The effects of dynamic soil–structure interaction (SSI) have been extensively studied in the last few decades, and proper analysis for the linear elastic case in frequency domain has been established successfully. However, SSI is rarely considered in the design of building structures, and instead, buildings are frequently analyzed using a rigid base assumption and quasi-static loading conditions that ignore SSI and its dynamic nature. Acknowledging these shortcomings, the National Institute of Standards and Technology (NIST) published in 2012 a set of recommendations on time-domain analyses of SSI for building structures compatible with standard finite element packages for consideration in engineering design. The so-called NIST GCR 12-917-21 report introduced a major simplification to enable frequency domain tools to be implemented in time domain analyses. That is, replacing the frequency-dependent soil impedance functions by a single-valued functions read at the flexible-base structure frequency; This work seeks to quantify the accuracy of this simplification considering fully coupled two-dimensional (2D) finite element models (FEM) as the reference. Using a Bayesian approach based on ensemble Kalman inversion (EnKI) and a range of numerical simulations of soil–foundation–building interaction, we estimate the optimal frequency that can be used to estimate soil impedance for time domain analyses; and we evaluate the improvement that the corresponding impedance offers relative to the full FEM results when compared to time domain analyses performed in accordance to the NIST recommendations outlined above.

在过去的几十年中，对动力土－结构相互作用（SSI）的影响进行了广泛的研究，并且已经成功建立了对频域中线性弹性情况的正确分析。但是，在建筑结构设计中很少考虑使用SSI，而是经常使用刚性基础假设和忽略SSI及其动态特性的准静态荷载条件来分析建筑物。认识到这些缺点，美国国家标准技术研究院（NIST）于2012年发布了一套有关SSI时域分析的建议，以对与标准有限元软件包兼容的建筑结构进行工程设计。所谓的NIST GCR 12-917-21报告引入了一项重大简化，以使频域工具能够在时域分析中实施。也就是说，用在柔性基础结构频率处读取的单值函数代替频率相关的土壤阻抗函数；这项工作试图以完全耦合的二维（2D）有限元模型（FEM）为参考来量化这种简化的准确性。使用基于整体卡尔曼反演（EnKI）的贝叶斯方法和一系列的土-基础-建筑物相互作用数值模拟，我们估算了可用于估算时域分析中土体阻抗的最佳频率。