Automatic chemistry reduction strategies for transportation fuel combustion modeling.

Perrine Pepiot (MAE, Cornell)

Computational Fluid Dynamics (CFD) has emerged as a powerful tool that can help design and optimize energy conversion systems. However, improving the predictability of numerical models requires the accurate description of the underlying chemical processes. The comprehensive chemical description of hydrocarbon fuel combustion kinetics now typically involves hundreds or thousands of individual chemical species, which are created or destroyed through an intricate network of chemical reactions. This translates into extremely large systems of coupled stiff ODEs (kinetics only) or PDEs (when coupled with a flow), making it a major challenge to use of this chemical knowledge in turbulent combustion models because those usually become intractable if more than tens of species are considered. In this context, the first part of this talk describes the design and implementation of automatic multi-stage reduction strategies to generate compact kinetic models for transportation fuel combustion, focusing on graph-based techniques. A second section will illustrate how these reduction algorithms can be combined with an adaptive strategy in the context of Probability Density Function methods to further reduce the computational burden associated with the chemical description in large-scale turbulent flame simulations.