Evaluation of the Impact of Incident Wavefield Modeling on Soil-Structure Interaction of Buildings Using Broadband Physics-Based 3D Earthquake Simulations

Abstract

Despite the complexity of real earthquake motions, the incident wavefield excitation for soil-structure interaction (SSI) analysis is conventionally derived from one-dimensional site response analysis (1D SRA), resulting in idealized, decoupled vertically incident shear and compressional waves for the horizontal and vertical components of the wavefield, respectively. Recent studies have revealed potentially significant deviation of the 1D free-field predictions from the actual three-dimensional (3D) site response and obtained physical insights into the mechanistic deficiencies of this simplified approach. Particularly, when applied to vertical motion estimation, 1D SRA can lead to consistent overprediction due to the refraction of inclined S waves in the actual wavefield that is not correctly accounted for in the idealized vertical P wave propagation model. However, in addition to the free-field site response, seismic demands on structures and non-structural components are also influenced by the dynamic characteristics of the structure and SSI effects. The extent to which the utilization of vertically propagating waves influences the structural system response is currently not well understood. With the recent realization of high-performance broadband physics-based 3D ground motion simulations, this study evaluates the impact of incident wavefield modeling on SSI analysis of representative building structures based on two essential ingredients: (1) realistic spatially dense simulated ground motions in shallow sedimentary basins as the reference incident motions for the local SSI model and (2) high-fidelity direct modeling of the soil-structure system that fully honors the complexity of the incident seismic waves. Numerical models for a suite of archetypal two-dimensional (2D) multi-story building frames were developed to study their seismic response under the following incident wavefield modeling conditions: (1) SSI models with reference incident waves from the 3D earthquake simulation, (2) SSI models with idealized vertically incident waves based on 1D SRA, and (3) conventional fixed-base models with base translational motions from 1D SRA. The impact of these modeling choices on various structural and non-structural demands is investigated and contrasted. The results show that, for the horizontal direction, the free-field linear and nonlinear site amplification and subsequent dynamic filtering of the base motions within the structure can be reasonably captured by the assumed vertically propagating shear waves. This leads to generally fair agreements for structural demands controlled by horizontal motions, including peak inter-story drifts and yielding of structural components. In contrast, vertical seismic demands on structures are overpredicted in most cases when using the 1D wavefields and can result in exacerbated structural damage. Special attention should be given to the potentially severe vertical floor accelerations predicted by the 1D approach due to the combined effects of fictitious free-field site amplification and significant vertical dynamic amplification along the building height. This can pose unrealistic challenges to seismic certification of acceleration-sensitive secondary equipment necessary for structural and operational functionality and containment barrier design of critical infrastructures. It is also demonstrated that vertical SSI effects can be more significant than those in the horizontal direction due to the large vertical structural stiffness and should be considered in vertical floor acceleration assessments, especially for massive high-rise buildings.