PhD Final Defense – Dylan Centella Raef

Modeling 3D Wave Propagation Effects in Seismic Site Response Analysis

Advisor: Professor Youssef M.A. Hashash

Abstract

Site response analysis (SRA) is used to estimate how local soil and geologic conditions modify earthquake ground motions, which is critical for seismic hazard evaluation used in engineering design. One-dimensional (1D) SRA remains the state of practice and research because it is simple, robust, and computationally efficient. However, by assuming laterally homogeneous soil conditions and vertically propagating shear waves, 1D SRA cannot fully represent three-dimensional (3D) wave propagation effects and 3D soil stress–strain behavior. Comparisons with borehole array recordings have shown that 1D analyses fail to adequately simulate ground motions at more than 50% of the sites investigated, suggesting that multidimensional wave propagation may contribute to the observed discrepancies between simulations and measurements. This dissertation evaluates the role of 3D wave propagation effects in SRA by developing and applying a framework for linear and nonlinear 3D SRA at two well-instrumented borehole array sites: the Treasure Island Downhole Array (TIDA) in California and the Delaney Park Alaska Digital Array (DPDA) in Alaska.

The study first develops and validates a linear 3D SRA framework using low-intensity recordings, for which soil response is expected to remain approximately elastic. Pseudo-3D shear-wave velocity (VS) models developed using the H/V geostatistical approach are implemented in a high-performance finite element platform for each site. Across both sites, simulations that include wave passage effects best reproduce measurements by reducing the fundamental mode overprediction observed in 1D analyses. The validated models are then used to quantify 3D effects through smoothed effective amplitude spectrum (EAS)-based and square-root-sum-of-squares (SRSS) response spectra (RS)-based 3D/1D scaling factors, defined as the ratio of surface response from a 3D simulation to that from a companion 1D column at the same location.

The framework is then extended to nonlinear conditions using the I-soil constitutive model and risk-targeted Maximum Considered Earthquake (MCER)-level input motions developed for each site. Relative to the linear results, nonlinear 3D/1D scaling factors become broader, less sharply peaked, and shifted toward lower frequencies, consistent with period lengthening. The strong directional dependence observed under linear simulations is also substantially reduced once nonlinear soil behavior is mobilized. A parametric study further shows that apparent wave passage velocity and propagation direction control the magnitude, frequency dependence, and spatial distribution of 3D effects under low-intensity shaking, whereas input motion intensity becomes the dominant controlling parameter under nonlinear conditions. A targeted subset of high-resolution simulations confirms that the original model resolution produces consistent results within the intended frequency range. Differences are primarily limited to higher frequencies where the higher-resolution model can more effectively propagate seismic energy.

The developments presented in this dissertation demonstrate a practical framework for evaluating nonlinear 3D site response analysis by combining a cost-effective site characterization approach with a high-performance finite element platform. The framework incorporates laterally variable subsurface conditions, wave passage effects in the input motions, and nonlinear soil behavior under 3D stress states. The results show that 3D SRA generally predicts higher surface response than 1D SRA. Under low-intensity shaking, these differences are more pronounced near shallowing bedrock areas. Moreover, wave passage strongly influences 3D effects under low-intensity shaking but this influence decreases under stronger shaking. As shaking intensity increases, nonlinear soil behavior becomes the controlling factor and increases the differences between the 1D and 3D results. These results provide evidence of the need for evaluating 3D site effects as part of the site-specific seismic hazard assessment.