Optimisation of injection-related parameters by using the NLPQL algorithm for a marine medium-speed diesel engine

The match of an injector with the combustion chamber was studied through computational fluid dynamics simulation under four different engine loads. Four design parameters including the start of injection, spray angle, injector protrusion length and swirl ratio were examined. The Latin hypercube together with a non-linear programming by quadratic Lagrangian algorithm were used in the optimisation. Comparisons were made in the engine loads in terms of the optimisation history, objectives, sub-objectives and design parameters. The commonalities of the design parameters of the optimums were summarised. Additionally, a detailed combustion process comparison was conducted on the same engine loads (100% and 25% engine loads) between the optimum and baseline designs, respectively. Finally, the effects of the design parameters on the objective were investigated by the response surface methodology. The results indicate that the non-linear programming by quadratic Lagrangian method is an effective algorithm to spot the optimums with the best trade-off between nitrogen oxides and soot emissions. The optimisation process presents better qualities at 100% and 75% engine loads than in the case of 50% and 25% engine loads. The design parameters of the optimum under each engine load have something in common, namely, that they all prefer late injection, low swirl, large injection angle and slightly smaller nozzle protrusion length. Besides, start of injection and swirl ratio have larger influence on the objective as opposed to the nozzle protrusion length and spray angle. The large start of injection together with the small swirl ratio can reduce the objective significantly and vice versa. A large nozzle protrusion length with a small spray angle contributes to the reduction of the objective and so does the combination of a small nozzle protrusion length with a large spray angle.