Answers are grounded in this paper and cite their sources.
Assesses emissions and performance of biodiesel blended with nanocatalysts, indicating improved sustainability.
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.
No takes yet. Share an insight, caveat, or question.
Helan et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: