Artificial neural network approach for prediction of thermal behavior of nanofluids flowing through circular tubes

• Two layers MLPNN estimates convective HTC of nanofluids flowing in circular tubes.
• Different ANNs types are developed and their performances have been compared.
• The best ANN model is selected based on its performance through some error indexes.
• The optimal model shows AARD% 04 2.41 and MSE = 1.7 × 10-5.
• Performance of the proposed model is compared with some conventional correlations.

This paper presents the best artificial neural network (ANN) model for the estimation of the convective heat transfer coefficient (HTC) of nanofluids flowing through a circular tube with various wall conditions under different flow regimes. The parameters of the ANN model are adjusted by the back propagation learning algorithm using wide ranges of experimental datasets. The developed ANN model shows mean square error (MSE) of 1.7 × 10− 5, absolute average relative deviation, percent (AARD%) of 2.41 and regression coefficient (R2) of 0.99966 in modeling of overall experimental datasets of convective HTC. The predictive performance of the proposed approach is compared with some reliable correlations which have been proposed in various literatures. The superior performance of the proposed model with respect to other published works has been found through the comparison of results.

Figure optionsDownload as PowerPoint slide