Assessment of a Differential Subgrid Stress Model for Large-Eddy Simulations of Turbulent Unconfined Swirling Flames

Swirling flames are widely used in engineering to intensify mixing and stabilize combustion in gas turbine power plants and industrial burners. Swirling induces new instability modes, leading to intensification of coherent structures, asymmetric geometry, vortex core precession, and flame oscillations. Large-Eddy Simulation (LES) has the capability to furnish more accurate and reliable results than the simulations based on Reynolds-averaged Navier-Stokes equations (RANS). Subgrid-scale models in LES need to describe the backscatter (local transfer of kinetic energy from small scales to larger scales) that is intensified in swirling flames. In this paper, the Differential Subgrid Stress Model (DSM), previously developed by the authors, is assessed using an experimental database from Sydney University on swirl-stabilized turbulent unconfined non-premixed methane-air flame. Regime without vortex precession is simulated numerically using the DSM and Smagorinsky subgrid-scale model. Experimental measurements of mean velocity, profiles of mass fractions, and temperature are used for comparison with the simulation data. The standard Smagorinsky model is considered the basic approach. Differences in the flow field statistics obtained in both subgrid-scale LES models are analyzed and discussed. The importance of taking the backscatter into account is highlighted.