Modelling Unsteady Processes with the Direct Simulation Monte Carlo Technique

摘要

Over the past 40 years, the Direct Simulation Monte Carlo (DSMC) technique has been developed into a flexible and effective solver for flow problems in the rarefied to near continuum regime. However, even with modern parallelised code, the efficient computation of unsteady near-continuum flows, which are important in processes such as Pulsed Pressure Chemical Vapour Deposition (PP-CVD), remains a challenge. We have developed an unsteady parallel DSMC code (PDSC) utilising advanced features such as transient adaptive sub-cells to ensure nearest neighbour collisions and a temporal-variable time step to reduce computation time. This technique is combined with a unique post-processor called the DMSC Rapid Ensemble Averaging Method (DREAM) which reduces the statistical scatter in the data sets produced by PDSC. The combined method results in a significant memory and computational reduction over ensemble averaging DSMC, while maintaining low statistical scatter in the results. The unsteady code has been validated by simulation of shock-tube flow and unsteady Couette flow, and a number of test cases have been demonstrated including shock impingement on wedges. The technique is currently being used to model the development of an underexpanded jet in a PP-CVD reactor.