(341a) Multi-Environment Probability Density Function Method for Modeling Turbulent Combustion Using Detailed Chemistry

The effect of turbulence on chemical reactions is known to be important in many practical combustion systems. There are only a few established models that can capture turbulence-combustion interaction in CFD codes, and all of these models are either very expensive (e.g. transport probability density function model (PDF)) or limited in what types of flames can be analyzed (e.g. assumed-shaped PDF). The multi-environment PDF (MEPDF) method retains many of the desirable properties of the transport PDF method, including the ability to treat the chemical source term exactly and address the nonlinear interaction between turbulence and finite rate chemical reactions with great accuracy.

Originally developed by the chemical engineering community to model simple chemical reaction processes (e.g., mixing tanks), the work described here represents the first to make use of the MEPDF method for combustion applications using realistic combustion chemistry. A CFD modeling tool has been developed which incorporats the MEPDF model into REI's three-dimensional adaptive mesh refinement (AMR) turbulent flow solver. The new modeling tool was benchmarked against other numerical solutions and experimental data for a series of well-characterized bluff-body stabilized turbulent flames. A 19-species augmented reduced mechanism based on quasi-steady state assumption was used in the simulations. The comparisons demonstrated that the MEPDF based CFD tool probvides much improved accuracy with respect to models commonly used in industry such as the assumed-shaped PDF mode, and the preditions of CO and NOx are comparable to more complex methods, such as the transport PDF method, while requiring only a fraction of the computational cost. Finally, predicted flame characteristics and emission levels using this new modeling tool for low NOx burners in a test furnace and a full scale ethylene cracking furnace are discussed and compared with measured and observed data.