Dns evaluation of chemistry models for turbulent, reacting, and compressible mixing layers

In this paper, we use the direct numerical simulation (DNS) approach to carry out fundamental studies that compare the performance of an eight-step/seven-species chemistry model and a model consisting of twenty-five steps and twelve species. The basic H₂-Air reaction is used, because of the relevance to combustion in turbomachinery and the scramjet engines, and results are presented for M_c=0.8, where M_c is the convective Mach number. Various definitions of the layer growth rate are used in an attempt to differentiate compressibility effects of high-speed from those of chemical reaction. The standard vorticity thickness definition, which is biased toward the hydrodynamic part of the flow, shows a similar evolution for both chemistry models and the non-reacting case. However, three other definitions, which depend on the extent of chemical reaction, show considerably different temporal evolution for the two chemistry models. In general, the twenty-five step model shows more vigorous reaction. Also observed for the 25-step system is the extraction of energy from the system, which is manifested as lower-than-upstream temperatures away from the centerline of the layer. We attribute this to the endothermic nature of some of the reaction steps. Examination of the anisotropy tensor shows no effect of chemistry or its modeling. However, no particular attention was paid to the integral length scale, i.e., the normalized initial correlation between scalar values in adjacent nodal points of the grid.