High-fidelity machine learning modelling of GO–Al₂O₃ assisted ethanol–MBD20 dual-fuel CI engine combustion

Klinmalee, Tossaphorn and Elumalai, P.V. and Alwetaishi, Mamdooh and Hussain, Fayaz and Chan, Choon Kit and Reddy, M. Venkateswar and Khan, Sher Afghan and Keçebas, Ali (2026) High-fidelity machine learning modelling of GO–Al₂O₃ assisted ethanol–MBD20 dual-fuel CI engine combustion. Applied Thermal Engineering, 291 (NA). pp. 1-31. ISSN 1359-4311 E-ISSN 1873-5606

	PDF - Published Version Restricted to Repository staff only Download (17MB) | Request a copy
Preview	PDF - Supplemental Material Download (161kB) | Preview

Abstract

The implementation of ethanol-assisted dual-fuel operation in conjunction with nanoparticle-enhanced biodiesel combustion has garnered significant interest as an effective method for improving the efficiency and reducing the emissions of compression ignition (CI) engines. Most of the existing studies focus on these methods independently, thereby challenging the integration of data-driven optimisation with sustainability assessments. This experimental study examines the performance, emissions, and sustainability characteristics of a single-cylinder dual-fuel compression ignition engine with 20 vol% Chlorella vulgaris biodiesel (MB20), 5–15% ethanol energy share, and hybrid graphene oxide–aluminium oxide (GO–Al₂O₃) nanoparticles at 25–100 ppm. Of all the fuel Among the formulations analysed, MB20GA25E15 performed best, with a maximum brake thermal efficiency of 34.8% (6.1% over diesel) and 4.2% lower brake-specific fuel consumption. The CO levels were reduced by 18 to 22%, the concentrations of HC by 15 to 19%, and the smoke opacity measurements by 12 to 18%. However, a minor increase in NOₓ emissions (3–7%) was observed, associated with higher cylinder temperatures and additional available oxygen. Nine machine learning models were created to generalise and interpret the trends seen experimentally. Gradient boosting was the model that predicted the most accurately. The SHAP-based output analysis showed that engine load, ethanol share, and nanoparticle dosage are the main contributors to the results. Based on the Pugh matrix sustainability assessment, MB20GA25E15 is the best fuel blend, providing a balance among efficiency, emissions, cost, and environmental benefits. A practicable strategy for achieving sustainable operation is proposed: a CI engine consisting of a detailed framework based on experimental methods, machine learning, and sustainability principles for applying an ethanol-assisted microalgae biodiesel dual-fuel CI engine equipped with hybrid GO–Al₂O₃ nanoparticles.

Item Type:	Article (Journal)
Uncontrolled Keywords:	Dual-fuel CI engine, Energy efficiency, Microalgae biodiesel (MBD20), Ethanol share, Hybrid nanoparticles (GO–Al₂O₃), Machine learning prediction, SHAP explainability, Sustainability assessment
Subjects:	T Technology > TJ Mechanical engineering and machinery > TJ163.26 Energy conservation
Kulliyyahs/Centres/Divisions/Institutes (Can select more than one option. Press CONTROL button):	Kulliyyah of Engineering > Department of Mechanical Engineering Kulliyyah of Engineering
Depositing User:	Prof. Dr. Sher Afghan Khan
Date Deposited:	26 Feb 2026 09:24
Last Modified:	26 Feb 2026 09:24
Queue Number:	2026-02-Q2258
URI:	http://irep.iium.edu.my/id/eprint/127569

Actions (login required)

View Item