An analysis of the convergence of the direct simulation Monte Carlo method

JOURNAL OF COMPUTATIONAL PHYSICS

卷:289

Article

Galitzine, Cyril¹  Boyd, Iain D.¹ 

[1] Univ Michigan, Dept Aerosp Engn, Ann Arbor, MI 48109 USA

关键词: DSMC;    Direct simulation Monte Carlo;    Particle method;    Error;    Convergence;    Correlation;   

DOI  :  10.1016/j.jcp.2015.02.039

来源: Elsevier

【 摘 要 】

In this article, a rigorous framework for the analysis of the convergence of the direct simulation Monte Carlo (DSMC) method is presented. It is applied to the simulation of two test cases: an axisymmetric jet at a Knudsen number of 0.01 and Mach number of 1 and a two-dimensional cylinder flow at a Knudsen of 0.05 and Mach 10. The rate of convergence of sampled quantities is found to be well predicted by an extended form of the Central Limit Theorem that takes into account the correlation of samples but requires the calculation of correlation spectra. A simplified analytical model that does not require correlation spectra is then constructed to model the effect of sample correlation. It is then used to obtain an a priori estimate of the convergence error. (C) 2015 Elsevier Inc. All rights reserved.

【 授权许可】

Free