Development of an AMR Octree DSMC approach for shock dominated flows

Key strategies used in the development of a scalable, three-dimensional direct simulation Monte Carlo (DSMC) program are described. The code employs an Octree based adaptive mesh refinement (AMR) that gives flexibility in capturing multi-scale physics. It is coupled with a robust cut-cell algorithm to incorporate complex triangulated geometries. With the use of distributed memory systems and Message-Passing-Interface (MPI) for communication, the code is potentially scalable. However, to simulate continuum-like conditions involving multi-scale physics, better scalability that has yet been achieved is desirable. The thesis identifies two main performance bottlenecks in simulating at continuum-like conditions, first, improving the scalability of the code for more than 128 processors by reducing the communication and evenly balancing the computational load, and second, improve the algorithmic performance of the code by eliminating the expensive recursive tree traversal inherent in Octree based mesh structure. In order to resolve the first issue sophisticated graph-partitioners have been used, however, without success. The thesis also explains the special considerations required for embedded geometries in a parallel computational environment. An efficient algorithm is discussed that allows for the checking of particle-surface interaction only if they are close enough to the geometry. The code calculates various surface coefficients and employs the Borgnakke-Larsen continuous relaxation model to simulate inelastic collisions of diatomic molecules. Finally, these strategies and models are validated by simulating hypersonic flows of argon and nitrogen over a hemisphere and double-wedge configuration and the solutions are compared with the results obtained from an older DSMC code known as SMILE