Experimental and ANN-based analysis of performance, combustion, and emission characteristics of a CI engine fueled with waste plastic oil–diethyl ether–diesel blends

This study differs fundamentally from prior investigations on WPO-diesel and WPO-DEE blends by combining combustion-resolved experimentation with predictive modeling, thereby advancing WPO utilization from empirical testing toward optimization-oriented engine integration. It examines the performance, combustion, and emission characteristics of a single-cylinder variable compression ratio (VCR) diesel engine fueled with ternary blends of diesel, waste plastic oil (WPO), and diethyl ether (DEE). WPO was produced via catalytic pyrolysis of LDPE waste and blended with diesel at 15%, 20%, 25%, and 30% by volume, while DEE was maintained at a constant 10% to improve ignition quality, volatility, and atomization. Engine tests were performed at a constant speed of 1500 rpm under variable loads ranging from 2 to 12 kg to evaluate the influence of blend composition and operating conditions on brake thermal efficiency (BTE), brake specific fuel consumption (BSFC), combustion development, and regulated emissions (CO, HC, NOx, CO₂). The D65B25DE10 blend (65% diesel, 25% WPO, 10% DEE) demonstrated the best overall performance among the tested fuels, achieving a 22.22% reduction in CO and an 11.88% reduction in HC emissions compared with diesel, although BTE decreased by 6.93% and BSFC increased by 6.03% at full load. Combustion analysis revealed extended ignition delay and higher peak cylinder pressure for higher-WPO blends, while DEE improved vaporization and supported more complete oxidation. To complement the experimental work, a feed-forward artificial neural network (ANN) model with a 6-12-6 architecture was developed using blend ratio, load, compression ratio, and speed as inputs to predict BTE, BSFC, and emissions. The ANN achieved strong correlation with experimental data (R2 > 0.97), confirming its suitability for performance prediction and blend optimization. The combined experimental and computational approach offers a comprehensive framework for evaluating WPO-based fuels, extending beyond previous binary blend studies by revealing the synergistic effects of DEE in ternary blends and establishing a robust ANN model for predictive optimization. This methodology demonstrates the potential of WPO-based fuels to reduce fossil diesel dependence while promoting sustainable waste-to-energy utilization.