Prediction of the Performance and emission characteristics of diesel engine using diphenylamine Antioxidant and ceria nanoparticle additives with biodiesel based on machine learning

Abstract

This study explores the impact of incorporating antioxidants diphenylamine (DPA) and nanoparticle ceria (CeO2) into Jatropha biodiesel (B30) blend on engine performance and exhaust emissions. The fuel blends utilized in this study consists of diesel, B30, B30 with 100 ppm of antioxidant diphenylamine (B30+DPA100), and B30 with 50 ppm of antioxidant diphenylamine and 50 ppm of nanoparticle ceria (B30+DPA50+CeO250). The experiments were designed using the design of experiment methodology, and they were conducted using various fuels to assess both engine performance and exhaust emissions characteristics. Further, machine learning algorithms (multilayer perceptron, random forest regression, and K-nearest neighbors) has been employed to develop a model for accurately predicting experimental outcomes. The K-nearest neighbors model surpassed the multilayer perceptron and random forest regression models, demonstrating a higher coefficient of determination value and precise outcome predictions. The experimental results reveal that adding antioxidants diphenylamine and nanoparticles ceria at 50 ppm to B30 significantly reduced nitrogen oxides emissions. Compared to B30, B30+DPA50+CeO250 showed a 6.35 % decrease in brake specific fuel consumption and an 8.68 % reduction in nitrogen oxides emissions. However, there was a slight increase of 5.74 % in brake thermal efficiency. Additionally, B30+DPA50+CeO250 exhibited a 2.54 % reduction in maximum cylinder pressure compared to B30.