Numerical and Neural Network Analysis of Natural Convection from a Cold Horizontal Cylinder above an Adiabatic Wall

Abstract

Free convection around cold circular cylinder above an adiabatic plate at steady-state condition has been investigated both numerically and by artificial neural networks. There is a growing demand for a better understanding of free convection from a horizontal cylinder in the areas like air cooling, refrigeration and air conditioning system, etc. Governing equations are solved in some specified cases by finite volume method to generate the database for training the neural network in the range of Rayleigh numbers of 10 5 to 10 8 and a range of cylinder distance from adiabatic plate (L/D) of 1/4, 1/2, 1/1, 3/2 and 4/2, thereafter a Multi-Layer Perceptron network is used to capture the behavior of flow and temperature fields and then generalized this behavior to predict the flow and temperature fields for other Rayleigh numbers. Different training algorithms are used and it is found that the back-propagation method with Levenberg-Marquardt learning rule is the best algorithm regarding the faster training procedure. It is observed that ANN can be used more efficiently to determine cold plume and thermal field in less computational time and with an excellent agreement. From obtained results, average Nusselt number of the cylinder investigated to study the effect of adiabatic wall on the isothermal cylinder. It also observed that in spaces farther than L/D = 3/2, average Nusselt number is almost constant, so the affect is renouncement and it works like a cylinder in an infinite environment.