Large-eddy simulation of lab-scale ethylene buoyant diffusion flames: Effects of subgrid turbulence/soot production interaction and radiation models

International audience; Large eddy simulation of the 15-kW ethylene buoyant diffusion flames burning under O2/N2 oxidizers with oxygen concentrations, XO2 , of 0.209, 0.168 and 0.152, investigated experimentally at FM Global, are performed by using a flamelet/presumed filtered density function (FDF) combustion model, an acetylene/benzene-based soot production model and the rank-correlated full-spectrum k (RCFSK) with non-grey soot as radiative property model. The presumed FDF considers the spatial intermittent nature of soot and the coupling between soot quantities and mixture fraction. Subgrid-scale (SGS) absorption turbulence/radiation interaction (TRI) is neglected whereas the filtered emission term is modelled from the presumed FDF approach. Model predictions reproduce with fidelity the flame structure, the soot production dynamics and the radiative loss for the three flames. Model results show that soot and radiating gas have a similar contribution to radiation for the flame burning in air and the contribution of radiating gas becomes increasingly dominant with reducing XO2 due to a decrease in soot volume fraction (SVF) throughout the overall flames. A sensitivity study is performed on the influence of the modelling approach of SGS turbulence/soot production interaction (TSI) and radiative heat transfer. Model predictions show that neglecting the coupling between soot quantities and mixture fraction overestimates substantially soot oxidation, resulting in significantly lower soot production and radiative loss. For the present LES where more than 80% of temperature variance is resolved, disregarding the SGS emission TRI has also a significant impact with an overprediction of mean and rms SVF by a factor of two. On the other hand, the approximation of grey soot is valid for these lab-scale flames characterized by moderate optical thicknesses and the use of a non-grey weighted-sum-of-grey-gases (WSGG) model coupled to the grey soot approximation represents an alternative to the full-spectrum k-distribution models.