Predicting Common Rail Direct Injection (CRDI) engine metrics using nanoparticle-enhanced pongamia pinnata biodiesel with machine learning

Abstract

The effective use of fossil fuels in compression ignition (CI) engines results in hazardous pollution into the atmosphere. Many scholars explored using biodiesel and its blends to replace fossil fuels and reducing hazardous emissions. The current study extends preparation of biodiesel through transesterification process from the karanja oil feed stock as an alternative energy for twin cylinder, Common Rail Direct Injection (CRDI), CI engines. Moreover, various types of metal and non-metal-based nanoparticles (NPs) are dispersed in karanja oil based B20 (20% of biodiesel is blended in 80% of diesel) sample in the presence of QPAN80 surfactant. To prepare stable and homogeneous nano assisted fuel blends the optimum ratio of 1:4 (NPs to surfactant) is used. The Diesel R-F Simulation software is used to find the calibration of CI engine using diesel fuel. The cylinder pressure (CP), brake thermal efficiency (BTE) are found to be 5% variation, while the brake specific fuel consumption (BSFC) is with in the 10% range. The performance parameters are found to be enhanced BTE increased by 9.8%, and BSFC is decreased by 18.7% for B20CNT50 mix at maximum brake power (BP) as compared to B20 blend. Further, the emission characteristics including carbon monoxide (CO), carbon dioxide (CO2), hydrocarbon (HC), and smoke are found to decrease by 7.71, 11.2, 10.2, and 5.71% for B20CNT50 mix at higher BP as correlated to D100 sample. Nevertheless, the nitrogen oxides (NOx) emission is lowered by 10.9% for B20CNT50 blend than the B20 mix at maximum BP. These findings disclosed that the B20CNT50 sample offers higher performance and lower emission characteristics and potential to use in CRDI, CI engines without engine modifications. Additionally, machine learning algorithms such as support vector regression (SVR), and random forest (RF) are implemented in accurately predicting engine performance and emission characteristics by analyzing relationships between input and output parameters. Their ability to model these interactions enables optimized engine design and reduced experimental cost.