PRELIMINARY CALCULATIONS OF RADIONUCLIDE RELEASE AND TRANSPORT FROM A REPOSITORY FOR SPENT NUCLEAR FUEL IN SWITZERLAND

J. Schneider and P. Zuidema
Nagra, Switzerland

P. Smith
Safety Assessment Management Limited, UK

P. Gribi and M. Niemeyer
Colenco Power Consulting AG, Switzerland

ABSTRACT

The results of a preliminary modelling study of radionuclide release and transport from a hypothetical Swiss repository for spent nuclear fuel are presented. The spent-fuel inventory is based on a consideration of the anticipated arisings from nuclear power generation in Switzerland. A part (at least) of these arisings will be reprocessed, with the resultant vitrified high-level waste returned to Switzerland. The decision as to whether to reprocess the remainder of the arisings is currently left open. Reprocessed waste has been considered in detail in an earlier study and the present paper is limited to the direct disposal of spent fuel. A repository concept, models and data for the release and transport of radionuclides are presented. Key uncertainties are identified, such as the evolution of redox conditions in the near field. Radiolysis may result in oxidising conditions near to the fuel surfaces, but model results show that, provided oxidising conditions are confined within the bentonite buffer surrounding the waste canisters, calculated doses are low and are insensitive to the exact position of the redox front. Higher doses are calculated in the extreme case of a redox front extending into the host rock, but, even in this case, the Swiss regulatory guideline of 0.1 mSv a^-1 is not exceeded. Finally, first calculations for the direct disposal of mixed-oxide fuels are presented.

INTRODUCTION

The Swiss electricity utilities have placed contracts with BNFL (UK) and COGEMA (France) for the reprocessing of about one third of the total arisings of spent fuel from nuclear power generation, assuming a 120 GWa (electric) scenario. The options for the remainder include (i) reprocessing and (ii) direct disposal of the spent fuel, thus leaving the arisings from 80 GWa of power generation as the theoretical maximum for direct disposal. Throughout this work, it is therefore assumed that an amount of unreprocessed spent fuel corresponding to this theoretical maximum will be disposed in the repository.

On the assumption of complete reprocessing, Nagra has recently performed an integrated analysis of the disposal of high-level waste in the crystalline basement of Northern Switzerland: the Kristallin-I safety analysis (1). A key feature of the safety-assessment component of Kristallin-I is the model chain that describes quantitatively how radionuclides migrate from the repository through the geosphere to the biosphere. The results are expressed as annual individual dose to an average member of a critical population group. This dose can then be compared to the limit of 0.1 mSv a^-1 specified by the Swiss regulatory guidelines (2).

No comparable work has yet been performed for the co-disposal of high-level waste from reprocessing and spent fuel. Nagra are in the process of acquiring the scientific understanding, computational tools and data that will be required in order to analyse the safety of such disposal. The modelling study presented in this paper, which considers only the direct disposal of spent fuel, represents a part of this process. The discussion of the repository concept, models and data focus on the near field, since the treatment of the geosphere and biosphere is identical to that of Kristallin-I and is discussed at length in (1).

A PROVISIONAL REPOSITORY CONCEPT

A provisional repository concept has been adopted for the calculations, in which spent fuel is packaged in canisters and emplaced co-axially in a horizontal system of tunnels deep underground in the crystalline basement of Northern Switzerland. The canister material and lifetime (i.e. minimum time at which a canister could first be breached, allowing water to access the fuel) are currently under discussion and modelling studies for a wide range of possible lifetimes will provide input for an eventual decision. For the preliminary safety calculations presented in this paper, a 10⁴ year lifetime is assumed, corresponding, for example, to the "conservative" value used for an ACP (advanced cold process) copper-steel canister in the safety analysis of a Finnish spent-fuel repository (TVO-92, (3)). Highly compacted bentonite is used as a backfill material, providing a physical and chemical buffer between the canisters and the host rock.

THE SAFETY-ASSESSMENT MODELS AND DATA

The following describes, in broad terms, the main features of the conceptual model for spent-fuel dissolution and radionuclide transport in the repository near field. Key uncertainties, and their treatment in the safety assessment, are also given.

