Future Groundwater Resources at Risk (Proceedings of the Helsinki Conference, June 1994). IAHS Publ. no. 222, 1994. 37 Assessment of releases from a nuclear waste repository in crystalline rock AIMO HAUTOJÀRVI & TIMO VBENO Technical Research Centre of Finland (VTT), PO Box 208, SF-02151, Espoo, Finland Abstract Spent fuel from the Olkiluoto nuclear power plant in Finland is planned to be disposed of in a repository to be constructed at a depth of about 500 m in crystalline bedrock. When the disposal canister is damaged, radionuclides may escape from the repository and eventually reach the biosphere by dissolution in the groundwater and transport via fractures in the rock. The main mechanisms affecting the migration of dissolved species in crystalline rock are advection, hydrodynamic dispersion, diffusion into the microfissures and pores of the rock matrix surrounding water-conducting fractures, and sorption on fracture surfaces and fillings as well as on the inner surfaces in the rock matrix. The effect of preferential pathways, or channelling, is taken into account in the modelling. The outcome from the TVO-92 safety analysis is that even with the concept of strongly channelled flow of groundwater, the geosphere significantly lowers the release rates of radionuclides from the repository into the biosphere. INTRODUCTION According to present estimates, a total of 1840 TU of spent fuel will be accumulated during the assumed 40-year lifetime of the TVO I and TVOII reactors at the Olkiluoto nuclear power plant in Finland. Before disposal the spent fuel will be cooled at least 15 years in an interim store which has operated at Olkiluoto since 1987. The spent fuel will be encapsulated in composite copper and steel canisters in a facility built on the site where the repository is located. The Advanced Cold Process (ACP) canister design consists of an inner container of steel as a load-bearing element and an outer container of copper to provide a shield against corrosion. The repository will be excavated in crystalline bedrock at the depth of about 500 m. Canisters will be emplaced in vertical holes drilled in the floors of horizontal deposition tunnels. The annulus between the canister and the rock is filled with compacted bentonite to ensure favourable physical, chemical and hydraulic conditions for long periods of time. The TVO-92 safety analysis by Vieno et al. (1992) of spent fuel disposal is based on updated technical plans for disposal presented by Teollisuuden Voima Oy (TVO) (1992a) and on the results of the preliminary site investigations carried out by TVO (1992b) at five sites in Finland between 1987 and 1992. All the areas consist of Precambrian crystalline bedrock. Investigation program for each area consisted of geological, geophysical, geohydrological and geochemical studies and measurements conducted both on the surface and deep in the boreholes. Field investigations were followed by interpretation and evaluation work.
38 Aimo Hautojàrvi & Timo Vieno FLOW OF GROUNDWATER Detailed groundwater flow analyses have been performed for all five areas as a part of site characterization studies. The aim has been to evaluate the present-day, natural regime of groundwater flow and, along with the studies on geology and geochemistry, to provide a comprehensive and reliable picture of the characteristics of the sites. The groundwater flow modelling was based on the continuous porous medium approach which is the only practicable method for large scale modelling. Fracture zones were modelled by two-dimensional structures between the three-dimensional elements representing the rock matrix as described by Taivassalo et al. (1993). The results of the groundwater analyses show that groundwater flow deep in the bedrock is minute and concentrates in fracture zones with high hydraulic conductivity. Outside the fracture zones, the average flux in the rock at a depth of 500 m is 0.01 to 0.1 1 m' 2 a" 1. The results are compatible with the observation that the hydraulic gradient deep in the Finnish bedrock is typically about 1% (10 m km" 1 ) or less, and that the average hydraulic conductivity of intact rock is about 10" 10 m s" 1 at all investigation sites. The flow rates in the fracture zones are governed by their transmissivity, which varies a great deal. At a depth of 500 m, the transmissivity of local fracture zones ranges typically from 10" 7 to 10" 5 m 2 s" 1. If the gradient is 1%, the flow rate across the entire width of the fracture zone is approx. 0.03 to 3 m 3 a" 1 for each metre perpendicular to the gradient. In the safety analysis, we need to evaluate the groundwater flow situation in the repository itself and in the surrounding rock as well as flow paths from the repository into the biosphere. The safety analysis much focuses on unfavourable extreme cases. The tunnels and shafts of the repository, as well as the disturbed rock zone formed around them as a consequence of excavation and heat generation may significantly affect groundwater flow in the vicinity of the repository. If the hydraulic conductivity of the repository is high and it is intersected by two or more conductive fracture zones, a U-tube flow situation may be formed. Moreover, due allowance must be made for the possibility that the boundary conditions affecting the present flow regime may undergo major changes in the long term due to e.g. glaciations. The groundwater flow calculations carried out by Koskinen et al. (1993) in the safety analysis were made assuming that the repository would be located at the Veitsivaara site. This made it possible to make due allowance for the true threedimensional structure of the bedrock at a real site. Veitsivaara was selected for the reference site because at the time when the analysis was begun in the spring of 1991, the site investigations and their reporting had proceeded furthest at the Romuvaara and Veitsivaara sites. Of these two alternatives, Veitsivaara was selected because it was considered as a "challenging" target owing to its geometrically complex array of fracture zones. The repository was intentionally placed in a location were it was intersected by two fracture zones. The transmissivities of the repository (including that of the disturbed rock zone) and the intersecting fracture zones were varied in a wide range in the sensitivity analyses. On the basis of these analyses, conservative conceptual models and data were selected for the release and transport analyses of radionuclides. Also other data used in the safety analysis has been chosen in such a way that it covers the conditions at all the five sites.
