I. G. McKinley
Nagra
Wettingen, Switzerland

M. Toyota
Japan Nuclear Fuel Limited
Tokyo, Japan

Many published concepts for disposal of high-level radioactive waste have been intended primarily to demonstrate the feasibility of safe management of such wastes and hence tended to be somewhat idealised and/or over-designed. As national programmes mature and move towards implementation, there is a corresponding move towards optimisation of designs in order to incorporate experience gained over the last few decades and to take better consideration of the practicalities of quality assurance during operation of the repository.

During such optimisation studies, opportunities arise to increase the robustness of the design by reducing uncertainties in long-term performance. This can involve, for example, adapting designs in order to eliminate potential disruption scenarios. Canister sinking, excessive stress build-up due to the expansion of canister corrosion products and backfill erosion can be effectively avoided by manipulating the backfill composition / mineral chemistry or by using a layered backfill with varying rheological and swelling properties. Optimisation must also take the practicalities of repository construction and engineered barrier system emplacement into account. Economic factors must also be considered; costs should be minimised as long as such reduction does not compromise safety requirements.

Such optimisation will be illustrated for the case of vitrified HLW disposal concepts considered in both Switzerland and Japan. These desk studies lead on to suggestions for experimental and work studies to fine-tune design variants and assess their relative pros and cons.

Many geological disposal concepts for high-level waste (HLW) have been extensively analysed over the last couple of decades and demonstrated to offer extremely high levels of safety. Due both to the inherent limitations in the geological formations available in some countries and the difficulties in detailed geological characterisation, many countries have evolved disposal systems thaqt place considerable weighting on the performance of the engineered barrier system (EBS). In Japan and Switzerland, similarities in waste form and geological/environmental constraints have led to very similar reference repository systems being adopted - involving encapsulation of vitrified HLW in thick steel canisters which are emplaced in tunnels, backfilled with compacted bentonite (Fig. 1).

Although these reference disposal systems have been the focus of extensive studies to demonstrate their safety [e.g. Nagra, 1985; PNC, 1992; Nagra 1994] and form the basis of very extensive R&D programmes, they are acknowledged not to be optimised from the points of view of construction feasibility, performance or cost. Recently, the aspect of practicality has received more attention in both concept evaluations [Toyota, 1995a] and demonstration studies [Huertas and Santiago, 1997]

Based on such considerations, procedures for optimisation have been presented which are intended to preserve the clear safety case which has already been demonstrated, but increase practicality and remove some possible open questions with regard to long-term safety [McKinley, 1997]. A further advantage of this option is that costs are also reduced. It is increasingly realised that such studies are important, as projects move towards implementation [McCombie, 1998].

In this paper, the initial optimisation studies presented by McKinley [1997] are extended and refined significantly with the specific aim of increasing transparency of the safety case.

The key features of the original Nagra/PNC disposal systems (Fig. 1) which give rise to their "robust" [cf. definition of robustness in McCombie et al., 1991] performance are as follows:

• Bentonite is a relatively simple, natural material; its main mineralogical component - montmorillonite- is often found as an infilling mineral in surrounding rock thus supporting predictions of its long-term stability based on theoretical, laboratory and natural analogue studies
• Compacted bentonite swells rapidly in contact with water - thus sealing fabrication gaps and, possibly, any localised inflow points in the surrounding rock; for the expected hydrogeological conditions in the host rock and the waste thermal output, complete saturation of the bentonite would take several hundred years thus delaying canister contact with groundwater
• Saturated bentonite displays plastic behaviour, acting as a mechanical buffer to protect the canister from minor rock movements
• The hydraulic conductivity of compacted bentonite is extremely low, ensuring that solute transport occurs predominantly by diffusion
• Diffusion of most dissolved species through compacted bentonite is strongly retarded due to sorption processes; the very small pore size also ensures that movement of colloidal species is effectively prevented
• Bentonite acts as a chemical buffer, in particular ensuring for long times that porewater pH remains in the slightly alkaline range and conditions are chemically reducing, even if there are uncertainties in the evolution of the chemistry of surrounding groundwater
• Temperatures within the EBS will remain below 100°C for a significant outer annulus of bentonite - ensuring that cementation effects will be negligible; temperatures above 100°C in inner zones persist only for a few decades, during which bentonite saturation has proceeded to a limited extent only (both factors which may greatly constrain possible microbial activity in this zone)
• Corrosion of steel is well understood and a minimum lifetime of 1000 years can be supported by extensive theoretical, laboratory and natural analogue arguments; in reality, lifetimes may exceed tens of thousands years in this diffusion-dominated environment
• Even after mechanical failure, remnant canister and canister corrosion products will act as a massive redox buffer - consuming any radiolytic oxidants produced after water contacts the waste (shielding by the thick layer of steel ensures that radiolysis is negligible as long as the canister is intact)
• The glass matrix corrodes slowly and relatively homogeneously - spreading radionuclide releases over a period >10⁵ years; releases may be further constrained by the low solubility of many radionuclides in the water present within the EBS.

