Computational modelling of dynamic soil-structure interaction in shear wall buildings with basements in medium stiffness sandy soils using a subdomain spectral element approach calibrated by micro-vibrations

Abstract

This paper presents a strategy for modelling dynamic soil-structure interaction (DSSI) using the spectral element method (SEM) with a Discontinuous Galerkin approach, calibrated by micro-vibrations. The proposed methodology allows not only to adjust the vibration frequencies of the structure but also the observed vibration modes. First, models of two structural shear wall buildings with basements in medium dense sandy soils are developed to estimate empirical modal characteristics and calibrate the structural subdomain and low-strain site properties. Convenient 3D arrays of multiple seismic sensors are used to obtain the environmental vibrations measurements. Afterwards, an optimization process is conducted to calibrate volumetric models of structures. This optimization is performed by preserving the most relevant modal frequencies and shapes to achieve an equivalent dynamic response. Finally, structural models are placed into a neighbouring soil model (soil subdomain), approximating nonlinear soil behaviour by an equivalent linear strategy. Using this complete soil-structure interaction model, relevant engineering performance parameters are assessed via simulations of buildings subjected to a plane wave excitation. The results show the significant effect DSSI have in shear-wall buildings with basements and the importance of considering the flexibility of the foundation in the interpretation of the results. In general, results indicate that DSSI effects are strongly dependent on the input frequency content, which might cause a reduction of the inter-story drifts. Furthermore, a significant period lengthening of the studied structures up to 47% is found, as well as a considerable decrease in story shear up to 220% and a maximum lateral roof displacement reduction of 34% when compared against fixed base referential responses.