Implications of rock structure on the performance in the near field of a nuclear waste repository

The performance of a deep repository for nuclear waste relies heavily on its potential to retain radionuclides at source. This is achieved through a durable waste package and favourable chemical conditions under which the most dangerous nuclides are extremely difficult to dissolve in groundwater. The persistence of these favourable conditions is, however, partially controlled by the groundwater flow in the near field and its capability to transport radionuclides released from the waste form. Most source-term models used in performance assessment assume a simple conceptual model, with a deposition hole intersected by a single fracture with the release controlled by diffusion in the buffer and the chemical conditions at the source. This conceptual model is certainly a simplification of reality, where the deposition hole is intersected by an irregular network of discrete fractures with spatially varying properties and where the fractures close to the deposition hole and the tunnel are altered by excavation effects. In terms of flow modelling, spatially varying discrete fracture systems can be simulated using a number of methods including stochastic discrete fracture networks models and stochastic continuum models. This choice of models leads to uncertainty, as does incomplete knowledge of the parameters that characterize the chosen model. For modelling migration through the geosphere, the largest uncertainties are those in groundwater flow distribution and in the flow-wetted surface area. Sensitivity analyses with existing source-term models suggest that near-field release is sensitive to flow and to fracture geometry in an intermediate range only, but this result may be an artefact due to the simplifications made. The adoption of a more realistic geometrical decription of the near-field rock may lead more useful description of which near-field rock properties affect source-term release.