Extensive analysis of this concept [PNC, 1992; Nagra, 1994] has indicated that most radionuclides would decay to insignificance within the EBS and that the few long-lived isotopes released would have very low concentrations. Nevertheless, some open questions remain which, even if rather unlikely, lead to scenarios in which performance can be significantly degraded (e.g. due to bentonite erosion, perturbation due to gas generation, stress build-up due to canister corrosion products, canister sinking). Recent analyses have suggested modifications of the EBS which would preserve the key safety features identified above but eliminate (or greatly decrease the likelihood) of such perturbations [Toyota, 1995b; McKinley 1997]. This paper extends past work by focusing on optimisation which improves the transparency of the safety case. A fundamental assumption in this analysis is that a modular construction technique is used which allows EBS units to be assembled in a quality assured manner and emplaced in prefabricated form. Such technology allows multiple component EBS systems to be considered - which would be impractical for an in-situ construction approach. The engineering, operational and quality assurance aspects of modular EBS fabrication and emplacement are discussed in a companion paper [Toyota and McKinley, 1998].

The basic components of the proposed EBS module (Fig. 2) are:

• An external, thin, steel handling shell
• A thin layer of geotextile material
• An annulus of 1: 5 compacted bentonite / sand mixture (density ~2 Mg/m³ )
• A pure, highly compacted (density 1.8 Mg/m³) bentonite annulus
• An inner sand or sand/bentonite layer
• A steel canister with a corrosion-resistant coating or cladding
• Vitrified HLW in a thin steel fabrication container.

Considering these components in turn, their contributions to the overall safety case are:

1. Outer steel shell: main role is for handling of the modular package: provides residual redox buffering and sorbing capacity and delays bentonite wetting
2. Outer geotextile layer: serves only to better spread the wetting front after failure of the outer shell, ensuring that complex stress fields do not build up due to heterogeneous swelling of the compacted bentonite (this application has been demonstrated in the FEBEX mock-up experiment run by ENRESA [Huertas and Santiago, 1998])
3. Bentonite / sand layer: swelling, plastic, highly sorbing layer but with better rigidity, thermal conductivity and erosion resistance than pure bentonite (excludes possible backfill erosion or canister sinking)
4. Pure bentonite: swelling, plastic, highly sorbing layer which acts as a filter for colloidal materials (allows only diffusion of radionuclides in pure solution)
5. Inner sand or sand/bentonite layer: high porosity buffer for any gas produced by steel corrosion - also may provide void space for expansion of solid corrosion products
6. Steel canister: steel provides mechanical strength and, when clad by a highly corrosion- resistant layer, provides containment for >~ 10⁴ years and redox buffering thereafter.

The outer steel shell of the design is internally supported and should be able to resist the slow build-up of hydrostatic pressure as the tunnel resaturates. The shell will eventually fail due to either mechanical pressure (e.g. due to creep of the walls or failure of any tunnel lining involved) or corrosion. In any case, for most packages, lifetimes of decades to centuries can be reasonably expected, which would cover the duration of the highest near-field temperatures.

After the outer shell is penetrated, water will pass through the geochemically inert geotextile layer and begin to wet the sand/bentonite layer. As the hydraulic conductivity of the latter decreases considerably as it wets and swells, the geotextile layer should encourage more even wetting of the inner layers. Complete saturation of the inner bentonite and sand layers would take decades to hundreds of years and may be constrained by water supply rates for low conductivity host rocks.

The key rôle of the pure, highly compacted bentonite is to act as a colloid filter. Sand/bentonite mixtures also have low hydraulic conductivities and can significantly contribute to radionuclide retention but with the additional advantages of high resistance to erosion (outer layer), higher thermal conductivity and lower tendency to creep (inner layer - to prevent canister sinking). The inner layer may, indeed, be prepared with a very high sand/bentonite ratio (~10:1, say) or even be pure sand. This would encourage the accumulation of any hydrogen gas resulting from anaerobic steel corrosion (or radiolysis) in this layer - feeding back to reduce corrosion rates or release rates of radionuclides from the waste form. A small amount of a suitable ferrous mineral (e.g. siderite) could be added to the sand or sand/bentonite mixtures to provide further redox buffering capacity (eliminating pessimistic redox-front penetration scenarios) and to enhance uptake of some radionuclides.

This concept involves a high strength inner steel canister (10 - 20 cm thick) which is coated or clad to minimise corrosion. Very hard, low corrosion coatings with material such as tungsten carbide or diamond have been developed for many industrial applications and procedures such as ion-beam assisted deposition or chemical vapour deposition allow possibilities of cost-effective application of layers up to ~1 - 5 mm thick. Alternatively, a thin layer (2 - 3 mm) of Ti alloy could be used to provide the required corrosion resistance. Although the fabrication technology is certainly not yet proven for these applications, the great chemical stability of such materials is well established from their application to extreme industrial environments and their natural analogues. This design allows good mechanical performance to be provided by a well known material (steel) but the overall safety principle of corrosion allowance (from the Nagra/PNC disposal system) to be replaced by corrosion resistance. The reduced thickness of the canister minimises potential potential problems due to swelling of any corrosion products eventually formed and an even partially complete coating would reduce gas production rates. Filling void spaces within the canister and the inner glass fabrication container may allow thinner canisters to be used by reducing mechanical strength requirements [cf. Toyota & McKinley, 1998]. It should be emphasised that performance assessment demonstrates that canister longevity is not a critical part of the safety case unless lifetimes can be guaranteed beyond ~10⁵ years. Even without providing such guarantees, the indication that many (or most) canisters will probably not fail over very long timescales is of considerable value in discussions with the general public. After failure, canister corrosion products will buffer redox conditions to ensure low solubilities of most relevant radionuclides.

The dimensions of the various components of the optimised EBS are summarised in Table 1. These component thicknesses are not independent; a thicker canister (~ 20 cm) may require a thicker pure bentonite layer to allow expansion of corrosion products while a thinner canister allows either simple reduction in the pure bentonite thickness or replacement of some part of the pure bentonite with bentonite sand mixture.

The basic concept outlined above is being further analysed in order to specify the dimensions of individual components and choose between design variants (e.g. material for canister corrosion protection). An outline of the key components of this investigation programme is provided in Table 1. Further studies related to determining practicalities of fabrication and tailoring of design and handling to specific host rocks are described elsewhere [Toyota and McKinley, 1997].

The initial scoping calculations and desk studies will lead on to an associated experimental study programme to demonstrate the concepts involved and validate the predicted behaviour of the system. Long-term, large scale and/or in-situ tests may be needed but, as the concept relies mainly on materials with very well known properties, the requirements for R&D should be relative modest.

Although the prime aim of this work is to derive a system which is optimised with regard to long term safety and practicality of implementation, potentially large savings in materials and operational costs could be obtained. Scoping calculations indicate that such savings on the concept outlined above would be much greater than the costs of the associated R&D work involved to implement it. Financial aspects will, however, be considered in further detail after the technical details of design and implementation have been better established.

The proposed 2nd generation design promises increased performance, with reduced uncertainties with also simplified emplacement, improved quality assurance and reduced costs [cf. Toyota & McKinley, 1998]. Modular EBS construction allows a series of simple materials with complementary properties to be combined under well defined conditions. This multiple barrier system combines the key safety features of the original design with additions which exclude potential perturbation scenarios. Although performance is provided by well known materials, use of modern materials/coating technology is suggested to provide a cost-effective, corrosion-resistant layer on the canister. A programme of R&D to refine this concept has been outlined which may lead on to large scale, field demonstration of its feasibility.

1. Huertas F., Santiago J.L., 1998: The FEBEX Project - General Overview; in MRS Proceedings, Scientific Basis for Nuclear Waste Management XXI, Davos, Switzerland (in press)
2. McCombie, C., Zuidema, P., McKinley, I.G., 1991: Sufficient validation - The value of robustness in performance assessment and system design; Validation of geosphere flow and transport models (GEOVAL), OECD, Paris pp 598-610
3. McCombie C., 1998: R&D in support of repository implementation - Do we need any more?; in MRS Proceedings, Scientific Basis for Nuclear Waste Management XXI, Davos, Switzerland (in press)
4. McKinley I., 1997: Engineering for robustness: An approach to optimising HLW disposal concepts; Waste Management, 17, pp1-8
5. Nagra, 1994: Kristallin-I: Safety Assessment Report; Nagra Technical Report series NTB 93-22, Nagra, Wettingen, Switzerland
6. Nagra, 1985: Projekt Gewähr 1985: Nuclear waste management in Switzerland - Feasibility studies and safety analysis - Summary; Nagra Projekt Gewähr Report series NGB 85-09, Nagra, Baden, Switzerland
7. PNC, 1992: Research and development on geological disposal of high-level radioactive waste - First Progress Report; PNC TN 1410 93-059, Power Reactor and Nuclear Fuel Development Corporation (PNC), Tokyo, Japan
8. Toyota, M., 1995a: The study on the disposal concept of high level radioactive waste - The investigation of near-field behaviour and layout of the repository (in Japanese), Ph.D. thesis, university of Tokyo, Japan
9. Toyota, M., 1995b: Mechanical interaction within near-field behaviours of the repositories; Proc. 6^th Int. Conf. High-Level Radioactive Waste Management, ANS/ASCE, Symp., pp694-698
10. Toyota, M., McKinley, I.G., 1998: Optimisation of layout, engineered barrier system fabrication and emplacement for HLW repositories; Proc. Int. HLW Management Symposium, Las Vegas, in press

BACK