Modeling the effect of adding oxygen functionalized multi-walled carbon nanotube to diesel, biodiesel, and bioethanol fuel blends on performance and emission of a diesel engine using artificial neural network

Blended biofuels such as biodiesel and bioethanol besides adding carbon nanotubes as catalysts to diesel fuel significantly improve engine performance and reduce emissions. In this study, biodiesel (5%) was initially added to the diesel fuel to evaluate engine performance and its emissions. The studied fuels were prepared as MWENT-COOH nanoparticles (30, 60 & 90 ppm) were added to the fuels of B5 (5% biodiesel and 95% diesel) and E6 (6% bioethanol and 94% diesel) and E3 (3% bioethanol and 97% diesel). Experiments were performed in triplicates and a multilayer feed-forward back-propagation (FFBP) artificial neural network (ANN) was used for modeling the obtained results. Fuel type, engine speed, density, viscosity, the thermal value of the fuel, inlet manifold pressure, fuel consumption, exhaust gas temperature, oil temperature, oxygen in the exhaust gases, relative humidity and pressure of inlet air were considered as independent or inlet layer parameters. Output layer parameters included engine performance and emission. The results represented the decrease in emissions of CO and unburned hydrocarbons and specific fuel consumption as well as an increase in nitrogen oxides emissions. Considering the R^2 and MSE, the ANN model based on the sigmoid learning function was introduced as the optimal one in comparison to the linear and hyperbolic tangent networks. The values of the regression coefficient (R^2) were also obtained for training, evaluation, and testing of the optimal network model. In conclusion, it can be mentioned that ANN was the most effective model in simulating the obtained data and investigating the sensitivity coefficient.