Evaluation of simulated ground motions using probabilistic seismic demand analysis: CyberShake (ver. 15.12) simulations for Ordinary Standard Bridges

There is a need for benchmarking and validating simulated ground motions in order for them to be utilized by the engineering community. Such validation may be geared towards a specific ground motion simulation method, a target engineering application, and a specific location; the validation presented herein focuses on a bridge engineering application in southern California. Catalogs of simulated ground motions representing a 200,000-year forecast are selected from the Southern California Earthquake Center CyberShake version 15.12 database for five sites in Southern California (~20,000 unscaled ground motions per site). They are used in Non-Linear Time History Analysis (NLTHA) of four Ordinary Standard Bridge structures. For each site, these data are used to obtain simulation-based Engineering Demand Parameter (EDP) hazard curves. These are compared against EDP hazard curves that are constructed using conventional methods based on empirical models, i.e., using recorded ground motions through Incremental Dynamic Analysis and integration over the Intensity Measure (IM) hazard curve. The two sets of simulation-based and conventional EDP hazard curves are compared at various return periods. To further account for the differences between simulated and recorded ground motions, direct comparisons are also made between IM hazard curves for simulated and recorded catalogs, as well as the EDP versus IM data obtained from NLTHA of the bridges. We observe that CyberShake simulates motions that yield similar EDP values compared to empirical data for shorter return periods. For longer return periods, however, EDPs from the simulation-based analysis tend to be lower than the EDPs obtained from utilizing recorded ground motions for short-period bridges, while the opposite is the case for long-period bridges. It is recommended that validation efforts go beyond IM levels and also include comparisons of the relations between IMs and EDPs. Finally, site-specific relations are proposed that correlate the ratio between the two types of EDPs (simulation-based and conventional) with the hazard level, shallow site condition, and site basin depth