Data-driven prediction of long-term thermal performance for steel pipe heat exchanger (SPHX) energy piles

Determining the thermal performance of ground heat exchangers (GHEXs) remains a critical challenge in the design of ground source heat pump (GSHP) systems. Among various GHEX types, the steel pipe heat exchanger (SPHX) energy pile is an innovative solution that utilizes steel pipes as both the primary reinforcement and heat exchangers, replacing conventional deformed rebars. However, its practical implementation has been hindered by the absence of a reliable method for predicting its thermal performance. In this study, an artificial neural network (ANN)-based prediction model was developed to estimate the thermal performance of SPHX energy piles. A computational fluid dynamics (CFD) model was formulated using in-situ thermal performance test (TPT) results, and a numerical database was established by considering various influential factors, such as the thermal conductivity of concrete and ground formations, the flow rate of the working fluid, and the initial temperature of the ground formations. These datasets were utilized to train the ANN model. The developed ANN model exhibited high accuracy in predicting the average heat exchange amount of SPHX energy piles, with an average error of 1.53 %. Furthermore, the model enabled the evaluation of the long-term thermal performance of SPHX energy piles based on the observed linear correlation between the average heat exchange amount and the operation time.