Validation and extension of synthetic ground motion simulation procedures

Abstract

For several years now, synthetic ground motions have been increasingly gaining attention in the field of earthquake engineering. Synthetic ground motion time series can be utilized in a variety of engineering applications including dynamic analysis and probabilistic seismic hazard analysis. However, these synthetic ground motions must capture the important characteristics and natural variability of recorded motions. Several ground motion simulation models and methods have been proposed, ranging from source-based deterministic to site-based stochastic and hybrid models. This study focuses on site-base stochastic models, which are more appealing to practicing engineers, and is comprised of two distinct studies related to synthetic ground motion simulation procedures. The first study is a validation of different site-based models for simulation of far-field ground motions. Correlations between spectral acceleration values at different periods should be properly represented when computing structural response. Synthetic far-field ground motions using the site-based model proposed by Rezaeian and Der Kiureghian are found to exhibit correlations between spectral periods that are higher than those of the recorded motions and higher than the empirical model of Baker and Jayaram. Variations of this model to reduce the correlations between the spectral periods are also explored. In this study, the effect of these higher correlations on structural response is examined by developing a structural model exhibiting higher mode effects and subjecting it to three different sets of ground motions: one recorded and two synthetic. For this model, results show that higher correlations tend to produce lower variability of structural response which in turn results in non-conservative estimates of seismic risk. Properly accounting for these correlations is expected to improve structural variability estimates. The second study presents methods to simulate near-fault ground motions for randomized source and site characteristics. It exten