AI-enabled Thermo-GNN assessment of waste-derived nanocatalytic biodiesel for sustainable and low-emission diesel engine operation
International Journal of Low-Carbon Technologies · 27.01.2026 · 21
DOI: 10.1093/ijlct/ctag047 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Low-Carbon Technologies, online erschienen am 12.05.2026. Verfasst von Lionus Leo George Mary Helan und 5 weiteren Autor:innen. Die Arbeit nennt 78 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
This study develops a clean, high-efficiency combustion strategy for agricultural diesel engines by integrating spent coffee ground biodiesel (SCGB) with graphene oxide–zinc oxide (GO@ZnO) core-shell nanoadditives. A Physics-Informed Thermodynamic Graph Neural Network (Thermo-GNN) is introduced to predict performance and emissions behavior using sparse experimental data. SCGB was produced via acid–base transesterification, blended with diesel (B25–B100), and doped with 50 and 100 ppm GO@ZnO nanoparticles. Experiments were performed on a single-cylinder diesel engine at 1500 rpm. Thermo-GNN was trained on 15 full-load pressure traces using nine physics-based constraints, including closed-cycle energy conservation, ideal-gas consistency, ignition-delay regularization, and an oxygen-donation constraint. Accuracy was tested on interpolation and extrapolation regimes. GO@ZnO improved combustion, raising peak pressure by 8.9% and peak heat-release rate by 45% versus diesel. The highest brake thermal efficiency (BTE) was 33.5% (7.7% above diesel), and brake specific fuel consumption (BSFC) decreased by 7.1%. Emissions dropped by 27.5% for carbon monoxide, 30.8% for hydrocarbons, 17.1% for nitrogen oxides (NOx), and 65% for smoke with the optimal nanofuel blend. Thermo-GNN accurately reconstructed pressure traces (R2 > 0.9987) and predicted BTE, BSFC, NOx, and smoke, with mean absolute percentage errors of 2.3%, 2.7%, 7.1%, and 12.6%, respectively, even under double extrapolation. Attention and latent-variable analyses confirmed mechanistic learning of premixed-phase catalytic activity and lattice-oxygen donation. The combined SCGB-GO@ZnO fuel pathway and the Thermo-GNN framework provide a scalable, physically interpretable foundation for deploying digital twins in existing diesel engines.
Zitieren
APA
Helan, L. L. G. M., Chandran, T., Jayabal, R., Subramani, S., Sivanraju, R., & Gnanasekaran, A. P. (2026). AI-enabled Thermo-GNN assessment of waste-derived nanocatalytic biodiesel for sustainable and low-emission diesel engine operation. International Journal of Low-Carbon Technologies, 21. https://doi.org/10.1093/ijlct/ctag047
Vancouver
Helan LLGM, Chandran T, Jayabal R, Subramani S, Sivanraju R, Gnanasekaran AP. AI-enabled Thermo-GNN assessment of waste-derived nanocatalytic biodiesel for sustainable and low-emission diesel engine operation. International Journal of Low-Carbon Technologies. 2026;21. doi:10.1093/ijlct/ctag047
BibTeX
@article{Helan2026,
author = {Helan, Lionus Leo George Mary and Chandran, Thiyagarajan and Jayabal, Ravikumar and Subramani, Sekar and Sivanraju, Rajkumar and Gnanasekaran, Ashwin Prabhu},
title = {AI-enabled Thermo-GNN assessment of waste-derived nanocatalytic biodiesel for sustainable and low-emission diesel engine operation},
journal = {International Journal of Low-Carbon Technologies},
year = {2026},
volume = {21},
doi = {10.1093/ijlct/ctag047}
}
EUVIMED