Documentos de Académico
Documentos de Profesional
Documentos de Cultura
net/publication/238179031
CITATIONS READS
51 375
3 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Michel Ghosn on 21 March 2016.
Abstract: This paper presents the results of an analytical seismic fragility analysis of a typical steel highway bridge in New York State.
The structural type and topological layout of this multispan I-girder bridge have been identified to be most typical of continuous bridges
in New York State. The structural details of the bridge are designed as per New York State bridge design guidelines. Uncertainties
associated with the estimation of material strength, bridge mass, friction coefficient of expansion bearings, and expansion-joint gap size
are considered. To account for the uncertainties related to the bridge structural properties and earthquake characteristics, ten statistical
bridge samples are established using the Latin Hypercube sampling and restricted pairing approach, and 100 ground motions are simulated
numerically. The uncertainties of capacity and demand are estimated simultaneously by using the ratios of demands to capacities at
different limit states to construct seismic fragility curves as a function of peak ground acceleration and fragility surfaces as a function of
moment magnitude and epicentral distance for individual components using nonlinear and multivariate regressions. It has been observed
that nonlinear and multivariate regressions show better fit to bridge response data than linear regression conventionally used. To account
for seismic risk from multiple failure modes, second-order reliability yields narrower bounds than the commonly used first-order reliabil-
ity method. The fragility curves and surfaces obtained from this analysis demonstrate that bridges in New York State have reasonably low
likelihood of collapse during expected earthquakes.
DOI: 10.1061/共ASCE兲1084-0702共2007兲12:6共689兲
CE Database subject headings: Earthquakes; Bridge, Steel; Bridges, highway; Uncertainty principles; Seismic effects; Risk
management; New York.
Introduction generated analytical fragility curves for bridges in the central and
southeastern America.
Fragility analysis provides a means to assess the seismic vulner- In this paper, we present fragility curves for highway bridges
ability of bridges through a conditional probability that a structure in New York State and northeastern United States. Although
or structural component will meet or exceed a certain damage earthquakes in New York State are infrequent, they pose a signifi-
level for a given ground motion intensity. Fragility curves are cant hazard because of the risk of extensive damage to the exist-
derived either by empirical 共observed兲 data from past earthquakes ing aging bridge network. For example, an earthquake near
or by analytical methods. Empirical fragility curves have been Massena Center in New York, on September 5, 1944, had a mag-
developed by statistical analysis of postearthquake inspection data nitude of 6.0 共VIII兲 共Mitronovas and Nottis 1993兲. A repeat oc-
from several recent earthquakes, e.g., Loma Prieta, CA 共1989兲, currence of a same magnitude earthquake may cause tremendous
Northridge CA 共1994兲, and Kobe, Japan 共1995兲 共Basoz and damage to existing bridge network in the region. Existing fragility
Kiremidjian 1998; Shinozuka et al. 2000b; Yamazaki et al. 1999兲. curves that have been developed for other regions of the country
Analytical fragility curves are developed through nonlinear dy- cannot be directly applied to New York State and northeastern
namic analysis or quasistatic analysis when actual bridge damage United States because of differences in ground motion character-
and ground motion data are not available 共Shinozuka et al. istics, typical bridge types, and construction practices.
2000a兲. Hwang et al. 共2001兲, Choi 共2002兲, Choi et al. 共2004兲 have Fragility curves for typical multispan continuous steel high-
way bridges in New York State have been developed through
1 nonlinear time history analyses of three-dimensional finite-
Structural Engineer, Gilsanz Murray Steficek LLP, 129 West 27th
St., New York, NY, 10001; formerly, Graduate Student, The City College
element 共FE兲 model of a bridge. The FE model incorporates non-
of the City Univ. of New York, New York, NY 10031. linear stiffness modeling of columns, bearings, and abutments for
2
Professor, Dept. of Civil Engineering, The City College of the City input in the longitudinal direction. Although the development of
Univ. of New York, New York, NY 10031 共corresponding author兲. analytical fragility curves has been addressed by several research-
E-mail: Agrawal@ccny.cuny.edu ers 共Choi et al. 2004; Shinozuka et al. 2000a,b兲, we make several
3
Professor, Dept. of Civil Engineering, The City College of the City significant new contributions to the analytical fragility procedure,
Univ. of New York, New York, NY 10031. E-mail: ghosn@ccny.cuny.edu in addition to the development of fragility curves and fragility
Note. Discussion open until April 1, 2008. Separate discussions must surfaces for New York State.
be submitted for individual papers. To extend the closing date by one
1. Parametric analysis: Previous studies considered uncertain-
month, a written request must be filed with the ASCE Managing Editor.
