragilities of Bridge 1
0 2,000,000 4,000,000 6,000,000 8,000,000 (on the Sacramento River) are insensitive to regional flood hazard,
(a) Bridge Restoration Cost ($) whereas the fragility curves of the Bridge 2 piers (on the San Joaquin
River) gradually weaken with increasing flood hazard level. The
100 same observation is also made from the fragility surfaces and risk
No flood curves of these bridges. This observation is attributed to the fact that
the large-diameter pile shafts used as the foundation element for
Annual Probability of Exceedance

1.1-yr flood
10-1 2-yr flood Bridge 1 and the seismic design philosophy (ductile design) adopted
10-yr flood for this bridge considerably helped in minimizing the impact of re-
20-yr flood gional flood events on the seismic vulnerability of the bridge. For
10-2 50-yr flood Bridge 2, conversely, the increased exposed height of the bridge piers
100-yr flood with increasing scour depth makes the bridge more seismically vul-
nerable as flood hazard level increases.
-3
10 This research contributes to advancing the knowledge base used
to make decisions concerning the seismic safety of highway bridges
to be constructed in seismically active, flood-prone regions. For the
10-4 first time, real-life bridges are analyzed for the evaluation of fragil-
ity and risk under this regional multihazard condition. The results
obtained from this research facilitate understanding among govern-
10-5 ment agencies and state officials regarding the criticality of this
0 1,000,000 2,000,000 3,000,000 4,000,000 5,000,000
multihazard and its postevent consequences for bridges. The
(b) Bridge Restoration Cost ($)
research can be expanded to quantify uncertainties associated with
Fig. 13. Seismic risk curves: (a) Bridge 1; (b) Bridge 2 fragility and the risk curves of the bridges for the same multihazard
condition. This study can also serve as a foundation for future

© ASCE 04015063-14 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063

 research and actions on the development of strategic plans for repair Lagasse, P. F., Ghosn, M., Johnson, P. A., Zevenbergen, L. W., and
or retrofit prioritization under similar multihazard conditions. Clopper, P. E. (2013). “Risk-based approach for bridge scour predic-
tion.” Final Rep. Prepared for National Cooperative Highway Research
Program, Transportation Research Board, National Research Council,
Acknowledgments Washington, DC, Ayres Associates, Fort Collins, CO.
Lee, K. H., and Rosowsky, D. V. (2006). “Fragility analysis of woodframe
The authors are thankful to the Pennsylvania State University buildings considering combined snow and earthquake loading.” Struct.
Saf., 28(3), 289–303.
(PSU), University Park, Pennsylvania, for providing facilities to
Liang, Z., and Lee, G. C. (2013a). “Bridge pier failure probabilities under
conduct the research. This study was supported by the National combined hazard effects of scour, truck and earthquake. Part I:
Science Foundation through Grant No. 1131359. The authors are Occurrence probabilities.” Earthquake Eng. & Eng. Vib., 12(2), 229–240.
thankful to the California Dept. of Transportation for providing Liang, Z., and Lee, G. C. (2013b). “Bridge pier failure probabilities under
design drawings of the bridges. combined hazard effects of scour, truck and earthquake. Part II: Failure
probabilities.” Earthquake Eng. & Eng. Vib., 12(2), 241–250.
Mander, J. B., Priestley, M. J. N., and Park, R. (1988). “Theoretical stress-
Downloaded from [Link] by Bob Munlemvo Mayemba on 10/23/24. Copyright ASCE. For personal use only; all rights reserved.

