Estimation of critical state-related formula in advanced constitutive modeling of granular material

Different critical state-related formulas, for the critical state line and the critical state-dependent interlocking effect, have been proposed in constitutive modeling of granular material during last decades, which rises up a confusion on how to select an appropriate model in the geotechnical applications. This paper aims to discuss the selection of these critical state-related formulas and parameters identification. Three formulas of critical state line together with two formulas of critical state-dependent interlocking effect are combined to propose six elasto-plastic models. Drained and undrained triaxial tests on four different granular materials are selected for simulations. In order to eliminate artificial errors, a new hybrid genetic algorithm-based intelligent method is proposed and used to identify parameters and estimate simulations with minimum errors for each granular material and each model. Then, the performance of each CSL and each state parameter is evaluated using two information criteria. Furthermore, the performance was evaluated by simulating three footing tests using finite-element analysis in which the models are implemented. All comparisons demonstrate the incorporation of nonlinear critical state line combined with the state parameter e/e c in constitutive modeling can result in relatively more satisfied simulated results.