Reliability of Stress Induced Damage Predictions in Hard Rocks With Continuum and Discontinuum Numerical Modelling Approaches
Deep Geological Repository , Excavation Damage Zone , Brittle Fracture , Discrete Element Method , Numerical Modeling , Reliability Analysis
The use of underground repositories known as deep geological repositories (DGR) is globally accepted to be the best solution for long term storage of nuclear waste. In combination with different types of engineered barriers, the final layer is the host rock providing a natural geological barrier to radio nuclide migration. In the context of the Canadian nuclear waste management efforts, the emphasis has been on sedimentary and crystalline rock formations to host the DGR. Construction of the underground spaces within such hard rocks induces irreversible brittle damage around the excavation. The prediction of the Excavation Damage Zone (EDZ) is paramount for the design process of DGR, since it can provide a potential leakage pathway for contaminants. This thesis includes a discontinuum and continuum numerical study to improve the understanding of the fracture development process associated with EDZ in massive rock masses. A two-dimensional discontinuum grain-based modeling (GBM) approach is implemented to characterize the mechanics of EDZ by explicitly simulating the process of fracture initiation and propagation. While the potential of GBMs in capturing field scale behavior is largely unexplored, this study bridges the gap through the development of a new methodology for generation and calibration of large scale models, which when applied narrows the computation efforts. This research confirms the nature of key subdivisions within the generalized EDZ, which includes four zones depending on the intensity of damage, and provides new insights into fracture evolution behavior within these zones. Key impacts of rock type and stress field on the micro mechanics of crack damage within EDZ are illustrated. Capabilities and limitations of the continuum based modeling approaches to predict discontinuum nature of brittle failure are addressed. The predictive numerical tools implemented in this study rely firmly on rockmass strength properties, which are always subject to a great deal of uncertainty. Such uncertainty can pose a challenge for obtaining reliable predictions. Given the low-risk tolerance for DGR, it is essential to quantitatively describe the impact of these uncertainties on predicted EDZ. A reliability-based approach is implemented with both discontinuum and continuum models to address the issue of uncertainty in rock mass strength.