Efficient Simulation‐Based Seismic Hazard and Demand Assessment Leveraging Monte Carlo Catalogs and Reduced‐Order Modeling: Application to CyberShake

ABSTRACT Simulation‐based seismic assessments rely on large ensembles of ground motions to represent site‐specific hazards and evaluate structural performance, leading to high computational demands from both simulations and response evaluations. In practice, ground motion selection algorithms are used to extract a manageable subset for response evaluations, an arguably counterintuitive step in this framework with its own limitations. To address these challenges, this article introduces the MC‐ROM approach, a two‐step method that reduces the computational demands of simulation‐based assessments, specifically adapted for the CyberShake framework. First, the Monte Carlo catalog method limits the number of simulations by selecting a smaller subset of earthquake scenarios that capture the site‐specific hazard, reducing simulation time. Then, Reduced‐Order Models (ROMs) are constructed for intensity measures that characterize the seismic hazard, selecting a small, representative subset of ground motions from the Monte Carlo catalog—thereby reducing response evaluations and associated computational effort. The effectiveness of MC‐ROM is demonstrated through case studies at three CyberShake sites in Southern California. Hazard curves for elastic pseudospectral accelerations and inelastic demands generated using MC‐ROM are similar to those from full CyberShake simulations, with maximum relative errors less than 5% and 9% at a 2% probability of exceedance in 50 years, respectively. Using only 2000 ground motions, the approach demonstrates a 94.9% reduction in required simulation workload and a 99.9% reduction in response evaluations compared to CyberShake. Both methods yield similar hazard disaggregation patterns, demonstrating the MC‐ROM approach as an efficient tool for simulation‐based seismic assessments and an alternative ground motion selection method for seismic demand analyses.