Application of Artificial Neural Networks for Prediction of Natural Convection Heat Transfer from a confined Horizontal Elliptic Tube

In this paper, simulation of natural convection heat transfer from a confined isothermal horizontal elliptic tube based on artificial neural network is presented. The experimental work with changing of all variables is time consuming and expensive and usually has problems associated with it. Experimental data used for the neural network was obtained by a Mach-Zehnder interferometer. In this study, various parameters have been used. Tube axis ratio, wall spacing and rayliegh number are inputs and average nusselt number is desired output. Multi-layer feedforward network was applied to simulate the steady condition of heat transfer rate distribution in described geometry. Results of network have excellent agreement with experimental data. Therefore, the network is used to predict the unseen data points within the range of experimental results. Keywords—Neural network, Back propagation, Nusselt number, Rayliegh number, Elliptic tube.