A long-term operation methodology is proposed for hybrid ground source heat pump (GSHP) systems combined with cooling towers. In this methodology, the calculation conditions, which are the apparent effective thermal conductivity of the ground, cooling/heating load of the GSHP system, and the heat pump COP, are modified by using measurement data. Then, the long-term temperature variation in the ground surrounding the ground heat exchangers (GHEs) is predicted through a simulation, and the optimum control parameter (Set temperature of the cooling tower operation) is determined. The proposed methodology was applied to a hybrid GSHP system installed in an office building. From the results, it was concluded that the operation without the cooling tower operation was optimal. This was because the estimated apparent effective thermal conductivity of 4.5 W/(m・K) was large and the maximum cooling load was smaller than that predicted in the design phase. The result of monitoring for 8 years showed that although the total heat injection rate of GHEs was approximately five times the total heat extraction rate, the long-term increase in the inlet temperature of the primary side of the heat pump was 1-2℃ and the inlet temperature was less than 30℃. In addition, the long-term variation in the measured inlet temperature was similar to that of the inlet temperature calculated by the simulation. Therefore, it was confirmed that the proposed methodology was effective.

The scale of ground heat exchangers, where there is groundwater flow, can be scaled by considering the heat transfer by groundwater flow in the design stage of the shallow geothermal system. The calculation of the temperature of ground heat exchangers field considering groundwater advection is one of the most noteworthy challenges in the simulation of shallow geothermal systems. This paper proposes a method to calculate the temperature of ground heat exchangers field with groundwater flow. The method calculates the temperature response function of the ground heat exchangers field with groundwater flow, which has not been solved theoretically, by using an artificial neural network. For the advection–diffusion problem in the potential flow field around an infinite circular cylinder (moving infinite cylindrical source problem), a parameter study was conducted using numerical calculations by the finite volume method. The results of the calculations were learned by an artificial neural network, and the artificial neural network regression model reproduced the temperature response function of the moving infinite cylindrical source problem. The results of the trained artificial neural network regression model in comaprison with the numerical calculation results were demonstrated 9.67 × 10^-4 RMSE , 0.0212 maximum absolute error, and 7.76 % maximum relative error in dimensionless temperature. The time required to calculate the hourly temperature response function for 20 years using the artificial neural network regression model was 54.9 s , which indicates a significant reduction in computation time compared with the numerical analysis method.