Abstract: Estimation of geometrical and hydraulic properties of rock fracture networks is of great significance for underground engineering construction and environment safety. Flow tests were conducted for 3D printed fracture network specimens in this study, in which each fracture had different orientations, lengths, rough surfaces and heterogeneous apertures. A numerical procedure was developed to simulate the fluid flow through the fracture network with the same geometrical properties as the specimen. Effects of surface roughness, aperture, hydraulic gradient and flow direction on the fluid flow through fracture networks were systematically investigated. Results show that as the flow rate increases, the relationship between the pressure gradient and the flux through the fracture network transits from a linear relationship to a nonlinear one, and the nonlinear relationship can be well fitted by Forchheimer’s law. The critical hydraulic gradient in different flow directions ranges from 0.015 to 0.195. The ratio of the pressure drop induced by the inertia force to the total pressure drop increases with the increasing hydraulic gradient, while the rate of increase gradually decreases. When the hydraulic gradient is close to 1.0, the pressure drop induced by the inertial force accounts for 68.5% of the total pressure drop. The topology of the fracture network determines the overall connectivity of the model, and the heterogeneous aperture allows the fluid to preferentially flow through the channels with greater permeability and smaller resistance within the connected fractures. Therefore, the ratio of areas of the main flow channels to the total area of fracture planes is smaller than 41%. The permeability of fracture networks decreases with the increase of the surface roughness of the fracture, but the decrement is reduced by increasing the fracture aperture. This study provides reliable laboratorial and numerical methods to analyze flow properties through rough fracture networks.

Key words: rock mechanics, fracture network, 3D printing, fluid flow, surface roughness, aperture heterogeneity