Heat recovery efficiency optimization of High-Temperature Aquifer Thermal Energy Storage system in naturally fractured reservoirs: A combined multi-physics modeling and regression prediction method

Abstract

High-Temperature Aquifer Thermal Energy Storage (HT-ATES) system holds significant potential for addressing the challenges of energy supply and demand management. However, the complexity of natural fracture networks within reservoirs poses a challenge for accurate simulations. In this work, a coupled thermal-hydraulic-mechanical (THM) model was developed to investigate the performance of HT-ATES system in naturally fractured reservoirs. The developed THM model was validated against analytical solutions, achieving mean errors of less than 4 %. The effects of varying parameters on heat recovery efficiency of HT-ATES system were examined. The results demonstrated that fracture number was the most influential factor on system performance, causing a variation in heat recovery efficiency of 38.58 %. More importantly, XGBoost algorithm was integrated with THM model to develop a surrogate model, which enabled the accurate regression prediction of the bottom-hole temperature/pressure evolutions as well as heat recovery efficiency, with five selected evaluation metrics demonstrating good performance, such as R² >0.99 and MAPE<0.3 %. Additionally, the developed surrogate model was able to effectively optimize the performance of HT-ATES system, with the optimal parameter combination having a maximum heat recovery efficiency of 69.8 %. This work provides insights into operational dynamics and contributes to efficiency improvement for HT-ATES system.