• Radionuclides are confined within the canister until the canister is breached.

  The time at which the canister is breached is uncertain and depends on design decisions. For preliminary safety calculations, it is assumed that all canisters fail simultaneously at 10⁴ years after emplacement. Conservatively, neither the radionuclide transport resistance nor the redox buffering capacity of the failed canisters are taken into account in this study.

• Actinides and most fission products are homogeneously distributed throughout the fuel matrix. There are, however, concentrations of some fission products in the grain-boundaries of the fuel and in the gap between the fuel and Zircalloy cladding. Furthermore, activation products, resulting from exposure to neutrons in the reactor, are present in the cladding and other structural parts of the fuel assemblies.

  The reference-case radionuclide inventory for preliminary spent-fuel calculations is taken from the TVO-92 safety analysis (3) (1.6 t_HM of UO₂ fuel per canister, with a burn-up of 36 GWd/t_HM), but with a total number of canisters (1620) based on a scaling of the TVO-92 inventory to a 80 GWa power-generation scenario. As a variation on this fuel type, mixed-oxide (MOX) fuel, with a burn-up of 55 GWd/t_HM, is also considered. The MOX inventory has been roughly estimated by Nagra using existing tools designed primarily for the characterisation of UO₂ fuel; the total number of canisters (3085) corresponds again to the arisings from 80 GWa of power generation, with the hypothetical assumption that only MOX fuel elements are used. A smaller amount per canister of 0.55 t_HM is assumed for MOX fuel because of the higher heat generation and constraints on the maximum temperatures that are presently set for the surrounding bentonite (4). The partitioning of the inventory between the matrix, the gap and grain-boundary regions is uncertain for some nuclides and is currently based on recommendations of (5).

• Following canister failure, water comes into contact with the waste. The following processes lead to the release of radionuclides:
  • There is a rapid release of the gap inventory, followed by a slower release of radionuclides segregated to grain-boundaries during fuel irradiation.

    These releases are modelled as being instantaneous.

  • The Zircalloy cladding and other structural parts begin to dissolve, releasing the activation products contained therein.

    The activation products are assumed to be released congruently with the dissolving cladding, at a fractional rate of 3 ×10^-6 a^-1. This value is based on a cladding corrosion rate of 2 ×10^-9 ma^-1 (5) and a typical thickness of the cladding of 0.75 mm for Swiss fuel elements.

  • The spent-fuel matrix begins to dissolve at a rate that is sensitive to redox conditions at the fuel surface, releasing the embedded radionuclides.

    There is uncertainty in the evolution of redox conditions at the fuel surface, and thus also in the spent-fuel dissolution rate. For preliminary spent-fuel calculations, the fractional dissolution rate is based on data from (3). In the case of UO₂ fuel, the rates are 5 ×10^-6 a^-1 for the time period 10⁴ a to 10⁵ a, 4 ×10^-7 a^-1 for the time period 10⁵ a to 10⁶ a and 2 ×10^-7 a^-1 beyond 10⁶ a. Rates that are a factor of five higher are assumed for MOX fuel, due to its approximately five times higher alpha activity and the correspondingly higher generation rate of radiolytic oxidants.

• Radionuclides diffuse in solution through the breached canister into the surrounding bentonite and thence into the host rock. Several elements reach saturation at low concentrations. The concentrations of isotopes of these elements are thus controlled by the elemental solubilities (shared between isotopes in proportion to their inventories). Diffusive transport is retarded by sorption on bentonite pore surfaces. The solubilities and sorption properties of some elements are sensitive to redox conditions.

  Sorption parameters are taken from (1), supplemented by data from (3) and (6). Solubilities are taken from (7). Uncertainty in the evolution of redox conditions within the near field leads to uncertainty in the solubilities and sorption properties of some elements. The effects of this uncertainty are scoped by the evaluation of a range of possible cases, as described below.

A new code (SPENT (8]) has been developed to model these near-field processes. The codes RANCHMD (9) and TAME (10) are used to model radionuclide transport through the geosphere and biosphere, respectively. The models and data for the geosphere and biosphere calculations correspond to the reference case in the Kristallin-I safety analysis (1).

