Performance and emission prediction of a tert butyl alcohol gasoline blended spark-ignition engine using artificial neural networks

This paper proposes the mathematical modelling using artificial neural network (ANN) for predicting the performance and emission characteristics of spark-ignition (SI) engine using tert butyl alcohol (TBA) gasoline blends. The experiments are performed with a four-stroke three cylinder carburetor type SI engine at three different revolution per minutes such as 1500, 2000, and 2500 with different blends ranging from 0% to 5% and at 10%. Experimental data are used for training an ANN model based on the feed-forward back-propagation approach for predicting the data at 6–9% with the same speeds. Results show that the blending of TBA with gasoline improves the emission characteristics compared with the gasoline. From the experimental testing data, root mean squared-error was found to be 0.9997% with the network 3-1-10. During this study, The ANN model accurately anticipates the performance and emissions of the engine.