Machine learning based analysis of diesel engine performance using Fe₃O₄ nanoadditive in sterculia foetida biodiesel blend

Abstract

Abstract This investigation examines the influence of Fe₃O₄ (magnetite) nano additions in sterculia foetida methyl ester (SME) mixtures on diesel engine performance, combustion, and emissions. SME was produced using transesterification and dispersed with surface modified Fe₃O₄ nanoparticles (NPs) employing probe-type ultrasonication to achieve uniform distribution. Engine tests were performed using pure diesel, a 25% SME blend-75% diesel (SME25), and Fe₃O₄ dispersed SME25 blends at concentrations of 50, 75, and 100 ppm. The results showed that the engine performance measures such as brake thermal efficiency (BTE) increased by 6.69% and specific fuel consumption (SFC) reduced by 7.23% for SME + 100Fe sample than SME25 mix. For the same blend, combustion metrics, such as cylinder pressure (CP) and heat release rate (HRR), increased by 4.46% and 24.21% respectively. Furthermore, at greater loads, the SME25 + 100Fe mix reduced carbon monoxide (CO), hydrocarbon (HC), nitrogen oxide (NOx), and smoke emissions by 23.92%, 22.42%, 5.38%, and 3.61%, respectively. A machine learning (ML) based computational model was created to predict engine performance and emission characteristics across various nano fuel blends. The model achieved great prediction accuracy, with correlation coefficients (R) ranging from 0.9973 to 0.99995 and Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE) values within acceptable bounds. The study confirms that Fe₃O₄ nano fuel blends improve engine efficiency, decrease emissions, and benefit from the integration of ML for accurate and data-efficient performance modelling. This technique is found to be potential for sustainable fuel technologies.