Micromechanical investigation of soil plasticity. Cambridge University, PhD thesis.

This research investigates the plasticity of crushable soil using the Discrete Element Method (DEM), with the computer code, PFC3D. DEM enables plastic soils to be modelled not as a continuum but as an aggregate of discrete grains. This study includes grain crushing as a crucial granular characteristic of soils. This is based on evidence from a literature study concerning soils crushing and an experimental study using a mini-oedometer installed with a digital camera. The experiments recorded real-time images of sands when compressed one-dimensionally. A relationship between grain crushing and macroscopic soil deformation was found. The DEM crushable soil was made of numerical agglomerates. Element test data was produced that covered a large range of stresses and densities, using five constant-value micro-parameters. Modelling the statistical strengths of these agglomerates during single particle crushing tests allowed the creation of an element having plastic behaviour comparable to the experimental data of a silica sand during both volumetric and shear loading. Breaking of agglomerates began at stresses less than one tenth of the statistical characteristic strength of individual grains during volumetric compression. Confining pressure could reasonably be normalised by the characteristic strength. Irrecoverable contraction of volume was always linked to particle breakage. Grain crushing was also seen to be responsible for the curved failure envelope of dilatant samples. The DEM simulations of generalised axi-symmetric stress-path directions created data that were identical in form to the laboratory triaxial tests which were originally used to support the various theories of plastic soil behaviour created by Roscoe, Rowe, Schofield, Burland, Bolton and others. The results replicated the partial successes of each approach and also demonstrated the dissonance between them which is evident in real soil data