Selection of Near-Fault Physics-Based Simulated Earthquake Ground Motions for Seismic Performance Assessment of Structures

Abstract

Selection of earthquake records for seismic performance assessment of near-fault structures is challenging because (1) earthquake-induced ground shaking in the near-fault region is highly sensitive to the fault rupture characteristics, seismic wave propagation patterns, and site conditions, and (2) field recordings of such near-fault shaking are relatively sparse. A common approach to representing the near-fault ground motion in engineering analysis is to explicitly consider records with strong directivity pulses (pulse-type records) in ground motion selection. In this study, we critically examine this approach by using broadband physics-based fault rupture simulations and high-performance computing tools to measure the bias associated with selecting earthquake records for the analysis of near-fault structures. Our findings reveal that the predicted structural demand distributions associated with selected ensembles of unscaled simulated records are not sensitive to the binary record classification based on the presence of strong directivity pulses. These findings support the hypothesis that selection of unscaled site-specific simulated ground motion records based on spectral shape eliminates the need to exclusively consider records classified as pulse-type.