Grain-Scale Numerical Modeling of Granular Mechanics and Fluid Dynamics and Application in a Glacial Context
The macroscopic behavior of granular materials is the result of the self-organizing complexity of the constituent grains. Granular materials are known for their ability to change phase, where each phase is characterized by distinct mechanical properties. This rich generic phenomenology has made it difficult to constrain generalized and adequate mathematical models for their mechanical behavior. Glaciers and ice streams often move by deformation of underlying melt-water saturated sediments. Glacier flow models including subglacial sediment deformation use simplified a priori assumptions for sediment rheology, which limit our ability to predict ice sheet dynamics in a changing climate.
In this talk I will present the soft-body Discrete Element Method which is a Lagrangian method I use in order to simulate the unique and diverse nature of granular dynamics in the subglacial environment. However, the method imposes intense computational requirements on the computational time step. The majority of steps in the granular dynamics algorithm are massively parallel, which makes the DEM an obvious candidate for exploiting the capabilities of modern GPUs. The granular computations are coupled to a fluid-dynamics solver in order to include grain-fluid feedbacks, which prove to be important for stick-slip behavior of glaciers.
All code is open source and freely licensed.