The manuscript for this paper was submitted for review and possible ties only in material properties and gap size between deck
publication on November 21, 2005; approved on May 4, 2006. This paper spans. In this paper, we carry out detailed parametric study
is part of the Journal of Bridge Engineering, Vol. 12, No. 6, November for the sensitivity of fragility to superstructure weight, con-
1, 2007. ©ASCE, ISSN 1084-0702/2007/6-689–699/$25.00. crete compressive strength, reinforcement yield strength,
Downloaded 26 May 2009 to 144.3.101.112. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
abutment wall–soil stiffness, friction coefficient of expansion
bearings and expansion-joint gap size and observe that un-
certainties associated with friction coefficient of expansion
bearings and superstructure weight have significant effects
on bridge ductility in addition to material strength and gap
size.
2. Analysis of uncertainties: Previous studies 共Karim and
Yamazaki 2001; Deodatis and Saxena 2000; Choi et al. 2004;
Hwang et al. 2001兲 assumed a set of constant values as ca-
pacity criteria and adopted the empirical value suggested in
HAZUS for the dispersion of seismic capacity and demand
关NIBS 共1999兲兴. These values are different in 1997 and 1999
versions of HAZUS and are not related to bridge types and
damage states. Actually, the capacity varies with structural
parameters, and the correctness of assumed dispersion will
largely influence fragility results. To overcome this limita-
Fig. 1. Configuration of a typical multispan continuous steel girder
tion, estimates of the dispersion in the ratios of demands to
bridge in New York: 共a兲 elevation; 共b兲 transverse cross section; and
capacities at different limit states are directly obtained based
共c兲 bearings
on the results of a regression fit of the data obtained from the
nonlinear analysis.
3. Regression analysis: Linear regression on bridge response for the connection between superstructure and supporting column/
quantities has been used in previous studies on fragility abutment bents, and the substructure. In the bridge model in Fig.
analysis. In this paper, we observe that quadratic regression 2, movement of the bridge in transverse direction is restrained by
provides a better fit on the bridge response data as compared steel expansion bearings or steel fixed bearings, although expan-
to the commonly used linear regression fit. Further, multi- sion bearings accommodate longitudinal movement. Because of
variate regressions of the fragility show further improvement restraint in transverse direction by bearings and in-plane rigidity
in the predictive model and provide fragility surfaces as a of the deck, seismic inertial forces are evenly transferred from
function of earthquake magnitude and epicentral distance. superstructure to the supports points on substructure bents.
4. Bridge fragility: Fragilities of individual components are Hence, transverse seismic behavior can be accurately modeled by
combined to develop fragility curves for the complete bridge using spine model in Fig. 2.
