An optimization study combining multidimensional CFD modeling and a global, evolutionary search technique known as the Genetic Algorithm has been carried out. The subject of this study was a 2-stroke, spark-ignited, direct-injection, single-cylinder research engine (SCRE). The goal of the study was to optimize the part load operating parameters of the engine in order to achieve the lowest possible emissions, improved fuel economy, and reduced wall heat transfer. Parameters subject to permutation in this study were the start-of-injection (SOI) timing, injection duration, spark timing, fuel injection angle, dwell between injections, and the percentage of fuel mass in the first injection pulse.
The study was comprised of three cases. All simulations were for a part load, intermediate-speed condition representing a transition operating regime between stratified charge and homogeneous charge operation. Case A involved permutation of all 6 parameters studied, while Case B involved 5 parameters that were subject to change, with the spray cone angle held constant at its experimental value. Case C focused on a single-injection strategy, unlike Cases A and B that allowed for split-fuel injections.
Four candidate optimum designs emerged from this optimization study, each offering distinct advantages and benefits over the baseline operating case. These benefits included reduced emissions of nitrogen oxides and unburned hydrocarbons, and improved fuel efficiency.
Injection angle was found to have an insignificant effect on engine performance at this operating condition. Some candidate optimum designs were obtainable with both single and split-fuel-injection strategies, while others were unique to the latter. Split-fuel-injection was found to be a versatile, and useful technique for enhancement of engine performance. Soot production was not taken into account in the present study, and could have brought about a different optimization direction, should it have been considered.