Performance analysis of the pillow plate channel using artificial neural network

Abstract In this work, an artificial neural network (ANN) is used for the performance analysis of pillow plate channels. The ANN model is trained using the values of the various geometric parameters of the channel, Reynolds number, heat transfer coefficient and specific pressure drop. The values of the heat transfer coefficient and specific pressure drop for each channel configuration are obtained by carrying out the numerical simulations for each channel. It is observed through the validation process that a well-trained ANN model is suitable for the performance analysis of the channel. A large data set is obtained after the performance analysis that can be further used to estimate the performance of the pillow plate channel for different geometric configurations. The validity of the ANN model is proved by carrying out a numerical simulation on a random channel configuration out of the large data set, and it is found that the specific pressure drops and heat transfer coefficient predicted by ANN is within 5% of the results obtained from the CFD analysis.