Entropy generation analysis and simulation of turbulent forced convection around tube with integral wake splitter using artificial neural network approach

Abstract

This paper deals with the investigations on entropy generation around a circular plain tube with integral downstream splitter plates with cross flow of air in a rectangular duct. This study is carried out in the Reynolds number range 49 × 103 to 227 × 103 on single plain cylinder and single cylinder of various splitter length-to-tube diameter ratios, L/D = 0, 1/3, 2/3, 1, 3/2, 2 and 5/2. Entropy generation characteristics are studied under constant heat flux and temperature conditions and simulated using the artificial neural network (ANN) algorithm. The analysis of the second law of thermodynamics, which considers both the heat transfer and pressure drop effects, indicated the addition of the plate decreases the entropy generation with increased plate length, the entropy generation number being more sensitive to the Reynolds number as plate length increases. entropy generation reduction was, in general, also evident with increased plate length. Results point towards tube with wake splitter being more efficient than plain tube based on entropy generation. The predicted values of the entropy generation number of augmented design, NS,a, using the ANN algorithm are compared with the calculated results. The results showed that the ANN model has good expectation performances with error approaches zero without conducting extra experiments consuming time, money and effort or solving nonlinear mathematical models. Owing to its performance, ANN is a good choice for modeling the entropy generation around a circular plain tube with integral downstream splitter plates.