Combined application of ANN prediction and RSM optimization of performance and emission parameters of a diesel engine using diesel-biodiesel-propanol fuel blends

Abstract

In this study, an artificial neural network (ANN) was used to estimated the performance and exhaust emission parameters of a diesel engine running on diesel, biodiesel, and propanol fuel mixtures. In addition, the parameters estimated by ANN were tried determining the optimum operating parameter by using Response Surface Methodology (RSM). In the experimental study, propanol was added in 3 different ratios (5%, 10% and 20%) into 100% diesel, 80% diesel and 20% biodiesel fuel blends. In addition, engine tests, were made at 5 different engine speeds with 400 min-1 intervals between 1000 min-1 and 2600 min-1 revolutions at full load. In addition, HC (Hydrocarbon), CO (Carbon Monoxide), NOX (Nitrogen oxides) and Smoke emissions were measured during in the working. ANN model was developed for estimation of engine output parameters depending on fuel mixture ratios and engine speed. In the ANN results, the regression coefficients (R2) of the proposed model were found to be between 0.924 and 0.99. When the obtained ANN results were compared with the experimental results, it was seen that the maximum mean relative error (MRE) was 6.895%. It has been shown that the applied model can predict with a low error rate. The RSM results showed that the optimum operating parameters were 2034-min-1 engine speed, 74.667% diesel, 11.36% biodiesel and 15% propanol fuel mixture. In addition, in the validation tests of the model where the desirability was 0.7833%, the highest error rate was obtained as 7.37% as a result of NOX. As a result of the study, it was seen that RSM supported ANN is a good method for estimating diesel engine parameters working with diesel/biodiesel/propanol mixtures and determining optimum operating parameters.