Influence of boundary conditions, specimen geometry and material heterogeneity on model calibration from triaxial tests

The recent capability of measuring full-field deformations using advanced imaging techniques provides the opportunity to improve the predictive ability of computational soil mechanics. This paper investigates the effects of imperfect initial specimen geometry, platen-soil and apparatus compliance, and material heterogeneity on the constitutive model calibration process from triaxial tests with nonlubricated platens. The technique of 3D-Digital Image Correlation (3D-DIC) was used to measure, from digital images, full-field displacements over sand specimen surfaces throughout triaxial compression tests, as well as actual specimen initial shape, and deformations associated with platen and apparatus compliance and bedding settlement. The difference between predicted and observed 3D specimen surface deformations served to quantify an objective function in the optimization algorithm. Four different three-dimensional finite element models (FEMs), each allowing varying degrees of material variability in the solution of the inverse problem, were used to study the effect of material heterogeneity. Results of the parametric study revealed that properly representing the actual initial specimen geometry significantly improves the optimization efficiency, and that accounting for boundary compliance can be critical for the accurate recovery of the full-field experimental displacements. Allowing for nonsymmetric material variability had the most significant impact on predicted behavior. A relatively high coefficient of variation in model parameters was found among a statistical ensemble of tests, underscoring the importance of conducting multiple tests for proper material characterization. Copyright (C) 2009 John Wiley & Sons, Ltd.