Numerical investigation of heat transfer in a tube equipped with twisted tape with different angles and under constant heat flux with copper nanofluid and evaluation of the results obtained using perceptron artificial neural networks

Abstract

This study was aimed at investigating the computational fluid dynamics to find the effects of different angles of the twisted tape as well as the copper nanofluid concentration changes on heat transfer performance (Nusselt number and convective heat transfer coefficient) in a tube containing nanofluid under a constant wall flux. This simulation was investigated in Reynolds number range of 2,000–10,000. Twisted tape at four angles (30°, 45°, 60°, and 90°) and nanofluid of concentrations 2% and 4% were examined inside a simple tube. Then, the data obtained from the numerical method for the Nusselt number were predicted using the perceptron artificial neural network with the Levenberg–Marquardt algorithm. In this network, Reynolds number, nanofluid volume fraction, and twist angles of twisted tape were considered as inputs and Nusselt number as output parameter. The most optimal increase in the convective heat transfer coefficient and Nusselt number at the twist angle of 30° with copper nanofluid of volume fraction 4% and Reynolds number 10,000 is 8899.77 and 246.191 W/m² K, respectively. The results obtained by artificial neural network showed that topology 1–10–3 had been very successful in predicting Nusselt number so that the values of correlation coefficient for training, validation, test data, and all data were 0.9994, 0.99661, 0.99319, and 0.99868, respectively, and the best performance of the network was 6.5537 which occurred in Epoch 12.

Keywords:

• Artificial neural networks
• heat transfer
• nanofluids
• numerical modeling
• twisted tape