Co-optimization and prediction of high-efficiency combustion and zero-carbon emission at part load in the hydrogen direct injection engine based on VVT, split injection and ANN

Abstract

Hydrogen fuel holds great promise for accelerating the energy sector's decarbonization efforts. However, hydrogen's unique properties pose challenges for hydrogen direct injection (HDI) engines, such as differences in airflow exchange capabilities for very low density and reduced combustion rates due to lean burn conditions. This study investigates the impact of combined variable valve timing (VVT) and split injection strategies on HDI engine combustion, efficiency, and emissions. Using Artificial Neural Network (ANN) models and experimental data, predictive models for Brake Thermal Efficiency (BTE) and Nitrogen Oxide (NOx) emissions were developed. Adjusting intake and exhaust valve timings improved BTE by up to 2.5 % and 1.8 %, respectively, while reducing NOx emissions by 48.9 % and 31.1 % under specific conditions. Optimizing split injection alongside VVT further increased BTE by 1.2 % and 0.9 %, but increased NOx emissions by 29.9 % and 26.5 %. The flexible application of VVT and split injection strategies aims to balance efficiency gains with emissions control. The ANN-based models demonstrated high accuracy (R2 > 0.974), proving reliable substitutes for real-engine testing in certain conditions. In conclusion, this research highlights the potential of VVT and split injection strategies to enhance HDI engine performance while mitigating environmental impact, supporting the transition to sustainable energy solutions.