Assessment of randomized Quasi-Monte Carlo method efficiency in radiative heat transfer simulations. (October 2019)

• Record Type: Journal Article
• Authors: Palluotto, Lorella; Dumont, Nicolas; Rodrigues, Pedro; Gicquel, Olivier; Vicquelin, Ronan
• Is Part Of: Journal of quantitative spectroscopy & radiative transfer. Volume 238(2019)
• Publication Date: 2019-10
• Subjects: Monte Carlo -- Quasi-Monte Carlo -- Low-discrepancy sequence -- Radiative heat transfer -- Combustion; Spectrum analysis -- Periodicals; Radiation -- Periodicals; Analyse spectrale -- Périodiques; Rayonnement -- Périodiques; Radiation; Spectrum analysis; Periodicals; 543.0858
• DOI: 10.1016/j.jqsrt.2019.07.013
• Languages: English
• ISSNs: 0022-4073

Abstract

Highlights: Randomized Quasi-Monte Carlo method is assessed in 3D configurations involving radiative heat transfer in participating media. The use of Randomized Quasi-Monte Carlo method reduces the computational time of the radiative heat transfer simulations of all the investigated cases compared to Monte Carlo method. Randomized Quasi-Monte Carlo is combined to a technique of importance sampling to further improve the efficiency. Abstract: Radiation can play a central role in turbulent reactive flows where heat transfer is enhanced in applications with high temperature and pressure. The Monte Carlo method is a successful technique to solve the radiative transfer equation accurately with relative ease while retaining detailed properties. However, its drawback is associated to a slow convergence rate. One strategy to improve the efficiency of Monte Carlo method consists in replacing the pseudo-random sequences with an alternative sampling: the low-discrepancy sequences. The introduction of such sequences in Monte Carlo leads to Quasi-Monte Carlo methods. Their advantage lies in a higher convergence rate compared to MC methods which have however not been assessed in 3D participating media. Additionally, in order to get an error estimation which is necessary in practical applications, a randomization of Quasi-Monte Carlo is needed (Randomized-QMC). Such Randomized-QMC methods have not been considered for simulations of radiative heat transfer in participating media before. In the present study, Monte Carlo and Randomized Quasi-Monte Carlo methods are assessed in terms of efficiency and computational cost in radiative heat transfer simulations of three practical 3D configurations. Comparisons in terms of local standard deviation, convergence rate, and final computational cost show that Randomized Quasi-Monte Carlo outperforms Monte Carlo in all the investigated cases.