Improved methods for characterizing fractured media have led to detection of fractures at multiple scales. As fluid flow is highly influenced by the fractures, appropriate modeling of flow in fractured porous media is essential for accurate prediction of the energy extraction in geothermal reservoirs. However, an exact fine-scale model where all fractures are resolved is not feasible from a computational perspective. As a remedy we present an efficient multiscale method, based on representing flow in large-scale fractures explicitly in the computational model, whereas flow in small-scale fractures and the porous media is upscaled. In contrast to traditional upscaling approaches we keep the link to the fine-scale model and can thus compute approximate fine-scale solutions, which can be utilized in fine-scale heat transport simulations. Furthermore, the accuracy of the approximate solution can be improved by applying the upscaled model as a preconditioner in a convergent iterative framework. Our methodology is demonstrated by considering synthetic examples mimicking an EGS reservoir, based on data on hydraulic aperture, mean length and number of fractures corresponding to different fracture types identified at the EGS Soultz Site (Alsace, France), involving both deterministic and stochastic realizations of fractures at multiple scales. The quality of the approximate solution is tested by considering heat transfer on the fine-scale as well as using a novel upscaled transport models resembling the MINC approach, but featuring improved flux calculations. Most notably, the transfer functions between the continua are explicitly calculated using the fine-scale description, and no heuristic terms appears in the model.
展开▼