Seismic Vulnerability of the Highway System

Task B1-3: Fragility Curves for Seismically Retrofitted Bridges and Effect of Retrofit on Transportation Network Performance

Subject Area: Loss Estimation Methodologies
Research Year 3

Principal Investigator and Institution

M. Shinozuka, University of California, Irvine


In Research Year 2, the effect of bridge seismic retrofitting by means of steel column jacketing on fragility curves of bridges and highway networks was assessed via Task B1-3. This was done under the assumption that the totality of the bridges under consideration form a statistically homogeneous population. In Research Year 3, this assumption will be removed, and the population of bridges will be divided into a number of bridge classes with each class possessing a similar set of structural and other characteristics. Bridges will be categorized depending on whether they are comprised of single or multiple spans; they are built on soft, medium, or hard soils; and whether their skew is light, moderate, or severe. This will result in a combination of 18 bridge classes.


Fragility curves for retrofitted bridges will be estimated by making use of the correlation that exists between fragility parameters (e.g., medians and log-standard deviations) and hysteretic parameters (e.g., yield point; or first and second elastic moduli in a bilinear moment-curvature (or rotation) curves of the columns). This task will establish the correlation and develop the fragility curves of the bridges retrofitted on the basis of the corresponding curves associated with bridges not retrofitted for each bridge class. Then, using all these fragility curves, a network analysis will be conducted to evaluate the effect of retrofit by computing increased amount of drivers' delay as the measure of degree of deterioration of the network subjected to a specific OD characteristics of the region.

For the 18 bridge classes to be considered, empirical fragility curves will be developed based on the Northridge earthquake damage data for bridges which were not retrofitted with steel or other column jackets. Fragility curves for the bridges retrofitted by steel jackets will then be obtained by the correlational analysis method developed in Task B1-3 during Research Year 2. These two sets of fragility curves (bridges both unretrofitted and retrofitted) will be used in the network analysis to evaluate the performance of highway networks (e.g., in the Los Angeles area) and to identify the difference in the performance when the bridges are assumed to constitute a homogeneous sample and when they are classified into nonhomogeneous classes. It is quite possible that the nonhonogeneous population, which is much more realistic, can predict a significantly lesser degree of performance.


A technical report describing the analytical procedures used to develop the fragility curves for each bridge class, the analysis of network performance, and the estimation of retrofit impacts on network performance. The retrofitted bridge fragility curves can then be implemented in future versions of REDARS.

Technical Challenges

The correlation between fragility and hysteretic parameters may be rather fuzzy.

return to Task Statement List