An Irregularly Portioned FDF Solver for Turbulent Flow Simulation

A new computational methodology is developed for large eddy simulation (LES) with the filtered density function (FDF) formulation of turbulent reacting flows. This methodology is termed the "irregularly portioned Lagrangian Monte Carlo finite difference" (IPLMCFD). It takes advantage of modern parallel platforms and mitigates the computational cost of LES/FDF significantly. The embedded algorithm addresses the load balancing issue by decomposing the computational domain into a series of irregularly shaped and sized subdomains. The resulting algorithm scales to thousands of processors with an excellent efficiency. Thus it is well suited for LES of reacting flows in large computational domains and under complex chemical kinetics. The efficiency of the IPLMCFD; and the realizability, consistency and the predictive capability of FDF are demonstrated by LES of several turbulent flames