Application of artificial neural network and response surface methodology for predicting and optimizing dual-fuel CI engine characteristics using hydrogen and bio fuel with water injection

In this work, the experiments were conducted to investigate the effect of using hydrogen (H2)andLemon Grass Oil (LGO) as a partial substitute fuel for diesel in a single-cylinder Compression Ignition (CI) engine. Initially, the experiment was conducted with neat diesel and then the engine was modified to operate under dual-fuel mode by inducting H2with intake manifold and LGO as direct injection (DI) fuel. The energy share of H2 was varied from 5 to 10% in this research work. Water (W) was mixed (3% volume) with diesel-LGO blends and its effect on oxides of Nitrogen (NOx) emissions was investigated. An ANN model has been developed to predict the correlation between engine output responses and input factors (load, LGO and hydrogen) using a standard back-propagation algorithm. Response surface methodology is concerned to optimize the engine input parameters in order to minimize the emissions and maximize the thermal efficiency. It was observed that the BTE got increased to 31% for 10%H2 + LGO25 whereas it was 24% for neat LGO and 30% for 10%H2 + LGO25 + 3%W modes. The cylinder pressure increased up to 63 bar for 10%H2mode than neat LGO operation. The HRR, underH2 premixed combustion model was observed to be 79 kJ/CA deg in 10%H2, which is comparatively higher than that of all the LGO operations. The maximum NOx emission was found for 10%H2 + LGO25% at 700 ppm peak load, whereas it reduced to 620 ppm for water-emulsified fuel.