Heat exchangers are widely used to transfer heat between two or more fluids in so many industrial applications. In general, the thermal performance of any thermal unit, such as heat exchangers, highly depends on the thermal characteristics of its sub-components, e.g., tubes and tube bundles. In the other words, a more accurate prediction of tubes in a complex thermal unit such as the shell and tube heat exchangers can lead to a more correct prediction of the proposed heat exchanger. There are various methods and approaches to improve the thermal analysis and calculations of tubes, e.g., the artificial neural network (ANN) and the computational fluid dynamics (CFD). The main goal of the present study is to incorporate the ANN and CFD methods and present a novel approach, which can reduce the time and computational costs for the thermal calculation of tubes and tube bundles. In this study, the numerical simulations are carried out for a tube and the achieved numerical solutions are suitably handled to train the network in ANN. The present ranges of working conditions are 3,000 < Re < 25,000 and 7 < Pr < 22. The ANN is developed using the Feedforward method in Python software. To increase the accuracy, a backpropagation approach is also used here. The results of ANN generated data are compared with the numerical solutions. The differences are less than 10%. In addition, the results of the ANN are compared with available experimental correlations. The conclusion is that there is a good agreement among them.