Technical Report NTB 92-04

This report represents the methodology and results of hydrogeological synthesis of crystalline rocks in Northern Switzerland. This synthesis concludes the regional investigation program KRI-I that was conducted by Nagra (Swiss National Cooperative for the Disposal of Radioactive Waste) during 1981 – 1993. The principal objective of this program is to assess the suitability of the crystalline basement as a host rock for a deep-seated repository for high-level and long-lived intermediate-level radioactive waste (HLW/ILW).

The present report provides a comprehensive hydrogeological characterization of the crystalline basement. The characterization study plays a key role in assessing the repository safety and performance, given that any release of radionuclides from a repository and further transport through the geological environment (geosphere) will occur by groundwater. The final objective of the characterization study is to provide a representative description of processes and parameters that control the flow system between the repository and the biosphere.

The hydrogeological characterization study of crystalline rocks was performed in 1991 – 1993 and included the following steps:

• compilation and a consistent interpretation of hydraulic borehole data
• development of the hydrogeological conceptual model of crystalline rocks
• analytic estimates of effective hydraulic properties of the defined conceptual units
• modeling and evaluation of regional flow
• modeling and evaluation of local-scale flow in a selected area of interest
• stochastic modeling of fracture flow at repository (block) scale and derivation of relevant input parameters for repository-safety assessments

The results of this multi-level synthesis study were presented in a series of Nagra internal reports. They all are integrated in this report which represents a complete background information on the hydrogeology of crystalline rocks in the basement of Northern Switzerland. The overall synthesis of all investigations performed in the scope of KRI-I program is presented in THURY et al. (1994).

Nagra's area of interest is a strip of crystalline basement between the deep Permocarboniferous Trough in the south and the German border in the north.

This area is characterized by Nagra's deep boreholes at the sites of Kaisten, Böttstein, Leuggern and Siblingen. Because the basement is covered by Mesozoic sediments throughout the investigation area, no direct surface observation is available. Instead, studies on crystalline outcrops in the adjacent Southern Black Forest are used to complement the scarce information from boreholes.

The first step of hydrogeological characterization is the development of a conceptual model that provides a simplified but consistent description of flow through crystalline rocks. Based on the synthesis of experimental data, the crystalline basement of the investigation area is subdivided into conceptual domains. Each domain is deemed to represent a homogeneous volume of rock that can be characterized by a single mean property. The combined framework of these domains accounts for the observed spatial heterogeneity of crystalline rocks. In terms of the large-scale flow, the basement is separated into large subvertical faults acting as water conduits and irregular blocks of relatively undisturbed rock. Based on generally observed tendency, the latter are further subdivided into an upper, relatively high-permeable crystalline unit and a lower, relatively low-permeable unit. An earlier geological synthesis suggested that potentially suitable crystalline rocks occur in two sub-regions of Northern Switzerland, identified as area West (Kaisten – Leuggern – Böttstein) and area East (Siblingen). These sub-regions are also characterized by different hydraulic properties. The lower crystalline unit in area West was identified as the potential host formation for a deep repository designated the Lower-permeability domain. In area East, which is characterized only by one borehole at Siblingen, no such crystalline-rock domain of similarly favorable properties was observed. The fundamental question is here, whether the data from the Siblingen borehole are representative for the area of interest, or whether they rather represent local hydrogeologic conditions that are specific for that site only. The conceptual model of sub-regional flow considers both alternatives as equally plausible.

At a much smaller scale of a potential repository (block scale), the flow through an undisturbed block of the low-permeable domain is controlled by a variety of discrete discontinuities. These planar features are identified as inflow points in hydraulic borehole tests and are generalized as transmissive elements in the conceptual model. They are characterized by statistical properties derived or inferred from borehole and outcrop observations.

The modeling of large-scale groundwater flow was used to confirm and improve the general understanding of the regional flow system in Northern Switzerland.

At this scale, the crystalline rocks were approximated by an equivalent-porous medium (EPM). Model results were compared with observed heads and hydrochemical evidence to test the plausibility of several conceptual hypotheses. The outcome is a set of concise statements, weighed hypotheses and alternatives that express the current conceptual understanding of the regional hydrodynamic system. The robustness of the simulated hydrodynamic system was also tested by inserting discrete faults and by including long-term scenarios that consider neotectonic changes in the topography and lateral displacements of river channels.

At a smaller scale, the local flow system in area West was evaluated by means of a hybrid or "double-porosity" model. This approach allows modelling explicitly the large water-conducting faults and the intervening blocks of undisturbed crystalline rocks, represented as EPM. The development of a new mesh generator was required to handle the complex geometry of arbitrarily oriented faults. A simplified geometric fault model designed by structural geology served as basis for the numerical grid. Multiple geometric variants of the hybrid model were implemented to cover the uncertainty related to the geometry and frequency of the hydraulically active faults. The two bounding patterns, referred to as the full scenario and sparse scenario, were assumed to cover the full range of possible model results and were subjected to a comprehensive analysis. The principal objective of the simulations is to evaluate the impact of major water-conducting faults on the local flow field in the intervening blocks of undisturbed crystalline rocks. The model results are evaluated from the perspective of needs for the repository safety analysis. The principal output consists in 1) the distribution of fluxes between the crystalline rock blocks and large faults, 2) the distribution of hydraulic gradients within the potential host rock as function of the fault frequency, and 3) the direction and path length of flow between the disposal location and the closest high-permeability medium. The simulated mean absolute gradients vary between 0.01 (full scenario) and 0.05 (sparse scenario). The direction and path length of flow within the low-permeable host rock depends on the size of the undisturbed block. It is shown that local flow systems are preserved only in large blocks. The comparison of model results with borehole observations suggests that not all faults in the geometrical network, such as identified by the structural geology, are hydraulically active. In other words, the conditions simulated in the sparse scenario (large blocks) are more plausible than those in the well-connected network of the full scenario.

The next level of modeling deals with characterization of flow through a typical block volume of low-permeable crystalline rocks surrounding a hypothetical repository drift. According to the conceptual model, the advective flow and transport at this scale is entirely controlled by conductive fractures, termed transmissive elements. Because the hydraulic and geometric properties of these features are not known explicitly, they are incorporated into a statistical framework. These are the discontinuum models that simulate the advective flow through a fractured medium, where the contribution of the matrix flow is negligible. The models are generated as complex stochastic networks based on statistical input data derived from borehole observations.

The simulation of fracture networks encompassing a hypothetical repository serves to provide the following input for repository-performance analysis:

• Frequency, trace length and transmissivity of intersected fractures. These parameters in combination with hydraulic gradients (obtained from the local-scale model) allow estimating the volumetric flow through a repository.
• Effective hydraulic conductivity of the modeled block. This parameter can be used as a feedback to characterize the block matrix in the hybrid models.

Evaluation of block-scale flow was performed for typical volumes of the lower crystalline rock unit in area West and East, with the final goal to provide quantitative input data for safety analysis calculations. The results are the water flow through a repository and its distribution among single transmissive elements, and the dilution potential of flow passing the repository within the near-surface aquifers and rivers in each area. Area East is less well-characterized and a substantial uncertainty is related to the evaluation of results at the block scale. Two separate datasets are derived for the safety analysis in area East, considering the two possible alternatives regarding the representativeness of Siblingen observations.