Professional Documents
Culture Documents
Abstract: Damage to skewed bridges in recent earthquakes has reinforced the potential vulnerability of these structures. The effect of skew
Downloaded from ascelibrary.org by Universidad De Guadalajara on 12/15/23. Copyright ASCE. For personal use only; all rights reserved.
angle on a bridge’s fragility could vary for different bridge types, ages, or geometric configurations. This paper conducts a probabilistic seismic
assessment of skewed bridge performance, focusing on single-frame concrete box-girder bridge subclasses. The effect of skew angle on bridge
seismic fragility is investigated for bridges with single- or two-column bents, integral or seat-type abutments, and minimal or significant levels of
seismic design. Component and system-level damage states consistent with HAZUS-MH definitions are also explored in this study. The results
reveal that older bridges, which are more likely to experience higher damage states, are particularly susceptible to column damage and are not
sensitive to skew. Similarly, the presence of integral abutments in newer bridges significantly reduces the vulnerability and minimizes the impact
of the skew angle on bridge fragility. For new bridges with seat-type abutments, the bridge skew angle has a significant effect on component and
system fragility for both single- and two-column bent bridges. For these subclasses, HAZUS-MH skew factors are found to reasonably estimate
the shift in median value fragility from their straight counterparts. DOI: 10.1061/(ASCE)CF.1943-5509.0000435. © 2014 American Society of
Civil Engineers.
Author keywords: Skewed bridges; Concrete box girder; Seat type; Integral; Fragility.
deformation before sliding. The bearings are modeled to incorporate and backfill and also between the deck and backwall, as well as
elastic perfectly plastic behavior in which the coefficient of friction rotating the bents foundation.
between the concrete and neoprene and the shear modulus of the
Treatment of Uncertainties and Ground
elastomeric pads are assumed as random variables (Table 3). Details
Motion Characteristics
on bearing models can be found in Nielson (2005). Pounding be-
tween deck and abutments in seat-type abutment bridges is captured One of the primary intentions of a fragility analysis is to capture the
by zero-length elements distributed along the expansion joint normal uncertainties inherent in seismic performance assessment and
to the face of the deck using the bilinear model recommended by quantify probabilistically the potential for bridge component and
Muthukumar and DesRoches (2006). In order to consider the effects system damage (Padgett and DesRoches 2008). Different sources,
of the skew angle on the seismic response of bridges, various skew such as bridge modeling, ground motion characteristics, or bridge
angles between 0 and 55 are included. Changing the skew angle of geometries may contribute to uncertainties. The geometric uncer-
bridges causes various changes in the bridge geometry, such as tainties in the subclasses of bridges were considered by generating
changing the contact area and related elements between the backwall bridge models with various span lengths, column heights, and deck
Pile axial stiffness Lognormal l 5 6:09 or 6:39 a z 5 0:3 kN=mm=pile (personal communication,
2011)
Abutment passive initial Uniform l 5 14:5 u 5 29 kN=mm=m Shamsabadi et al. (2010)
stiffness
Damping Normal m 5 0:045 s 5 4:3 Fang et al. (1999);
Bavirisetty et al. (2000)
Abutment gap Normal m 5 40:2 s 5 19 mm Based upon inventory
review
Mass Uniform l 5 0:9 u 5 1:1 Nielson (2005)
Loading direction Uniform l50 u 5 2p radian
a
Parameters are for old and new bridges, respectively.
