Calibrating frequency-independent soil–foundation parameters for time-domain substructure soil–structure interaction analysis

This study introduces a framework for deriving calibrated frequency-independent soil-foundation (CFISF) parameters for time-domain substructure soil-structure interaction (SSI) analysis. While various time-domain approximations of the frequency-dependent impedance function have been proposed to facilitate SSI consideration in seismic analysis and design, their practical implementation often faces a critical trade-off: oversimplification compromises accuracy, while excessive complexity hinders practical application. To resolve this limitation, we propose a framework to derive CFISF parameters that can accurately simulate the time-domain response of soil-structure systems while maintaining ease of implementation. The framework comprises three key steps. First, dimensional analysis establishes the functional forms of the relationships linking the CFISF parameters to the soil-structure system parameters. Second, a Bayesian model inversion technique is employed to estimate CFISF parameters for a broad range of soil-structure model configurations, generating a comprehensive dataset of soil-structure model parameters and their corresponding estimated CFISF parameters. This step involves: (1) simulating the time-domain soil-structure system response under white-noise excitation using the direct approach, which serves as the ground truth, and (2) estimating the CFISF parameters by minimizing discrepancies between the direct and substructure model responses. Third, regression analysis is applied to the dataset to determine the unknown coefficients of the functional forms derived from dimensional analysis, yielding explicit equations for the CFISF parameters. To evaluate the performance of CFISF parameters beyond the calibration dataset, we conducted an assessment study on four soil-structure systems with varying geometries and material properties. The results demonstrate that the CFISF parameters enable substructure models to closely replicate direct model responses, outperforming traditional time-domain impedance function approximations. This framework equips the practicing community with a reliable tool for incorporating SSI effects into seismic design, enhancing both the accuracy and efficiency of SSI modeling and analysis.