Simulation of a turbulent non-premixed flame using conditional source-term estimation with trajectory generated low-dimensional manifold

Conditional source-term estimation (CSE) is a method to close the mean chemical reaction source-term in an averaged transport equation. It is used with a trajectory generated low-dimensional manifold (TGLDM) to simulate a turbulent non-premixed flame. Integral equations are inverted for two progress variables, 〈YCO2|ξ〉〈YCO2|ξ〉 and 〈YH2O|ξ〉〈YH2O|ξ〉, by assuming spatial homogeneity in the conditional averages. Using these two progress variables, the conditional source terms of temperature and other scalars are interpolated from the TGLDM table and mapped back into the physical space to be substituted into the transport equations. Solving a transport equation using a source-term interpolated from the TGLDM is found to improve the prediction of NO over simply interpolating the mass fraction of NO directly from the TGLDM. This method has been applied in a large eddy simulation (LES) of a turbulent non-premixed flame. Both GRI-Mech 3.0 and GRI-Mech 2.11 are found to be able to predict the temperature and major species well. However, only GRI-Mech 2.11 gives an acceptable prediction of NO. It is found that major species can be interpolated from the TGLDM table which can significantly reduce the computational cost.