Explicit Filtering LES for turbulent premixed flames

Abstract

Large eddy simulations of a premixed, turbulent, methane-air round jet flame at an equivalence ratio of 0.8 and unburnt gas temperature of 800 K are performed using Explicit Filtering LES (EFLES) method. The nominal inflow Reynolds number is 1500 and flame falls in thin reaction zone regime of turbulent premixed combustion. EFLES is formally derived from the approximate deconvolution method for LES computations and has been successfully applied to non-reacting flow computations in the past. This work is the first application of the method to a fully compressible, multi-component reacting flow problem. A 13-species reduced chemical mechanism with species transport is used to solve for the mass fraction fields. A flame sensor is used to automatically detect steep species gradient within the flame and then adaptively reduce the filter order based on local resolution of flame scales. EFLES results are compared with a fully resolved DNS obtained with the same flow solver on a much finer grid without any turbulence model. LES predicts a slightly shorter mean flame height and thicker flame brush when compared to the DNS due to the lack of fine scale turbulence and thicker reaction zones in the former. EFLES solutions for time averaged statistics of velocity, temperature and major species mass fractions show good qualitative and quantitative agreement with DNS results. Slight quantitative differences are observed in other transported species fields on going downstream the dump plane, but the qualitative match with DNS is reasonable. Profiles of mean species mass fractions conditioned on progress variable from the LES are seen to closely agree with those from the DNS, thereby, showing that the differences observed in physical space are purely due to a broader flame brush being captured in the LES. The conditional net heat release rate and species reaction rates also agree quantitatively with the DNS. Therefore, the results obtained from this work are encouraging and show that EFLES is a viable method for further study in the context of turbulent reacting flow LES.