Probabilistic analytical benchmarking of source-based and site-based ground motion simulation models

Scarcity of the recorded ground motions and their unavailability to meet specified design scenarios have led to an increased use of the simulated ground motions in performance-based earthquake engineering. This paper offers a benchmark study aimed at validating various ground motion simulation models for engineering practice. The ground motion simulation models used in this study cover both site-based and source-based techniques aimed to benchmark their performance using response of single- and multi- degree-of-freedom (SDOF and MDOF) systems to seismic excitation. Source-based ground motion simulation predicts time series using models that explicitly incorporate the physics of the earthquake source and the resulting propagation of seismic waves; Southern California Earthquake Center (SCEC) broadband simulations used in this study represent source-based ground motion simulation technique. Site-based ground motion simulation techniques, on the other hand, use statistical approaches without necessarily solving the mathematical notions that describe the physics of source dynamics and wave propagation to generate ground motion time series. Ground motions simulated from three historic events (Northridge, Loma Prieta, and Landers) are used and applied to a set of SDOF (non-deteriorating), and MDOF (Box-Girder Seat-type bridge) models and their responses are compared with what is observed using recorded ground motions. Evaluation of the efficiency and sufficiency of the synthetic ground motions show that they are in general a good representation of recorded ground motions. Ground motions obtained via the source-based simulation approach provide a more accurate estimation of RotD50 spectral acceleration () while site-based simulations provide a better representation of Arias Intensities (I0). Issues related to variability in response of structures using recorded and simulated ground motion sets are presented.