1. Introduction
The production of energy from nuclear fuels inherently generates radioactive waste as a by-product, and thus the requirement for long-term disposal of this waste is an inevitable consequence. Geological disposal of radioactive waste is universally favoured with many nations choosing either clay-based host formations [
1,
2] with an additional clay backfill material occupying the void space around the waste, salt formations or crystalline host rocks, for example the Forsmark Spent Fuel Repository which has a granodiorite host rock with a clay backfill material [
3,
4]. Clay formations are favoured by a number of European countries for the disposal facility, with research laboratories in France, Switzerland and Belgium studying the Callovo-Oxfordian Claystone [
5], the Opalinus clay [
6,
7] and the Boom clay [
8,
9,
10] respectively [
11,
12]. Currently, the only operational geological disposal facility for intermediate- to high-level waste is the Waste Isolation Pilot Plant (WIPP) in a salt formation in Carlsbad, New Mexico; many other proposed sites globally are in the conceptual, planning or approval stage [
1,
3,
4,
13]. The waste is either placed in large boreholes in the floor of deposition boreholes (“in-floor”) or the waste is stacked in the tunnels themselves (“in-tunnel”), with a clay backfill material occupying the void space [
14]. Bentonite is commonly proposed for use in the engineered barrier system (EBS) and as a backfill material in many proposed facilities [
15,
16,
17] because of its high swelling capacity, very low permeability, low diffusion rate, high sorption capacity and self-sealing characteristics [
12,
18].
Radioactive wastes produce and emit heat as they decay, and the amount generated, by high-level wastes especially, can be significant. Depending on the form of the radioactive waste and the amount of time that has passed before the waste is placed in the disposal facility, each canister of high-level waste (HLW) could emit around 800 W of heat, assuming a vitrified HLW package with a volume of 2.04 m
3, a 1.6 kg mass of uranium and a total activity content of 6.3 × 10
3 TBq [
19]. Self-heating due to radioactive decay can produce significant amounts of energy. This heat can have an adverse effect, with thermal loading leading to mechanical failure or chemical degradation of the barrier or host rock [
20,
21]. Additionally, the thermal gradient through the host-rock could lead to alterations in groundwater pressure gradients and flow, or could have an effect on the near-field mass transport mechanisms and chemical evolution [
14]. Chemical alteration of bentonite can take place through ion exchange reactions and mineral dissolution-precipitation, both of which could modify the chemical and physical properties of the clay barrier or host rock permeability [
22,
23].
The hydraulic conductivity of bentonite has been found to increase with increasing temperature up to 80 °C [
24,
25]. This has been attributed to the reduction in water viscosity with increasing temperature [
25]. However, Zihms and Harrington [
26] found that the sensitivity of bentonite permeability to temperature could not entirely be explained by changes in water viscosity. Repository designs will need to address possible variations in the surrounding temperature of the disposal site and ensure that the maximum design temperature is not exceeded [
27,
28]. Studies have found that at temperatures below 100 °C, the sealing performance of clays is not significantly affected [
26,
29,
30]. However above this temperature, illitisation and mechanical degradation of the clay could lead to a reduced performance [
12]. In addition, the permeability of the clay has been observed to increase with increasing temperature [
29]. The maximum operational temperature for an EBS, therefore, is an important design consideration [
28]. However, this must be balanced with the reality that the maximum operational temperature will have a bearing on the repository footprint (and cost) in terms of the spacing between waste packages, the distance between the disposal galleries and the overall volume of material stored [
14]. Radioactive waste must spend a minimum time at the ground surface before permanent disposal can take place, so that the waste can cool and the risk of wider distribution of migrating radioactive particles is lowered. Repository design must take account of the need to quickly achieve safe and permanent geological disposal, and reduce the time spent by the waste at ground surface, without increasing risk. The UK surface storage period for vitrified HLW is 50 years [
28]. Minimising the repository footprint is also an important economic consideration, with cooler waste allowing smaller spacing between canisters. Gaining an improved understanding of the response of the engineered barrier system to higher temperatures may allow the radioactive waste to be disposed of more quickly. In addition, studying bentonite under an increased thermal load will provide further insight into the upper limit of acceptable temperatures for waste disposal.
