A finite element modeling of drained triaxial test on loose sand using different constitutive models

Numerical modeling of soil can be used as a complementary or alternative method for laboratory tests. Therefore, in the simulations of geotechnical problems, properly using constitutive models (such as the cap model) requires accurate calibration of the model parameters. In the present study, the draining behavior of Nevada sand at a relative density of 40% was evaluated using Mohr-Coulomb (MC), Drucker-Prager (DP), and modified Drucker-Prager/cap (MDPC) models in Abaqus and compared with laboratory data. In this context, the conventional triaxial compression (CTC) technique was used for finite element modeling of selected drained triaxial tests by keeping the radial stress constant and increasing the axial stress. Based on the results, MC and DP constitutive models, where the behavior of the materials is linear elastic-perfectly plastic, at high confining pressures, due to the inability to simulate soil hardening, showed significant differences with the experimental results. In other words, with increasing confining pressure, the behavior of sand tends to harden, and the ability of the MDPC model, which has a hardening function based on volumetric plastic strain, increased in simulating sand behavior. Proper determination of cap parameters can have a significant effect on the results. In the present study, cap hardening parameters for Nevada sand have been determined based on experimental data.