Comparison of total and effective stress-based constitutive models for modeling strain-softening of fine-grained soils

抄録

Infrastructure damage attributed to strain-softening of clayey soils has been documented in many case histories involving seismic loading. For critical infrastructure projects, potential deformations due to seismic loading are increasingly being analyzed using numerical modeling approaches, such as nonlinear deformation analyses. These analyses rely on material models which can represent the aspects of soil behavior important to the problem being analyzed. This study examines the ability of the effective stress-based PM4Silt constitutive model to represent cyclic-softening in single element direct simple shear simulations and field scale simulations of the Fourth Avenue landslide. Guidance is provided regarding calibration of PM4Silt and the sensitivity of the solution to the input parameters is explored. The results using PM4Silt are then compared to a total stress-based constitutive model to understand how differences in constitutive model complexity affect the results at both the element scale and field scale. Mesh dependency of the solution using both models is explored with and without regularization approaches. PM4Silt is shown to reproduce the deformations observed at the Fourth Avenue Landslide. Effective stress-based constitutive models are found to be more appropriate than total stress-based models when element level response or a more complete estimation of deformations patterns, both inside and outside the failure mass, is required. Total stress-based models are found to be adequate if the user is primarily interested in whether failure will occur. Mesh dependency is observed in the PM4Silt model, but these effects are reduced by using a displacement-based calibration procedure.