The aim of this study is to experimentally investigate the response of compact MX80 bentonite during a thermal loading cycle up to 200 °C, extending the temperature range that has previously been considered (e.g., [
25,
26]). While repeated thermal cycling is highly unlikely to occur in-situ, it has been examined as part of this study to assess permanent changes to the bentonite thereby gaining a broader understanding of its response to thermal loading. Two complementary experimental configurations were used: the first configuration had a constant volume boundary condition (simulating the deposition hole characteristics in a hard-rock geological setting) and the second used an isotropic confining pressure that allowed volume change (reflecting disposal in a soft-rock geology). The two configurations offered different advantages. The constant volume boundary configuration was instrumented with up to 3 porewater pressure transducers and 5 load cells measuring the porewater pressure and swelling pressure of the clay respectively. However, due to the nature of the boundary condition within a constant volume test, the development of even a very small gap between the clay and the side of the vessel could lead to the production of a high permeability pathway around the outside of the clay. To address this issue a second apparatus was constructed in which the sample was isotropically stressed, allowing volume change while preventing fluid migration around its periphery. The disadvantage of the isotropic system was that the surface of the clay could not be instrumented. In this paper, we combine new measurements on the sensitivity of permeability to temperature with a reanalysis of previous data (constant volume radial flow (CVRF) experiment CVRF-1) originally presented by Zihms and Harrington [
26]. The results of four experiments presented here provide details of the intrinsic permeability, the total stress, the swelling pressure and the porewater pressure during thermal loading of compact bentonite. Comparisons between the experimental results from the different experimental configurations are made and the observations provide an important insight into EBS behaviour under a range of temperature conditions.
4. Discussion
The data from the four experiments showed that the processes occurring within the apparatus over time are complicated and need to be carefully deconvolved. The intrinsic permeability of the bentonite measured in the four tests showed a general trend for decreasing permeability with increasing temperature (
Figure 11); the ISO-1 data showed this trend most clearly. The CVRF-1 data did not show a clear trend of either increasing or decreasing permeability with increasing temperature, although the data from CVRF-1 were capped at temperatures below 120 °C, which could have limited the extent of the trend seen in the data. Above approximately 60 °C, the CVRF-2 and ISO-2 experiments showed the general trend for permeability decrease with temperature increase, suggesting a reduction in the porosity of the clay. Although in the CVRF tests, the sample volume could not increase beyond the limits imposed by the pressure vessel, it was possible for the sample volume to alter slightly within the vessel. It can be inferred therefore that the pore volume within the bentonite changed after each increase in temperature, which would have had the effect of altering the fluid pathways in the clay, shown by the increase and decrease in the permeability measured.
The degree of consolidation of the starting material is likely to have a significant control on the thermal behaviour of the bentonite. Temperature controlled triaxial testing on Kaolin between 22 and 90 °C has shown that the stress history affects whether contraction or dilation are observed on heating; normally consolidated samples contract with increasing temperature, whilst over-consolidated samples dilate [
47]. Cekerevac and Laloui [
47] also showed that the amount of contraction observed increases with increasing temperature, and this effect was more pronounced above 60 °C. The samples used in this study were close to the pre-consolidation stress and are therefore more likely to exhibit contractional rather than expansional behaviour under a thermal load. In the context of this study, the preconsolidation stress is assumed to be equivalent to the compaction force necessary to achieve the required swelling pressure. Given that three of the four test samples were taken from the same bentonite block, it is difficult to envisage significant variation between the properties in the block and it seems unlikely that this is the primary cause for any discrepancy in the data from the different tests. As such, further work is required to map permeability as a function of temperature, compaction and dry density, especially at higher temperatures. However, the variation in permeability, once corrected for viscosity and density (
Figure 3B,C), remains well within 1 order of magnitude and is therefore not of primary concern from the perspective of disposal of radioactive waste material with temperatures up to 150 °C.
