On the use of numerical model reduction techniques in chemically reacting flows

Simulation of detailed multicomponent reactive flow processes becomes quite expensive and, despite the advent and advancement of digital computers, still engulfs an inconceivably large amount of time. Chemistry-based model reduction aims at breaking down the dimensionality of an underlying combustion model such that its simulation ideally takes considerably less time. In this thesis we examine a few practical model reduction approaches which bear the potential to be used in the context of chemically reacting flows. This examination serves also as a fairly broad overview of numerical methods needed for efficient model reduction approaches. Furthermore, we propose a novel, heuristic model reduction procedure that hinges less on iterative features and thus is less affected by convergence issues. For the nonlinear scenarios considered in this thesis, the running times obtained with the new method in conjunction with the reduced problems are quite promising when compared to the simulation costs of the respective full models. The qualitative studies carried out here shed light on the direction in which future model reductions for chemically reacting flows may proceed. We expect the novel method to become a viable alternative in the field of reduced combustion simulation