The Discrete Element Method (DEM) is used to explore the highly nonlinear dynamics of a granular bed when exposed to stress conditions comparable to those at the bed of warm-based glaciers.

In the DEM, the material is simulated on a grain-by-grain basis, and defining the micromechanical properties of the inter-particle contacts parameterizes the model. For validating the numerical approach, the macromechanical behavior of the numerical material is compared to the results from successive laboratory ring-shear experiments. Overall, there is a good agreement between the geotechnical behavior of the real granular materials and the numerical results. The materials deform by an elasto-plastic rheology under the applied effective normal stress and horizontal shearing. The peak and ultimate shear strengths depend linearly on the magnitude of the normal stress by the Mohr-Coulomb constitutive relationship.

The numerical approach allows for a detailed analysis of the material dynamics and shear zone development during progressive shear strain. We demonstrate how the shear zone thickness and dilation increase with the magnitude of the normal stress. The stresses are distributed heterogeneously through the granular material along stress-carrying force chains. Between the force chains are the volumetrically dominant inactive zones. Overall, the force chain orientation is parallel to the maximum compressive stress.

The data-parallel nature of the basic DEM formulation makes the problem ideal for utilizing the high arithmetic potential of modern general-purpose GPUs. Using the Nvidia CUDA C toolkit, the algorithm is formulated for spherical particles in three dimensions with a linear-elastic soft-body contact model. We have coupled the DEM model to a model for porewater flow, and we present early results of particle-porewater interactions. The two-way mechanical coupling is used to investigate pore-pressure feedbacks, which may be very important for the dynamics of soft-bedded glaciers.