system using multiple failure modes theory. Second-order re- Steel bearings in the bridge in Fig. 2 are modeled using mul-
liability bounds are shown to provide narrower bounds on tilinear plastic link elements. The bottom segments of the col-
fragility than first-order reliability bounds used in previous umns where plastic hinges are likely to occur are modeled using
studies. bilinear plastic elements. The abutments are modeled using beam
elements supported on springs. The expansion joints at abutment
are modeled with gap element. The bases of the column bents are
Bridge Characteristics and Modeling assumed to be fixed for a continuous spread footing on rock. Due
to the symmetry, seismic analysis can be performed indepen-
It has been observed from the 2003 Bridges Inventory data pro- dently in the two horizontal directions. The dominance of the
vided by the New York State Deptartment of Transportation that longitudinal motion during multi-axis earthquake is apparent for
58.2% of the highway bridges in New York State are multibeam this bridge type from the restraints provided by the bearings/ the
or girder 共slab-girder兲 types. Steel multispan simply supported abutments, and the column boundary conditions, which is fixed–
bridges constitute 70%, and steel multispan continuous bridges free in longitudinal direction and fixed–fixed in the transverse
account for 27% of all multispan slab-girder bridges. The fragility direction. This observation has also been verified through simpli-
analysis for steel multispan simply supported bridges is addressed fied analytical modeling and detailed simulation results. Hence,
in a separate study 共Pan 共2006兲兲. Continuous steel bridges typi- longitudinal motion of the bridge is most critical direction causing
cally have 2 or 3 span with maximum span lengths between 30 damage to bridge components. A brief description on nonlinear
and 40 m. A hypothetical typical continuous steel plate girder modeling of column elements, abutment springs, gap elements
bridge shown in Fig. 1 has been selected on the basis of typical and bearings is presented in the following.
design details for each component obtained from the review of
bridge design drawings. The superstructure of the bridge consists
Columns
of 240 mm thick continuous cast-in-place composite concrete
deck supported on six steel plate I-girders. The girders are sup- For longitudinal motion, each column acts as a vertical cantilever
ported on two reinforced concrete three-column bents. The end beam. Plastic hinges can only form at the bottom of the columns
bearings are seated on the abutments with U-shaped wing walls. for this type of bridge, as pinned joints with bearings 共both fixed
A three-dimensional finite-element model of the bridge using and expansion兲 can accommodate rotations between the cap beam
SAP2000 共9.1.1.V, Computers and Structures, Inc., Berkeley, and the girders. The nominal moment–curvature diagram for each
Calif.兲 is shown in Fig. 2. The deck and steel girders are com- column section is determined using the BIAX program 共Wallace
bined together and modeled as one line of elastic beam elements, and Ibrahim 1996兲. Elastic-perfectly plastic idealization for the
as it presents effective stiffness and mass distribution characteris- moment–curvature relationship is assumed, as shown in Fig. 3.
tics of the bridge. The bridge superstructure itself is expected to The equivalent yield curvature y is found by extrapolating the
remain essentially elastic during typical earthquake ground mo- line joining the origin and the point corresponding to the first
tions for New York State. Nonlinear behavior has been considered yield of a reinforcing bar, up to the nominal moment capacity M n,
Downloaded 26 May 2009 to 144.3.101.112. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Fig. 2. Three-dimensional finite-element model of the bridge using SAP2000
Gap Elements
The nominal gap between ends of the deck and the abutments is
76.2 mm 共3 in.兲. The gap element of SAP2000 is utilized to
Fig. 3. Moment–curvature diagram of piers account for the possibility of pounding when the longitudinal
Downloaded 26 May 2009 to 144.3.101.112. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Fig. 4. Load–response for abutment in longitudinal direction Fig. 5. Sensitivity analysis of bridge parameters for PGA= 1.0 g
deformations close either of the two gaps. The SAP2000 gap 共1996兲 have recommended the use of one bilinear truss and two
element is set as a “compression-only” connection such that the link elements in parallel to model bearing behavior. The bearings
element does not apply any resistance before the closure of the models are incorporated in SAP2000 by a trilinear force–
gap. When the gap between the deck and abutment closes, pound- displacement relationship shown in Fig. 2 with k2 = 0.5k1,
ing occurs. The infinite stiffness of gap element can be assumed at k3 = 0.04k1, u1 = 0.4 mm, and u2 = 0.87 mm for the displacement
the contact location 共Priestley et al. 1996兲. After the gap closes, range smaller than 20 mm.