widths. In this study, the abutment backwall height was assumed to Probabilistic Seismic Demand Models and Capacity
be constant in all bridges and a mean value of 3.5 m was defined, Limit States
based on the plans reviewed. For the single-column bent bridges, the
deck widths had a range of 8–16 m, whereas for the two-column bent First, probabilistic seismic demand models (PSDMs) were de-
bridges the range considered was 9–24 m. These estimates were veloped to characterize the probability density functions for com-
based on a review of plans and engineering judgment. ponent demands as a function of ground motion intensity, selected as
A Latin hypercube sampling (LHS) technique (Ayyub and Lai peak ground acceleration (PGA), based on the findings of previous
1989), commonly adopted for variance reduction, was used in this studies (Padgett et al. 2008). The component responses considered
study to generate the bridge samples. The span length and column for fragility analyses are listed in Table 4. For each subclass of
height were considered as dependent variables of the deck width and bridges separate probabilistic seismic demand models were con-
were defined based on 10 specified deck widths in each bridge subclass structed from the peak component responses of the 100 simulations
generated by the LHS technique. The span lengths varied from 16.8 to for that design era, abutment and bent type, and skew angle of
60.2 and 14 to 62.7 m, while the column heights varied from 4.1 to 10.8 interest. The form of the demand model follows that proposed by
and 4.2 to 12.6 m in the single- and two-column bent bridges, re- Cornell et al. (2002) and can be represented in power model
spectively. For other uncertainties related to bridge modeling such as
variations in material properties, mass, and damping, quasi-Monte SD ¼ aIMb (1)
Carlo sampling (Morokoff and Caflisch 1995) was used. Quasi-Monte
Carlo sampling is a technique used to sample random variables where a and b 5 coefficients obtained using regression analysis; and
governed by probability density functions and is much more accurate IM 5 intensity measure. The conditional seismic demands or
than other sampling techniques. Quasi-Monte Carlo methods use PSDM, are modeled using a lognormal distribution as follows:
" #
quasi-random (also known as low-discrepancy) sequences instead of
lnðdÞ 2 lnðSD Þ
random or pseudorandom sequences. In Table 3 all of the considered P½D $ d j IM ¼ 1 2 F (2)
modeling parameters and their probability distributions are presented. bDjIM
The nonlinear time history analysis of the bridges was conducted
using four bins of 20 typical nonnear-fault records identified by where F½ 5 standard normal cumulative distribution function;
Krawinkler et al. (2003) from the Pacific Earthquake Engineering SD 5 median value of the seismic demand [Eq. (1)]; and bDjIM
Research (PEER) database and one bin of 20 near-fault California 5 lognormal SD (dispersion) of the demand conditioned on the IM,
records from the SAC Steel Project (1997) database. These ground which is also estimated from the regression analysis. To calculate the
motions have been adopted and described in detail in past probabilistic system-level fragility of a bridge, Nielson and DesRoches (2007a)
seismic response assessment studies (Mackie and Stojadinovic 2001; proposed the development of a joint PSDM. In this study, damage to
Shafieezadeh et al. 2012). The adoption of 100 records reflecting the bridge components was first assessed and then system-level
a range of ground motion realizations exceeds the recommended fragility curves were developed based on the damage states de-
minimum number of recorders suggested by Nielson and Mackie scribed in HAZUS-MH (FEMA 2005) corresponding to slight,
(2009). Each set of orthogonal horizontal components of ground moderate, extensive, and complete damage. Hence, the capacity
motions was randomly paired with a bridge sample, producing limit states relating to the performance of each component in the
a total of 100 nonlinear dynamic analyses for each bridge subclass. bridge were identified by drawing upon previous studies as well as
The vertical components of ground motions were not considered in on the analytical models of bridge components used in the analysis.
this study. Table 5 presents the limit states adopted, in which four levels of
Shear key seat type, old (mm) 1:0 3 dv 0.25 7:0 3 dv 0.25 14:0 3 dv 0.46 22:0 3 dv 0.46
Shear key seat type, new (mm) 1:0 3 dv 0.25 8:0 3 dv 0.25 17:0 3 dv 0.46 29:0 3 dv 0.46
Unseating, old (mm) N/A N/A N/A N/A N/A N/A 400 0.46
Unseating. new (mm) N/A N/A N/A N/A N/A N/A 600 0.46
Note: Abut-a 5 abutment active deformation; Abut-p 5 abutment passive deformation; Abut-t 5 abutment transverse deformation; Brg-Long&Trans
5 elastomeric bearing deformation in the longitudinal and transverse directions; du 5 shear key displacement corresponding to ultimate force in the backbone
curve of the shear key; dv 5 shear key displacement corresponding to yield point in backbone curve of shear key; ymax 5 maximum soil displacement capacity.
