SCALAR DISSIPATION AND MIXING IN TURBULENT REACTING FLOWS

The mixing term that appears in probability density function-based formulations of the statistical behavior of turbulent reacting flows was studied with the use of direct numerical simulations of an irreversible, second-order, isothermal chemical reaction between initially segregated reactants in decaying, homogeneous turbulence. Three-dimensional, time-dependent pseudospectral calculations with 64(3) Fourier modes were made as well as were Monte Carlo simulations of the joint concentration pdf equations for coalescence-dispersion (C/D) models. It was found that the development of the scalar microscales depends largely on the initial concentration distributions (both isotropic and nonisotropic distributions were used) and is relatively independent of the reaction rate constant. Joint concentration probability density functions (pdf's) were compared to predictions of C/D models that make use of parameters evaluated from the nonreacting or inert system. Although the models predict proper variance decay when the microscales are specified, the shapes of the pdf's are not accurately predicted, in that the models predict too rapid a growth of probability near the mean value.