Evaluation and Comparison of two Frequently Used Cyclic Constitutive Models for Sandy Soils

Selection of an appropriate soil constitutive model for advanced numerical analyses requires thorough understanding of behavior of constitutive models. Here, SANISAND and HS-SMALL models which are advanced widely available soil models for cyclic behavior of granular soils are evaluated. These models commonly used in engineering analyses. The behavior of these two models in cyclic and half-cycle triaxial tests in drained and undrained conditions are compared with laboratory results. Comparisons show that both models simulate monotonic tests accurately. However, comparisons in cyclic loadings with small and large strain amplitudes show neither of them can be expected to describe actual behavior of soils in all loading conditions. SANISAND in cyclic loading with large strain amplitude is reasonably match with actual behavior and can simulate the liquefaction phenomena, but in reloading paths in half-cycle drained triaxial test, shows discrepancies. HS-SMALL is less accurate in cyclic triaxial tests with large strains due to the perfect plastic mechanism of this model. HS-SMALL simulates damping and stiffness degradation of the soil with strain with relatively good accuracy in small strains, but fails to simulate volume changes in drained test or pore pressure generation in undrained tests, properly. Finally, both models can be recommended where soil elements undergo mainly monotonic paths. SANISAND is recommended for problems consisting cyclic loads with relatively large strain amplitude and HS- SMALL for use in simulations with cyclic loads with small strain amplitude where simulation of excess pore pressure generation is not of primary concern. These models are not recommended for half-cyclic loadings.