the abutment is assumed to be rigidly connected to bridge decks
in axial direction and begins to contribute in resisting the longi-
tudinal displacements. Parametric Analysis
Downloaded 26 May 2009 to 144.3.101.112. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
4. The abutment wall–soil stiffness is not a sensitive parameter represent the 28-day strength. As the concrete ages, the strength
for this type of bridge for PGA up to 1.0 g. increases with time. This factor has not been considered in the
5. Friction coefficient of expansion bearing is a significant pa- analysis.
rameter. Neglecting the effect of friction resistance is not a
suitable simplification. The second column bent may be sub-
Gap Size
ject to larger share of seismic forces in superstructure be-
cause of larger friction coefficient of the high type rocker Considerable variations in temperature conditions will cause large
bearing. deviations in gap size from its nominal value. The average annual
Based on previous observations, it is observed that bridge re- mean temperature in New York State ranges from 40° F 共4.44° C兲
sponse is sensitive to uncertainties associated with friction coef- in the Adirondacks to 55° F 共12.78° C兲 in the New York City area.
ficient of bearings, reinforcement yield strength, superstructure Actual temperature varies from −5 to 90° F 共−20.56 to 32.22° C兲.
weight, gap size and concrete compressive strength 共in the order Hence, the temperature variation ⌬T, that is the difference be-
of decreasing importance兲. Hence, uncertainties associated with tween temperature during a seismic event and the one during
these parameters must be considered in developing the fragility construction, can be modeled as a random variable in the range
curves. from −95 to 95° F 共−70.56 to 35° C兲. For the purpose of estimat-
ing the gap size when a seismic event occurs, the temperature
variations ⌬T are considered to follow normal distribution, with
Bridge Structure Uncertainties mean = 0 and standard deviation = 58 to obtain ⌬T samples in
the temperature range from −95.4 to 95.4° F 共−70.78 to 35.22° C兲
Latin Hypercube sampling 共LHS兲 approach is used 共Wyss and corresponding to 5–95% cumulative probability.
Jorgensen 1998兲 to account for uncertainties in parameters The gap between the deck and each abutment is assumed to be
discussed earlier. In this approach, probability distributions are 76.2 mm 共3 in.兲 for as-built conditions. The change of gap size is
assumed for each parameter, and the probability density function ␣⌬TL. Coefficients of thermal expansion ␣ for steel and concrete
of each random variable is divided into a histogram with equal are in the range of 共6.1– 6.7兲 ⫻ 10−6 / ° F 共1.1– 1.2兲 ⫻ 10−5 / ° C兲
probability intervals graded linearly corresponding to cumulative and 共4.1– 7.3兲 ⫻ 10−6 / ° F 共7.4– 13兲 ⫻ 10−6 / ° C兲, respectively. For
distributions varying from 5 to 95%. a composite deck, the coefficient of thermal expansion ␣ of the
composite superstructure can be calculated as
f y: 共1 2 3 4 5 6 7 8 9 10兲
Concrete Compressive Strength
Concrete compressive strength of 20.7 MPa 共3.0 ksi兲 is the stan- f ⬘c : 共2 7 10 8 1 4 9 3 5 6兲
dard design strength for bridges built in New York State during
1970s and 1980s. According to MacGregor 共1992兲, 3.0 ksi
concrete can be assumed to have strength that follows a normal W: 共9 4 3 5 7 2 10 1 6 8兲
probability distribution with mean value 27.2 MPa and a standard
deviation of 4.24 MPa 共COV= 0.16兲. It is noted that these tests ⌬T: 共9 2 10 1 4 5 3 6 8 7兲
Downloaded 26 May 2009 to 144.3.101.112. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Fig. 6. Flow chart for ground motion generation for fragility analysis
b: 共5 1 7 9 8 6 4 3 10 2兲 共4兲
Each column of the matrix in Eq. 共4兲 represents one bridge
sample. For example, the first sample will be associated with 1st,
Fig. 7. Steps in development of analytical fragility curves/surfaces
2nd, 9th, 9th, and 5th bin of histograms for f y, f ⬘c , W, ⌬T, and b,
respectively.
