A Parallel DSMC Implementation Including Adapative Mesh Refinement and Cut-Cell Algorithms

Abstract

Parallel implementation of a three-dimensional direct simulation Monte Carlo (DSMC) code is outlined that runs on distributed memory systems using the Message-Passing Interface (MPI). The code employs an embedded 3-level Cartesian mesh, accompanied by a cut-cell algorithm to incorporate triangulated surface geometry into the adaptively rened Cartesian mesh. Two separate data structures are proposed in order to separate geometry data from cell and particle information. The geometry data structure requires little memory so that each partition in a parallel simulation can store large portions of the entire mesh. This has the potential to decrease inter-processor communication for particle movement and for adaptive mesh renement in large parallel simulations. Parallel strategies are detailed for the main DSMC loop, adaptive mesh renement, and the cut-cell routine. Parallel scalability is tested on a simple test case of free-stream ow. For a xed problem size, linear scaling is demonstrated until the number of simulation particles per processor falls below approximately 625,000. However, when the problem size is increased with the number of processors (constant granularity), linear scaling is demonstrated up to 128 processors.