Multi-objective optimization of HDPE pyrolysis oil–diesel blends across variable compression ratios: an integrated experimental and ensemble learning approach

The growing concerns over fossil fuel depletion and plastic waste have encouraged the use of waste-derived fuels. In this study, waste plastic oil (WPO) derived from high-density polyethylene (HDPE) was used as a substitute for diesel fuel in a compression-ignition (CI) engine with variable compression ratios (CRs). WPO was produced at temperatures between 400 °C and 500 °C with a clay catalyst, achieving yields of 65%–70% of liquid fuel with a higher calorific value of 40,950 kJ/kg and density of 860 kg/m 3 . Blends between 0% and 40% were tested at CR 15 to 18 with engine loads 0–12 kg at 1500 rpm. The brake thermal efficiency (BTE) of the WPO blends was higher than diesel fuel, reaching a maximum experimental efficiency of 38.70% with the B40 blend at CR 18, compared to 33.33% for baseline diesel. At full load, the brake-specific fuel consumption (BSFC) for the optimal B40 blend was recorded at 0.260 kg/kWh. The CO emissions of the WPO blends were comparable to diesel at higher CRs. As the engine load increased, a monotonic increase in NO x emissions was observed, although the WPO blends consistently maintained lower overall CO 2 outputs than the baseline diesel fuel. Random forest models were deployed to forecast the behavior of the CI engine operating on WPO blends, ultimately demonstrating superior predictive accuracy compared to artificial neural networks (ANNs). The non-dominated sorting genetic algorithm II (NSGA-II) optimization identified CR = 17, B40 blend at full load (12 kg) as the optimal operating condition, yielding a BTE of 38.47%, brake power (BP) of 3.44 kW, CO of 0.020 vol%, and NO x of 310 ppm, representing the most favorable combined performance-emission trade-off among all investigated operating matrices.