Finite element implementation, validation, and deep foundation application of a bounding-surface plasticity model.

Abstract

The Bardet model requires nine parameters in order to describe the behavior of a soil. This thesis presents methods of determination of these parameters and the model sensitivity with respect to variation in values of its parameters. This sensitivity analysis illustrates the relative importance of each one of the model parameters in the overall performance of the model. The model parameters are grouped into three categories based on their relative effect on the model performance. The calculation of pore water pressure is achieved by considering an extremely small compressibility for the pore water and soil particles. The pore water pressure is expressed as the product of the volume change and the combined bulk modulus of soil particles and pore water. For drained condition, the combined bulk modulus takes the value of zero. For undrained conditions, it approaches infinity and the soil becomes incompressible. Verification of implementation of the model in the finite element program is carried out. Finite element calculations are made for soil behavior in different laboratory tests and compared with results of calculations made by integrating the constitutive equations numerically. Four test conditions are considered in the verification process. These are, monotonic drained, monotonic undrained, cyclic drained, and cyclic undrained cases. In the first stage of validation, the performance is investigated when the constitutive model is used to simulate stress-strain and strength of cohesionless soils tested under variety of test conditions. Comparisons are made between laboratory test results and the variety of test conditions. Comparisons are made between laboratory test results and the corresponding model simulations. The behavior of Sacramento River sand, Fuji River sand, a crushed quartz sand, and a sandy silt, are considered in a number of conventional and non-conventional stress-path and strain-path tests. Comparisons between the model simulations and the measured responses of these soils demosntrate a very good performance of the model. In the second stage of validation, analyses of three boundary value problems are performed. The first, is the behavior of a model-scale footing resting on the surface of a sandy silt. The second, is the behavior of 14/25 Leighton Buzzard sand in the Cambridge Simple Shear Device. The third, is the analysis of a six metre wide foundation resting on the ground surface for three different conditions of the supporting soil. (Abstract shortened by UMI.)