Probabilistic seismic hazard analysis for spatially distributed infrastructure considering the correlation of spectral acceleration across spectral periods

Abstract

Regional seismic hazard analyses are necessary to assess the infrastructure performance within a region and ensure that mitigation funds are utilized effectively by probabilistically considering the suite of potential earthquake events. This research aims to efficiently represent the regional seismic hazard through a compact set of seismic inputs in the form of spectral acceleration (SA) maps by considering the spatial cross-correlation of SA at a wide period range. The SA maps can then be used to probabilistically estimate the performance of a portfolio of spatially distributed structures with different fundamental periods. Efficient representation reduces the number of required SA maps to decrease the computational demands of the infrastructure performance analysis in the subsequent steps. The added dimension of the between-period spatial SA correlation exacerbates the challenge of effectively simulating and selecting a set of SA maps to reproduce the hazard curves particularly at long return periods. Two approaches are proposed to generate an optimal set of SA maps: (a) a simulation-based methodology that uses state-of-the-art variance reduction methods and (b) a simplified methodology that aims to increase the ease of use and reduce the computational demands of the simulation. The two approaches are implemented and compared using the city of San Francisco as a case study to illustrate their feasibility. The simplified approach increases the scalability of the methodology to larger study areas at the expense of reduced accuracy in terms of seismic hazard curve and SA correlation errors.