For the artificial ground freezing (AGF) projects in highly permeable formations, the effect of groundwater flow cannot be neglected. Based on the heat transfer and seepage theory in porous media with the finite element method, a fully coupled numerical model was established to simulate the changes of temperature field and groundwater flow field. Firstly, based on the classic analytical solution for the frozen temperature field, the model’s ability to solve phase change problems has been validated. In order to analyze the influences of different parameters on the closure time of the freezing wall, we performed the sensitivity analysis for three parameters of this numerical model. The analysis showed that, besides the head difference, the thermal conductivity of soil grain and pipe spacing are also the key factors that control the closure time of the frozen wall. Finally, a strengthening project of a metro tunnel with AGF method in South China was chosen as a field example. With the finite element software COMSOL Multiphysics^® (Stockholm, Sweden), a three-dimensional (3D) numerical model was set up to simulate the change of frozen temperature field and groundwater flow field in the project area as well as the freezing process within 50 days. The simulation results show that the freezing wall appears in an asymmetrical shape with horizontal groundwater flow normal to the axial of the tunnel. Along the groundwater flow direction, freezing wall forms slowly and on the upstream side the thickness of the frozen wall is thinner than that on the downstream side. The actual pipe spacing has an important influence on the temperature field and closure time of the frozen wall. The larger the actual pipe spacing is, the slower the closing process will be. Besides this, the calculation for the average temperature of freezing body (not yet in the form of a wall) shows that the average temperature change of the freezing body coincides with that of the main frozen pipes with the same trend.

Compared to porous media, fractured aquifers are generally characterized by a more pronounced hydraulic heterogeneity. To describe hydraulic properties of fractured subsurface, investigation methods such as hydraulic tests, tracer tests and hydrogeophysical tests have been widely used. In recent years, thermal tracer tests are obtaining more attention because thermal response signals can be easily and economically obtained at a high resolution, e.g. using distributed temperature sensing (DTS) systems. Some studies have even employed the travel-time-based thermal tracer tomography (TTT) to reconstruct the aquifer heterogeneity (Somogyvári M. et al., 2016; Somogyvári M. and Bayer P., 2017). In this study, we further develop and apply the TTT method for a field scale investigation of the hydraulic properties at a geothermal test site in Göttingen, Germany, equipped with five instrumented experimental wells. Presently, using travel-time-based thermal tracer tomography to describe the hydraulic connectivity or conductivity is limited to the condition that the heat transfer must be convection dominated. Thus, the field experiments have to be divided into two steps. A full length well warm water injection test is firstly conducted to obtain information about the basic hydrogeological conditions, such as the fracture insertion depth and the connectivity between the wells. Subsequently, four multilevel thermal tracer tests are performed. The temperature changes in all five wells are recorded using a DTS system. Finally, based on the travel-time-based inversion method, the hydraulic conductivity distribution of the fractured aquifer can be obtained. Preliminary test results showed that the orientation of transmissive fractures is mainly along the E-W direction at our test site. Given the good hydraulic connectivity, the first thermal tracer tomographical tests in a fractured aquifer were performed between two wells positioned along this direction. As next, we will work on the reconstruction of the fracture distribution between those two wells.

Shallow geothermal energy has been popularly used for house heating and cooling by combining borehole heat exchangers (BHEs) with a heat pump. The efficiency and sustainability of BHE systems are receiving increasing attention in recent years. Groundwater flow, among the hydrogeothermal properties of the subsurface, is considered to be a critical positive factor in maintaining BHE efficiency, especially when the groundwater flux is greater than 10-7 m/s. However, in practice, exact information on local groundwater flow is typically not available. Rough estimates may introduce significant uncertainty in the design an performance of BHE systems. In this study, we implemented an uncertainty analysis of groundwater flow impacts on the seasonal coefficient of performance for a heat pump (SCOP). We first built a numerical model based on an existing geothermal site using FEFLOW. Groundwater flow parameters including groundwater level, flow direction, and groundwater flux within the layered subsurface were assumed to follow a Gaussian distribution in which the mean value is determined from regional hydrogeological conditions. Afterward, we selected the input data sets about groundwater flow by using Sobol sequence sampling, and then employed the selected data sets and the corresponding simulated SCOP in a random forest regression for training and predicting. Preliminary results show that groundwater flux and flow direction have a significant impact on SCOP. The impact of groundwater flow direction is strongly correlated with the arrangement of the BHEs.