Chemistry accelaration - Reacting Flow Lab

The goal of this research is the development of Lagrangian stochastic models of acceleration in homogeneous turbulence (with and without mean deformations) that agree with the available data from DNS and experiments. We have identified a class of second-order Markovian stochastic models (conditional on a log-normal representation for the Lagrangian dissipation) that are exactly consistent with Gaussian (one-point, one-time) velocity statistics and conditionally Gaussian acceleration statistics (the Reynolds (2003) model belongs to this class). Preliminary calculations of two-time cross-correlations (below) using the Reynolds (2003) model and a cubic acceleration model (an easily identifiable nonlinear model in the class) show that the former has an unphysical cusp at the origin (as opposed to smooth behavior generated by the latter). DNS data are going to be used as a discriminator among different stochastic models in the class.