A regularized deconvolution model for sub-grid dispersion in large eddy simulation of turbulent spray flames

Abstract

In this study, a sub-grid scale (SGS) dispersion model is developed for simulating turbulent spray flames. This model employs a regularized deconvolution method (RDM) to determine the sub-grid fluctuations from the filtered velocity field. The model is examined in simulations of a turbulent counterflow n-dodecane spray flame, and benchmarked against a simplified Langevin model (SLM) and simulations without dispersion closure (NOM). To identify conditions that characterize the importance of the spray-flame coupling, a regime diagram is constructed from physical arguments and theoretical analysis by considering idealized conditions. Based on the proposed criteria, a range of operating conditions is considered, involving different Stokes- and Reynolds numbers as well as inert and reacting conditions. A priori DNS analyses are performed to compare the probability density functions (PDFs) of slip velocities. Significant discrepancies between models are observed for smaller Stokes numbers. Subsequently, large eddy simulations (LES) are performed and effects of mesh resolution on statistical flow-field quantities are investigated. Qualitative and quantitative differences are observed for coarse-grid simulations with larger droplets, where a double-flame structure is predicted by DNS and LES-RDM, whereas a single flame is predicted by LES-NOM and LES-SLM. For the finer grid, all three LES models predict the double-flame structure. Turbulence modulation by the spray is expected to play a role in creating different flame topologies, and the RDM-based SGS-closure model is shown to provide improved predictions for the dispersed and carrier phases for a wider range of grid resolution.