Optimization of a high-speed direct-injection diesel engine at low-load operation using computational fluid dynamics with detailed chemistry and a multi-objective genetic algorithm


Abstract

A passenger car high-speed direct-injection diesel engine operating at low-load conditions in the modulated kinetic combustion mode was optimized using a multidimensional computational fluid dynamics code and a multi-objective genetic algorithm. Spray targeting, piston bowl geometry, and swirl ratio were optimized. Since the combustion is mainly kinetics controlled, detailed chemistry was considered through a recently developed adaptive multi-grid chemistry (AMC) model. The numerical results from the AMC model, including the pressure and pollutant emissions, were first validated on the baseline engine for a parametric sweep by comparison with the results from the standard KIVA-CHEMKIN model. The AMC model was found to give consistent results with the KIVA-CHEMKIN model, with a computational cost that is less than half that of the KIVA-CHEMKIN model. Optimal designs from the optimization were also validated using the full KIVA-CHEMKIN model and were found to reduce the fuel consumption and/or pollutant emissions. Start-of-injection timing was found to be the primary parameter influencing the fuel consumption and soot emissions for the engine operating in the low-load condition. Later injection benefits the fuel consumption and soot reduction. However, further retardation of the injection timing leads to reduced combustion efficiency and even misfire, and results in higher unburned hydrocarbon emissions. Different piston bowl shapes have different responses to bulk flow motions and the resulting geometry-generated turbulence will affect soot formation and oxidation.