Technical Report NTB 13-02

An Assessment of the Impact of the Long Term Evolution of Engineered Structures on the Safety-Relevant Functions of the Bentonite Buffer in a HLW Repository

Bentonite is important as a near-field buffer and backfill for a SF/HLW repository in Opalinus Clay in Switzerland, with desirable properties of this (compacted) material including swelling, low solute transport rates, inter alia. Intrinsic to these properties is that they should be preserved in the long-term, i.e. over safety-relevant timescales (up to a million years). Although evidence from natural analogues can be used to demonstrate the potential stability of bentonite in the repository environment, there are a number of processes which could potentially perturb its long-term performance, such as thermal gradients from the decay heat of waste packages and chemical gradients due to the presence of thermodynamically unstable materials (steel, concrete) used as engineering structures.
 

These potential interactions of bentonite with engineered components (canister metals, concrete tunnel liner, steel support mesh, and arches, plus transport rails) in a geological repository for SF/HLW in Opalinus Clay have been assessed. These interactions are likely to be strongly non-linear, with a complex interplay between fluid transport, clay ion exchange and dissolution, secondary mineral growth, and consequent changes in physical properties (porosity, perme­ability, swelling pressure) of the clay. Despite the differing nature of the chemical inter­actions of bentonite with concrete and steel, it is considered that the timescales of alteration will be very similar due to the similar rates of key processes in the bentonite (montmorillonite dissolution at alkaline pH and the growth of zeolite and sheet silicate minerals). Although timescales of a million years are considered for this alteration, it is envisaged that near-field evolution will be curtailed well within this timeframe by mass transport constraints (porosity decreasing to zero) or mass balance limitations (reactants completely consumed). The time-dependent nature of some of these processes means that not all are readily amenable to study by conventional laboratory experimental procedures. Consequently, the results of published reaction-transport simulations have been important in defining key properties at performance assess­ment-relevant timescales.
 

This evaluation suggests that for an optimistic estimate of bentonite alteration at 100 ka limited by mass transport constraints (porosity decreasing to zero), there will be a thin (0.05 m thick; 1 vol.-% total bentonite) alteration layer around the canister, derived partly through thermal redistribution of minerals and aqueous solutes, and partly due to interaction of the steel canister with bentonite. This results in a thin zone with zero porosity and zero swelling pressure (mont­morillonite totally altered) around the canister, but with a hydraulic conductivity the same as the initial value (potential minor fracturing is assumed to cancel out the effects of decreased porosity). The mineralogical composition of the thin zone is assumed to consist of a thin layer of calcite, gypsum/anhydrite and magnetite on the canister, with montmorillonite in the altered bentonite replaced by Fe-silicates such as cronstedtite, berthierine and chlorite. Beyond this inner alteration zone is an annulus of 0.68 m (92 vol.-% total) of unaltered bentonite.
 

The potential interaction of metallic engineered structures other than the canister with bentonite is relatively minor. Even using a mass balance constraint with no kinetic or mass transport limitations for conversion of montmorillonite to chlorite, then only 2 vol.-% of the total bentonite in the canister zone would be transformed by reaction with these structures. Removing the rails prior to closure would reduce this effect to 1 vol.-%.
 

At the external margin of the bentonite, alteration of the buffer and sealing element adjacent to the concrete liner is optimistically estimated to be 0.02 m in thickness (4 vol.-% total) at 100 ka after closure (note that there is no concrete along the sealing element interface, so this estimate is unrealistic). The hydraulic conductivity of this zone is estimated to be decreased in com­pari­son with the initial state. The porosity and swelling pressure of this zone are likely to decrease to zero over a timescale of a few hundreds to a thousand years due to alteration of mont­morillonite. The mineralogical composition of this zone is likely to be characterised by a sequence of calcite, C-(A)-S-H minerals, Ca-zeolites, sepiolite and saponite clays, with C-S-H minerals forming nearest the cement contact, and other minerals such as zeolites and clays forming further towards the canister.
 

The concrete liner itself may degrade via conversion of portlandite and C-S-H gel to ettringite with a consequent increase in porosity. Reaction-transport simulations suggest that porosity may increase a few percent over 100 ka as an optimistic estimate of degradation.
 

For the sealing element, there are no canisters present, but there are steel reinforcement arches, along with steel mesh, steel anchors and steel rails. Using the conversion of montmorillonite to chlorite mass balance constraint described above, a maximum of 5 vol.-% of bentonite could be transformed to non-swelling silicates. Removing the rails prior to closure would reduce this effect to 4 vol.-%.
 

For a pessimistic estimate of bentonite alteration at 100 ka after closure, it is assumed that the canister is entirely corroded and degraded to iron oxyhydroxides and in the surrounding bentonite, montmorillonite is considered to be totally converted to non-swelling Fe-silicates such as cronstedtite, berthierine and chlorite (mass balance estimate). This transformation is expected to preserve the original porosity and hydraulic conductivity, but to decrease swelling pressure to zero. This altered zone is estimated to be 0.45 m thick and to extend to the bentonite zone altered by interaction with the concrete liner.
 

Similarly, alteration of the buffer and sealing element adjacent to the concrete liner is pessimistically estimated to be 0.2 m thick (35 vol.-% total bentonite) at 100 ka after closure. The porosity and hydraulic conductivity of this zone are estimated to be unchanged with regard to the initial state, but with zero swelling pressure due to removal of montmorillonite. The mineralogical composition of this zone is likely to be characterised by a sequence of calcite, C‑(A)-S-H minerals, Ca-zeolites, sepiolite and saponite clays, with C-S-H minerals forming nearest the cement contact, and other minerals such as zeolites and clays forming further away. A zone of exchange of Ca for Na ions in montmorillonite a few tens of cm thick will advance in front of the mineral dissolution-precipitation reactions. Corresponding estimates for an OPC concrete liner indicate that the amount of bentonite mass altered is increased by a factor of 2.5 and alteration thicknesses are increased by a factor of between 2.5 and 3 for the tunnel dia­meters considered (~ 60 vol.-% total bentonite).
 

The concrete liner is estimated to be totally degraded in terms of physical properties as a pessi­mistic estimate at 100 ka after closure, but a zone of ettringite, calcite and tobermorite may exist marking its former presence.
 

The state of the bentonite barrier at 1 Ma after closure is estimated to be similar to that at 100 ka, albeit with more crystalline degradation products of clay and concrete.
 

The interactions of a copper canister with bentonite are predicted to be restricted to minor amounts of cation exchange in montmorillonite (Cu for Na), resulting in no changes to safety-relevant properties over the lifetime of the repository.