CALCULATIONAL CASES

The rate of fuel-matrix dissolution and, for some radionuclides, solubility limits and sorption parameters, are known to be dependent on redox conditions. These are also thought to be key parameters controlling near-field performance. The most significant processes determining redox conditions within and around the repository following canister breaching are:

• the generation of oxidants and reductants by radiolysis of water in contact with fuel surfaces (principally -radiolysis for times beyond about 1000 years),
• the recombination of oxidants and reductants,
• the inventory of reductants (including the fuel itself) in the repository and in the host rock,
• the transport and reaction properties of oxidants and reductants.

Some radiolytic reductants are relatively inert (e.g. H₂) and may migrate away from the fuel surfaces without recombining with oxidants, resulting in (i) net oxidising conditions near the fuel surfaces (at least initially, when the generation rate by radiolysis is high) and (ii) the establishment of a redox front at some distance away from these surfaces. Note that the position of this redox front will be time dependent.

Rather than attempt to model the position of the redox front as a function of time, four alternative cases are defined, in which time-independent redox conditions are assumed, in order to evaluate the sensitivity to this source of uncertainty. The cases are illustrated in Fig. 1. CASE 1, in which reducing conditions are assumed throughout the near field (including the fuel surface) is the most optimistic from the point of view of safety, but is considered difficult to support on the basis of current information and is not therefore evaluated. The cases for which calculations are performed are:

CASE 2: The redox front penetrates only a small distance into the bentonite (compared to its total thickness): sorption parameters and solubility limits are based on reducing conditions.

CASE 3: The redox front penetrates a considerable distance into the bentonite, but not beyond: sorption parameters for the bentonite are based on oxidising conditions and solubility limits are based on reducing conditions.

CASE 4: The redox front penetrates into the host rock: sorption parameters and solubility limits in the near field are based on oxidising conditions.

Where redox-sensitive radionuclides migrate across the redox front, from a region where oxidising conditions prevail (generally higher solubilities) to a region where reducing conditions prevail (generally lower solubilities), the resulting precipitates may take the form of colloids. If the redox front is located in the bentonite (CASES 2 and 3), the fine structure of the bentonite pores prevents the migration of these colloids and only solute transport needs to be considered. If the redox front penetrates into the geosphere (CASE 4), colloids may be transported through fractures in the host rock and, if they are prevented (due, for example, to their size) from diffusing into the stagnant pore water of the adjacent wall rock, then the radionuclides associated with these colloids would not be subject to the retarding effects of matrix diffusion and sorption on matrix pore surfaces. The distances over which such colloids would be transported through the host rock are highly uncertain and two extreme situations, within CASE 4, are evaluated in order to investigate the effects of this uncertainty:

CASE 4a: Colloids either do not form, or are transported only a short distance through host-rock fractures before destabilising. Only solute transport is modelled in the geosphere.

CASE 4b: Colloids are transported, without either filtration or retardation, through host-rock fractures. Consequently, to model this case, all radionuclides escaping from the near field are transferred instantaneously to the biosphere.

CASES 2, 3 and 4a/b are evaluated for UO₂ spent fuel. From a consideration of the inventory of near-field reductants, which is large in comparison to the integrated production of radiolytic oxidants (see, e.g., (11)), CASE 2 is assumed to be the most likely and is taken as a reference case for studies of sensitivity to fuel type (both UO₂ and MOX fuels are considered). However, the realism of CASE 2 depends not only on the inventory of reductants but also on their accessibility and on the kinetics of the relevant redox reactions. CASES 3 and 4a/b are therefore currently not excluded.

Fig. 1. Calculational cases to evaluate the sensitivity of the annual individual dose to the position of a redox front.

RESULTS

UO₂ Spent Fuel; The Reference Case

Fig. 2 shows the calculated dose as a function of time for a UO₂ spent-fuel repository with redox conditions defined according to CASE 2 (reference case). The following observations are made:

• The total dose curve displays three maxima: the first (and highest) due principally to ¹²⁹I shortly after the canisters are breached (10⁴ a), the second (about an order of magnitude lower) due to ¹³⁵Cs at about 3 ×10⁵ a, and the third (less than a factor of two smaller than that of ¹³⁵Cs) due to ²³⁷Np at about 7 ×10⁶ a.
• The relatively early and high ¹²⁹I dose is due mainly to its presence in the instant-release fraction of the fuel inventory, to its high solubility and to the assumption that it is non-sorbing in the near-field and only weakly sorbing in the geosphere.
• The highest dose maximum is more than two orders of magnitude below the Swiss regulatory guideline of 0.1 mSv a^-1.

