A comparative study of seismic fragility estimates using different numerical methods

Abstract

Seismic fragility estimations constitute the primary part of the probabilistic seismic safety assessment of important structures, such as nuclear containments, dams, important bridges, etc. The seismic fragility of a structure is expressed through a family of 'fragility' curves, which plot the conditional probabilities of failure against varying intensities of the seismic hazard. The failure probability of the structure can be defined for multiple limit states. Over the last three decades, various numerical techniques have been developed to estimate the seismic fragility of structures. Among these, three prominent methods selected for the present study are the ones proposed by Kennedy et al. [1], Shinozuka et al. [3] and Ellingwood et al. [4]. These methods vary in terms of the numerical analysis of analytical data in estimating the seismic fragility of a structure. The present work applies these three different fragility analysis techniques to the seismic fragility estimation of a nuclear containment structure and compares the results obtained from the three different methods. The inner containment structure of an Indian PHWR is used for this case study. The results obtained show that the method proposed by Shinozuka et al. provides the most accurate fragility estimations; however, it is also the most computation intensive. The conventional method proposed by Kennedy et al. provides the least accurate results and needs to use an updated uncertainty database.