LS-DEM, was recently developed to overcome the challenges with capturing particle shape in the classical Discrete Element Method (DEM), which is used to model discrete particle interactions of granular materials. The key advance in the method is using level sets to characterize the shape morphology. This method is able to take advantage of 3D x-ray computed tomographic (XRCT) images, which provide exact granular morphology. Level sets are grids with values representing the distance to a surface. Negative values represent the inside of a surface, while positive values represent the outside of a surface. Additionally, level sets have fast particle interaction calculations due to the signed distances to the surface. Essentially, two particles will be in contact if the union of their level sets have negative values that overlap. The main drawback to using level sets is the large amount of memory that is required per grain, however, this can easily be overcome with standard domain decomposition techniques implemented in high performance computing environments.

Description:

This project will focus on the development of a high performance computing implementation of LS-DEM. A base software package is currently underdevelopment, which the research fellow's work will augment. Specific work on the project can be tailored to the fellows interest and could include among other possibilities implementing numerical time stepping methods, creating routines for classic geomechanical tests, developing multi-physics algorithms (i.e. thermal models), or high performance optimization of existing code on clusters.