Assessment of performance and exhaust emission quality of different compression ratio engine using two biodiesel mixture: Artificial neural network approach

In this work, the thermal performance characteristics and emissions of VCR, one cylinder compression ignition (CI) engine fuelled with a mixture of two biodiesel in a blend with diesel has been assessed by using artificial neural network (ANN). Further, the two biodiesel considered are Pongamia and Jatropha mixed in different volume ratio, i.e. 25:75, 50:50, 75:25 these mixtures are called M1, M2, and M3. The mixtures are used to prepare the various blends operated with diesel fuel, which are utilized in the experimentation. The engine experimental data required for the training and validation of ANN model are obtained through the VCR engine operated with pure diesel and blends of two biodiesel mixture as a fuel at different load and compression ratio. To train the ANN model, mixture ratio, blend ratio, load and compression ratio (CR), are selected as the inputs and output variables are brake specific fuel consumption (BSFC), exhaust gas temperature (EGT), brake thermal efficiency (BTE), carbon monoxide (CO), hydrocarbons (HC), nitrogen oxides (NOx), smoke density. Different architectures of ANN are trained by varying the number of hidden neurons in the hidden layer and corresponds to the minimum mean square error (MSE) for validation data for selecting the optimum architecture to estimate the parameters. The thermal performance and emissions of VCR engine estimated by using proposed ANN model are found to be quite close to experimental values with reasonable accuracy as the correlation coefficient is ranging from 0.97 to 0.99. (C) 2020 The Authors. Published by Elsevier B.V. on behalf of Faculty of Engineering, Alexandria University.

Automated summary

In this study, the thermal performance and emissions of a VCR diesel engine using a blend of two biodiesels in varying ratios were assessed using artificial neural network (ANN). By training the ANN model and selecting the optimal architecture, the estimated parameters closely matched the experimental values with high accuracy.