Citation

Al Awadh, Mohammed and Goh, Michael Kah Ong (2026) RSM optimisation and explainable machine learning of a CNG–diesel–Nannochloropsis B10/B20 Fueled PCCI engine with hybrid nanoparticles. Applied Thermal Engineering, 298. p. 131019. ISSN 1359-4311

Text 7.pdf - Published Version Restricted to Repository staff only Download (90MB)

Abstract

Engines that use compression ignition have a constant trade-off between soot and nitrogen oxides emissions, which makes it difficult to reduce both at the same time with traditional combustion methods. A technique called Premixed Charge Compression Ignition (PCCI) assisted could be a good solution, but the efficiency of this method is very sensitive to the reactivity of the fuel used. New fuel strategies are needed for this reason. This paper presents the first study on using hybrid nanoparticle-enhanced Nannochloropsis biodiesel and CNG (Compressed Natural Gas) in dual-fuel PCCI combustion, along with RSM Optimisation, machine learning explainability, and sustainability assessment based on the Pugh matrix. A comprehensive experimental and numerical study of diesel–biodiesel–nanoparticle–CNG hybrid fuel blends for improving performance, emissions, and sustainability of compression ignition engines The GO–CeO₂ and Al₂O₃–TiO₂ nanoparticles (25–100 ppm) and the CNG energy shares (0–30%) were tested with B10 and B20 biodiesel blends. The results demonstrated a significant performance increase. There was an increase in brake thermal efficiency (BTE) from 32.7% for D100 to 35.3% for optimised nanoparticle–CNG blends, while the brake specific energy consumption (BSFC) decreased by 8–14%. While NOx levels increased mildly, it reduced CO, HC and smoke by 40–55%, 30–45% and 20–35%, respectively. The significant interactions between fuel blend and nanoparticles at the highest order (combination) were identified by Response Surface Methodology (RSM), and the combination exhibiting the lowest viscosity was B20 + 50 ppm of GO–CeO₂ + 10% CNG, which was predicted to give the highest efficiency. Elevated predictive accuracy (R2 = 0.98–0.995) for performance and emission outputs was achieved using 3 machine learning models (CatBoost, Random Forest, and XGBoost), with strong predicted–actual agreement fanfare. Shapley additive explanations (SHAP) analysis indicated that output variation was primarily driven by load, nanoparticle concentration, and biodiesel fraction. We created a GUI for real-time prediction and blending for users. A sustainability analysis using a Pugh matrix yielded that B20 + GO–CeO₂ + CNG was the pathway with the most sustainability. Hybrid fuel strategies are preferred for next-generation CI engines because of their many technical, environmental, and computational advantages, such as those demonstrated in this study.

Item Type:	Article
Uncontrolled Keywords:	PCCI engine, Nannochloropsis biodiesel, CNG dual-fuel, Hybrid nanoparticles, Machine learning, SHAP explainability
Subjects:	Q Science > QA Mathematics > QA71-90 Instruments and machines
Divisions:	Faculty of Information Science and Technology (FIST)
Depositing User:	Ms Suzilawati Abu Samah
Date Deposited:	04 May 2026 02:52
Last Modified:	07 May 2026 09:24
URII:	http://shdl.mmu.edu.my/id/eprint/15832

Downloads

Edit (login required)