Model-prediction and optimization of the performance of a biodiesel – Producer gas powered dual-fuel engine

• Published in: Fuel (Guildford) 2023-09, Vol.348, p.128405, Article 128405
• Main Authors: Sharma, Prabhakar, Sharma, Avdhesh Kr, Balakrishnan, Deepanraj, Manivannan, Arthi, Chia, Wen Yi, Awasthi, Mukesh Kumar, Show, Pau Loke
• Format: Article
• Language: English
• Subjects: Artificial neural network; Biomass gasification; Emission; Green fuel; Machine learning
• ISSN: 0016-2361; 1873-7153
• DOI: 10.1016/j.fuel.2023.128405

•A dual-fuel combustion engine powered with biodiesel/diesel pilot and producer gas.•An MLP-ANN model using a 3–10-6 topology of three layers was successfully developed.•Combustion-emission characteristics were model-predicted and optimized.•A considerable reduction in exhaust emissions was discovered.•In a lab-based test, the model's errors were determined to be less than 5% Diesel engines have been blamed for harming the environment owing to toxic emissions that raise glasshouse gas (GHG) levels. This study intends to model-forecast and improve the emission and combustion parameters of a dual-fuel combustion engine fueled by biodiesel/diesel pilot and producer gas (PG), which burn cleaner than fossil diesel. To assure a high part of locally producible green fuel, biodiesel was made from waste cooking oil, and PG was made from Babool waste wood. Experiment data were obtained at varied engine loads, fuel injection timings, and pilot fuel mix ratios. Using the experimental data, a multi-layer perceptron architecture was used to create an Artificial Neural Network (ANN) based prediction framework to predict outcomes such as brake thermal efficiency, brake-specific energy consumption, oxides of nitrogen, carbon monoxide, unburnt hydrocarbons, and peak in-cylinder pressure. Statistical measures for the predictive model viz., R (0.964 – 0.998) and R2 (0.9292 – 0.996), and root mean square error (0.008 – 2.185) prove the model’s robustness. Using the desirability technique, the trade-off analysis between efficiency and emission showed that 74.37% engine load, injection timing of 27 degrees crank angle before top dead center (°CA bTDC), and a 20% biodiesel/diesel blend were the ideal operating conditions. An experimental investigation confirmed the prediction errors were fewer than 5%. Using this reliable hybrid approach of predicting and optimizing the performance of a dual-fuel engine, a considerable reduction in exhaust emissions with an acceptable level of engine performance was discovered, which would reduce the negative impact on the environment.