Assessment of releases from a nuclear waste repository 39 RELEASE FROM THE REPOSITORY INTO THE ROCK If the conditions in the geosphere in the vicinity of the repository do not change drastically and if no major disruptive event hits the repository, the copper-steel canisters remain intact for millions of years and no significant amount of radioactive substances will ever escape from the repository. However, in order to analyze the performance of the other parts of the multibarrier system consisting of engineered and natural release barriers, the canister was assumed to simply "disappear" in the reference scenario of the safety analysis. In the deposition holes canisters are surrounded by compacted bentonite which swells when absorbing water. The hydraulic conductivity of compacted bentonite is very low and the dominant transport mechanism in it is diffusion. Release rates of radionuclides from a damaged canister into the rock are affected by several factors, the most important ones being the fraction of radionuclides in the fuel-cladding gap of the fuel rods, the degradation rate of the fuel matrix, solubilities of elements, diffusion and retardation of nuclides in the bentonite buffer, and the flow of groundwater in the rock around the deposition hole. On the basis of scoping analyses where steady-state as well as transient mass flows in the deposition hole were examined, a greatly simplified conceptual near-field transport model was constructed for the reference scenario (Fig. 1): The physical containment provided by the canister is assumed to disappear 10 000 years after the sealing of the repository. The "disappearing" time of the canister is varied in the sensitivity analyses. Radionuclides are released from the fuel and other parts of the fuel assembly directly into a thin layer of bentonite at the level of the upper end of the canister. From the source area nuclides can diffuse upward and downward in the bentonite. There is 3.7 m of bentonite below and 1.5 m above the source compartment. The diameter of the deposition hole is 1.5 m. The interface between the near- and far-field models is at the upper end of the bentonite, at a distance of 1.5 m from the source compartment. A zeroconcentration boundary condition is applied at the interface, and the radionuclides zero concentration boundary at the interface of near-field and far-field 1.5 m bentonite 3.7 m bentonite reflecting boundary Fig. 1 The conceptual near-field transport model.
40 Aimo Hautojàrvi & Timo Vieno are assumed to be transferred without any resistance into the groundwater flowing in the disturbed rock zone below the tunnel. This assumption corresponds to a case where there is a strong flow of groundwater in the disturbed rock zone below the tunnel floor. - The other boundaries of the deposition hole are assumed to be reflecting for radionuclides. The above conceptual model and the conservative data (D e in the bentonite is 1 10" 10 m 2 s" 1 m applied in the reference scenario result in that the steady-state mass low from the deposition hole into the rock is equivalent to a flow rate of 3.7 1 a" 1. MIGRATION IN THE GEOSPHERE The most important phenomena affecting the migration of dissolved species in fractured rock are: the flow of the groundwater in the fracture system, sorption on fracture surfaces and fillings, and matrix diffusion, i.e. diffusion from the water-conducting fracture into the microfissures and pores of the surrounding rock, reviewed recently by Valkiainen (1992). The channelling phenomenon, which may result in that a large proportion of the water moving in the rock is flowing along narrow channels in few, well-conducting fractures, may significantly speed up migration of species in crystalline rock. In such a case, the velocity of the groundwater in the channels is high and the surface area available for sorption and matrix diffusion is small. Channelling and the existence of fast flow paths depend greatly on the local heterogeneity of the rock and are difficult to investigate. In TVO-92, the conceptual model of migration of radionuclides in the geosphere is based on an assumption of strongly channelled flow of groundwater. Matrix diffusion is the only dispersing and retarding phenomenon taken into account. According to the general principles of the safety analysis, we are mainly interested in the fastest transport routes from the repository into the biosphere. When analysing the migration of radionuclides on these fast routes, we intentionally ignore longitudinal dispersion, diffusion into the stagnant pools of water in the fracture, and sorption in fracture fillings. The rock matrix around the water-conducting fracture is modelled as a porous medium. The migration of a stable specie along the fracture can hence be described by a pair of equations which are analogous to the equations describing the conduction of heat from a flowing fluid into a solid material. For the case where a constant water phase concentration C 0 is prevailing at the inlet of the fracture beginning from t = 0, we get the water phase concentration at the distance of L in the fracture C f (L,t) = C 0 erfc[wr 1/2 ] (*) with u as a parameter describing the transport properties of the flow system where