Damage States
Simulation of Ground Motions The vulnerabilities during earthquakes include damage to the
critical components of columns and bearings. It has been ob-
To perform the fragility analysis, various levels of ground mo- served from numerical simulation results that the maximum shear
tions have to be considered to evaluate the probability that the force in the columns is always far below their shear capacity.
bridge will be subject to certain damage states within a given Hence, column fails when the flexural ductility capacity is
return period. These ground motion records must be representa- reached during an earthquake ground motion. The typical inelas-
tive of the seismic characteristics of the site. There are only two tic behavior of concrete bridge piers is shown in Fig. 3 where
existing representative suites of rock motions with return periods critical points are associated with the curvatures ⬘y , y, d, u.
of 500 and 2500 years that are applicable for the New York City These critical points are related to the types of damage. ⬘y indi-
region 共Weidlinger Associates 1998兲. Because of the scarcity of cates the initiation of yielding. The bifurcation point at y is
actual earthquake data, artificial earthquake records must be gen- associated with the formation of plastic hinge in the column. The
erated using the method of probabilistic seismic hazard analysis strength degradation occurs at d, when the maximum moment is
共PSHA兲. Fig. 6 presents a flow chart describing the process used reached. Crushing of concrete occurs at ultimate curvature u
for ground motion generation. The interactive deaggregation pro- when the strain in the concrete is equal to cu, which accounts for
grams of the U.S. Geological Survey 共USGS兲 are used to perform the effect of confinement 共Mander et al. 1988兲
PSHA and deaggregation 共http://neic.usgs.gov/neis/general/
seismicity/usគeast.html兲. The dominant events, defined by 1.4s f yhsu
cu = 0.004 + 共5兲
magnitude–epicentral distance pairs, can then be identified from f ⬘cc
their level of contribution. Following the recommendations of where s = volumetric ratio of the confining steel; su = strain at the
Harmsen et al. 共1999兲, a set of events is selected by choosing the maximum tensile stress of the transverse steel, which is com-
共M w , R兲 pairs, which either have the largest hazard contribution or monly taken as 0.15 for Grade 40 reinforcement, and 0.12 for
would produce target hazard levels. Then, synthetic ground
Grade 60 reinforcement; and f ⬘cc = compressive strength of the
motions PGA larger than 0.1 g are simulated through the SGMS
confined concrete. The critical curvatures can be obtained using
program 共Halldorsson and Papageorgiou 2004兲 by defining the
the program BIAX. Fig. 8 shows the values of ⬘y , y, d, u
magnitude 共M w兲, Epicentral distance R, and soil class. A suite of
obtained for the columns of these ten bridge samples. The capac-
one hundred ground motions at varying hazard levels are gener-
ity difference among samples is due to the variation in material
ated for fragility analysis. General properties of these records are:
strength and dead loads.
共1兲 moment magnitude range: 5.5–7.5; 共2兲 hypocentral distance
Potential damage to fixed high type bearing can be associated
range: 10– 120 km; and 共3兲 soil categories: NEHRP B-C bound-
with bond failure of the high strength anchor bolts and rocking of
ary 共V100 = 760 m / s兲.
the bearing on the bedding material, which would start when the
displacement exceeds 20 mm 共Mander et al. 1996兲. This displace-
ment threshold is significantly above the values observed in the
Fragility Analysis 100 dynamic analyses performed. Hence, this mode of failure is
ignored in the fragility analysis. For expansion bearings, instabil-
The procedure for fragility analysis is described in the flow chart ity may cause the superstructure to move a considerable distance
in Fig. 7. Each of the 10 bridge samples generated previously is and fall off the bridge seat in the worst case. High type rocker
matched with 10 earthquake samples to generate 100 earthquake- bearings would overturn about the edge of the bottom of the
bridge samples for nonlinear time history analysis. rocker 共by rolling off the rocker on the masonry plate兲, when the
Downloaded 26 May 2009 to 144.3.101.112. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Fig. 8. Pier curvature capacities at four damage states for ten
samples
Pf = P 冋 册
Sd
Sc
艌1 =1−
冉
ln共1兲 −
=
冊 冉冊 共6兲
type of multiple failure modes, different failure modes are corre-
lated. The first-order upper and lower bound estimates of the
