Temperature and mole fraction of nitrous oxide from simulation of a methane diffusion flame using comprehensive chemistry. Predicting the formation of pollutants in flames requires detailed modeling of both the carbon chemistry in the flame and nitrogen chemistry that leads to pollutants. These simulations, performed using a parallel adaptive mesh refinement algorithm for low Mach number combustion, use a comprehensive chemical mechanism that models the behavior of 65 species and 447 reactions. The computed results show excellent agreement with experimental data and provide additional insight into the mechanisms of pollutant formation.

Research Objectives
The long-term goal of this project is to develop high-resolution adaptive methods for partial differential equations and to implement the resulting methodology into production-quality software tools that are broadly applicable to a number of DOE Office of Science research programs. The current focus applications are magnetic fusion, accelerator design, and turbulent reacting flow. Of these applications, computational reacting flow is the most mature, with a goal of developing a high-fidelity numerical simulation capability for turbulent combustion processes such as those arising in furnaces and engines.

Computational Approach
The principal computational tools for this project will be based on the adaptive mesh refinement algorithm developed by CCSE and ANAG. The methodology is based on a block-structured refinement approach that allows computational effort to be focused in regimes of the flow where it is required. For reacting flow simulations, we use a version of this methodology developed for low Mach number combustion. The algorithm uses a fractional step discretization that easily facilitates the inclusion of complex kinetics mechanisms as well as differential diffusion and radiative heat transfer. The key issue in modeling turbulent reactions is the interplay between kinetics and the small-scale turbulent eddies in the flow.

Accomplishments
During combustion of coal and biomass fuels, fuel-bound nitrogen compounds are volatized and released with combustible gas, potentially leading to enhanced NOx formation. Laboratory experiments at the Technical University of Denmark are being used to study this phenomenon. In the experiment, ammonia is added to an inflow methane stream to model the effects of fuel-bound nitrogen. To simulate the experiment computationally, we incorporate a set of detailed methane mechanisms describing the nitrogen chemistry, and NO formation in particular. These comprehensive mechanisms contain as many as 65 chemical species and 447 fundamental reactions. We were able to show that by modeling the detailed kinetics in an adaptively refined diffusion flame, we could accurately predict the NO produced by the flame as a function of inlet ammonia.

We have begun to study turbulence-flame interaction in three dimensions to determine the effect of the turbulence on the average flame speed and to assess how turbulence modulates flame chemistry. We precomputed a field of isotropic decaying turbulence in the reactant stream. This field was then allowed to convect into the initially steady premixed hydrogen flame. The complete hydrogen mechanism was simulated (9 species, 27 reactions). The adaptive algorithm is set up to track the flame front and regions of strong vorticity, locally refining the base 32 x 32 x 64 grid by up to a factor of 4. We are currently beginning the analysis of these results.

Significance
The modeling of turbulent fluid flow, even in the non-reacting case, remains one of the great scientific challenges. For realistic combustion scenarios, the picture becomes more complex, because small-scale turbulent fluctuations modify the physical processes such as kinetics and multiphase behavior. Developing techniques that accurately reflect the role of small-scale fluctuations on the overall macroscopic dynamics would represent a major scientific breakthrough.

Publications
J. B. Bell, M. S. Day, A. S. Almgren, M. J. Lijewski, and C. A. Rendleman, "A parallel adaptive projection method for low Mach number flows," in Proceedings of the ICFD Conference on Numerical Methods for Fluid Dynamics, University of Oxford, March 26-29, 2001.

J. B. Bell, M. S. Day, A. S. Almgren, M. J. Lijewski, and C. A. Rendleman, "Adaptive numerical simulation of turbulent premixed combustion," in Proceedings of the First MIT Conference on Computational Fluid and Solid Mechanics, June 11-15, 2001.

P. McCorquodale, P. Colella, and H. Johansen, "A Cartesian grid embedded boundary method for the heat equation on irregular domains," Lawrence Berkeley National Laboratory Report LBNL-47459 (2001).

http://www.seesar.lbl.gov/AMR/index.html