Skip to main content
SLU:s publikationsdatabas (SLUpub)

Sammanfattning

Field data of physical properties in heterogeneous crystalline bedrock, like fracture zones, fracture connectivity, matrix porosity and fracture aperture, is associated with uncertainty that can have a significant impact on the analysis of solute transport in dractured rock. The purpose of this study is to develop a performance assessment (PA) model for analyses of radionuclide transport in the geosphere, in which the model takes into account both the effect ot heterogeneities of hydrological and geochemical rock properties. By using a travel time description of radionuclide transport in rock fractures, we decompose the transport problem into a one-dimensional mass transfer problem along a distribution of transport pathways and a multi-dimensional flow problem in the fractured bedrock. The hydraulic/flow problem is solved based on a statistical discrete-fracture model (DFM) that represents the network of fractures around the repository and in the surrounding geosphere. A Monte Carlo technique reflects the fact that the representation of the fracture network is uncertain. If the flow residence time PDF exhitibts multiple peaks or in a another way shows a more erratic hydraulic response on the network scale, the three-dimensional travel time approach is superior to a one-dimensional transport modeling. Examples taken from SITE 94, a study performed by the Swedish Nuclear Power Inspectorate, showed that such cases can be found in safety assessments based on site data. The solute transport is formulated based on partial, differential equations and perturbations (random spatial variability in bedrock properties) are introduced in the coefficients to reflect an uncertainty of the exact appearance of the bedrock assodiated with the discrete data collection. The combined approach for water flow and solute transport, thereby, recognises an uncertainty in our knowledge in both 1) bedrock properties along individual pathways and 2) the distribution of pathways. Solutions to the central temporal moments of the residence time probability density function (PDF) fór solutes in both one and three dimensions are derived in closed form for a solute Dirac pulse. The solutions are based on a model that takes into account advection along the network of fracture planes, diffusion into the rock matrix and sorption kinetics in the rock matrix. The most relevant rock properties including fracture aperture and several matrix properties as well as flow verocity are assumed to be spatially random along transport pathways. The auto-covariance function representing the spatial variability in a rock property is also separated in terms of local variations, within individual fractures, and regional/global variations in a network of fractures. Analyses indicate that the regional/global variation probably dominates over the local variation due to the longer correlation lengths. This may have implications for planning of data collection, in whick it is likely that one should pay more attention to the large-scale variations in bedrock properties. However, site specific data (e.g. on the variance) is needed also on the single fracture-scale to be able to draw general conclusion on this issue. Furthermore, measured geochemical data from Äspö Hard Rock Laboratory in Sweden is used to exemplify how the discrete-fracture and the solute transport models are coupled. Experimental studies based on rock samples taken at Äspö Hard Rock Laboratory, reveal that crystalline bedrock can possess a marked heterogeneity of various physical and geochemical properties even on a fracture-scale. By inserting measurement values in the solutions we can conclude that the heterogeneity of the rock properties in single fractures contributes to increasing significantly both the variance and the skewness of the residence time probability density function for a pulse travelling in a fracture. The Äspö-data suggest that the bias introduced in the variance of the expected value of the residence time PDF for radionuclides by neglecting the heterogeneity of the rock properties is very large for fractures thinner than a few tenths of a millimetre. This bias would be even larger if the large-scale variation in bedrock properties on the network-scale was also accounted for

Publicerad i

SKI rapport
2004, nummer: 2004:14
Utgivare: SKI

SLU författare

UKÄ forskningsämne

Miljö- och naturvårdsvetenskap

Permanent länk till denna sida (URI)

https://res.slu.se/id/publ/4336