Analysis of local entropy generation is an effective means to investigate sources of efficiency loss in turbulent combustion from the standpoint of the second law of thermodynamics. A methodology, termed the entropy filtered density function (En-FDF), is developed for large eddy simulation (LES) of turbulent reacting flows to include the transport of entropy, which embodies the complete statistical information about entropy variations within the subgrid scale. The modeled En-FDF contains a stochastic differential equation (SDE) for entropy which is solved by a Lagrangian Monte Carlo method. In this study, a numerical study has been done on effectiveness of SDE to model entropy variation using a partially stirred reactor (PaSR). This provides a computationally affordable case to compare different effects of entropy generation source terms and fine tune mixing coefficients. In this equation, turbulent mixing is modeled with Interaction by Exchange with the Mean (IEM). Combustion source terms are provided by direct integration of a GRI3.0 mechanism for methane/air system. Evolution of entropy was calculated from stochastic model and then compared with the one obtained directly by integrating the chemical mechanism. It was shown that results of both calculations have very good agreement versus different mixture fractions.
Entropy filtered density function for large eddy simulation of turbulent
reacting flows
