A Sensitivity Analysis for Sparse-Lagrangian MMC in Simulations of a n-dodecane Reacting Jet

This paper presents a detailed sensitivity analysis of the sparse-Lagrangian multiple mapping conditioning (MMC) model to different parameters in simulations of n-dodecane flame A which is adopted by the engine combustion network (ECN). The model is fully coupled with a large eddy simulation (LES) approach. A gas-jet model is used for the fuel injection. The MMC-LES model is first examined for a non-reacting case and the sensitivity of the results to variations in the inlet turbulence intensity are examined. It is found that the mixture fraction profiles agree well with the experimental data. The vapour penetration is overpredicted but there is significant improvement by increasing the turbulence intensity of the inlet jet from 10% to 15%. The model sensitivities to inlet turbulence intensity, mixing model parameters and chemical kinetics is then investigated for reacting cases. Simulations are performed at various levels of ambient oxygen (13% - 21%). The turbulence intensity effects on the computed ignition delay times are insignificant while it is found that lower inlet turbulence improves the lift-off lengths. The MMC-LES relies on a mixing localness parameter, f_m, with lower values enforcing greater localness in mixture fraction space. Three different values are tested here. A lower f_m provides a better prediction of the ignition delay times while the effects on lift-off lengths are not significant. Both 106-species and 54-species kinetics mechanisms are used. The results reveal a strong influence of the chemistry model with the 54-species scheme leading to significant improvements in both the ignition delay times and lift-off lengths.