Effects of combustion models on soot formation and evolution in turbulent nonpremixed flames

Due to the complex multiscale turbulence-chemistry-soot (TCS) interactions in sooting flames, developing predictive models remains a formidable challenge even with the improved accuracy of the large eddy simulation (LES) approach. LES-based soot models have three main components: a) models for gas-phase chemistry and precursor evolution, b) models for soot particle dynamics, and c) models for subfilter scale TCS interactions. The focus of this work is this latter aspect, for several recent studies have shown that subfilter correlations between the gas-phase and the soot particles are of primary importance in affecting predictability. To this end, a consistent LES/probability density function (PDF) approach with detailed chemistry (denoted as full transport and chemistry (FTC)) and state-of-the-art soot models is used to accurately characterize subfilter TCS interactions. The statistical soot model is based on the Hybrid Method of Moments (HMOM), considering detailed nucleation, condensation, coagulation, surface growth, oxidation, and fragmentation processes. To study the sensitivity of soot predictions to combustion model, tabulated chemistry based on the radiation flamelet/progress variable (RFPV) approach is accounted for in the LES/PDF framework. The Delft-Adelaide natural gas jet flame is simulated to investigate the effects of combustion models on soot predictions. It is found that the location of inception and soot evolution are rather sensitive to the combustion model. Compared to the PDF/RFPV model, the PDF/FTC model improves the simulation results and predicts lower soot volume fraction with soot formation and growth further downstream. Accounting for detailed subfilter TCS interactions through the PDF/FTC model suppresses the contributions of aromatic-based soot growth (nucleation and condensation), while the contribution of acetylene-based surface growth is significantly enhanced and even dominates over condensation at downstream locations. These results imply that the choice of combustion model has a significant impact on the characterization of subfilter TCS interactions due to the strong coupling between turbulence, soot, and chemistry. The tabulated chemistry approach probably cannot capture the high sensitivity of soot formation and growth phases to the history of turbulent gas-phase compositions encountered.