Experimental Evaluation of Simplified Models for Seismic Soil–Pile–Structure Interaction: Limitations of Rigid‐Base and Equivalent SDOF Approaches

ABSTRACT The prediction of seismic responses in soil–pile–structure (SPS) systems presents a significant challenge due to the complex effects of soil–structure interaction (SSI). Conventional design methods, such as the rigid base assumption and lumped mass (single‐degree‐of‐freedom, SDOF) modeling, often oversimplify these interactions, potentially leading to unsafe or overly conservative designs. This study presents a series of 1‐g shaking table tests conducted to experimentally investigate the seismic behavior of SPS systems, with results interpreted at prototype scale. Time‐frequency analysis using frequency sweep tests and the Stockwell transform was employed to systematically identify modal characteristics and higher‐mode effects. The experiments were conducted on dry sand with varying relative densities and subjected to a range of peak base accelerations (PBAs). Dynamic responses, including natural period shift, damping ratio variations, base shear forces, and inter‐story drifts, were systematically compared across rigid base, SDOF, and fully modeled SPSs. The findings reveal that simplified models significantly overestimate structural stiffness and underestimate seismic demands, particularly under higher PBAs. Specifically, the SDOF method was found to underestimate the resonant period of the SPS, limiting its ability to accurately simulate dynamic behaviors. In addition, the base shear forces measured from pile‐supported systems were higher than rigid base systems because of the amplification by SSI. The findings provide a direct, experimental quantification of the discrepancies between simplified models (SDOF and rigid base) and a realistic MDOF system. This study's primary contribution is the experimental validation that highlights the conditions under which simplified models can be non‐conservative, providing critical data for practicing engineers to refine their design assumptions.