Filtered density functions from direct numerical simulation of a reactive jet in cross-flow

Abstract

Direct numerical simulation (DNS) with multi-step hydrogen-air chemical kinetics is used to investigate the nearfield of a flame stabilized above a reactive jet in cross-flow (JICF). JICF configurations are typically used where rapid mixing is desirable; classical applications are fuel injection nozzles and dilution holes in gas turbine combustors. Due to the computational cost of DNS, approximate solution methods such as large-eddy simulation (LES) are essential to parametrically study the effect of changing fuel jet configurations on the far field, but these methods require submodels capable of accurately capturing the near-field flame stabilization for success. By incorporating a wealth of turbulencechemistry interactions (between the flame and vorticity generated by the jet shear layer instability as well as product recirculation by a large counter-rotating vortex pair), this DNS is exceptionally well suited to exploration of unclosed terms in LES formulations such as the chemical source-term. One quantity of direct relevance to several models for stratified combustion, such as the Bray-Moss (BM) model and doubly-conditional source-term estimation (DCSE), is the filtered density function for the mixture fraction ? and partially premixed progress variable c. Empirical extraction of the filtered density functions of progress variable and mixture fraction at two representative locations demonstrates the complexity of approximating these two functions from a one- or two-parameter functional form.