Large Eddy Simulation of a Swirl-Stabilized Flame

Abstract

Swirl is used in a wide range of combustions systems such as engines, furnaces, gasifiers, and boilers, to enhance mixing, stabilize flames, and reduce pollutant emissions. Numerical modeling of swirling flows remains a challenging task, since there may exist complex recirculating flow patterns and flow instabilities associated with vortex breakdown, precessing vortex core, and jet precession. In swirling flames, the situation becomes more complex because the unsteady heat release can add other modes of instability. The origins and nature of these instabilities are still not well understood despite many experimental and numerical studies have been conducted in the area. The Sydney swirl burner flame series provide an excellent platform for validating numerical methods for turbulence-chemistry interactions and have been target flames for the TNF workshop series. The burner has well-defined boundary conditions and comprehensive experimental data sets have been documented for different fuel compositions and flow conditions. Compared with the piloted and bluff-body stabilized flames, swirl-stabilized flames pose an additional challenge to numerical modeling because of the complex flow patterns and inherent flow instabilities. In this study, a large eddy simulation (LES)-based multi-environment turbulent combustion model is used to model the Sydney swirl burner flame SMH1. The multi-environment filtered density function model (MEFDF) depicts the filtered density function (FDF) as a weighted summation of a small number of multi-dimensional Dirac delta functions in composition space. It is derived from the transport FDF equation using the direct quadrature method of moments (DQMOM). The MEFDF method with multiple reactive scalars retains the unique property of the joint FDF model of treating the chemical source term exactly. A 19-species mechanism reduced from GRI-Mech 2.11 is employed for chemical kinetics. The in situ adaptive tabulation algorithm (ISAT) is used to speed-up the evaluation of the chemical source term. The predicted radial profiles of the axial velocity, azimuthal velocity, mixture fraction, temperature, and species mass fractions of CO2, CO, and NO are in reasonable agreement with the experimental data. It has been found that, compared with the experimental data, the profiles of the temperature and species mass fractions shifted slightly outward in the radial direction at downstream locations and NO mass fraction is slightly over-predicted at most locations. Further work will be needed to find out possible reasons for these discrepancies.