The flow through fractured rock has been intensively studied in recent decades, in part due to plans in many countries to site repositories for high level nuclear waste in deep geologic formations. These systems are often fractured and allow water to flow through a network of connected conduits. Most model-based studies of flow in crystalline fractured rock adopt a discrete fracture network concept. Such studies often employ simplifications, including the assumption that each fracture is represented with effectively homogenous properties within the surface of individual fractures. Therefore, it is of interest to study flow through realistic rough-surfaced fractures to further develop understanding of the applicability of homogenisation assumptions at the single fracture scale.
This study uses 3D printing to create different fracture samples representing both smooth-surfaced constant (effectively homogenous) and rough-surfaced varying aperture fractures. The use of 3D printing allows for the identical geometries for both the fracture sample and numerical mesh to be prepared. Laboratory flow tests over a range of hydraulic gradients were conducted through the printed samples. Discharge was numerically simulated both using the cubic law homogenisation assumption as well as the Navier Stokes equations using a numerical CFD solver. Inertial forces influence the discharge and have more importance on varying aperture geometry, therefore the transition to non darcian flow in constant apertures occurs at a higher hydraulic gradient. The predicted discharge from the CFD solution was within 5% of the experimental discharge through the varying aperture sample for hydraulic gradients of 0.01 and 0.05. Thereby, this study demonstrates that 3D printing can be a useful tool for accurately reproducing rough-surfaced fractures for the purpose of conducting flow experiments, and specifically for evaluating applicability of the Darcy flow regime in rough surfaced discrete fractures in a consistent computational – experimental framework.