Assessment of releases from a nuclear waste repository 41 e is the porosity of the rock matrix (-), D e is the effective diffusion coefficient from the fracture into the rock matrix (m 2 s" 1 ), R is the retardation factor of the nuclide in the rock matrix (-), W is the total width of the flow channels per rock area (m m" 2 ), q is the flux of the groundwater (Darcy velocity) in the rock (m 3 mf 2 a" 1 ). The transport of a dissolved species is reduced when u is increased. For a non-sorbing species R p = 1, and for a moderately or strongly sorbing species R p ~ K d pje p, where K d is the volume-based distribution coefficient and p s is the density of the rock. We hence get u = \e D 1 1/2 WL (3) non-sorbing ip e \ "nnn-qnrhincr e n U A V-'J q ~, r ^ H/2 sorbing «[D e K dp^ *L (4) H The porosity of the rock matrix, e p, has thus not a very significant effect on the migration of a sorbing species, the influence being incorporated in the diffusion coefficient D e. From the point of view of migration of dissolved species, the most important (but unfortunately rather poorly known) quantity describing the transport properties of a flow path in fractured rock is WL WL t w ~q vw2b v ~2b~ v where W is the total width of the flow channels per rock area (m mf 2 ), L is the length of the flow path (m), q is the flux (Darcy velocity) of the groundwater (m 3 m~ 2 a" 1 ), v is the average velocity of the groundwater flowing in the fractures (ma -1 ), 2b v is the average volume aperture of the fractures (m), t w is the groundwater transit time (a). The higher the transport resistance WLIq, the weaker is the transport of dissolved species. WL is the "flow wetted surface", i.e. the rock surface in direct contact with groundwater flowing in the channels; dispersion and retardation due to the matrix diffusion is enhanced with an increasing WL. On the basis of the results of the groundwater analysis and estimated channelling properties (the W parameter) of the virgin rock, the disturbed rock zone below the tunnel floor, and fracture zones, the total transport resistance WLIq of the flow path from the repository into the biosphere is taken to be 10 4 a m" 1 in the reference scenario. This corresponds, for example, to a situation where the canister is located next to a fracture zone intersecting the repository and the flow is directed upward in the fracture zone (Fig. 2). In the migration codes, we cannot use WLIq directly as an input parameter because the codes employ more "conventional" parameters such as the length of the flow pathline L, the velocity of the groundwater v, and the aperture of the fractures 2b v. For the TVO-92 safety analysis, these parameters have been fixed in such a way that we get the chosen value for our primary input parameter WLIq:
42 Aimo Hautojàrvi & Timo Vieno rock zone fracture zone Fig. 2 An example of a fast transport route from repository into the biosphere. - the length of the flow pathline L = 400 m, - the velocity of the groundwater v = 80 m a" 1, - the volume aperture of the fractures 2b v = 0.5 mm. These data produce WL/q = 10 4 a m" 1 and a groundwater transit time t w = 5 years (cf. equation (5)). The value of the transport resistance is varied in the sensitivity analyses which include also a "short-circuit" case where WL/q = 240 3 a nr 1 the corresponding transit time of the groundwater being no more than one year. The groundwater transit times applied in the migration analysis are very short when compared to the estimated mean ages of deep groundwaters which are typically in the order of thousands or tens of thousands of years according to Lampén & Snellman (1993). There are, however, significant uncertainties in estimating the actual age of the groundwater using, for instance, the carbon-14 method. Furthermore, the groundwater analyses often give indications of mixing of groundwaters of different origins and ages. The potential existence of fast flow paths is further supported by the fact that traces originating from the Chernobyl fallout have been observed by Ittner et al. (1991) at a depth of over 100 m in packed-off sections of an artesian borehole. DISCUSSION Heterogeneous properties of the fracture system and the fractures themselves may result in that a major part of the groundwater flow takes place in a few fractures and in small parts of the fracture volumes. In a sparse system of flow channels the pathways do not interact with each other within reasonably short transport distances. Different channels may have, of course, very different transport properties. Within a channel the main mechanisms affecting the transport of dissolved species are advection, dispersion, matrix diffusion and chemical interactions. In the TVO-92 safety analysis, the conceptual model of transport in the geosphere is based on the assumption of a strongly channelled flow of groundwater. The
