Empirical study on butanol-ethanol-gasoline blends using Artocarpus heterophyllus peel resource for eco-friendly gasoline engine application

• Published in: Process safety and environmental protection Vol. 191; pp. 2222 - 2236
• Main Authors: Ashwin, Raghavan, MohanRaj, Thangavelu
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.11.2024
• Subjects: Artificial Neural Network; Artocarpus heterophyllus peel; Combustion; Emission characteristics; Open ECU; Prediction; Artocarpus heterophyllus peel; Artificial Neural Network; Prediction; Combustion; Performance; Emission characteristics; Open ECU
• ISSN: 0957-5820
• DOI: 10.1016/j.psep.2024.09.090

Since using current engine fuels contributes to climate change and global warming, there is intense competition to provide an ecologically friendly and less harmful substitute fuel. It proved to be rather feasible to mix engine-designed fuel with alcohol fuel. In this work, the peel of Artocarpus heterophyllus is used to produce the bioethanol. In three ratios—10 % butanol-ethanol + 90 % petrol (BEG10), 20 % butanol-ethanol + 80 % petrol (BEG20), and 30 % butanol-ethanol + 70 % petrol (BEG30)-butanol-ethanol and petrol were mixed in the present work. Using a four-stroke, spark ignition (SI) engine one-cylinder, the effect of the butanol-ethanol-gasoline combination on operational properties was examined and studied. Three test fuels (BEG10, BEG20, and BEG30) were studied in the engine speed range of 2500 rpm to 4500 rpm. The best practical responses found were brake thermal efficiency (43.21 %) and brake-specific fuel consumption (0.23 kg/kWh). The findings showed that more significant concentrations of ethanol-butanol improve the performance characteristics. In the end, engine emission parameters such as hydrocarbon (40 %), carbon monoxide (25 %), and nitrogen oxide (5.88 %) were reduced by the butanol-ethanol combination with petrol. Artificial neural network (ANN) models of multiple regression are used to forecast the engine performance parameters. An ANN model is trained using a database produced from the experimental findings. It shows that the suggested artificial neural network model can provide a precise correlation between input variables and output replies. The input parameters like Speed, Load, and Blend were predicted using the identified model. The R2 (0.9638, 0.9735) value of BTE and BSFC obtained indicated that the neural network approach model generated was more accurate for the prediction process. ANN application is thus a superior forecasting tool for SI engine performance characteristics. [Display omitted]