Performance prediction of HCCI engines with oxygenated fuels using artificial neural networks

Butanol and ethanol are promising conventional fuel alternatives particularly when utilized in advanced combustion mode like homogeneous charge compression ignition (HCCI). This study investigates the performance and emission characteristics of HCCI engines fueled with oxygenated fuels (i.e. butanol and ethanol). The investigation is done through a combination of experimental data analysis and artificial neural network (ANN) modeling.This study uses HCCI experimental data to characterize variations in seven engine performance metrics including indicated mean effective pressure (IMEP), thermal efficiency, in-cylinder pressure, net total heat released, nitrogen oxides (NOx), carbon monoxide (CO), and total hydrocarbon (THC) concentrations. Two types of ANNs including radial basis function (RBF) and feedforward (FF) are developed to predict the seven engine performance metrics. The experimental data at 123 HCCI operating points from two different engines are collected to validate the ANN models. The validation results indicate both RBF and FF models can predict HCCI engine performance metrics with less than 4% error for butanol and ethanol fueled engines. The results show that the FF neural network models are advantageous in terms of network simplicity with fewer required neurons but need twice as much training time compared to the RBF models