Assessment of releases from a nuclear waste repository 43 migration of radionuclides has been described in terms of simple one-dimensional paths. Only matrix diffusion has been taken into account as dispersing and retarding phenomenon. The effect of matrix diffusion depends on the intrinsic properties of the rock matrix [~ (D e K d p s ) in ] and even stronger on the flow parameters [~ WL/q]. Due to the large uncertainties associated to the channelling phenomenon and the characteristics of groundwater flow in the fracture system, it is presently not possible to exclude the existence of a few extremely fast flow paths from the repository into the biosphere or into a well drilled in the vicinity of the repository. Very slow and long flow paths also exist, but they are not very interesting from the safety analysis point of view because actually "nothing" is transported into the biosphere through them. The outcome from the TVO-92 safety analysis is that even with the concept of strongly channelled flow, the geosphere has a significant effect on the release rates of moderately or strongly sorbing radionuclides into the biosphere. Dispersion caused by matrix diffusion lowers also peak release rates of weakly sorbing nuclides released rapidly from the fuel-cladding gaps of the fuel rods in a damaged canister. The overall results of the safety analysis attest that the planned disposal system fulfils the safety requirements. The spent fuel, the buffer and the geosphere restrict efficiently the release of radionuclides even if the canister is initially defective or is broken soon after the sealing of the repository. No extraordinary characteristics are required from a site in crystalline bedrock to warrant the long-term safety of a deep repository for spent nuclear fuel. The main roles of the geosphere in the disposal system are to protect the waste from human and other external actions and provide stable mechanical and chemical conditions for the repository. Locating of the repository within a site and properties of the rock in the vicinity of the repository (e.g. the disturbed rock zone around the excavated tunnels) are more essential than the general geology or rock type of the area. It is important to characterize in detail the site where the repository will be excavated, so that fracture zones can be taken into consideration when constructing the repository. In the TVO-92 safety analysis, all site-specific data has been selected in such a way that in reality a more favourable environment can certainly be found for the repository at each of the five investigation sites. Suitable places for the repository can be found at each of them. In December 1992, Teollisuuden Voima Oy decided to continue site investigations at three sites. The sites are Kivetty in Konginkangas, Romuvaara in Kuhmo and the site of the Olkiluoto nuclear power plant in Eurajoki. There are less uncertainties in the conceptual bedrock models and the explorability of these sites is more favourable than of the two other sites preliminarily investigated. More detailed site investigations will be carried out at the chosen three sites between 1993 and 2000. The site of the repository will be selected in 2000. REFERENCES Ittner, T., Gustafsson, E. & Nordqvist, R. (1991) Radionuclide content in surface and groundwater transformed into breakthrough curves. A Chernobyl fallout study in a forested area in Northern Sweden. Swedish Nuclear Fuel and Waste Management Co (SKB), Technical Report 91-28. Koskinen, L., Hautojàrvi, A. & Vieno, T. (1993) Groundwater flow modelling for the TVO-92 safety analysis of spent fuel disposal. Proc. HLRWM'93 (Las Vegas, April 1993), 1595-1601. Lampén, P. & Snellman, M. (1993) Summary report on groundwater chemistry. Nuclear Waste Commission of Finnish Power Companies, Report YJT-93-I4.
44 Aimo Hautojàrvi & Timo Vieno Taivassalo, V., Koskinen, L. & Meling, K. (1993) Groundwater flow analyses in preliminary site investigations - Modelling strategy and computer codes. Nuclear Waste Commission of Finnish Power Companies. (To be published). TVO (1992a) Final disposal of spent nuclear fuel in the Finnish bedrock - Technical plans and safety assessment. Nuclear Waste Commission of Finnish Power Companies, Report YJT-92-31E. TVO (1992b) Final disposal of spent nuclear fuel in the Finnish bedrock - Preliminary site investigations. Nuclear Waste Commission of Finnish Power Companies, Report YJT-92-32E. Valkiainen, M. (1992) Diffusion in the rock matrix - A review of laboratory tests and field studies. Nuclear Waste Commission of Finnish Power Companies, Report YJT-92-04, Vieno, T., Hautojàrvi, A., Koskinen, L. & Nordman, H. (1992)TVO-92 safety analysis of spent fuel disposal. Nuclear Waste Commission of Finnish Power Companies, Report YJT-92-33E.