Résumé
Geothermal energy harvesting requires extraction and injection of geothermal fluid. Doing so in an optimal way requires a quantitative understanding of site-specific heat transfer between geothermal fluid and the ambient rock. We develop a heat transfer particle-tracking approach to model that interaction. Fracture-network models of heat transfer in fractured rock explicitly account for the presence of individual fractures, ambient rock matrix, and fracture-matrix interfaces. Computational domains of such models span the meter scale, whereas fracture apertures are on the millimeter scale. The computations needed to model these multi-scale phenomenon can be prohibitively expensive, even for methods using nonuniform meshes. Our approach appreciably decreases the computational costs. Current particle-tracking methods usually assume both infinite matrix and one-dimensional (1D) heat transfer in the matrix blocks. They rely on 1D analytical solutions for heat transfer in a single fracture, which can lead to large predictive errors. Our two-dimensional (2D) heat transfer simulation algorithm is mesh-free and takes into account both longitudinal and transversal heat conduction in the matrix. It uses a probabilistic model to transfer particle to the appropriate neighboring fracture unless it returns to the fracture of origin or remains in the matrix. We use this approach to look at the impact of a fracture-network topology (e.g. the importance of smaller scale fractures), as well as the matrix block distribution on the heat transport in heterogeneous fractured rocks.