Our latest paper has been accepted for publication in Applied Thermal Engineering 🎉

Excited to share that our paper on machine learning-assisted ammonia engine combustion has been accepted in Applied Thermal Engineering, a Q1 journal! This work bridges experiments and AI to optimize pre-chamber design for cleaner combustion.

---

• Abstract: Transportation contributes significantly to global CO₂ emissions (24% of energy-related emissions), necessitating carbon-free energy solutions for net-zero goals. However, hydrogen (H2) is a promising candidate, its storage and transportation challenges persist. ammonia (NH3), a carbon-free alternative, faces limitations such as low burning velocity and narrow flammability, requiring further optimization. This study combines experimental and machine learning (ML) approaches. Stage-I experimentally investigates the impact of pre-combustion chamber (PCC) on four different nozzle geometries and slit/non-slit configurations on combustion at varying equivalence ratios. Stage-I is further divided into two categories Case-I and Case-II. In Case-I, nozzle bodies without slits were compared, with the 10.03 diameter achieving the highest combustion pressure (1.31 MPa at ER 0.8). In Case-II, the similar diameter was investigated with/without slits, showing a 14% performance improvement with slits. Stage-II employs Random Forest (RF), Artificial Neural Network (ANN), and Deep Neural Network (DNN) models to predict the heat release rate (HRR). Since, RF illustrates low error among experimental values, and it acts as a benchmark for further predicting error of evaluated by ANN and DNN at hidden ER. Root Mean Square Error (RMSE) and R² (coefficient of determination) for ANN and DNN have been illustrated with values of 23.10, 0.87 and 45.24, 0.51 respectively. Since ML values provides better fitment with experimental data, future studies may prioritize ML models, cutting down on expensive experiments and time consumption.