Comparative study of homogenization techniques for evaluating the bearing capacity of bimsoils under shallow foundations

Abstract

Bimsoils, characterized by discrete blocks within a finer-grained matrix, pose challenges in evaluating the bearing capacity of shallow foundations due to their spatial variability. To address this, a binary random field coupled with a Finite-Element model is used to simulate the variability of bimsoils. This study focuses on investigating the effect of the blocks spatial fraction (γ), the isotropic and anisotropic spatial correlations, and the undrained shear strength (cu) ratio between the matrix and the blocks. The results reveal that the coefficients of variation (CV) for the bearing capacity (qu), at a given γ, reach nearly 20% due to the diverse spatial configurations. This dispersion is attributed to the development of distinct failure mechanisms. However, optimizing the local average area used to evaluate γ can help reduce these CV values. The average qu for a given γ can be accurately determined using the Bruggeman symmetric effective medium (BEM) equation, ensuring safe design compared to traditional homogenization techniques. The BEM equation considers γ and the cu ratio, providing an accurate estimation of bearing capacity for an equivalent homogeneous model suitable for probabilistic analyses.