References strain model for confined concrete.” J. Struct. Eng., 10.1061

/(ASCE)0733-9445(1988)114:8(1804), 1804–1826.
AASHTO. (2010). AASHTO LRFD bridge design specifications, 5th Ed., Marshall, C. (2014). “California cities seek united front to counter earth-
Washington, DC. quake and flood threats.” h[Link]
Alipour, A., Shafei, B., and Shinozuka, M. (2013). “Reliability-based cali- /19/californian-cities-san-francisco-threat-quake-floodingi (Apr. 3, 2014).
bration of load and resistance factors for design of RC bridges under Megally, S. H., Silva, P. F., and Seible, F. (2002). “Seismic response of sac-
multiple extreme events: Scour and earthquake.” J. Bridge Eng., 10 rificial shear keys in bridge abutments.” Structural Systems Research
.1061/(ASCE)BE.1943-5592.0000369, 362–371. Project, Rep. No. SSRP-2001/23, Univ. of California, San Diego, La
API (American Petroleum Institute). (2003). Recommended practice for Jolla, CA.
planning, designing and constructing fixed offshore platform-working Mosher, R. L. (1984). Load transfer criteria for numerical analysis of axial
stress design, Washington, DC. loaded piles in sand—Part II: Load pile capacity curves for steel and
Arneson, L. A., Zevenbergen, L. W., Lagasse, P. F., and Clopper, P. E. concrete piles, U.S. Army Engineering and Waterways Experimental
(2012). “Evaluating scour at bridges.” Hydraulic Engineering Circular Station, Automatic Data Processing Center, Vicksburg, MS.
(HEC) No. 18, Publication No. FHWA-HIF-12-003, Federal Highway Muthukumar, S. (2003). “A contact element approach with hysteresis
Administration, U.S. Dept. of Transportation, Washington, DC. damping for the analysis and design of pounding in bridges.” Ph.D. dis-
Aviram, A., Mackie, K. R., and Stojadinovic, B. (2008). “Effect of abutment sertation, Georgia Institute of Technology, Atlanta.
modeling on the seismic response of bridge structures.” Earthquake Eng. OpenSees (Open System for Earthquake Engineering Simulation) 2.4.0
& Eng. Vib., 7(4), 395–402. [Computer software]. Pacific Earthquake Engineering Research Center,
Banerjee, S., and Prasad, G. G. (2013). “Seismic risk assessment of rein- University of California, Berkeley, CA.
forced concrete bridges in flood-prone regions.” Struct. Infrastruct. Eng., PEER (Pacific Earthquake Engineering Research Center). (2013). “NGA
9(9), 952–968. database.” h[Link] (Feb. 13, 2013).
Bozorgzadeh, A., Megally, S. H., Ashford, S., and Restrepo, J. I. (2007). Prasad, G. G., and Banerjee, S. (2013). “The impact of flood-induced
“Seismic response of sacrificial exterior shear keys in bridge abutments.” scour on seismic fragility characteristics of bridges.” J. Earthquake
Rep. No. SSRP-04/14, Univ. of California, San Diego, La Jolla, CA. Eng., 17(6), 803–828.
Caltrans. (1994). “Bridge bearings.” Chapter 7-1, Caltrans bridge memo to Ramanathan, K. N. (2012). “Next generation seismic fragility curves for
designers, California Dept. of Transportation, Sacramento, CA. California bridges incorporating the evolution in seismic design phi-
Caltrans. (2000). “Bearings.” Section 14, Bridge design specifications, losophy.” Ph.D. dissertation, Georgia Institute of Technology, Atlanta.
California Dept. of Transportation, Sacramento, CA. Ries, K. G. (2007). “The National Streamflow Statistics program: A com-
Caltrans. (2010). Caltrans seismic design criteria version 1.6, California puter program for estimating streamflow statistics for ungaged sites.”
Dept. of Transportation, Sacramento, CA. Chapter 6 of Book 4, U.S. Geological Survey techniques and methods,
Caltrans. (2012). “2012 construction statistics.” h[Link] USGS, Reston, VA.
/esc/estimates/Construction_Stats_2012.pdfi (Mar. 17, 2014). Shamsabadi, A., Rollins, K. M., and Kapuskar, M. (2007). “Non-
FEMA (Federal Emergency Management Agency). (2009a). Flood insur- linear soil-abutment-bridge structure interaction for seismic per-
ance study, Shasta County, California, and incorporated areas, U.S. formance-based design.” J. Geotech. Geoenviron. Eng., 10.1061
Government Printing Office, Washington, DC. /(ASCE)1090-0241(2007)133:6(707), 707–720.
FEMA (Federal Emergency Management Agency). (2009b). Flood insur- Shinozuka, M., Feng, M. Q., Lee, J., and Naganuma, T. (2000).
ance study, San Joaquin County, California, and incorporated areas, “Statistical analysis of fragility curves.” J. Eng. Mech., 10.1061
U.S. Government Printing Office, Washington, DC. /(ASCE)0733-9399(2000)126:12(1224), 1224–1231.
FEMA (Federal Emergency Management Agency). (2013). HAZUS MR4 USGS. (2013a). “National seismic hazard mapping project.” hhttp:
technical manual, Dept. of Homeland Security, Emergency Preparedness //[Link]/hazardtool/i (Dec. 18, 2013).
and Response Directorate, FEMA Mitigation Division, Washington, DC. USGS. (2013b). “National water information system.” h[Link]
FHWA (Federal Highway Administration). (2006). “Seismic retrofitting .[Link]/usa/nwis/peaki (Mar. 24, 2013).
manual for highway structures: Part 1—Bridges.” Rep. No. FHWA- Wang, Z., Duenas-Osorio, L., and Padgett, J. E. (2014a). “Influence of scour
HRT-06-032, U.S. Dept. of Transportation, McLean, VA. effects on the seismic response of reinforced concrete bridges.” Eng.
Ghosn, M., Moses, F., and Wang, J. (2003). “Design of highway bridges for Struct., 79, 202–214.
extreme events.” National Cooperative Highway Research Program Wang, Z., Padgett, J. E., and Duenas-Osorio, L. (2014b). “Risk-consistent
(NCHRP) Rep. 489, Transportation Research Board, National Academies calibration of load factors for the design of reinforced concrete bridges
Press, Washington DC. under the combined effects of earthquake and scour hazards.” Eng.
Interagency Advisory Committee on Water Data. (1982). “Guidelines for deter- Struct., 79, 86–95.
mining flood flow frequency.” Bulletin 17B of the Hydrology Subcommit- Wang, Z., Song, W., and Li, T. (2012). “Combined fragility surface analysis
tee, Office of Water Data Coordination, U.S. Geological Survey, Reston, VA. of earthquake and scour hazards for bridge.” Proc., 15th World Conf. on
Konstantinidis, D., Kelly, J. M., and Makris, N. (2008). “Experimental Earthquake Engineering, Lisbon, Portugal.
investigations on the seismic response of bridge bearings.” Earthquake Zhou, Y., Banerjee, S., and Shinozuka, M. (2010). “Socio-economic effect of
Engineering Research Center Rep. 2008-02, Univ. of California, seismic retrofit of bridges for highway transportation networks: A pilot
Berkeley, CA. study.” Struct. Infrastruct. Eng., 6(1–2), 145–157.

© ASCE 04015063-15 J. Bridge Eng.

J. Bridge Eng., 2016, 21(3): 04015063