Parallel Direct Simulation Monte Carlo and its application to flows in micro channels

Abstract

This chapter presents parallelization of Direct Simulation Monte Carlo (DSMC) to compute flow in micro-channel. DSMC is a particle-based method widely used to simulate flow of gases. Parallelization of DSMC has become important task to improve calculation efficiency. Most of its applications include dilute gases, although recently DSMC has found new applications in the field of hydrodynamical and chemical fluctuations and also in the modeling of gas flows in microelectromechanical systems (MEMS). The gas flow in micro-channels requires methods and descriptions, which are valid for dense gases and high Knudsen number. The flow rate, for channel and for wedge, show similar decreasing trend with increasing Knudsen number form 0.1 to 0.5. The DSMC is based on the uncoupling deterministic motion of particle and stochastic collisions. Therefore the parallelization approach is divided in two sections: deterministic motion of particles and stochastic part, related to collisions of particles. Parallelization of deterministic part of the algorithm revolves around domain decomposition and issues connected with communication between processor. For the stochastic part, the choice of efficient parallel random number generator is important.