A computational and experimental study of spark ignition engine combustion

This work focuses on aspects of combustion in a spark ignition engine. A pent-roof research engine was used to generate an experimental data set which was combined with a preexisting data set from a disc-chamber research engine. The combined dataset was used to refine a thermodynamic spark ignition engine combustion code which could operate in either a three-zone entrainment and bum up, or a two-zone direct combustion, configuration. The pent-roof engine was skip-fired to ensure residual gases were purged and care was taken to ensure that the thermodynamic state and chemical composition of the intake mixture were well defined. The combustion chamber, which featured near complete optical access, was illuminated using a sheet of laser light. Mie scattered laser light from fine seed particles was recorded allowing the position of the flame front to be tracked. These images were then ensemble averaged and combined to give a three-dimensional reconstruction of the mean combustion progress variable field for three different engine speeds. As the flame approached the combustion chamber walls it was found to decelerate. A relationship between the burning velocity of an unconstrained flame and a flame approaching a wall was derived which agreed well with the experimental results. This relationship was incorporated into the engine simulation code and found to improve greatly the predictions of the two-zone combustion model. New flame acceleration and laminar burning velocity submodels reported in the literature were added to the engine simulation code. The suitability of these models for simulating spark ignition engine combustion was evaluated using the disc and pent roof experimental data sets and model constants adjusted to optimise the performance of each submodel. Although there were substantial differences between the individual submodels, over the range of operating conditions for which experimental data was available changes to the submodel used had a negligible effect on the combustion model predictions. The work concludes with an evaluation of the performance of the three-zone combustion model for simulating a pent-roof engine. The model was modified for the pent-roof engine to include suitable assumptions for turbulence and the position of the centre of the flame. Using constants which were chosen to fit data recorded in the disc chamber engine, the model predictions for the pentroof engine were comparable in accuracy to predictions for the disc roof engine. The model was incorporated in a commercially available manifold gas dynamics simulation software to allow the predictive simulation of complete engine cycles.