Performance evaluation of borehole thermal energy storage systems: A sensitivity, stochastic, and optimization approach

The performance of a Borehole Thermal Energy Storage (BTES) system is evaluated through sensitivity, stochastic, and optimization analyses. The assessment considers seven critical parameters spanning design, operational, and geological aspects. The combinations of these factors are strategically selected utilizing the Box-Behnken design of the experiments. Subsequently, a commercial simulation tool, which models double U-tube borehole heat exchangers (BHE), is employed to generate data on the BTES system's performance metrics such as roundtrip efficiency (RTE). The gathered data is used for training four types of surrogate models: response surface model (RSM), artificial neural network, support vector machine, and random forest. These models are capable of predicting the complex behaviors of the BTES system efficiently for several other combinations of those factors saving time and computation typically associated with three-dimensional simulations. Among surrogate models, the response surface model demonstrates the best performance in fitting the data. For stochastic analysis with RSM, one million BTES scenarios were generated using Monte Carlo simulations by combinations of the seven factors, revealing RTE values ranging from 11.1% to 98.5% over 10 years. The sensitivity analysis shows that spacing has the greatest impact on RTE, followed by volumetric flow rate ratio, BHE length, and then discharging temperature.