High-performance Computing of Hypersonic, Shock-shock Interactions using Kinetic, Particle Approaches

Deborah Levin, University of Illinois at Urbana-Champaign

Usage Details

Recent experiments performed at the University of Illinois provide flow visualization and heat flux of hypersonic flows over a double-wedge configuration. The modeling of such three-dimensional flows using kinetic, particle methods such as direct simulation Monte Carlo (DSMC), provides the highest fidelity in understanding thermochemical non-equilibrium processes when extremely complex shock interactions are present. Although the DSMC algorithm is massively parallel, it is still computationally intensive. We have developed a new code that takes advantage of Adaptive Mesh Refinement techniques using unstructured octree grids to maximize the placement of computational particles in flow regions where the collision frequency is highest. The parallelization of our new Scalable Unstructured Gas dynamics Adaptive Refinement (SUGAR) code has been analyzed and found able to handle hundreds of billions of particles. However only a petascale asset such as Blue Waters will make such simulations feasible.