With ever more stringent emissions and performance regulations, more emphasis and efforts have been made in accurate modeling of the combustion process and engine-out emissions in engine development. However, accurate modeling of the combustion process requires detailed chemistry. Highly detailed mechanisms typically include hundreds of species and thousands of reactions, and solution of such reaction set has been one of the largest bottlenecks in numerical modeling of the IC engine with CFD. Typically, the accuracy in chemistry modeling is sacrificed by reducing the mechanism size for the sake of computational efficiency. In this study, a lookup-table based approach is applied for modeling the combustion process in an HCCI engine. Instead of solving chemistry on-the-fly during the CFD simulations, the chemistry is solved for possible combination of thermodynamic and mixing conditions. The turbulence-chemistry interaction is considered using a flamelet approach. Then, the solution is stored in a table, such that chemistry information can be retrieved during the CFD simulation. The lookup-table method, referred to as Flamelet Generated Manifold (FGM), provides significant runtime reduction in CFD simulations with high fidelity chemistry modeling.
The FGM model was applied to a canonical HCCI experiment from Sandia National Laboratory. The experiment examined the effect of different levels of fuel stratification on ignition and combustion of a gasoline HCCI engine. The different levels of stratification were generated by controlling the amount of directly injected fuel. This case has been highly challenging for modeling using traditional modeling approaches. With FGM, it was possible to use the most detailed reaction mechanism to describe the chemistry as completely as one can. The effect of different surrogates on modeling results was investigated as well. It was found that the one proposed by Gauthier showed the most promising results in reproducing the highly complicated combustion with partial fuel stratification.