CFD numerical investigation of piston shape and fuel injection strategies for enhanced ammonia/diesel dual fuel combustion

The urgent need to decarbonize the transport sector has accelerated the development of carbon-free fuels and innovative combustion strategies. Among these alternatives, ammonia (NH 3 ) emerges as a promising vector thanks to its zero-carbon nature and excellent storage properties. However, its low reactivity and high auto-ignition temperature pose significant challenges for the engine performance and control its exhaust emissions. This study focuses on the improvement of the performance and exhaust emissions of NH 3 /diesel fuel combustion in dual fuel engine. In first step, a 3D-CFD model is carried out to investigate the NH 3 stratified injection effect, compared to its induction into the engine with the intake air (experimental baseline case). In second step, the adopted configuration is combined with the piston bowl geometry effect for further enhancement in NH 3 /diesel fuel combustion. To reduce 3D-CFD computational cost, an artificial neural network (ANN) metamodel has been developed as substitution tool. A multi-objective optimization is then performed using NSGA-II algorithm for maximizing combustion efficiency while minimizing NO x as well as unburned NH 3 emissions. The results show that split stratified NH 3 injection, in conjunction with a redesigned piston bowl, significantly improves flame propagation and in-cylinder flow mixing. The best configuration achieved 11% increase in combustion efficiency, 64% reduction in unburned NH3, and 40% decrease in NO x emissions, compared to the baseline case. In absolute terms, the optimized configuration achieves a combustion efficiency of approximately 89%, an unburned NH 3 emission level of 1.8E-06 kg, and a NO x emission level of 2.2E-07 kg. This integrated approach presents a robust solution for enabling high NH 3 substitution ratios in compression-ignition engines while addressing the key challenges of combustion stability and pollutant formation.