Experimental investigation and artificial neural network modelling of waste plastic oil–carbon nanotube blends in a compression ignition engine

The present work investigates the production, characterization, and engine application of waste plastic oil (WPO) derived via catalytic pyrolysis using natural zeolite, along with performance enhancement through carbon nanotube (CNT) additives. The produced fuel exhibited diesel-like characteristics, with density increasing from 832 to 934 kg/m3 and cetane number improving by ~10% with increasing WPO fraction, indicating favorable ignition quality. Engine experiments conducted on a single-cylinder CI engine revealed that brake thermal efficiency (BTE) increased with load for all fuels, reaching 30.26% for diesel at full load. Neat WPO (WP100) showed a reduction of ~12.9% in BTE, whereas optimized blends significantly recovered efficiency. Notably, the WPO10+50 CNT blend achieved 29.98% BTE, representing only ~0.9% deviation from diesel, along with a ~4.4% reduction in BSFC compared to base WPO blends. Emission analysis demonstrated substantial improvements, where CO emissions reduced by up to ~44.7% and HC emissions by ~80–85% for WPO-based fuels compared to diesel, due to enhanced oxidation and oxygenated fuel characteristics. Combustion analysis indicated improved heat release rate (~52–53 J/°CA) and higher peak cylinder pressure (~74 bar) for CNT blends, approaching diesel-like combustion behavior. An artificial neural network (ANN) model developed using the trainscg algorithm with logsig–tansig transfer function achieved superior prediction accuracy (R = 0.9815, MSE = 0.0019, NSE = 0.9917), with prediction deviations within ±1–2% for most parameters. Overall, the study establishes that optimized WPO–CNT blends can achieve near-diesel performance with significantly improved emission characteristics, supported by robust ANN-based predictive modeling.