Table 5. Descriptions of Component-Level Damage Assigned to Each System Damage State Consistent with HAZUS-MH (FEMA 2005) Qualitative Damage
States
Component-level limit state
Limit state Primary component Secondary component System-level limit state
Slight DC1 DC2 DS1
Moderate DC2 DC3 DS2
Extensive DC3 DC4 DS3
Complete DC4 N/A DS4
Note: DS 5 damage state.
damage are considered for all components, termed DC1–DC4. The (2008), 150 and 300% of shear strain were assumed for moderate
columns were assumed to be controlled by flexural failure and two (DC2) and complete (DC4) damage states, respectively; following
different behaviors, strength degrading and ductile, were assumed Padgett (2007), 100 and 200% of shear strain were assigned to slight
for old and new seismic design era bridges, respectively. The (DC1) and extensive (DC3) damage states, respectively. In the new
strength-degrading behavior assumed for poorly confined columns bridges with newer and thicker bearings characterized by higher
and their related limit states (displacement ductility) were developed shear deformation strength, the shear deformation dominated the
following the recommendations of Hwang et al. (2000) for slight to limit states up to extensive damage while sliding in bearings dictated
extensive damages and the FHWA (1995) recommendations for the bearing limit states for complete damage, and is defined as half of
complete damage. For ductile columns, the statistical analyses of the bearing support length. For shear keys, the limit states were
Berry and Eberhard (2003) on the PEER (2000) column databases identified based on the mechanical model proposed by Megally et al.
were used for moderate and complete damage states. Banerjee and (2001). The shear key limit states were taken as a function of the
Shinozuka (2008) calibrated analytical models with empirical field displacements corresponding to the first yield of the shear key
data from large-scale shaking table tests on a two-span bridge at the backbone curve, which is defined as slight damage in seat-type
University of Nevada (Reno, Nevada). According to their findings, bridges. In the integral bridges, the shear key limit states were
extensive damage to columns is approximately 65% of complete defined as a function of the ultimate capacity on the backbone curve
damage. The displacement ductilities were converted to curvature of the shear key, which is defined as the moderate damage state. It
ductility as a common limit state (Nielson and DesRoches 2007a, b; should be noted that these points are variable depending upon
Padgett and DesRoches 2008) based on the relationships presented abutment width, number of piles, and strength of material (concrete
by the FHWA (2006). Unique estimates of column yield curvature and steel) because the shear key design is a function of substructure
were derived for each bridge sample to estimate the analogous capacity. Deck unseating is the only limit state that is defined just in
curvature ductility demand. the complete damage state (DC4) and varies from 400 mm in old
Limit states for the active abutment deformations were adopted bridges to 600 mm in new bridges based on bridge plans. The
based on the approach outlined by Nielson (2005), which relates component limit states for both the integral and seat-type bridges can
approach settlement to active deformation. According to Choi be found in Table 4.
(2002), in the passive and transverse directions the limit states are The probability distributions of the component capacities were
taken as a function of maximum displacement of soil because they all assumed to follow a lognormal distribution. According to column
change with increasing abutment width as a result of geometric test data and regression analyses conducted by Berry and Eberhard
sampling or increased skew and soil characteristics. Bearing ca- (2003), coefficients of variation equal to 0.46 and 0.25 were sug-
pacity limits were assumed to be dictated by shear in all damage gested for cover spalling and bar buckling in columns, respectively;
states for the old bridges because the shear limits were found to be therefore, for columns, a dispersion of 0.44 was adopted for DC2,
less than the sliding deformation limit. Following Aviram et al. while 0.25 was used for DC4. The recommendations of Nielson
value of the capacity at component damage level i; bCi 5 logarithmic and the potential for deck unseating. However, columns are the
SD for the capacity at level i; bD 5 logarithmic SD of the demand exception because they have higher median values (lower fragility)
from the regression analysis; and PGA 5 IM conditioned exceed- in two-column bent bridges as a result of lower curvature ductility
ance probability. demand of columns in two-column bent bridges in comparison with
The components are divided into two categories, primary and sec- columns in single-column bent bridges. It is obvious from Fig. 3(b)
ondary, based on their contribution to the bridge system damage states. that the effect of the skew angle on abutment fragilities is significant.
