Modeling and Control of Heat Transfer in a Single Vertical Ground Heat Exchanger for a Geothermal Heat Pump System

The Ground Heat Exchanger (GHE) is regarded as the most critical component of a Geothermal Heat Pump System (GHPS) due to its direct contact with the Earth and ability to benefit from its relatively steady temperature. A GHE can attain the maximum benefit from the Earths heat when the water temperature reaches the ground, which occurs when the water velocity is moderate, allowing the heat exchanger to balance. Optimally, controlling water velocity is crucial such that the water reaches the desired temperature. Therefore, in this dissertation, I proposed a novel mathematical model of heat transfer behavior between the water inside the underground pipe and the surrounding ground for heating and cooling modes in a GHE. The proposed dynamic model was applied to three case studies of GHPS at Oakland University, the University Politècnica de València, and Oklahoma State University in heating and cooling modes to assess its validity and further enhance the performance of the GHE by determining the optimum velocity range. The results revealed the optimal water velocity ranges for three GHPSs. Model Predictive Control (MPC) was designed to optimize the GHE’s output temperature by controlling the water velocity, which can reduce the power consumption used for the water circulation pump and therefore maximize efficiency. In this dissertation, I also introduced an Improved Grey Prediction Model (IGM (1,1)). The proposed IGM (1,1) model was based on optimizing the current predicted value by subtracting the error prediction between the previous accumulated time response of the GM (1,1) model and the previous background value throughout the prediction length. The IGM (1,1) model was applied to perform the GHPS’s output temperature prediction eight hours in advance at Oklahoma State University, the University Politècnica de València, and Oakland University, respectively. Thus, the IGM (1,1) model outperformed the traditional GM (1,1) model for all used datasets.