Computational Modeling of Strain Localization in Frictional Granular Materials

<p> This research project addresses fundamental processes associated with the deformation and failure of granular materials; specifically strain localization, and the role that it plays in earthquake-induced ground failures. Experimental observations suggest that strain localization is an integral part of the ground failure mechanism, even in soil that has liquefied as result of strong ground shaking. The goal is to quantify this, based on existing experimental data, in order to develop a micromechanically-based unified computational constitutive model to analyze and quantify experimental observations. The model has broad application to many problems in granular materials, in particular to ground failure in earthquakes. The program involves the creation of a science-based computational model through the following steps: 1. Starting with the micro-scale and grain-to-grain interaction, to explicitly relate fabric and its evolution to macroscopic backstress and its evolution. 2. Based on grain-to-grain contact forces and the distribution density function of the contact normals (which characterizes the fabric), obtain explicit expressions for the resistance to the dilatant shearing observed and quantitatively documented in photoelasic experiments. 3. Based on the dilatant sliding model at the meso-scale, develop explicit constitutive relations involving just a few well-defined constitutive parameters. 4. Using an existing extensive experimental database, relate the constitutive parameters to test data. 5. Develop computational algorithms based on the plastic predictor-elastic corrector concept. 6. Using the constitutive algorithm in DYNA3D, quantify experimental results on shearbanding. This also will verify the effectiveness of the model for application to other problems. </p>