Fig. 2. Annual individual dose from a repository for UO₂ spent fuel as a function of time with redox conditions defined according to CASE 2 (reference case). The dose curves for individual radionulides that are most important in defining the shape of the total dose curve are also shown.

Sensitivity to Redox-Front Position

Fig. 3 shows the calculated dose as a function of time for a UO₂ spent-fuel repository with redox conditions defined according to CASE 2 (reference case), CASE 3 (near-field sorption parameters based on oxidising conditions), CASE 4a (near-field solubilities and sorption parameters based on oxidising conditions) and CASE 4b (as CASE 4a, but with the hypothetical assumption of instantaneous radionuclide transfer from the near field to the biosphere):

• CASES 2 and 3 give very similar calculated doses, indicating a lack of sensitivity to near-field sorption parameters, within the range defined by uncertainty in sorption due to uncertainty in redox conditions. Of the radionuclides that define the positions and magnitudes of the dose maxima, only ²³⁷Np is redox-sensitive in its sorption properties, and the magnitude of the ²³⁷Np maximum is constrained by Np solubility.
• CASE 4a, on the other hand, displays higher calculated doses. The radionuclides that dominate dose are those from the 4N+3 chain (²³⁹Pu, ²³⁵U, ²³¹Pa, ²²⁷Ac). The height of this maximum is determined by the release and transport properties of ²³⁵U, which was modelled assuming both a reduced near-field sorption parameter value (by a factor of 250) and an increased solubility limit (the solubility limit of U is not reached under oxidising conditions (7)) with respect to the reference case. Nevertheless, even in this unlikely case, the highest dose maximum is still about one order of magnitude below the Swiss regulatory guideline of 0.1 mSv a^-1.
• CASE 4b (hypothetical direct release from the near field to the biosphere) displays a similar dose maximum to CASE 4a, but the maximum is now shifted towards earlier times. CASE 4b represents an upper bound, in terms of dose, for the consequences of colloid-facilitated geosphere transport.

Fig. 3. Annual individual dose (summed over all radionuclides) from a repository for UO₂spent fuel as a functio of time in CASE 2, CASE 3 and CASE 4a/b (see Fig. 1 and text for a definition of cases).

Sensitivity to Fuel Type

Fig. 4 shows the calculated dose for a hypothetical MOX spent-fuel repository, with redox conditions defined according to CASE 2. The following observations can be made:

• With respect to UO₂ spent fuel (Fig. 2), the most noticeable difference is the higher (by about a factor of five) ¹²⁹I peak for MOX spent fuel. This can be traced back to the increase in the instant-release fraction from 5% (UO₂) to 20% (MOX) (5), combined with the high solubility and the assumption that this radionuclide is non-sorbing in the near field and only weakly sorbing in the geosphere.
• In other respects, the dose curves for MOX and for UO₂ spent fuel are remarkably similar. Reasons for this finding include the similar total inventory of key radionuclides and their similar (fuel dissolution rate, instant release fractions) or identical behaviour in the near-field, geosphere and biosphere for the two fuel types.
• The maximum dose (summed over all radionuclides), which is dominated by ¹²⁹I, is still well over an order of magnitude below the Swiss regulatory guideline of 0.1 mSv a^-1.

Fig. 4. Annual individual dose from a repository for MOX spent fuel as a functi of time with redox conditions defined according to CASE 2. The dose curves for individual radionuclides that are most important in defining the shape of the total dose curve are also shown.

DISCUSSION AND CONCLUSIONS

A modelling study has been presented considering direct disposal of spent fuel in a hypothetical repository deep underground in the crystalline basement of Northern Switzerland. The calculations described in this paper are preliminary and further work will be required, for example:

• to finalise repository design and, in particular, a canister design based, in part, on lifetime requirements;
• to obtain more realistic estimates of the expected Swiss inventories for UO₂ and, in particular, MOX spent fuel with different burn-ups;
• to assess the evolution of redox conditions in the near field due to radiolysis after canister failure and the resulting effects on radionuclide sorption, solubility limits and fuel dissolution rates.