probability of failure ignore the correlation between these failure
where and = mean value and standard deviation of ln共Sd / Sc兲 modes and can be expressed as 共Cornell 1967; Kottegoda and
obtained from regression analysis. By using quadratic regression, Rosso 1997兲
Downloaded 26 May 2009 to 144.3.101.112. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Fig. 11. Fragility curves for system reliability due to multiple failure
modes
p共Fi 艚 F j兲 = p 冉 Sdi
Sci
艌 1,
Sdj
Scj
冊 冉冉 冊
艌 1 = p ln
Sdi
Sci
艌 0,
ln冉 冊 冊
Sdj
Scj
艌0 共12兲
Downloaded 26 May 2009 to 144.3.101.112. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Fig. 14. Contour map of bridge piers for probability of exceeding y
observed that the r2 values are larger for the multivariate regres-
sion in M w and R than those obtained for the linear and quadratic
regressions as functions of PGA. The fragility for specific
damage-state can be computed using Eq. 共6兲 by expressing , the
mean value of ln共Sd / Sc兲, in the form of Eq. 共15兲. Figs. 13共a–d兲
show fragility surfaces of bridge piers for damage states related to
exceeding the curvatures ⬘y , y, d, u. It is observed that the
fragility increases with decreasing R and increasing M w. Within a
practical scope of M w 艋 7.5 and R 艌 10 for this region, the maxi-
mum fragility for exceeding curvatures ⬘y and y could reach
Fig. 12. Regression analysis of ln共demand/capacity兲 versus ln共M w兲 100%, while the maximum fragilities for exceeding curvatures d
and ln共R兲 共km兲: 共a兲 pier curvature ⬘y ; 共b兲 pier curvature y; 共d兲 pier and u are only 88 and 9%, respectively. The scope of M w and R
curvature d; 共e兲 pier curvature u; 共e兲 high type rocker bearing that would lead to a certain failure probability can be determined
displacement limit= 153 mm; and 共f兲 low type sliding bearing from the contour plots. For example, it is observed from Fig. 14
displacement limit= 102 mm that there are three contour curves for the 10, 50, and 90% prob-
ability of exceeding the curvature y. The hatched area in Fig. 14
gives the range of earthquake magnitudes and epicentral distances
that may cause the formation of plastic hinges in piers with more
than 50% probability. Alternatively, given an earthquake with
known magnitude, the fragility can be determined by cutting a
section through the fragility surface to obtain a fragility curve for
different distances to the epicenter. For example, the curves in
Fig. 15共a兲 is obtained by cutting sections in fragility surface of
Fig. 13共b兲 at M w = 7.4, 7.0, 6.6, and 6.2. Fig. 15共a兲 shows that the
probability of forming a plastic hinge in a pier decreases sharply
with increasing distance and bridges farther than 50 km from epi-
center are quite safe. This distance is relatively shorter than that
Downloaded 26 May 2009 to 144.3.101.112. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Fig. 16. Fragility surfaces for instability of expansion bearings
observed for sites in the western United States. This happens concrete compressive strength and gap size. These two im-
because the high frequency contents in earthquakes in northeast- portant factors have not been considered in previous studies.
ern United States die off with distance more rapidly than low 2. Linear, quadratic, and multivariate regressions are conducted
frequency contents observed in earthquake records from western on the ratio of seismic demand to capacity ln共Sd / Sc兲 to ac-
United States. Contour plots can also be used to determine dam- count for the uncertainties in these variables simultaneously.
age probabilities for different earthquake magnitudes at a particu- In previous studies, uncertainty in demand was considered
lar site as shown in Fig. 15共b兲. The complete system damage through regression while the uncertainty in capacity was as-
could either be due to bearing instability or/and column collapse. sumed based on engineering judgment.
Figs. 16共a and b兲 show fragility surfaces for bearings with a maxi- 3. Comparisons of results of three types of regressions show
mum probability of 58 and 68% for high type rocker and low type that the multivariate regression provides the best fit in terms
sliding bearings, respectively. Similarly, the combined fragility of the coefficients of determination. For fragility versus PGA
surface for the system collapse can be obtained using the second- curves, quadratic regression is significantly superior to linear
order bounds, as shown in Fig. 17. regression. Using linear regression for strong earthquakes
共⬎0.6 g for New York State兲 is not reasonable because of
nonlinear trend of ln共Sd / Sc兲 for large PGAs.