The primary components are defined as those that affect the vertical
stability and load-carrying capacity of the bridge. Extensive or complete
damage to these components may lead to closure of the bridge. Column
damage and deck unseating belong to this category, while shear key,
bearing, and abutment are defined as secondary components. To dis-
tinguish between component and system limit states, the four damage
levels for the bridge system adopt the HAZUS-MH terminology (slight,
moderate, extensive, and complete), and corresponding descriptions of
the component-level damage assigned to each system damage state are
presented in Table 5, where DC 1–4 do not correspond directly to the
slight to complete damage states. Instead, primary components con-
tribute to the slight to complete damage states, while secondary com-
ponents contribute to the earlier damage states. The system fragility
curves were derived considering the correlations between components
using the approach adopted in previous studies (Nielson and DesRoches
2007a; Padgett and DesRoches 2008). The system fragility curves for
each system-level damage state (DS) can be calculated as follows:
8
>
> N
>
>P [ Eprimary_i DCj j IM for j ¼ 4
>
>
>
>
i51
>
< N
DSj j IM ¼ P [ Eprimary_i DCj j IM
>
> i51
>
>
>
>
>
> M
>
:þP [ Esecondary_m DCjþ1 j IM for j # 3
m51
(4)
bridges is small because of the low ductility capacity in the columns, column bent new bridge with integral and seat-type abutments are
which dominates bridge behavior. In bridges with integral abut- summarized in Fig. 7 for a 45 skew angle at the moderate damage
ments, increasing the skew angle may slightly decrease the bridge state. The same trend in relative vulnerability of components can be
fragility because of the higher passive pressure at the abutments. The found for other subclasses and damage states. Fig. 7 reveals that in
effect of skew angle in new bridges is presented in Fig. 6(b) by most of the components the integral bridges are less vulnerable than
comparing the median values of their fragility at the moderate and the seat-type bridges because of the lower displacement experienced
in seismic excitations in integral bridges. The abutment is the only
exception, which is a function of the dynamic behavior of the system.
Unlike seat-type bridges, integral bridges have a monolithic abut-
ment with no gap between the deck and abutment. As a result, the soil
and abutments engage during the seismic response of an integral
bridge for small displacement, while in the seat-type bridges this
engagement is postponed until the closure of the gap. Therefore, for
a specified bridge displacement, abutments and soil experience
higher deformation in integral bridges and may have increased
vulnerability. This passive abutment vulnerability of integral bridges
is countered by the decrease in vulnerability of other components, as
shown in Fig. 7.
damage states. Fig. 8 reveals that the application of the HAZUS-MH This paper studies the effect of skew angle on the fragility of single-
skew modification factor in single- and two-column bent seat-type frame concrete box-girder bridges for various abutment types and
bridge classes of new design details can predict the skew angle effect bridge ages, considering single- and two-column bridges. The im-
on the fragility quite well. It can be concluded from Fig. 9 that pact of skew is investigated for columns, abutments, bearings, shear
Fig. 8. Comparison of the median value from the fragility analysis and Fig. 9. Comparison of the median value from the fragility analysis and
from applying the HAZUS-MH skew modification factor to the zero- from applying the HAZUS-MH skew modification factor to the zero-
skew fragility from the analysis for various skew angles in the new seat- skew fragility from the analysis for various skew angles in the old seat-
type (a) single- and (b) two-column bridges at moderate and extensive type (a) single- and (b) two-column bridges at moderate and extensive
damage states damage states
Fig. 10. Comparison of the median value from the fragility analysis and
Fig. 11. Comparison of the median value from the fragility analysis and
from applying the HAZUS-MH skew modification factor to the zero-
from applying the HAZUS-MH skew modification factor to the zero-
skew fragility from the analysis for various skew angles in the (a) old and
skew fragility from the analysis for various skew angles in the (a) old and
(b) new single-column integral bridges at slight and moderate damage
(b) new two-column integral bridges at slight and moderate damage
states
states
girder bridges because several subclasses are not highly affected Gardoni, P., Der Kiureghian, A., and Mosalam, K. M. (2002). “Probabilistic
capacity models and fragility estimates for reinforced concrete columns
by skew. If applied, a conservative estimate of bridge fragility can be
based on experimental observations.” J. Eng. Mech., 10.1061/(ASCE)
anticipated. However, the existing skew modification factors pre- 0733-9399(2002)128:10(1024), 1024–1038.
sented in HAZUS-MH do a satisfactory job of reflecting the impact of Hwang, H., Liu, J. B., and Chiu, Y.-H. (2000). “Seismic fragility analysis of
skew on the fragility of the subclasses most affected by skew, which highway bridges.” MAEC RR-4, Center for Earthquake Research and
are the new era single-frame bridges with seat-type abutments. Information, Univ. of Memphis, Memphis, TN.
