Clay and granitic units are potential host media for future repositories for used nuclear fuel. The report addresses the representation and characterization of flow in these two media within numerical process models. In low permeability crystalline rock, flow is primarily in relatively sparse networks of fractures. Discrete fracture network (DFNs) models are an approach to representing flow in fractured rock that explicitly represents the geometry and flow properties of individual fractures. A new approach places constraints on the generation of the DFN, and permits creation of high-quality computational meshes that enable accurate flow and transport solutions on the generated DFNs. The DFN capability is extended for use of highly parallel software to solve for groundwater flow on the unstructured DFN meshes and to combine DFN meshes with traditional volume meshes representing backfilled tunnels or boreholes. A new approach to the tracking of particles representing radionuclide mass through the DFN, is also described.
Water flow in clay media is another important process. The observed relationship between water flux and hydraulic gradient can be highly non-linear. To capture this non-Darcian flow behavior, a new relationship between water flux and hydraulic gradient (summarized in a previous report) is shown to be consistent with experimental observations for both saturated and unsaturated conditions.
Flow in the excavation damage zone of a generic repository is shown, as the result of non-Darcy flow, to be extremely small such that solute transport is diffusion dominated rather than advection dominated. It is also shown that measurements of relative permeability in unsaturated conditions may be significantly in error unless the non-Darcian flow behavior is taken into account.
Data from the Bedrichov Tunnel Experiment in the Czech Republic are evaluated as are available data from the KURT generic underground research laboratory in Korea. Additionally, information derived from a literature survey are summarized on upscaling selected properties to the spatial scales relevant for performance assessment modeling.