Evaluation of extrapolation ability of artificial neural network modeling on the heat transfer performance of a finned heat pipe

Experimental and numerical analysis have been conducted to examine the heat transfer characteristics of a finned heat pipe based on thermal resistance networks. The present numerical analysis also reports the enhancement of heat transport of heat pipe using the fins. The key simulation parameters considered were three types of fins with circular, square, and hexagonal shapes, the fin length in the range from 19.05 mm to 38.1 mm, the number of fins in the range from 5 to 20, and the fin thickness in the range from 0.25 mm to 1 mm. The heat transfer rate shoots up by 44.7 % in the case of finned heat pipe when compared with the baseline model with respect to the variation in the simulation parameters. An artificial neural network, which is one of the machine learning methods, was used to predict the heat transfer performance obtained from thermal resistance analysis of the finned heat pipe. This paper introduces a novel approach by developing an ANN model that maintains high accuracy over a broader range of operational conditions. The optimized ANN model could predict the heat transfer performance of the finned heat pipe with reasonable accuracy. In addition, the heat transfer rate of the finned heat pipe could be predicted accurately from extrapolated and interpolated data using the optimized ANN model. © The Korean Society of Mechanical Engineers and Springer-Verlag GmbH Germany, part of Springer Nature 2024.