Development of a Tool to Calibrate Vertical Ground Heat Exchanger Model Using Optimization

This paper presents a calibration process and its tool of the vertical ground heat exchanger model used in a building energy simulation program, Energyplus. To adequately analyze the performance of the system, calibration of the system model is crucial. The calibration procedure is to estimate input data of the simulation that match the results of the simulation with measured data by an inverse method. The vertical ground loop heat exchanger consists of ground and borehole systems. The thermal properties of the borehole system usually can be found from manufacturer’s data. However, the thermal property of the ground is hard to evaluate. In this paper, an evaluation tool of the thermal properties of the ground around the borehole is developed using Matlab. This tool consists of three submodels. The first one is a G-function curve fit model which represents the relationship between variation of thermal conductivity and g-function values. The second model is the vertical ground loop heat exchanger model which predicts the return water temperature from a ground loop heat exchanger using the short time response factor method. The vertical ground loop heat exchanger model in Energyplus is converted to Matlab code and integrated into the calibration model for this research. The last sub-model is the optimization model that uses the Nelder and Mead simplex optimization scheme to find parameters which minimize the difference between the simulation results and the field measurement data. This tool estimates the ground thermal propertiesusing an optimization scheme based on data collected from field measurement. Far field ground temperature and the ground thermal conductivity are estimated to be used as input data of the vertical ground loop heat exchanger model in Energyplus. This program is validated using a case study which is performed for an actual building, ZOE which is located in the University of North Texas and its system. 2 weeks’ measurement data were compared with the simulation result. The average deviation between the simulation result and measurement data for 2 weeks is 0.27 °C.