Advances in Discrete Element Modelling using GPUs
Please login to view abstract download link
Simulation of granular flow using the Discrete Element Method (DEM) is a crucial tool that is used in the pharmaceutical industry to gain understanding of processes, giving additional insights into the behavior of the material under different conditions, leading to improved designs and optimizations. Typically, large amounts of particles are needed to accurately model real-world problems. Together with small time steps, the problem becomes very expensive in terms of computational effort. This work presents the evolution of modern Graphics Processing Units (GPUs), and compares theoretical compute capabilities with actual gains in simulation performance we have observed using our code XPS. For this purpose, we analyze models of various pharmaceutical processes, considering cohesive powders, non-spherical particle shapes like tablets, contact models with contact history, and intra-particle coating using efficient algorithms from the field of computer graphics. Finally we show how the GPU advances the DEM.
