Report No.: CCEER-13-3


Authors: AmirHormozaki, E., Pekcan, G. and Itani, A.

Date: February 2013

Sponsoring Agency: Federal Highway Administration (FHWA)

Performing Organization:
Department of Civil Engineering/258
University of Nevada, Reno
Reno, NV 89557


Reconnaissance studies following recent significant earthquakes have revealed the seismic vulnerability of curved bridges. Curved steel bridges are widely utilized as important components of transportation networks across the United States especially when geometry and curve alignment of a highway requires a curved structure. However, there are a few studies regarding the seismic vulnerability of horizontally curved steel girder bridges as one of the most critical components of transportation networks Comprehensive and rational evaluation of seismic vulnerability of transportation networks can be achieved through methodological Seismic Risk Assessment (SRA). Fragility curves which express the conditional probability of reaching or exceeding a predefined damage state are essential as input to any SRA tools such as HAZUS and REDARS for the seismic vulnerability evaluation. This study aims to fill in the gap on the current state-of-the-knowledge in the seismic response and fragilities of horizontally curved bridges with steel I-girders.

The main contribution of this study is to develop fragility curves for horizontally curved< steel girder highway bridges in the United States. The entire study has been performed on both seismically and non-seismically designed bridges which have major differences in column confinement, type of bearings, and abutment support length. The proposed grillage model in this study substantially reduced the time to run a large number of nonlinear response history analyses. A survey of this class of bridge across the United States is performed to establish the distribution of various geometric parameters. Sampling on these geometric parameters along with other predominant material, analysis and component properties obtained from a sensitivity analysis led to a set of statistical bridge samples. The component fragility curves are then developed based on the results
of nonlinear response history analyses of benchmark bridges. They revealed columns and bearings at abutments in both radial and tangential directions to be the most critical components. The system fragility curves are subsequently developed for generic curved bridges independent of the central angle. This representation of fragility curves is expected to be useful when the specific central angles are not quantified (e.g. National
Bridge Inventory). Non-seismically designed bridges are recognized to be significantly more vulnerable than seismically designed bridges due to their inadequate column confinement and inadequate support length provided at the abutments. In order to establish the effect of central angle on the seismic vulnerability of curved bridges, fragility curves as a function of the central angle were also presented. In addition, fragility curves were developed for two intensity measures, namely PGA and Sa1.

It was concluded that Sa1 is an efficient and proficient intensity measure with lower dispersions, lower modified dispersions and higher R2, while PGA is more practical due to higher slopes of probabilistic seismic demand models. A set of functional relationship between central angle and the median values of the fragility curves was proposed based on regression analyses. Another curvature index named torsion index was also found to be a significant parameter on seismic vulnerability of this class of bridge and also a better curvature-representing index than the central angle for seismically designed bridges. Furthermore, suitability of the maximum central angle below which the curved bridges can be analyzed as their straight counterparts presented in design codes and guidelines is investigated. Accordingly, it was concluded that while the 30-degree central angle adopted in AASHTO Seismic Guidelines (2007) is reasonable, 90-degree angle required by AASHTO LRFD (Interim 2008) is controversial. Following a comparison of the fragility curves developed in this study versus their HAZUS counterpart, it was found that the HAZUS fragility curves underpredict the seismic vulnerability of this class of bridges mainly due to two reasons: 1) As it was established in this study, the seismic demand on various components and overall vulnerability of curved bridges increases with increasing central angle. However, HAZUS does not include the central angle to categorize and characterize curved bridges. This introduces significant inaccuracies in the predicted vulnerabilities when fragility curves in HAZUS are used for curved bridges, 2) the underlying assumption in the development of fragility curves implemented in HAZUS was that the system fragility is governed by the columns, however, other bridge components have significant influence on the overall system fragility as it was demonstrated in this study.