Toward confident regional seismic risk assessment of spatially distributed structural portfolios via entropy-based intensity measure selection

Abstract

Intensity measure (IM) selection is a crucial step in regional seismic risk assessment (RSRA) of spatially distributed structural portfolios. In order to facilitate more confident regional seismic risk estimates, this study proposes an entropy-based IM selection methodology, offering the first systematic and quantitative regional-level IM selection approach. By conceptualizing the spatially distributed structural portfolio as an integrated multi-response structural system, the joint entropy of the system’s unconditional seismic demands is leveraged as an IM evaluation criterion. Owing to the adaptation of a newly developed advanced IM co-simulation method and multivariate surrogate demand modeling techniques, this entropy-based IM selection approach is able to holistically incorporate uncertainties rising from the spatial IM random field, structural parameters, and surrogate demand models, during the course of uncertainty propagation in RSRA. The efficacy of the proposed methodology is demonstrated along with practical heuristics for alleviating the computational burden, based on a hypothetical highway bridge portfolio. Different application cases in the context of RSRA are considered, including pre-event RSRA considering a single scenario-earthquake as well as a stochastic earthquake catalog, and post-event RSRA considering record updating. The results consistently highlight the significance of the proposed IM selection method in facilitating more confident regional seismic risk estimates. Moreover, this study also provides valuable insights into record updating in reducing the level of uncertainty of the spatial IM random field, and its implication on IM selection in post-event RSRA.