The CVRF-2 and ISO-2 tests produced a high permeability measurement at 200 °C; whilst for the ISO-2 test this is an experimental artefact, for the CVRF-2 test, this is a true response of the experiment. At 200 °C with the backpressure at 1 MPa and assuming a linear pressure gradient, the fluid present in the last 1/5th of the sample length (1.2 cm), closest to the outflow end, may have experienced a phase change to steam. This may have contributed to the higher permeability values recorded at this temperature. However, because the inflow permeability is also much higher and fluid entering the sample at 200 °C would have remained a liquid, the phase change alone cannot explain the increase in permeability at this temperature. Conservatively, if we assume that the phase change affected only the second half (lengthways) of the sample, this would have only resulted in a factor 2 increase in permeability. More likely, the sample will have undergone thermal contraction (as discussed above e.g., [
47]) sufficient to generate one or more high conductivity pathways. Once high flow-rates had been observed at 200 °C in the CVRF-2 test, the experimental temperature was reduced to 150 °C to see if the sample would recover before the test was terminated; the flow-rates quickly returned to those prior to the final temperature increment, suggesting that the majority of the observed behaviour at this high temperature was recoverable (
Figure 10). However, the sample in this test only experienced the 200 °C temperature for 5 days, which would not have been long enough to induce any thermally derived alteration in the bentonite, such as remineralisation changes to the structure and fabric of the sample [
48]. In addition, our tests used de-ionised water as the injection fluid and the amount of mineral alteration, and consequent changes in thermo-hydro-mechanical (THM) behaviour, may change with the composition of the permeant.
The stress changes shown in the load cell data from the CVRF experiments (
Figure 5 and
Figure 7) suggest that the bentonite experiences a thermally induced consolidation, reducing the ease with which fluid can migrate through the sample. During Phase 2 of both CVRF tests after each increment in temperature, the load cells recorded a sharp increase in total stress that gradually relaxed to a well-defined asymptote. Up to around 120 °C, there was a general positive correlation between the maximum value of the spike in total stress and the temperature step, thereafter the relationship between these parameters becomes unclear, possibly in part due to the thermal compliance of the apparatus (
Figure 5,
Figure 6 and
Figure 7). In addition, the load cell data showed a thermal dependency for total stress, whereby the axial total stress value, once it reached an asymptote, increased after each temperature increase up to 150 °C (
Figure 7). The asymptotic value of radial stress showed very little change after each temperature increment until the 150 °C increment where both load cells showed a decrease in value. The two isotropic experiments showed different behaviours. For ISO-1, as the temperature increased the permeability decreased, indicating the occurrence of thermal consolidation as with the CVRF-2 test. For ISO-2, below 60 °C, increasing temperature resulted in greater permeability; above 60 °C and below 170 °C there was a decrease in permeability before the apparatus failed. A caveat should be added that it is unlikely that the tests were conducted at true hydraulic equilibrium. Although the stress data reach an asymptote after a few weeks (
Figure 7), the low permeability nature of this material along with the high swelling capacity means that it could take months, or even years for these samples to reach hydraulic equilibrium [
49] which should be acknowledged when interpreting thermo-hydro-mechanical-chemical (THMC) data.
The asymptotic values after each temperature increment recorded by each load cell were not the same indicating that the clay was not exerting a homogeneous swelling pressure on the vessel in all directions. In both tests, during the hydration phase the axial load cells recorded a lower value than the radial cells (except radial 4 in CVRF-1) but during Phase 2 where the temperature increased, the axial load cells recorded larger values than the radial load cells. The primary contribution to the spike in stress is caused by thermal expansion of the porewater, the thermal expansion coefficient of which is much larger than the thermal expansion of the apparatus, indicating that the interface between the clay and the apparatus should remain competent. However, the slight increase in apparatus volume may help to explain why the negative trend in permeability is less obvious in the data in the CVRF tests.
