Optimization of performance and emission of a diesel engine fueled with isopropyl alcohol Blends: A comparative ANN-GA and RSM-HCO application

Abstract

This study focuses on optimizing the operating parameters of a diesel engine fueled by diesel-isopropyl alcohol blends, with the aim of enhancing engine performance and minimizing emissions. Using data collected from 144 experimental runs, predictive models were developed utilizing artificial neural networks (ANN) and response surface methodology (RSM). The models exhibited impressive accuracy, with an average absolute percentage error below 2 % for each parameter and an R2 value exceeding 0.99. Subsequently, these models were optimized using genetic algorithm (GA) and hill climbing algorithm (HCO) to identify the optimal engine operating conditions. The results indicate that both ANN-GA and RSM-HCO models exhibit satisfactory accuracy in predicting engine parameters. The ANN-GA model demonstrated an average deviation of 2.9 % from experimental data, while the RSM-HCO model exhibited an average deviation of 5.4 %. Both optimization results indicated that, in an emissions-focused approach, desirability above 0.9 could be achieved with isopropyl alcohol-diesel blends and high engine speeds, such as 2600–2700 rpm, with a 3.5 bar engine load, resulting in low emission values and high engine performance.