Effect of Ground Motion Suite Size on Uncertainty Estimation in Seismic Bridge Fragility Modeling

Abstract

Seismic fragility modeling of highway bridges is becoming a well developed and accepted field among engineers. Within any fragility estimate exists many sources of uncertainty. One source of uncertainty which is key to robust fragility modeling is the aleatory uncertainty which exists within the ground motion suite itself. One obvious question an analyst must ask pertains to the appropriate number of ground motions which must be used to appropriately capture this aleatory uncertainty. This study looks to add to the existing knowledge base, established by previous research, by explicitly looking at the resulting uncertainty estimates as the size and make-up of the ground motion suite is altered. To explore this issue of ground motion suite size four bridge models which are representative of bridges in the Central and Southeastern United States (CSUS) are developed. Median value estimates for all bridge modeling parameters are utilized and held constant and no uncertainty in component capacities is used in an effort to decouple the aleatory uncertainty in the ground motions from the epistemic uncertainty present in typical modeling parameters. Utilizing a suite of 160 recorded ground motions as a pool, numerous subsets of this suite are created and used to generate fragility curves. To keep variation in median and dispersion estimates reasonable this preliminary study recommends that a suite of 80 or more ground motions be used. Such information is essential as fragility modeling gets closer to being an everyday reality for the practicing engineer.