Further evidence for the thermal contraction of the clay is provided by
Figure 12, which shows a trend for decreasing average effective stress with increasing temperature, measured by the load cells in the CVRF tests. These data (
Figure 12) complement the permeability data (
Figure 11) and suggest that a general contraction of the bentonite may be occurring. Other studies have investigated the effect of consolidation on saturated soils, finding that low over-consolidation causes contraction whilst high over-consolidation causes expansion up to a specific temperature, followed by contraction for temperatures above this point (fractionated illitic material and glacial lake clay, normally consolidated [
50]; low porosity clay [
51], MC clay (similar to Kaolin) and bentonite [
52]; Boom Clay, Pasquasia Clay and Spanish Clay [
53]; Boom Clay [
54]; Kaolin [
55]). The thermal history of the bentonite in a repository could therefore have an important effect on its operational performance. This is demonstrated by the lowering of the average effective radial stress value for the same temperature over repeated thermal cycles (
Figure 12A), which could indicate that thermal hardening of the CVRF-1 sample may have occurred. Thermal loading can lead to both elastic and plastic changes in the clay; the irreversible alteration is seen as hysteresis in
Figure 12 and is equivalent to the clay’s thermal memory. Thermal hardening is a phenomenon that has been observed as a result of thermal cycling in unsaturated bentonite [
56], illite [
33] and saturated Boom clay [
54], and increases the yield stress of the material [
56,
57]. The 150 °C measurement of the CVRF-2 sample (
Figure 12B) suggests that the same phenomenon may have been observed if the thermal cycle of the CVRF-2 sample had been continued. However, the lack of any obvious hysteresis in the axial effective stress measured in the CVRF-1 sample (
Figure 12A) or in the permeability measurements (
Figure 11) suggests that the majority of the deformation experienced by the clay was reversible. A study on the effect of very low temperatures on the performance of bentonite found that above 0 °C the clay swelling pressure decreased as the temperature increased, with the effect more pronounced for higher clay densities [
58]; this was attributed to two different components contributing to the swelling: entropy due to mixing and hydration of the interlayer space. However, the temperature range for these results was 0 to 20 °C and it is unclear whether this trend would continue at higher temperatures. Additional data is required to explain this discrepancy; however, the salient result is that the permeability change is relatively small over the range of temperatures of interest.
In addition to the thermal contraction of the clay, the steel pressure vessel may thermally expand by as much as 0.9% of its volume (see
Appendix B), causing a corresponding decrease in the density of the bentonite as its volume expands by an equal amount. In the isotropic tests, where boundary stress is actively applied to the sample, expansion of the water with increasing temperature leads to drainage of water from the sample. This in turn leads to consolidation of the clay as the confining fluid presses on the sample, explaining the more marked decrease in permeability with increasing temperature shown by the isotropic tests. It is not possible to decouple the contribution of thermal contraction of the clay, the density decrease of the clay and thermal expansion of the steel and water without using an alternative material to construct the apparatus. However, by conducting experiments using an isotropic confining pressure apparatus where clay contraction does not impact interface sealing, we provide additional evidence for the thermal response of the clay at the higher temperatures. Further experimentation is required and development of an apparatus constructed from INVAR steel, which has a lower thermal expansion coefficient than 316 stainless steel, would significantly reduce experimental complications due to thermal compliance of the apparatus.
These experiments have shown that the permeability of compact MX80 bentonite is sensitive to both the thermal loading and the nature of the boundary condition imposed. Zihms and Harrington [
26] showed that MX80 bentonite has a thermal memory that affects the response of the clay to imposed stress; this has also been demonstrated by the results from the CVRF-2 test. The data from CVRF-2, ISO-1 and ISO-2 all showed a general trend for decreasing permeability with increasing temperature; this effect was more marked in the isotropic tests where the volume of the sample was free to change. Thermal expansion of the water within the bentonite will drive fluid from the sample at higher temperatures; in the isotropic tests, the sample was then forced to contract by the confining pressure, leading to the lower permeability values recorded than in the CVRF tests. However, the hydraulic properties were not greatly affected by temperature changes below 150 °C, with changes primarily attributed to the thermal dependence of viscosity.