Conclusions 4. Failure modes of individual components, i.e., instability of
bearings, and pier collapse, are combined to obtain first and
This paper presents an improved approach for developing analyti- second order bounds on the fragility of system failure. The
cal fragility curves for a typical multi-span continuous steel high- median PGAs are 0.60, 0.66, 1.07, and 1.19 g for the limit
way bridge. Analytical fragility curves developed in this paper are states corresponding to first yielding of longitudinal rein-
likely to be used extensively by state Departments of Transporta- forcement, formation of the plastic hinge, start of strength
tions and other transportation agencies in the northeastern United degradation in bridge columns and bridge collapse.
States region. Besides this, the proposed approach has the follow- 5. Fragility surfaces as functions of M w and R have been devel-
ing significant other advantages over existing approaches: oped. Using contour maps of a fragility surface, values of
1. The proposed approach uses uncertainties associated with earthquake magnitudes and epicentral distances can be deter-
friction coefficient of bearing and superstructure weight in mined for each probability of exceedance. For example, it is
addition to other factors, e.g., reinforcement yield strength, observed from fragility surface that the maximum fragility
could reach 100%, 100%, 88% and 88% for the four damage
states of bridge for M w 艋 7.5 and R 艌 10 km in northeastern
United States.
Further work on seismic fragility of typical multispan simply
supported steel highway bridge in northeastern United States will
be presented in future publications.
Acknowledgments
Downloaded 26 May 2009 to 144.3.101.112. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Principal, Weidlinger Associates, New York, during the course of Kottegoda, N. T., and Rosso, R. 共1997兲. Statistics, probability, and reli-
this research are sincerely acknowledged. ability for civil and environmental engineers, McGraw–Hill, New
York.
MacGregor, J. G. 共1992兲. Reinforced concrete design, 2nd Ed, Prentice-
Hall, Englewood Cliffs, N.J.
References Mander, J. B., Kim, D. K., Chen, S. S., and Premus, G. J. 共1996兲. “Re-
sponse of steel bridge bearings to the reversed cyclic loading.” Tech-
Ang, A. H., and Tang, W. H. 共1975兲. Probability concepts in engineering nical Rep. No. NCEER 96-0014, NCEER, Buffalo, N.Y.
planning and design, Vol. 1, Wiley, New York. Mander, J. B., Priestley, M. J. N., and Park, R. 共1988兲. “Observed stress–
Applied Technology Council 共ATC兲. 共1996兲. “Improved seismic design strain behavior of confined concrete.” J. Struct. Eng., 114共8兲, 1827–
criteria for California bridges: Provisional recommendations.” ATC- 1849.
32, Redwood City, Calif. Melchers, R. E. 共1999兲. Structural reliability analysis and prediction, 2nd
Basoz, N., and Kiremidjian, A. 共1998兲. “Evaluation of bridge damage Ed., Wiley, New York.
data from the Loma Prieta and Northridge, CA earthquakes.” Rep. No. Mitronovas, W., and Nottis, G. N. 共1993兲. “A historical look at earth-
MCEER-98-0004, Multidisciplinary Center for Earthquake Engineer- quakes in the northeast.” Proc., “Earthquakes in the Northeast-Are
ing Research, Buffalo, N.Y. We ignoring the Hazard?” Workshop on Earthquake Science and
California Department of Transportation 共CALTRANS兲. 共1989兲. Bridge Safety for Educators, Technical Rep. No. NCEER-93-0005, Buffalo,
N.Y.
design specifications manual, Sacramento, Calif.
National Institute of Building Sciences 共NIBS兲 and Federal Emergency
Choi, E. 共2002兲. “Seismic analysis and retrofit of mid-America bridges.”
Management Agency. 共1999兲. “Technical manual.” HAZUS99-SR2,
Dissertation, Dept. of Civil and Environmental Engineering, Georgia Washington, D.C.
Institute of Technology, Atlanta. Nowak, A. S. 共1993兲. “Calibration of LRFD bridge design code.”
