Implementation of a Nonlinear Soil Material Model into a FEA Framework for the Simulation of the Seismic Soil-Structure Interaction
Date Issued
November 1, 2019
Author(s)
Abstract
Taking into account the Soil-Structure-Interaction (SSI) effect when evaluating the response of structural systems is important as it provides more realistic behaviors than using the simplified approach of assuming that the base is fixed. Additionally, most of the international literature that investigate SSI, use linear material models to discretize the soil domain or springs. However, the previous assumption doesn’t take into account the damping effect of the soil and other 3D phenomena that occur when the foundation and soil interact. In this research work, a numerical investigation on the SSI effect is carried out on two Reinforced Concrete (RC) frames (L30 and H30) that were studied experimentally using fixed base boundaries. The two models differ in the ductility behavior, where L30 RC frame is reinforced in a way to develop a low ductile behavior compared to the H30 RC frame. Nevertheless, the soil domain herein, modeled with the Ramberg-Osgood constitutive model, which is a nonlinear material model integrated to a Finite Element Analysis (FEA) software for the needs of this research work. The Finite Element (FE) models that were constructed in this thesis are subjected to a nonlinear dynamic analysis. Twelve soil domains that have different material properties and depths are considered. The material properties of the different soil domains (hard, medium and soft soil) were taken according to ASCE7-10 standard. It is worth mentioning that 3D parametric hexahedral elements are used to discretize all the structural elements, including beams, columns, slabs and the soil continuum. Detailed reinforcement including shear and longitudinal rebars are modeled within the RC mesh as embedded elements in order to enhance the accuracy of the model and simulate the steel reinforcement in an exact manner. Finally, the dynamic response of the fixed and SSI models were compared.