A detailed multidimensional simulation for the cold start process of a gasoline direct injection engine using a fractal engine simulation model

Abstract

The cold start process for a gasoline direct injection (GDI) engine was studied through multidimensional simulations using a Fractal Engine Simulation (FES) model integrated into CONVERGE CFD. The simulations were focused on the very first firing cycle in the cold start process, as most of the hydrocarbon emissions derive from this cycle. The dramatically changing engine speed and low wall temperatures in the cylinder present challenges for simulation. It turned out that the FES model was able to predict the in-cylinder pressure traces for all four cylinders for the first firing cycle, and gave good agreement with the experimental measurements under these extremely transient conditions. More comprehensive analysis demonstrated the causes for the different behavior of the combustion in different cylinders: more injected fuel and a higher induced tumble ratio in cylinder 3, the first to fire, led to a higher average equivalence ratio and better fuel distribution, which resulted in the highest peak pressure; the worse fuel distribution from the lower tumble ratio, together with the lower normalized turbulence, caused weaker combustion in cylinder 4; the peak pressures were similar and on the low side for cylinders 2 and 1, mainly determined by the low average equivalence ratio. An improved strategy with multiple injections was proposed, with the first two in the intake stroke and two later injections in the compression stroke rather than one injection during each stroke. Although the total injected fuel was the same as the baseline strategy, more fuel evaporated due to the enhanced fuel evaporation from both droplets and wall films, resulting in a higher average equivalence ratio in the cylinder. The peak cylinder pressure and IMEP rose by 33.3% and 8.4%, respectively, using the 4-injection strategy compared to the baseline 2-injection strategy, and the 4-injection strategy was verified by the experiments, as well.