This study evaluates the impact of varying fuel injection timing (FIT) and dual-fuel modes on the performance and emissions of a compression ignition (CI) engine under different load conditions. The biodiesel used was derived from Chlorella protothecoides microalgae through a two-step transesterification process, and its elemental composition was characterized using gas chromatography-mass spectrometry (GC-MS). Acetylene gas was introduced into the engine intake manifold at a rate of 3 L per minute (LPM), while a blend of 20 % methyl ester from Chlorella protothecoides (B20 MEOA) served as the primary injected fuel. To predict engine performance and emission characteristics, advanced machine learning models were employed and evaluated using four statistical criteria, including R-squared, mean absolute error (MAE), and mean squared error (MSE). Experimental results indicated that the optimal configuration involved a dual-fuel mode combining B20 MEOA with acetylene gas and an advanced FIT of 25° before top dead center (bTDC). Performance analysis revealed that under all load conditions, the specific fuel consumption (SFC) decreased by 7.3 %, while brake thermal efficiency (BTE) increased by 1.6 % compared to conventional diesel. Emission testing showed a 7.6 % rise in nitrogen oxide emissions, alongside significant reductions in unburned hydrocarbons (12.5 %), carbon monoxide (25.6 %), and smoke intensity (7.5 %) relative to standard diesel operation. Optimization of the engine parameters ensured that key metrics, such as brake power (BP) and brake-specific fuel consumption (BSFC), remained within acceptable limits. The random forest model outperformed other machine learning models, demonstrating superior accuracy in predicting performance and emissions across all statistical measures. This study underscores the potential of combining advanced biodiesel blends with optimized FIT strategies to improve engine efficiency and emissions control, offering a promising approach for sustainable dual-fuel engine operations.
Read full abstract