Analysis of Thermosiphon Heat Pipe Performance Using an Artificial Neural Network

In this paper, a combined experimental and numerical investigation was conducted to understand the characteristic performance of a thermosiphon heat pipe. The influences of various parameters, such as working fluid and heat load, on the performance of a vertical copper thermosiphon heat pipe (HP) were investigated experimentally and numerically. An artificial neural network approach (ANN) was utilized to simulate the thermosiphon heat pipe and predict the heat transfer characteristics. The ANN study shows that, having an extended predict capacity, the difficult role of HP can be successfully represented by the nonlinear structure of this simulator. The predicted values of the mean square error of 0.098% were reached for the training while the error in the experiment data is about 6.25%. Based on the ANN model, there were satisfactory agreements and less recorded errors between predicted and observed values.