Development of a Hybrid CPU-GPU Computational Tool for Modeling Flows through Porous Media using Particle Methods

Deborah Levin, University of Illinois at Urbana-Champaign

Usage Details

The problem that we are focusing on is the simulation and modeling of chemically reacting flows through micro-porous material/structures. The understanding of the diffusion mechanism through the microporous structures and estimation of the heat transfer rate on the material surface from DSMC simulations can have a significant impact on the design of fuel cells, CO₂ removal membranes, as well as thermal protection systems used in space reentry vehicles. In our DSMC solver, the Barnes-Hut grid-free octree structure provides flexibility to handle highly irregular and complex immersed bodies. Attention is paid to key factors such as the size of the octree leaf node, which should be less than the local mean free path. To justify the decoupling of particle movement and collision processes in DSMC, the timestep is nearly one fifth of the mean collision time. Additionally, to account for the presence of complex geometries, ray-tracing modules are integrated to determine gas-surface interactions and compute volumes of the irregular gas phase regions.