Experimental and numerical consideration of the effect of CeO2 nanoparticles on diesel engine performance and exhaust emission with the aid of artificial neural network

Reduction of exhaust emission and fuel consumption is one of the most important challenges in the engine communities. One of the methods to overcome the issue is improving fuel by modification or reformulation of its composition. To this end, the experiments were conducted such that the power, emissions, and fuel consumption on a CI diesel engine were altered using fuel blend of diesel with nanoparticles. For this purpose, cerium oxide nanoparticles in 10-30 nm scale were used and added to the base fuel blend in rates of 10, 20, and 40 ppm. The results showed a significant reduction in NOx and HC and a slight increase in CO emissions as compared to pure diesel fuel. In addition, a slight decrease was observed in fuel consumption while the brake power exhibited no significant changes for this fuel blend. Three sets of input elements namely, BSFC, nanoparticle addition, and engine speed were considered whereas power, NOx, HC, and CO emissions are output parameters. The results, however, indicate that 12 neurons of hidden layer, together with application of Levenberg-Marquardt training rule led to the best network performance with the least MSE value of 0.000172. According to the current investigation, the network modeling succeeded in presentation of efficient interconnecting relation between nanoparticle impact in fuel with engine power and pollutant emissions. (C) 2016 Elsevier Ltd. All rights reserved.