Publication

Abstract : The automobile industry is facing more stringent emission norms and vastly escalating fossil fuel requirements. Sustainable alternative fuels for internal combustion (IC) engines have to be used. This study investigates the performance optimization and emission behavior of an IC engine fueled with Azolla pinnata biodiesel (AB) blends. The performance of 5%, 10%, and 20% AB blends (D95AB5, D90AB10, D80AB20) tested in the laboratory at a constant engine speed of 2200 rpm and at different load conditions. AB showed properties that were good for fuel and higher flashpoint (130 °C), density (860–880 kg/m³), and viscosity (4.8–5.2 cSt) besides a lower calorific value (39–41 MJ/kg). The experimental outcome showed an increase in brake thermal efficiency with an increase in the percentage of biodiesel, with D80AB20 reaching 7.7%, while brake-specific fuel consumption decreased by up to 20.5% at higher loads. The adoption of biodiesel blends almost always led to a significant drop in emissions, in particular of CO₂, ranging from 25–40% and HC from 50% for the lower blends, though NOₓ showed an increase owing to the oxygen content in the air. To complement the research carried out experimentally, ANN, kNN, RF, SVM, and XGB machine learning (ML) algorithms were used with the help of Python programming with the dataset divided into 80% for training and 20% for testing. XGB shone out by showing the best multi-output prediction capacity (R² up to 0.999), which underlines the significance of AI-based data modeling for accurate performance and emission forecasts of diesel engines.