Therefore, the results should be interpreted with caution; further iterations are clearly needed using more reliable data and models as indicated above. Nevertheless, the following statements can be made: (i) Even in the case that gives the highest dose (CASE 4b); i.e. where it is assumed that the redox front has moved beyond the bentonite into the adjacent host rock and where no credit is taken for geosphere retention, the Swiss regulatory guideline of 0.1 mSv a^-1 is not exceeded. (ii) The reference case for this study (CASE 2); i.e. where it is assumed that the redox front has penetrated only a small distance into the bentonite, leads to a total dose curve with a maximum that lies more than two orders of magnitude below the Swiss regulatory guideline of 0.1 mSv a^-1. (iii) First calculations for a hypothetical repository for MOX spent fuel yielded doses comparable to those obtained for a UO₂ spent-fuel repository (assuming redox conditions defined according to CASE 2 for both fuel types).

REFERENCES

1. NAGRA, "Kristallin-I Safety Assessment Report", Nagra Technical Report NTB 93-22, Nagra, Wettingen, Switzerland (1994).
2. HSK & KSA, "Guidelines for Swiss Nuclear Installations: Protection Objectives for the Disposal of Radioactive Wastes", Swiss Federal Nuclear Safety Inspectorate (HSK) and Federal Commission for the Safety of Nuclear Installations (KSA); HSK R-21/e, HSK, Villigen, Switzerland (1993).
3. T. VIENO, A. HAUTOJÄRVI, L. KOSKINEN and H. NORDMAN, "TVO-92 Safety Analysis of Spent Fuel Disposal", Report YJT-92-33E, VTT, Helsinki, Finland (1993).
4. R. SATO, T. SASAKI, K. ANDO, P.A. SMITH and J. SCHNEIDER, "Calculations of the Temperature Evolution of a Repository for Spent Fuel in Crystalline and Sedimentary Host Rocks", Nagra Technical Report, Nagra, Wettingen, Switzerland (in preparation).
5. L.H. JOHNSON, D.W. SHOESMITH and J.C. TAIT, "Release of Radionuclides from Spent LWR UO₂ and Mixed-Oxide (MOX) Fuel", Nagra Internal Report, Nagra, Wettingen, Switzerland (1994).
6. I.G. McKINLEY and J. HADERMANN, "Radionuclide Sorption Database for Swiss Safety Assessment", Nagra Technical Report NTB 84-40, Nagra, Wettingen, Switzerland (1985).
7. U. BERNER, "Eingabe-Werte für den Nahfeld-Code SPENT - Kommentare zu den Löslichkeitslimiten", PSI Aktennotiz, Paul Scherrer Institute, Würenlingen and Villigen, Switzerland (1996).
8. R. BROWN and M. WILLIAMS, "SPENT: A Source Term Model for Spent Fuel Assessment in STMAN", Nagra Internal Report, Nagra, Wettingen, Switzerland (1993).
9. A. JAKOB, J. HADERMANN and F. ROESEL, "Radionuclide Chain Transport with Matrix Diffusion and Non-Linear Sorption", PSI Report 54, PSI, Würenlingen and Villigen, Switzerland (1989) and Nagra Technical Report NTB 90-13, Nagra, Wettingen, Switzerland (1989).
10. R.A. KLOS, H. MüLLER-LEMANS, F. VAN DORP and P. GRIBI, "TAME - The Terrestrial-Aquatic Model of the Environment: Model Definition", PSI Report 96-18, PSI, Würenlingen and Villigen, Switzerland (1996) and Nagra Technical Report NTB 93-04, Nagra, Wettingen, Switzerland (1996).
11. P.A. SMITH and I.G. McKINLEY, "Production of Radiolytic Oxidants and Penetration of a Redox Front into the Bentonite in the Cases of UO₂ and of MOX Spent Fuel", Nagra Internal Report, Nagra, Wettingen, Switzerland (1995).