Performance of different constitutive soil models: from element tests to the simulation of vibratory pile driving tests

Abstract

The present study deals with the investigation of the applicability (by means of parameter calibration), robustness and prediction quality of advanced constitutive soil models for the numerical investigation of complex geotechnical problems. The range of available constitutive soil models extends from simple linear to time-dependent and hydromechanically coupled nonlinear modelling approaches. It is the user's task to select a constitutive model suitable for the problem at hand. This requires in-depth knowledge of the soil behaviour as well as the strengths and weaknesses of the available constitutive models, most of which have only been validated using element test simulations. The procedure from parameter calibration using laboratory tests under well-defined boundary conditions (element tests) to the simulation of boundary value problems is complex in many respects and is often not followed with advanced constitutive models due to the large number of parameters required and the necessary laboratory tests. In this paper, the prediction quality of three models, namely Hypoplasticity with Intergranular Strain, Sanisand and Hypoplasticity with Intergranular Strain Anisotropy is inspected. The investigation is carried out based on back-calculations of laboratory tests and a well-documented model test to evaluate their suitability in representing complex soil mechanical aspects, such as the material behaviour under cyclic loading, particularly pore pressure accumulation. The parameter calibration is performed both "manually" as well as with a specially developed automatic calibration software. Subsequently, model tests of vibratory pile driving in water-saturated sand are simulated using the previously calibrated parameters.