Application of a discrete-feature model to assess geological uncertainties in a regulatory context

A discrete-feature model was used to integrate deterministic and stochastic geo-scientific information, to support the Swedish regulatory authorities review of SR-Can. The discrete-feature approach was used to replicate the main geometric features of the proponents models of the two candidate sites, including: (1) deformation zones on scales of 1 km to 10 km which are treated deterministically, (2) deterministic rock domains, (3) discrete fractures on scales of 2 m to 1 km which are treated stochastically, (4) excavation-disturbed zones around repository tunnels, where the tunnels are adapted to the deformation zones, and (5) deposition holes which are adapted to each realisation of the stochastic fractures, to avoid fractures for which mappable characteristics would indicate unacceptable seismic or hydrologic risks. Application of this approach in the SR-Can review provided independent estimates of key safety parameters including utilisation factors (the percentage of usable deposition-hole positions in a given layout), distribution of groundwater flow to canisters, and retention properties and release points for discharge paths to the biosphere. These results were provided as input to a compartment model (AMBER) which was used for both deterministic and probabilistic risk calculations. Discrete-feature model variants were used to scope uncertainties in deformation-zone properties and boundary conditions, as well as the potential for enhanced transport due to spalling of deposition holes at one of the sites which has a strongly anisotropic stress field. This application showed that a discrete-feature model with realistic complexity can provide independent evaluation of key geo-scientific factors for safety, in the regulatory context. (authors)