Modeling and performance optimization of a direct injection spark ignition engine for the avoidance of knocking

The paper applies simulation techniques for the prediction and optimization of the thermo-fluid-dynamic phenomena characterising the energy conversion process in an internal combustion engine. It presents the development and validation of a 3D CFD model for a GDI optically accessible engine operating either under stoichiometric homogeneous charges or under overall lean mixtures. The model validation is realized on the ground of experimental measurements of the in-cylinder pressure cycle and of the available optical images. The model comprehends properly developed sub-models for the spray dynamics and the spray-wall interaction. This last is particularly important due to the nature of the mixture formation mode, being of the wall-guided type. In the stoichiometric mixture case, the possible occurrence of knocking is also considered by means of a sub-model able to reproduce the pre-flame chemical activity. The CFD tool is finally included in a properly formulated optimization problem aimed at minimizing the engine specific fuel consumption with the avoidance of knocking. The optimization, performed through a non-evolutionary algorithm, allows determining the best engine control parameters (spark advance and start of injection).