Hydrogen‐enriched algae biodiesel combustion in diesel engine: A high‐efficiency energy strategy enhanced by advanced graph neural network‐predictive modeling

Abstract The growing demand for carbon‐neutral fuels has driven increased research into hydrogen (H 2 )‐assisted biodiesel combustion. Engine performance, combustion, and emissions were studied using algae biodiesel blends with H 2 enrichment at 3 and 6 LPM. A graph neural network (GNN) model was also developed to link experimental dual‐fuel data with engine behavior predictions. Experiments of six biodiesel blend ratios and two H 2 flow rates were performed at five different loads (0–100%), evaluating performance, combustion, and emissions. Due to the lower calorific value of the fuel, the brake thermal efficiency (BTE) reduced by 6.1% with a higher biodiesel mixture, and 6 LPM H 2 enhanced the engine performance by 3.7% and compensated for the thermal energy loss. The H 2 enrichment enhanced peak pressure and heat release rate (HRR) by 6–6.4%, compensating for losses from biodiesel usage. Overall, nitrogen oxides (NO x ) emissions increased by 23.1% with B100 and 3.6% for 6 LPM H 2 addition. Hydrocarbons (HC) were reduced by 87.5%, carbon monoxide (CO) by 28.8%, and the total amount of smoke decreased by 27.1% with the B100 + 6 LPM H 2 condition. A 90‐node heterogeneous GNN using 38 physics‐informed features achieved R 2 >0.95 and RMSE <5% for five simultaneous outputs, effectively capturing nonlinear interactions between hydrogen and biodiesel. Overall, H 2 enhances the performance and clean‐burning potential of algae biodiesel, significantly reducing key pollutants while causing a modest increase in NO x . The developed GNN framework provides an efficient predictive tool for optimizing H 2 biofuel dual‐fuel engines and supports the advancement of low‐carbon combustion technologies.