Combustion and Flame, Vol.221, 352-363, 2020
Influence of gas radiative property models on Large Eddy Simulation of 1 m methanol pool fires
The main objective of this work is to analyze the effects of gas radiative property models on the radiative outputs and flame structure in a 1 m diameter methanol pool fire. Large Eddy Simulation (LES) are run with the non-adiabatic steady laminar flamelet (SLF)/presumed filtered density function (FDF) model to close subgrid-scale (SGS) turbulence-chemistry and turbulence-radiation emission interactions. The radiative transfer equation (RTE) is solved using the Finite Volume Method (FVM) with different angular meshes, including uniform angular dicretizations comprising 16 (polar angle) x 24 (azimuthal angle) and 48 x 96 control angles and the FTn scheme with n up to 48. Four gas radiative property models, namely the Rank Correlated Full Spectrum k-distribution (RCFSK), the non-grey Weighted-Sum-of-Grey-Gases (WSGG), and two versions of the grey WSGG based on different evaluations of the mean path length are assessed. Model predictions with the RCFSK reproduce with fidelity the available experimental data in terms of flame structure and radiative loss to the surrounding. The radiative heat transfer within the flame, where radiation is isotropic, is adequately resolved with the FT12 angular resolution whereas finer angular meshes with at least 32 segments to discretize the polar angle are required to predict accurately the vertical distributions of radiative flux outside the flame. The grey models fail to reproduce the radiative structure of the flame, predicting an optically thin flame instead of an optically intermediate medium. In an opposite way, although the non-grey WSGG overestimates both emission and absorption, it provides predictions in overall reasonable agreement with the RCSFK. (C) 2020 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
Keywords:Large methanol pool fire;Large eddy simulation;Radiative heat transfer;Gas radiative property models;Angular discretization