Efficient Multidimensional Simulation of HCCI and DI Engine Combustion with Detailed Chemistry

This paper presents three approaches that can be used for efficient multidimensional simulations of HCCI and DI engine combustion. The first approach uses a newly developed Adaptive Multi-grid Chemistry (AMC) model. The AMC model allows a fine mesh to be used to provide adequate resolution for the spray simulation, while dramatically reducing the number of cells that need to be computed by the chemistry solver. The model has been implemented into the KIVA3v2-CHEMKIN code and it was found that computer time was reduced by a factor of ten for HCCI cases and a factor of three to four for DI cases without losing prediction accuracy. The simulation results were compared with experimental data obtained from a Honda engine operated with n-heptane under HCCI conditions for which directly measured in-cylinder temperature and H₂O mole fraction data are available. The second approach to improve efficiency uses a recently developed a set of spray models which reduce numerical grid size dependencies; thus enabling the simulation of DI combustion on relatively coarse meshes to save computing time. The new spray models, including a gasjet model for the near nozzle droplet-gas momentum exchange calculations and advanced collision models, have also been implemented into the KIVA code. Reductions in computing time by a factor of ten without a significant loss in accuracy are realized for DI engine simulations through the combined use of the AMC model and the mesh-independent spray models on coarse computational meshes. Finally, the third approach combines the AMC model with the mesh-independent spray models and parallelizes the chemistry solver based on the computing load so that simulations can be performed on relatively coarse meshes on multiple processors (up to four). Computational times were reduced by a factor of more than twenty for DI engine when parallelizing with four processors.