Predictive Modeling and Optimization of Engine Characteristics with Biogas–Biodiesel-Powered Dual-Fuel Mode: A Neural Network-Coupled Box–Behnken Design

The current study seeks to predict and optimize the performance and emission characteristics of a dual-fuel diesel engine powered by biogas and biodiesel. A Box–Behnken design (BBD) was employed to generate an L₂₉ orthogonal array consisting of four factors and three levels. The experiments were conducted at a fixed engine load of 75% and a B20 (20% biodiesel and 80% diesel) pilot fuel blend, with varying compression ratios, fuel injection pressures, fuel injection timings, and biogas flow rates. An artificial neural network (ANN) model was developed using the collected data to estimate key performance and emission parameters. The ANN model exhibited high accuracy, with an overall correlation coefficient (R) of 0.9987 and a mean square error of 0.04. Multi-output RSM was employed to optimize the engine parameters, showing that a compression ratio of 16:1, an injection pressure of 246.08 bar, an injection timing of 26.05° bTDC, and a biogas flow rate of 0.36 kg/h were the optimal operating conditions for the engine. The corresponding optimal engine responses for brake thermal efficiency, brake specific fuel consumption, carbon monoxide, hydrocarbon, nitrogen oxides, and smoke emissions were calculated to be 28.25%, 0.30 kg/kWh, 3.00 g/kWh, 0.026 g/kWh, 4.42 g/kWh, and 42.85%, respectively. The experimental validation demonstrated a close agreement between the predicted results of the model and the actual experimental outcomes, with a maximum error of 7.7%. The ANN-coupled BBD approach represents a viable hybrid method for effectively modeling, predicting, and optimizing the performance of a dual-fuel engine.