Benchmarking ANN Setup Options for Trade-Off Functions of Heat Transferring Baffle Inserted Tubes

Artificial Neural Networks (ANNs) have been widely used for thermal analysis of heat exchangers during the last two decades. This present research uses artificial neural networks (ANNs) to determine Nusselt numbers and friction factors for nine different baffle plate-inserted tubes. MATLAB toolbox was used to search for better network configuration prediction by using commonly used multilayer feed-forward neural networks (MLFNN) with backpropagation (BP) learning algorithm with five different training functions with adaptation learning function of mean square error and TANSIG transfer function. In this research, eighteen data samples were used in a series of runs for each of nine samples of baffle-inserted tube. The uncertainties of experimental quantities were computed by using the method presented. The uncertainty calculation method used involves calculating derivatives of the desired variable with respect to individual experimental quantities and applying known uncertainties. Up to 70% of the whole experimental data was used to train the models, 15% was used to test the outputs and the remaining data points which were not used for training were used to evaluate the validity of the ANNs. The results show that the TRAINBR training function was the best model for predicting the target experimental outputs. The study concluded that artificial neural network methodology has been successfully applied to transient forced convective heat transfer to determine the time-averaged values of Nusselt number, friction factor, entropy generation number and irreversibility distribution ratio. It is obvious that all of the training functions are in good agreement with the experimental data set but the TRAINLM training function is the best training function for the prediction of output layer parameters.