On Numerical Simulations of Progressive Failure and Localized Deformation

Abstract

Soft sensitive clays are present in some of the most populated parts of Norway, and large slides in such clays have ahuge damage potential. It is thus important to understand their behavior and to be able to calculate their responses in connection to construction work. The strain-softening behavior of soft sensitive clays is well known. This behavior introduces some challenges to numerical modeling for softening materials, because strains then tend to concentratein thin zones, shear bands. The thickness of these zones is mesh dependent in standard Finite Element Method(FEM) calculations.One approach to overcome mesh dependency in FEM-calculations is to use rate dependent material models underdynamic loading conditions. This will in theory give a thickness of the shear band dependent on either acombination of parameters or a perturbation in the material, instead of dependency to the mesh spacing.A biaxial test is simulated under dynamic loading conditions in the finite element code Plaxis 2D. As material laws,the rate-dependent soil models Soft Soil Creep and n-SAC are used along with the rate-independent soil modelMohr-Coulomb.Results from the biaxial test are inconclusive both from the Soft Soil Creep model and the n-SAC model, while theMohr-Coulomb model as expected did not regularize the strain localization.The n-SAC soil model is also tested on stability calculations for a slope. In December 2006, a large landslide took place at Småröd, 90 km north of Göteborg, Sweden. This landslide was by an Independent Investigation Committee concluded to be caused by a large fill upslope, and the failure mechanism was addressed as progressive. According to progressive failure theory, this landslide can hence be classified as a downwards progressive landslide.Stability of the slope in Småröd is calculated using the soil model n-SAC, and variations of the simulations are performed in order to investigate effects of rate, updated geometry, mesh spacing and to see if the n-SAC model isable to capture the progressive failure mechanism.The n-SAC model is found capable of capturing the progressive failure mechanism as described in theory, but load capacity for the slope seems to be dependent of the mesh spacing. Dependency of the mesh spacing was also expected. Updated geometry seems to be essential for capturing the global failure mechanism experienced for the slope in reality. Updated geometry makes it also possible to capture the global failure mechanism for the rate independent non-softening material model Mohr-Coulomb as well. Rate effects in the stability simulations have notshowed conclusive results.Load capacity is in simulations for the Småröd slope found to be lower by all simulations using the n-SAC model,than in the simulations by the Independent Investigation Committee. The Independent Investigation Committee used both FEM with a Mohr-Coulomb soil model with Tresca failure criterion and Limit Equilibrium methods.