International Congress on Scientific Advances (ICONSAD'21), Balıkesir, Turkey, 22 December 2021 - 25 January 2022, vol.1, pp.801-816
In this study, a complete methodology is presented, from modeling to optimization of a DI Diesel engine operating with multiple pilot injections (MPI) strategy at low load. Optimum predictive combustion model parameters were found using in-cylinder pressure measurement data of only one operating point and validity of these parameters was verified with 12 experiment points which include changes in operating point, EGR fraction, rail pressure, timing and quantity of 1st and 2nd pilot injections. The effects of changing the timing and quantity of pilot injections on performance and emissions were investigated. Relative sensitivity analysis as well as Spearman correlation coefficients analysis was performed to show the effects of each multiple pilot injections parameter on performance and emission results. The 2nd pilot injection quantity is found to be the strongest multiple pilot injections parameter on performance and emission results, while the 2nd pilot injection timing has the least impact. Also, due to very high air-fuel ratio, no significant correlation or adverse effect was observed in soot formation. Non-dominated Sorting Genetic Algorithm–III, a very suitable search algorithm for multi-objective optimization study including four factors was performed in order to minimize NOx emissions while maintaining brake power. Using the genetic algorithm with crossover rate and mutation rate to search for optimum multiple pilot injections parameters, a 25% reduction in NOx emissions and a 0.7% increase in brake power were achieved.