Krawinkler, H., Medina, R., and Alavi, B. (2003). “Seismic drift and
References ductility demands and their dependence on ground motions.” Eng.
Struct., 25(5), 637–653.
AASHTO. (1998). LRFD bridge design specifications, Washington, DC. Mackie, K. R., and Stojadinovic, B. (2001). “Probabilistic seismic demand
Anderson, D. L., Mitchell, D., and Tinawi, R. G. (1996). “Performance of model for California bridges.” J. Bridge Eng., 10.1061/(ASCE)1084-
concrete bridges during the Hyogo-Ken Nanbu (Kobe) earthquake on 0702(2001)6:6(468), 468–480.
January 17, 1995.” Can. J. Civ. Eng., 23(3), 714–726. Mackie, K. R., and Stojadinovic, B. (2004). “Fragility curves for reinforced
Aviram, A., Mackie, K. R., and Stojadinovic, B. (2008). “Guidelines for concrete highway overpass bridges.” Proc., 13th World Conf. on Earth-
nonlinear analysis of bridge structures in California.” PEER Rep. 2008/ quake Engineering, International Association for Earthquake Engineering,
03, Pacific Earthquake Engineering Research Center, Univ. of Califor- Tokyo.
nia at Berkeley, Berkeley, CA. Mander, J. B., Kim, D. K., Chen, S. S., and Premus, G. J. (1996). “Response
Ayyub, B. M., and Lai, K. L. (1989). “Structural reliability assessment using of steel bridge bearings to the reversed cyclic loading.” Rep. No. NCEER
Latin hypercube sampling.” Proc., 5th Int. Conf. on Structural Safety and 96-0014, National Center for Earthquake Engineering Research, State
Reliability (ICOSSAR), Part II, ASCE, New York. Univ. of New York at Buffalo, Buffalo, NY.
Banerjee, S., and Shinozuka, M. (2008). “Experimental verification of McKenna, F., Scott, M., and Fenves, G. L. (2010). “Nonlinear finite-element
bridge seismic damage states quantified by calibrating analytical models analysis software architecture using object composition.” J. Comput.
with empirical field data.” Earthquake Eng. Eng. Vib., 7(4), 383–393. Civ. Eng., 10.1061/(ASCE)CP.1943-5487.0000002, 95–107.
Bavirisetty, R., Vinayagamoorthy, M., and Duan, L. (2000). “Dynamic Megally, S. H., Silva, P. F., and Seible, F. (2001). “Seismic response of
analysis” Bridge engineering handbook, W.-F. Chen and L. Duan, eds., sacrificial shear keys in bridge abutments.” Rep. No. SSRP-200l/23,
CRC, Boca Raton, FL. Dept. of Structural Engineering, Univ. of California at San Diego, San
Berry, M. P., and Eberhard, M. O. (2003). “Performance models for flexural Diego.
damage in reinforced concrete columns.” PEER Rep. 2003/18, Pacific Mid-America Earthquake Center (MAEC). (2007). “MAEViz software.”
Earthquake Engineering Research Center, Univ. of Washington, Seattle. Æhttp://http://mae.ce.uiuc.edu/software_and_tools/maeviz.htmlæ.
Brandenberg, S. J., Zhang, J., Kashighandi, P., Huo, Y., and Zhao, M. Morokoff, W. J., and Caflisch, R. E. (1995). “Quasi-Monte Carlo in-
(2011). “Demand fragility surfaces for bridges in liquefied and laterally tegration.” J. Comput. Phys., 122(2), 218–230.
spreading ground.” PEER Rep. 2011/01, Pacific Earthquake Engineering Muthukumar, S., and DesRoches, R. A. (2006). “Hertz contact model with
Research Center, Univ. of California at Los Angeles, Los Angeles. non-linear damping for pounding simulation.” Earthquake Eng. Struct.
