Validation of (not‐historical) large‐event near‐fault ground‐motion simulations for use in civil engineering applications

Author(s): Petrone, F; Abrahamson, N; McCallen, D; Miah, M | Abstract: Ground-motion simulations generated from physics-based wave propagation models are gaining increasing interest in the engineering community for their potential to inform the performance-based design and assessment of infrastructure residing in active seismic areas. A key prerequisite before the ground-motion simulations can be used with confidence for application in engineering domains is their comprehensive and rigorous investigation and validation. This article provides a four-step methodology and acceptance criteria to assess the reliability of simulated ground motions of not historical events, which includes (1) the selection of a population of real records consistent with the simulated scenarios, (2) the comparison of the distribution of Intensity Measures (IMs) from the simulated records, real records, and Ground-Motion Prediction Equations (GMPEs), (3) the comparison of the distribution of simple proxies for building response, and (4) the comparison of the distribution of Engineering Demand Parameters (EDPs) for a realistic model of a structure. Specific focus is laid on near-field ground motions (l10km) from large earthquakes (Mw7), for which the database of real records for potential use in engineering applications is severely limited. The methodology is demonstrated through comparison of (2490) near-field synthetic records with 5nHz resolution generated from the Pitarka etnal (2019) kinematic rupture model with a population of (38) pulse-like near-field real records from multiple events and, when applicable, with NGA-W2 GMPEs. The proposed procedure provides an effective method for informing and advancing the science needed to generate realistic ground-motion simulations, and for building confidence in their use in engineering domains.