Evaluating the site-specific applicability of one-dimensional seismic ground response analysis
One-Dimensional (1D) seismic ground response analysis is the most commonly performed analysis in geotechnical earthquake engineering. However, previous studies have shown a troubling fact that only a small fraction of sites are modeled well by 1D analysis. The objectives of this research are to assess the site-specific suitability of 1D analysis by identifying the issues that hinder the performance 1D analysis and to develop approaches to better match the observed sites response. The downhole array technique is used in this work to evaluate 1D analysis because it provides the most direct observations of how seismic waves are modified by the subsurface soil and rock. An important phenomenon in downhole array analysis is the potential presence of pseudo-resonances, which has not been effectively taken into account in previous studies and which affects the assessment of the accuracy of 1D analysis. The first part of this research provides insights into the cause and effect of pseudo-resonances and an approach is outlined to distinguish true-resonances from pseudo-resonances. The small-strain damping (D [subscript min] ) is a key parameter in linear ground response analysis and using laboratory-measured values tend to over-predict the response because it does not account for wave scattering present in the field. The second part of this research focuses on methods of increasing the D [subscript min] values in the profiles to better match observed site response, with the site response evaluated in terms of different ground motion characteristics. Alternatively, the randomization of shear wave velocity profiles is also assessed to provide more insights into the variable seismic properties at a site. A hypothesis that links the level of increased damping to the level of spatial variability in materials implied by the geologic conditions is proposed. To broaden the application of the 1D analysis, it is crucial to be able to identify sites that can be modeled accurately by 1D analysis. A taxonomy scheme is developed that classifies sites into different groups based on the similarity in their responses in terms of being modeled well by 1D analysis. This classification system is based on downhole array data but can be applied to non-downhole array sites. The taxonomy results presented in this study show that an increased portion of sites are suitable for 1D analysis.