Buckle, I. G. (1994). “The Northridge, California, earthquake of January 17, Dyn., 35(7), 811–828.
1994: Performance of highway bridges.” Technical Rep. NCEER-94- Nielson, B. (2005). “Analytical fragility curves for highway bridges in moderate
0008, National Center for Earthquake Engineering Research, State Univ. seismic zones.” Ph.D. thesis, Georgia Institute of Technology, Atlanta.
of New York at Buffalo, Buffalo, NY. Nielson, B. G., and DesRoches, R. (2007a). “Analytical seismic fragility
CALTRANS. (2007). “Reinforced concrete bridge capacity assessment training curves for typical bridges in the central and southeastern United States.”
manual.” Structure Maintenance and Investigations Rep., Sacramento, CA. Earthquake Spectra, 23(3), 615–633.
Cho, S., Eguchi, R. T., Ghosh, S., and Huyck, C. K. (2006). “REDARS Nielson, B. G., and DesRoches, R. (2007b). “Seismic fragility methodology
validation report.” Rep. No. MCEER-06-0007, Multidisciplinary Center for highway bridges using a component level approach.” Earthquake
for Earthquake Engineering Research, Buffalo, NY. Eng. Struct. Dyn., 36(6), 823–839.
Choi, E. (2002). “Seismic analysis and retrofit of Mid-America bridges.” Nielson, B. G., and Mackie, K. R. (2009). “Tracking uncertainties from
Ph.D. thesis, Georgia Institute of Technology, Atlanta. component level to system level fragility analyses through simulation.”
Choi, E., DesRoches, R., and Nielson, B. (2004). “Seismic fragility of Proc., 10th Technical Council on Lifeline Earthquake Engineering
typical bridges in moderate seismic zones.” Eng. Struct., 26(2), 187–199. (TCLEE) Int. Conf. on Structural Safety and Reliability, ASCE, Reston,
Cornell, C. A., Jalayer, F., Hamburger, R. O., and Foutch, D. A. (2002). VA.
“Probabilistic basis for 2000 SAC Federal Emergency Management Pacific Earthquake Engineering Research (PEER). (2000). “PEER strong
Agency steel moment frame guidelines.” J. Struct. Eng., 10.1061/ motion database.” Æhttp://peer.berkeley.edu/smcatæ.
(ASCE)0733-9445(2002)128:4(526), 526–533. Padgett, J. E. (2007). “Seismic vulnerability assessment of retrofitted bridges
Dutta, A. (1999). “On energy based seismic analysis and design of highway using probabilistic methods.” Ph.D. thesis, Georgia Institute of Tech-
bridges.” Ph.D. thesis, State Univ. of New York at Buffalo, Buffalo, NY. nology, Atlanta.
Ellingwood, B., and Hwang, H. (1985). “Probabilistic descriptions of re- Padgett, J. E., and DesRoches, R. (2008). “Methodology for the de-
sistance of safety related structures in nuclear plants.” Nucl. Eng. Des., velopment of analytical fragility curves for retrofitted bridges.” Earth-
88(2), 169–178. quake Eng. Struct. Dyn., 37(8), 1157–1174.
“Fractional order intensity measures for probabilistic seismic demand a typical California box girder bridge.” Proc., Technical Council on
modeling applied to highway bridges.” Earthquake Eng. Struct. Dyn., Lifeline Earthquake Engineering Conf., ASCE, Reston, VA.
41(3), 391–409. Yashinsky, M. (2010). “Performance of highway and railway structures
Shamsabadi, A., Kapuskar, M., and Martin, G. R. (2006). “Nonlinear during the February 27, 2010, Maule Chile earthquake.” EERI/PEER/
seismic soil-abutment-structure interaction analysis of skewed bridges.” FHWA Bridge Team Rep., Pacific Earthquake Engineering Research
Proc., 5th National Seismic Conf. on Bridges and Highways: Innovation Center, Univ. of California at Berkeley, Berkeley, CA.