A Fully Consistent Hybrid Les/Rans Conditional Transported Pdf Method for Non-premixed Reacting Flows

Abstract

A novel hybrid modeling approach that combines a finite-volume large eddy simulation (LES) with a Reynold-Averaged Navier-Stokes (RANS) conditional probability density function (PDF) method for turbulent non-premixed reacting flows is presented. The hybrid method incorporates the turbulent flow modeling of the LES with the chemistry–turbulence interaction modeling of the Lagrangian RANS-PDF. Specifically, the conditional composition RANS-PDF method is applied as the turbulence-chemistry closure for the LES. For a statistically steady-state case, the evolution of stochastic particles in a RANS framework reduces the computational effort, compared to the classical LES filtered density function (FDF), by performing the expensive chemical ODE integration at a lower spatial resolution. This approach is justified by the slow spatial variations of the reacting variables conditioned on the mixture fraction. Here, a consistent two-way coupling between the LES flow-field and the conditional composition RANS-PDF is formulated. The accuracy and the consistency of the method are established studying the Sandia flame series with detailed chemistry. Good predictions of the mean flow-fields and mean and conditional reactive scalars have been obtained, compared with the experimental data, with an overall decrease of the computational effort up to two orders of magnitude.