Advances In the Simulation of Detailed Chemistry In Reacting Flows

It is very difficult to simulate reacting flows in all but the simplest systems, since real-world chemistry is almost always complicated (many species and reactions) and the corresponding differential equations are stiff (due to short-lived reactive intermediates). Most commonly used approaches reduce the computational requirements by simplifying the chemistry without controlling the error. However, it is important to develop a better understanding of the tradeoff between error and CPU time.  We have tackled this problem with several different approaches, including chemistry model reduction (adaptive chemistry), species elimination (from both the chemistry and the transport), elaboration of the mathematical structure of common model reduction techniques, and the associated errors, and development of new numerical methods for performing the simulations (including porting CHEMKIN and the stiff solvers on to a GPU). Here we give an overview of our recent work, comment on the relative merits of the different approaches, and identify which are most promising.