Hydrogen-enriched dual-fuel CI engine fueled with Mahua biodiesel and hybrid nano-additives: Integrated experiments, explainable machine learning, and multi-objective optimization

Hydrogen-enriched dual-fuel compression-ignition (CI) engines are a potential pathway towards higher efficiency and lower carbon-intensive emissions. Studies conducted so far have considered hydrogen enrichment, biodiesel fuels, nano-additives, and data-driven optimization as separate entities; hence, there is no integration or comprehensive understanding about them, which leads to an efficiency-nitrogen oxides trade-off. This study presents an integrated experimental-machine learning-explainable artificial intelligence-multi-objective optimization framework for a hydrogen-assisted dual-fuel CI engine fueled with a Mahua biodiesel-diesel (B20) blend and hybrid nano-additives (Al2O3–TiO2 and CeO2-MWCNT, 50-100 ppm). Experimental results indicated that hydrogen enrichment hybridized with nano-additives improves brake thermal efficiency by 8-14% and reduces brake-specific fuel consumption by 10-18%. HC, CO, and smoke emissions are reduced by up to 35%, 32%, and 45%, respectively. There is a moderate increase in NOx by 12-28%. Machine-learning models achieved high predictive accuracy (R2 > 0.99). The XGBoost exhibited superior generalization. The SHapley Additive exPlanations analysis found that the dominant factors were engine load, the hydrogen energy share, and the concentration of nano-additives. The XGBoost-Multi-Objective Grey Wolf Optimizer (XGB–MOGWO) framework created Pareto-optimal solutions showing a strong and interpretable pathway for advancing trade-offs between efficiency and emissions in dual-fuel engines.

Text 125252.pdf - Published Version Restricted to Repository staff only Download (21MB) Official URL or Download Paper: https://linkinghub.elsevier.com/retrieve/pii/S0360...

Actions (login required)

View Item