Effects of Soil Nonlinearity on Physics-Based Ground Motion Simulation and Their Implications on 1D Site Response Analysis: An Application to Istanbul

Abstract

Previously, we have conducted a suite of 57 broadband physics-based ground motion simulations (GMSs) for a region in Istanbul, Turkey, in which soils were modeled as linear elastic materials. However, from a geotechnical earthquake engineering point of view, soil can indeed exhibit nonlinear behavior, especially in shallow crust with soft soil layers and when subjected to strong ground shaking induced by seismic waves, and hence affect the wave propagation and ground motions. To quantitatively investigate the effects of soil nonlinearity on ground motions, in this study, we select four representative earthquake scenarios and perform fully nonlinear broadband (0–8 Hz) GMSs using a 3D bounding surface plasticity model. In addition, utilizing the motions at the bedrock level from 3D simulations, we conduct 1D nonlinear site response analyses (SRAs) for 2912 sites with different bedrock depths and $V_s$ profiles. Results indicate that compared to 3D nonlinear simulations, the 3D linear cases can both amplify and de-amplify ground motion intensities, depending on the ground shaking levels, while the 1D nonlinear SRAs are inclined to yield over-estimations, especially for vertical motions. Twelve stations are also selected to further evaluate the applicability of 1D SRA when soil nonlinearity is considered. Some features in 1D soil profiles, such as $V_s$ reversal and deep bedrock depth, are shown to yield unreliably under- and over-estimations, and therefore dramatically influence the accuracy of SRA predictions.