Selection and Calibration of a Hydromechanical Model for Discrete Fracture Networks

Abstract

ABSTRACT: As part of a preliminary evaluation of a deep geological repository for spent nuclear fuel, the hydraulic properties of a discrete fracture network were calibrated using published permeability data from comparable sparsely fractured crystalline rock. These permeability data exhibit two primary trends: a relatively gradual reduction with depth below c. 300 m and a marked increase above c. 300 m depth. Respectively, these are hypothesized to be caused by progressive fracture closure with increasing confining stress and periodic shear dilation of fractures proximal to stress criticality due to long-term natural fluctuations in and redistribution of stress. The links between bedrock stress, fracture geometry, and fracture hydraulic properties are widely recognized, and many numerical models describe these relationships. Several combinations of models were considered: four models for the relationship between fracture stress and hydraulic aperture, two models to identify critically stressed fractures, and two models to calculate critically stressed apertures. Appropriate calibration permitted these models to correlate well with both the major trends in the data and additional subtleties, indicating they can capture the interactions of the modelled natural processes. In contrast to empirical fitting of observed data, this suggests the models can produce predictive results. 1. INTRODUCTION Nuclear power offers a reliable and low carbon source of energy. However, the safe disposal of the spent fuel is a significant associated challenge. The primary option for achieving safe disposal is the emplacement of spent fuel in tunnel systems hundreds of meters underground. Several countries are building or evaluating repositories in high strength, low permeability rocks such as granite. However, bedrock fractures (or joints) form planes of weakness and potential conduits for fluid flow. Consequently, understanding the fracture systems is critical to assuring both construction/operational safety and long-term resilience of waste containment. Rather than modelling the groundwater flow system as continuous with bulk flow properties, the faults and fractures can be modelled explicitly as a discrete fracture network (DFN) to provide a more geometrically realistic framework for assessing interactions of groundwater with engineered barriers, individual groundwater pathways, and water-rock interactions. The explicit representation of fractures provides a more natural conceptual model for the integration of structural data (e.g., spacings and orientations of geologically mapped/imaged/logged fractures) with hydraulic measurements and thereby assess the spatial continuity, variability and anisotropy of flow. Additionally, by conceptualizing hydraulic properties in respect of fracture geometrical properties and rock stresses, mechanistic predictions can be made for different structural settings, depths, during construction and in response to changes in loading (e.g., glaciations). The fractures seen in boreholes only represent a tiny fraction of those within the volumes that need to be modelled, and hence the measured properties are used as a basis for inferring statistics and correlations. From this, plausible realizations of fractures in the surrounding bedrock are extrapolated using stochastic modelling, and hence predictions of flow transport made with a DFN are intrinsically probabilistic.