A Hybrid-LES-RANS Transported PDF Combustion Model for Non-Premixed Flames

Abstract

The intention of this work is to combine Large-Eddy-Simulation (LES) for the prediction of flow and mixture fraction fields with a Reynolds-Averaged-Navier-Stokes (RANS) transported probability density function (TPDF) method for the prediction of turbulent non-premixed flames. The motivation for this work is based upon the property of LES to provide a better description of complex flow fields than most current RANS methods can offer, while TPDF-methods excel in predicting the reacting species fields. However, using the straight forward extension of PDF methods for LES, the filtered density function (FDF) approach requires a large number of PDF particles in each LES cell and is thus computationally expensive. Therefore, a method is proposed to use the time-averaged LES flow field, mixture fraction field and mixture fraction PDF as a turbulence model for a RANS TPDF method operating on a much coarser grid. A projection of the mixture fraction conditioned PDF to evaluate the instantaneous LES density field is proposed as coupling device. The reconstruction of mixture fraction PDF from a LES simulation and the coupling to the TPDF method in postprocessing mode is validated using the TNF Sandia D flame, showing good agreement with experiment.