Hydrogen-enriched nano-enhanced waste plastic pyrolysis oil blends: Combustion, emission, and machine learning-based optimization in a dual-fuel diesel engine

Abstract

This study investigates the combustion, performance, and emission characteristics of a hydrogen-enriched dual fuel engine operated with a nano-enhanced ternary fuel blend (diesel/waste-plastic-driven pyrolysis oil/diethyl ether). Experiments were conducted at a fixed engine speed of 1500 rpm, with a load variation ranging from 20% to 100%, a constant hydrogen flow rate of 3 LPM, and an Al₂O₃ nanoparticle concentration varying from 25 to 100 ppm. The nano-enhanced blends showed improved combustion and thermal efficiency compared to baseline blends (Diesel + H₂) operation. Under full load operating conditions, the P20DEE5NP100+H2 blend achieved an 11.5% improvement in brake thermal efficiency (BTE), while carbon monoxide (CO) and hydrocarbon (HC) emissions decreased by 49 % and 29 %, respectively, compared to the baseline blend. However, enhanced combustion intensity led to increased NOx emissions, indicating a trade-off between efficiency and NOx formation. Combustion analysis showed 5% high peak cylinder pressure and an advanced heat release rate with nanoparticle dispersion, which suggests reduced ignition delay and improved premixed combustion. To support experiments, five different supervised machine learning (ML) models were developed for multi-output prediction. The Extra Tree (ET) and Gradient Boosting (GB) models exhibited superior predictive capability, attaining normalized performance scores of 1.00 and 0.96, respectively. Multi-objective optimization using a desirability-based approach identified an optimal operating condition at 62.45% load with 100 ppm nanoparticle dispersion, yielding a composite desirability of 0.651 and a prediction error within ±3.1%. These findings show the potential of integrating waste-driven fuels, hydrogen enrichment, and data-driven modeling for sustainable engine operation.