Prediction of entropy generation for nanofluid flow through a triangular minichannel using neural network

• Heat transfer and second law of thermodynamics are evaluated for nanofluid flow.
• Heat transfer enhances by raising Re and concentration and reducing particle size.
• Friction entropy generation trend is in contrast with thermal entropy generation.
• By changing particle size, total entropy generation incorporates an optimum value.
• Entropy generation rates are modeled in terms of effective parameters using ANN.

Heat transfer characteristics and second law of thermodynamics are evaluated for the water–Al2O3 nanofluid flow in a triangular minichannel under constant wall heat flux. The effects of parameters such as Reynolds number, particle size, wall heat flux, and particle concentration on entropy generation rates are investigated. By increasing the Reynolds number and the concentration and by reducing the particle size, the convective heat transfer coefficient enhances. The thermal entropy generation rate decreases when the concentration and the Reynolds number are increased, while it increases when the heat flux and the particle size are increased. However, the frictional entropy generation incorporates a trend which is totally in contrast with the thermal entropy generation. The effect of Reynolds number change on the frictional entropy generation rate is more significant than that of concentration. At higher Reynolds numbers, changing of the concentration alters the frictional entropy generation rate more significantly. By changing the particle size, the total entropy generation rate incorporates a minimum value (i.e. optimum value). The Bejan number has large values near the walls, and the effect of changing the particle size on the Bejan number is more noticeable at the greater concentrations. In addition, at the cross sections near the inlet, the Bejan number is negligible in large parts of central regions of the channel, and increases along the channel length. Based on the data obtained, the model of entropy generation rates was developed in terms of effective parameters using Artificial Neural Network (ANN).

Figure optionsDownload as PowerPoint slide