The present study extends the recently developed KIVA-GA computer code to incorporate a generalized piston bowl geometry parameterization and multi-mode optimization. The new code was used to optimize the combustion chamber geometry of a small-bore automotive Diesel engine. The Genetic Algorithm (GA) merit function, which was calculated with a modified version of the KIVA-3V engine simulation code, included NOx, unburned HC, soot, and fuel consumption.
A novel parameterization was included in KIVA-GA that allows for a variable number of parameters to define the bowl shape. The in-house G-Smooth grid generation package was used to create the KIVA grids with a specified compression ratio and mesh resolution.
The improved KIVA-GA methodology was used to optimize engine emissions and performance simultaneously for two operating conditions. Six combustion chamber geometry parameters were included, along with start of injection (SOI) timing for each operating condition, injection pressure for each operating condition, swirl ratio (SR), and spray-included half angle. The predicted results show that improvements can be made in emissions without necessarily relying on emissions control strategies such as multiple injections and EGR.