Technical Report NTB 93-14
Methodology for Deriving Hydrogelogical Input Parameters for Safety-Analysis Models - Application to Fractured Crystalline Rocks of Northern Switzerland
Switzerland is one of many nations with nuclear power that is seeking to identify rock types and locations that would be suitable for the underground disposal of nuclear waste. A common challenge among these programs is to provide engineering designers and safety analysts with a reasonably representative hydrogeological input dataset that synthesizes the relevant information from direct field observations as well as inferences and model results derived from those observations. Needed are estimates of the volumetric flux through a volume of rock and the distribution of that flux into discrete pathways between the repository zones and the biosphere. These fluxes are not directly measurable but must be derived based on understandings of the range of plausible hydrogeologic conditions expected at the location investigated.
The methodology described in this report utilizes conceptual and numerical models at various scales to derive the input dataset. The methodology incorporates an innovative approach, called the geometric approach, in which field observations and their associated uncertainty, together with a conceptual representation of those features that most significantly affect the groundwater flow regime, were rigorously applied to generate alternative possible realizations of hydrogeologic features in the geosphere. In this approach, the ranges in the output values directly reflect uncertainties in the input values.
As a demonstration, the methodology is applied to the derivation of the hydrogeological dataset for the crystalline basement of Northern Switzerland. The various steps are highlighted in the following.
The hydrogeological conceptual model of the groundwater flow system in the crystalline basement of Northern Switzerland includes descriptions of the hydrogeologic framework and groundwater flow. The crystalline rocks (block elements) are separated by major water-conducting faults. Block elements contain small-scale water-conducting features whose transmissive properties decrease with depth, resulting in an upper higher-permeability domain (about 500 m thick) and an underlying low-permeability domain. In the region, groundwater flows from the major recharge area in the southern Black Forest (Germany) to the principal discharge areas along the Rhine River. On a much smaller scale, groundwater flow in the low-permeability domain occurs principally through a complex network of discrete water-conducting features or transmissive elements.
Regional- and local-scale numerical models of groundwater flow were developed to describe infiltration and exfiltration areas and general flow paths from potential siting areas to the exfiltration regions. The regional-scale model supports the conceptual model of regional flow. The local-scale model utilizes a hybrid modeling approach, in which each of the principal hydrogeologic units is treated as an equivalent porous medium, and the major water-conducting faults are described explicitly.
Model results indicate that groundwater flow characteristics at the local scale depend significantly on the frequency of major water-conducting faults. Two scenarios were tested, one with a high frequency of faults (full scenario) and one with a low frequency of faults (sparse scenario). Distributions of hydraulic gradient were generated for each scenario. Modeling at the block scale, or repository-tunnel scale, was conducted to evaluate the distribution of flow through transmissive elements as a function of their geometric and hydraulic properties. The goal is to assess the expected range of flow path lengths and flow rates from hypothetical repository tunnels to the major water-conducting faults, which are assumed to determine the layout of a repository. Discrete inflow points observed in boreholes are assumed to represent transmissive elements, the principal paths for water flow in the low-permeability domain.
The numerical discontinuum model NAPSAC was used to integrate the statistical information from boreholes into a stochastic framework of discrete transmissive elements. The NAPSAC model allows for the incorporation of all relevant geometric and hydraulic properties of the transmissive elements that are believed to affect the distribution of groundwater fluxes at the block scale. Uncertainty is taken into account by generating multiple realizations of the transmissive-element distribution. From these realizations, the intersections of simulated transmissive elements with hypothetical repository tunnels was directly determined. The following distributions of geometric and hydraulic properties were obtained for transmissive elements intersecting a tunnel segment: number, trace length, transmissivity, conductance, and effective flow path and path length from the tunnel segment to the boundaries of the block.
A major unknown is the size distribution of transmissive elements. On the basis of sensitivity analysis, the geometric and hydraulic characteristics of transmissive elements were determined to be relatively insensitive to size distribution. Nonetheless, the overall pattern of flow within a block depends on the connectivity of the transmissive-element network, which is a function of size.
To evaluate groundwater flow at the block scale, the geometric approach was developed. In this approach, gradients derived from the hybrid local-scale model were multiplied directly by the individual transmissive-element conductance to generate a range of possible volumetric fluxes within each transmissive element. Specific parameters derived are flow through an individual transmissive element, normalized flow through individual transmissive elements, total flow through a 500-meter tunnel segment, and total flow through a repository. Results of the geometric approach were determined to be conservative on the basis of a comparison with results of dynamic flow simulations, or dynamic approach, using the same geometric network.
The geometric and hydrogeologic results were transformed to the actual input values used in the performance assessment of crystalline rocks of Northern Switzerland. The geometric approach differs from the standard hydrodynamic modeling approach in that the former assumes that the driving force (gradient), as determined from the local-scale model, is not influenced by the variability of hydraulic parameters at the block scale. An advantage of the geometric method is that it is simple to implement and it produces demonstrably conservative values for groundwater fluxes. This approach is thus a satisfactory trade-off between a more complex approach (with realistic but uncertain results) and a simple approach (with more certain – in a conservative sense – results).