Bearing Capacity of Spatially Random Cohesive Soil Using Numerical Limit Analyses

Abstract

This paper describes a probabilistic study of the two-dimensional bearing capacity of a vertically loaded strip footing on spatially random, cohesive soil using numerical limit analyses (NLA-CD). The analyses use a Cholesky decomposition (CD) technique with midpoint discretization to represent the spatial variation in undrained shear strength within finite-element meshes for both upper- and lower-bound analyses and assumes an isotropic correlation length. Monte Carlo simulations are then used to interpret the bearing capacity for selected ranges of the coefficient of variation in undrained shear strength and the ratio of correlation length to footing width. The results are compared directly with data from a very similar study in which bearing capacity realizations were computed using a method of local average subdivision (LAS) in a conventional displacement-based finite-element method (FEM-LAS). These comparisons show the same qualitative features but suggest that the published FEM calculations tend to overestimate the probability of failure at large correlation lengths. The NLA method offers a more convenient and computationally efficient approach for evaluating effects of variability in soil strength properties in geotechnical stability calculations.