Influence of number of simulated particles on DSMC modeling of micro-scale Rayleigh–Bénard flows

Two-dimensional micro-scale Rayleigh–Bénard flows are investigated numerically using direct simulation Monte Carlo method. An enclosure of length-to-height aspect ratio of AS = 4 is taken to explore the influence of initial setting of simulated molecules. The simulation domain is divided into 81 × 21 sampling cells and the range of Rayleigh number from 3000 to 10 000 corresponds to the convection state. Cases of 8, 10, 12, 14, 16 and 24 simulated particles in each collision cell are examined. It is shown that flow patterns with three-, four- or five-roll modes may appear depending on the number of simulated particles.