화학공학소재연구정보센터
Combustion Science and Technology, Vol.186, No.4-5, 475-489, 2014
RADIATIVE HEAT TRANSFER THROUGH THE FUEL-RICH CORE OF LABORATORY-SCALE POOL FIRES
Radiative heat transfer calculations are conducted along the axis of six axisymmetric pool fires by using the exact line-by-line (LBL) method, the narrow band correlated k (NBCK) model, the full-spectrum correlated k (FSCK) model, the multi-scale full-spectrum k-distribution (MSFSK) model, and the wide-band model implemented in the fire dynamic simulator (FDS). The two baseline cases correspond to 34 kW and 176 kW methane pool fires generated on a burner of 0.38 m diameter. For each heat release rate, two other moderately and heavily sooting pool fires were generated by considering higher soot volume fractions while keeping temperature and gaseous species concentrations unaltered. For each radiative model, the corresponding absorption coefficients for carbon dioxide, water vapor, carbon monoxide, and methane were generated from the same high-resolution spectroscopic databases. Model results show that the contribution of carbon monoxide to the radiative intensity can be neglected, whereas that of methane increases with the heat release rate (HRR) and decreases as the soot loading increases. It is also found that the gray approximation for soot holds for the 34 kW pool fires and the weakly and moderately sooting 176 kW pool fires but ceases to be valid for the heavily sooting 176 kW pool fire. Concerning the accuracy of the different approximate radiative models, comparisons with the LBL solutions show that the NBCK model can be used as a reference if LBL solutions are not available. On the other hand, the FDS wide-band model fails in predicting accurately the radiative intensity through the fuel-rich core of pool fires. Finally, the FSCK provide predictions within 10% of LBL solutions with the exception of the heavily sooting 176 kW pool fire where the strong attenuation of radiation by methane invalidates the correlated assumption of the absorption coefficient. In this case, the MSFSK model must be considered, improving substantially the predictions of the FSCK.