Modelling the effect of recent stress history on the deformation of overconsolidated soils
The aim of the research was to study the behaviour of overconsolidated soils subjected to small changes of strain or stress appropriate to the investigation of ground movements around excavations, retaining walls or foundations, and to develop a constitutive soil model that can predict such behaviour. The principal feature of soil behaviour investigated was the effect of recent stress history, defined by 9 the angle of rotation between the previous and current stress path directions. Stress path triaxial tests were carried out on both reconstituted and undisturbed samples of speswhite kaolin and London clay. The tests, which followed on from previous work by Richardson (1988), examined details of the influence of recent stress history, which was found to have a significant influence on the stress-strain response of the soil for the current loading path. The data from the tests together with a re-evaluation of the existing experimental data and a limited investigation of the effect of recent stress history in 3D stress space, enabled the main features of the soil behaviour to be identified. The stress-strain response of the soil was found to be highly non-linear, inelastic and dependent on recent stress history; if the stress path rotation was 18O, i.e. a complete reversal, the soil stiffness was at a maximum and was at a minimum for no rotation. As the loading path continued the influence of the recent stress history gradually diminished until it was no longer evident. Recent stress history also affects strain paths and effective stress paths measured during drained and undrained loading respectively. The significance of mean effective pressure and overconsolidation ratio was also investigated. Soil models which can predict this behaviour need to include at least one kinematic yield surface which allows plastic straining inside the state boundary surface. A two-surface yield model of this type, formulated by Al Tabbaa (1987) was evaluated. The predictions of this model did not compare well with several important aspects of the experimental observations and it was shown that an additional kinematic surface is necessary to model non-linearity, inelasticity and the recent stress history effect. A new three-surface model based on the two-surface model was developed, within the framework of Critical State soil mechanics, which successfully predicts all the main features of the soil behaviour. It is described by eight parameters which are all soil properties and, with one exception, all have a clear physical meaning.