Performance Comparison of Two Monte Carlo Ray-tracing Methods for Calculating Radiative Heat Transfer

Abstract Monte Carlo ray tracing (MCRT) methods are important for solving the radiative transfer equation. However, the particular ray-tracing procedure used has a significant impact on the calculational performance. In this paper, we analyze and compare the performances of MCRT pathlength (PL) and collision-based (CB) methods for different surface and medium parameters and different degrees of uniformity. The results show that in a gray medium and with a radiation heat balance system, the PL method is superior when the surface emissivity is less than 0.15 for surface elements, and when the mean optical thickness per element (MOTE) is small and the emissivity is large for space elements; otherwise, the CB method is superior. Also the overall performance of the PL method is better than that of the CB method. However, PL method is more sensitive to non-uniformity of medium parameters than the CB method, and that the PL method is more sensitive to uniformity of the scattering coefficient than to the absorption coefficient. In addition, for a gray body and uniform optical parameters, reducing the number of grids improves the performance of the MCRT. And the performance index of the PL method is the best when the cutoff level of energy beam tracking is 10^-7 and 10^-6 for surface and space elements respectively. These quantitative conclusions will help in selecting suitable MCRT methods for engineering calculations, based on the physical properties of materials and conditions.