Choi, E., DesRoches, R., and Nielson, B. 共2004兲. “Seismic fragility of NCHRP Project No. 12-33, Univ. of Michigan, Ann Arbor, Mich.
typical bridges in moderate seismic zones.” Eng. Struct., 26共2兲, 187– Pan, Y. 共2006兲. “Seismic fragility and risk management of highway
199. bridges in New York state.” Ph.D. dissertation, City Univ. of New
Clough, C. W., and Duncan, J. M. 共1991兲. Foundation engineering hand- York, New York.
book, 2nd Ed., H. Y. Fang, ed., Chapman and Hall, New York, 223– Priestley, M. J. N., Seible, F., and Calvi, G. M. 共1996兲. Seismic design
235. and retrofitting of bridges, Wiley, New York.
Cornell, C. A. 共1967兲. “Bounds on the reliability of structural system.” J. Priestley, M. J. N., Seible, F., and Chai, Y. H. 共1992兲. Design guidelines
Struct. Div., 93共1兲, 171–200. for assessment retrofit and repair of bridges for seismic performance,
Deodatis, G., and Saxena, V. 共2000兲. “Effect of spatial variability of Rep. No. SSPP-92/-1, Dept. of Applied Mechanics and Engineering
ground motion on bridge fragility curves.” 8th Specialty Conf. on Sciences, Univ. of California, San Diego.
Probabilistic Mechanics and Structural Reliability 共CD-ROM兲, Shinozuka, M., Feng, M. Q., Kim, H., and Kim, S. 共2000a兲. “Nonlinear
static procedure for fragility curve development.” J. Eng. Mech.,
ASCE, Reston, Va.
126共12兲, 1287–1295.
Halldorsson, B., and Papageorgiou, A. S. 共2004兲. “Papageorgiou region
Shinozuka, M., Feng, M. Q., Lee, J., and Naganuma, T. 共2000b兲. “Statis-
specific ground motion simulations using the specific barrier model.”
tical analysis of fragility curves.” J. Eng. Mech., 126共12兲, 1224–
User’s manual of SGMS, Engineering Seismology Laboratory, State 1231.
Univ. of New York, Buffalo, 具http://civil.eng.buffalo.edu/EngSeisLab/ Wallace, J. W., and Ibrahim, Y. A. 共1996兲. “Strength analysis of rein-
Publications/SGMSv5/SGMSv5-User-Manual-Draft.pdf典. forced concrete sections program.” User’s manual for BIAX, Univ. of
Harmsen, S., Perkins, D., and Frankel, A. 共1999兲. “Deaggregation of California at Berkeley, Calif.
probabilistic ground motions in the central and eastern United States.” Weidlinger Associates. 共1998兲. “New York City seismic hazard study and
Bull. Seismol. Soc. Am., 89共1兲, 1–13. its engineering applications.” Final Rep., New York.
Hwang, H., Liu, J., and Chiu, Y. 共2001兲. “Seismic fragility analysis of Wyss, G. D., and Jorgensen, K. H. 共1998兲. A user’s guide to LHS: Sand-
highway bridges.” Technical Rep., Center for Earthquake Research ia’s Latin hypercube sampling software, Risk Assessment and Systems
and Information, Univ. of Memphis, Memphis, Tem. Modeling Dept., Sandia National Laboratories, Albuquerque, N.M.
Karim, K. R., and Yamazaki, F. 共2000兲. “Comparison of empirical and Yamazaki, F., Hamada, T., Motoyama, H., and Yamauchi, H. 共1999兲.
analytical fragility curves for RC bridge piers in Japan.” 8th Specialty “Earthquake damage assessment of expressway bridges in Japan.”
Conf. on Probabilistic Mechanics and Structural Reliability Optimizing Post-Earthquake Lifeline System Reliability, ASCE Proc.,
共CD-ROM兲, ASCE, Reston, Va. 5th U.S. Conf. on Lifeline Earthquake Engineering, Seattle, 361–370.
View publication stats Downloaded 26 May 2009 to 144.3.101.112. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright