Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
ADEQUACY OF TREATMENT OF THE PHYSICAL AND CHEMICAL STABILITY OF THE CANISTERS AND BUFFER Copper canisters enclosed in a buffer of compacted bentonite are the essential feature of the Swedish plan for disposal of spent nuclear fuel. The combination of canisters-plus-buffer is confidently expected to resist mechanical disturbance and chemical attack for at least a million years. If the analysis supporting this expecta- tion is sound, other parts of the disposal plan are of secondary importance: the bedrock in which the canisters will be placed serves only the functions of providing a stable and chemically benign environment and insuring against rapid transport of radionuclides in the very unlikely event of early canister failure. The central role of canisters-plus-buffer has not changed between KBS-2 and KBS-3. A few details have been altered: the thickness of canister walls has been reduced from 0.2 m to 0.1 m; filling the canisters with copper rather than lead is proposed as a possible alternative; and adding ferrous phosphate to the backfill as a reductant is no longer advocated. Performance estimates of the canisters and compacted bentonite have been supported by more elaborate analyses and laboratory testing. But the salient features of this part of the disposal plan remain the same, and the review of these features in the previous NRC report needs only minor supplementing here. The important task in the present review is to judge how well the newer work supports and remedies weaknesses in the technical basis for the original conclusions about durability of canisters and bentonite. Major considerations are the feasibility of canister construction, the mechanical behavior of canisters enclosed in compacted bentonite, the resistance of the canisters to corrosion and to stress-corrosion cracking, and the ability of the bentonite buffer to control access 30
31 of corrosive agents in the groundwater to canister surfaces. FABRICATION OF CANISTERS Can copper canisters of the proposed sort be fabricated using state-of-the-art technology? In the KBS-2 plan, canisters with walls 0.2 m thick were to be formed by forging cast ingots of oxygen-free high conductivity (OFHC) copper, establishing a suitable grain size, and machining both inside and outside to close dimensional tolerances. The thick walls were designed to minimize gamma-ray radiolysis of water at the canister surface and to insure against penetration by deep pit- corrosion. More recent studies have shown that radiolysis is adequately controlled by smaller thicknesses and that pit corrosion is slower than originally thought; the planned thickness has accordingly been cut to 0.1 m. The method of forming the canisters remains the same, and its feasibility has now been demonstrated at full scale in Swedish metal-working plants (KBS-3, p. 4:6/14; TR 83-20). In KBS-2, after fuel rods were placed in a canister, the space around them was to be filled with molten lead, and a copper lid in three layers was to be sealed in place by electron-beam welding. Use of this procedure with canister walls 0.2 m thick was beyond the state-of-the-art at the time of KBS-2, and some question remained about its feasibility; but now that the thickness is smaller, it is well within known industrial capability. All steps in the process have been demonstrated using simulated fuel rods, and procedures for evaluating the welds and for quality assurance in the whole operation have been spelled out and tested in practice (KBS-3, p. 10:5; TR 83-20, TR 83-25, TR 83-32). An attractive alternative process is described in KBS-3. Instead of molten lead, copper powder would be used to fill the remaining space in the canisters, after which they would be sealed and the copper powder compacted by hot-isostatic pressing. This process uses a combination of high temperature and isostatic gas pressure (usually an inert gas), and would produce, in effect, a solid mass of copper around the fuel rods. Hot-isostatic pressing is a widely used industrial process, and existing furnaces in Sweden are large enough to accommodate the 0.8-m diameter proposed for the canisters. So far as the panel is aware, no existing furnaces are tall enough to handle the 4.5-m
32 height of the canisters, but there is no known functional reason why these could not be built. Although not yet demonstrated at full scale, the proposal to fill the canisters with copper powder followed by hot-isostatic pressing seems reasonable. Whether this method or the molten lead method will ultimately be favored remains uncertain, but KBS scientists in conversation (September 1983) expressed a preference for the copper powder method. MECHANICAL CAPABILITY Is the mechanical capability of the canisters-plus-buffer sufficient to withstand remotely possible tectonic displacement? Canisters of either type, filled with lead around the fuel rods or filled with copper powder and isostatically compressed, are solid cylinders of metal 4.5 m long and about 80 cm in diameter. They will be placed in vertical holes, with a diameter of 1.5 m and a depth of nearly 8 m, drilled in the floors of tunnels in solid rock. The holes are to be lined with shaped blocks of compacted bentonite, and each hole will be capped with a cover of concrete or granite. The bentonite will swell as water seeps into the holes, eventually filling all open spaces and exerting an external pressure on the canisters estimated to be 10 to 15 MPa. (One megapascal (MPa) equals 10 bars equals 9.87 atmospheres equals 145 psi). The canisters plus bentonite, from a mechanical stand- point, will be effectively a part of the rock. A question then arises as to how effectively this combination would resist a displacement of the rock resulting from fault movement. The worst imaginable case would be a horizontal fault cutting many canisters in an array roughly at their mid-points. If the displacement is slow, the copper and bentonite would presumably be malleable enough to adjust to the deformation without serious rupture. But a sudden movement, as in an earthquake, might well fracture both the bentonite and the canisters and provide access for groundwater to the fuel rods of many canistersâjust as the opening of a fracture anywhere in the rock would furnish a new channel for groundwater flow. In the opinion of Roland Pusch, the Swedish authority on bentonite, this possibility is the most serious weakness of the KBS plan (Roland Pusch, University of Lulea, personal communication, 1983). For the thick-walled canisters of KBS-2, Pusch estimated that
33 a sudden displacement of not much greater than 0.03 m might be enough to cause serious damage; for the thinner- walled canisters of KBS-3, he would reduce the estimate to 0.01 m (TR 83-47). He hopes to devise a full-scale experiment for testing his estimates of the mechanical response to various kinds of deformation of canisters enclosed in bentonite. While agreeing with Pusch that uncertainty about the response of tightly held canisters to sudden displacement is a real weakness in the KBS program, the panel notes that the probability of rock movement in precisely the right orientation and of the right magnitude and sudden- ness to do appreciable damage is exceedingly slight. Earthquakes in Sweden are uncommon, and recorded rock displacements of the last million years that can reason- ably be ascribed to earthquake activity are nearly all along planes that are steeply inclined rather than hor- izontal. Steep or vertical displacement would be less damaging to individual canisters, and would affect only a small number. The possibility that earthquake movement might damage many canisters in a repository sometime in the next million years cannot be entirely discounted. Pusch is right in noting the possibility and seeking to bound its effects experimentally, but the panel regards the possibility as too remote to constitute a serious drawback to the KBS plan. CORROSION RESISTANCE Will the canisters have sufficient corrosion resistance to prevent contact of groundwater with waste for a million years? Copper is thermodynamically stable in pure water under the proposed repository conditions, but groundwater in Swedish bedrock is far from pure. Analyses of water samples from the sites under investigation show enough potentially corrosive solutes to make questionable the ability of the canisters to survive for very long times if they were to be placed simply in contact with moving groundwater. The argument for minimal corrosion over hundreds of thousands of years is thus based, not on the thermodynamics of an unreactive metal in contact with relatively benign groundwater, but on the control of possible corrosive agents by the bentonite buffer and by reactions of the groundwater with bedrock.
34 The argument in KBS-3 follows the same lines as in KBS-2 (KBS-3, p. 10:8/15). The water that eventually makes contact with canister surfaces is the groundwater commonly found at depth in granitic rocks, its composition only slightly modified by the bentonite buffer. It is kept practically stationary within the buffer, and motion of solutes to and from the canister surfaces is limited to slow diffusion through 0.35 m of bentonite. The principal oxidizing agents of concern are dissolved oxygen and hydrogen ion in the presence of sulfide (the hydrogen ion being capable of oxidizing copper if the reaction is driven by the extreme insolubility of cuprous sulfide). Dissolved oxygen will be significant for a time after repository closure, because of residual air in all unfilled openings and in interstices of the backfill; but after natural groundwater conditions are reestab- lished, dissolved oxygen content will be kept very low by reaction with minerals containing ferrous ion in the surrounding rock. Sulfide is low in the natural ground- water because of reaction with ferrous ion to form insoluble ferrous sulfide, but dissolved sulfide can form in appreciable amounts by bacterial reduction of sulfate if organic material is present as nutrient for the bacteria. Sulfate is one of the prominent ions in the groundwater (up to 15 mg/1), and enough organic matter is commonly present (2 to 7 mg/1) to be a potential source of 1 or 2 mg/1 of sulfide ion; additional sulfate, sul- fide, and organic matter may be present as impurities in the bentonite. Dissolved sulfide can be kept low by oxidative preheating of the bentonite to remove most of the sulfides and organic matter, by using bentonite with a low content of sulfate, and by slowing access of ground- water from the backfill in tunnels above the canisters by placing a cap (copper, granite, or concrete) on each canister hole. These methods of control are estimated to be effective enough to limit the amount of copper con- verted to oxide or sulfide in a million years to no more than 30 kg per canister (out of a total single canister weight of about 20,000 kg) (TR 83-24). Other possible oxidizing agents (sulfate acting directly, nitrate, nitrite, and hydrogen ion in the presence of chloride) are readily shown to be unimpor- tant, either because of low concentrations or because of extreme slowness of reaction at temperatures below 80Â°C. Another possibility is oxidation by products of radioly- sis of water at canister surfaces (oxygen and peroxides, produced together with hydrogen). Calculations show,
35 however, that the amount of gamma radiation penetrating 0.1 m of copper would be capable of forming only enough radiolytic oxidants to combine with about 10 kg of copper in a million years, even if no recombination of the radiolysis products is assumed. One significant difference between KBS-2 and KBS-3 is the planned mixing of ferrous phosphate with the backfill material in the former and its absence in the latter. Earlier, the addition was intended to ensure that ground- water in the repository would remain reducing. Omission of the reductant in KBS-3 is in part the result of cal- culations showing that ferrous ion derived from iron minerals in the bedrock is sufficient to maintain reduc- ing conditions, and in part a response to concern that addition of yet another chemical species to the reposi- tory environment would complicate its chemistry unduly (TR-83-36). In assessing the potential effects of corrosion on the copper canisters, the KBS-3 report consistently takes a conservative stance, generally more conservative than the stance in KBS-2. Examples of the conservatism are as follows: 1. Oxidation of copper by dissolved oxygen can proceed only at high redox potentials (ca. +50 mV), whereas oxidation by hydrogen ion in the presence of sulfide requires much lower potentials (less than -200 mV); yet both reactions are assumed to go to completion in estimating the corrosion rate. 2. No correction is made for the copper powder added to canisters made with the isostatic compression process; the additional copper would consume reactants if the outer part of a canister should be breached by corrosion. 3. In calculating the possible effect of sulfate on canister corrosion, no credit is taken for the ferrous ion in the bentonite in controlling the availability of sulfide as a corrodant. 4. Kinetic studies in KBS-3 were made for the most part with classical, well-founded procedures, but some nontraditional techniques were employed involving the concept of "equivalent flow." At the panel's request the results were recalculated by Hong Lee (University of Florida, personal communication, 1983) using classical diffusion models. The rates obtained by Lee are close to those estimated by Neretnieks (TR 83-24); but where differences exist, the Neretnieks model predicts slightly higher corrosion rates.
36 By way of contrast, the important effect of pit cor- rosion is treated less conservatively in KBS-3 than in KBS-2. The latter uses a very conservative value of 25 for the pitting factor; but the KBS-3 authors point out that, in recent metallurgical studies by the Swedish Corrosion Institute (TR 83-24), a factor of 5 has been shown to be more reasonable for the two kinds of copper (oxygen-free high-conductivity copper and phosphorus deoxidized copper) that are contemplated for use in the canisters. A maximum pitting factor of 5 is also indicated by examination of archeological specimens, native copper, and buried lightning-conductor plates (TR 83-24). In KBS-3, corrosion rates are calculated using both of these factors, but the rates obtained with the larger factor are described as unrealistically high. Overall, the calculations of corrosion rates in KBS-3 are similar to those in KBS-2 but are improved by use of new data and new calculational methods. In the panel's opinion the calculations are soundly based and use assump- tions with a safely conservative bias. The ability of canisters surrounded by compacted bentonite in Swedish bedrock to last for a million years and more seems adequately documentedâprovided, of course, that the canisters are made according to rigid specifications and that the subsurface environment is not subject to radical change. STRESS CORROSION CRACKING Will the canisters be immune to stress corrosion cracking for a million years? One corrosion-failure mechanism that in theory might be capable of causing a catastrophic breach is stress corrosion crackingâthe cracking of metal as a result of exposure of a susceptible material in a specific environ- ment to an enduring tensile stress. The possibility of damage by such a mechanism under conditions to be expected in a Swedish repository was investigated by the authors of KBS-2 and shown to be negligible. The reason is that the necessary enduring tensile stress will not exist because of the method of canister fabrication, and in any event could not be maintained for long in a metal as readily stress-recovery-annealed as pure copper. Even if a repository were to be invaded by seawater when the land subsided under the weight of a glacier, the necessary oxidizing conditions and concentrations of nitrogen
37 compounds would not be attained. After careful review, the NRC subcommittee agreed that a good case had been made in KBS-2 for the conclusion that the rather specific conditions needed for initiating stress corrosion cracking would not be found in a KBS repository (NRC, 1980). The NRC subcommittee raised a possible question on the grounds that newly reported research had suggested some tendency for cracking of high-purity copper in certain nitrogen-containing environments. More recent work by Benjamin et al. (TR 83-06), however, showed that the electrochemical conditions necessary to stimulate cracking of pure copper in dilute sodium nitrite solutions are so remote from conditions expected in a repository that they eliminate stress corrosion cracking of the canisters as a source of concern. BENTONITE BUFFER Can the bentonite buffer around each canister and the bentonite-sand backfill in tunnels and shafts be depended on to keep the movement of groundwater along canister surfaces very slow and to maintain conditions in the repository that will prevent or greatly inhibit corrosion? In both KBS-2 and KBS-3 the canisters are to be sur- rounded by shaped blocks of compacted bentonite*; and after the canisters are emplaced, a mixture of bentonite and sand is to be used as backfill in the tunnels and shafts. As groundwater penetrates the repository after closure, the bentonite is expected to swell and fill all vacant spaces, thus forming a mechanical cushion for the canisters and serving as a barrier to rapid groundwater flow. At the canister surfaces, the expanded bentonite will, supposedly, keep groundwater motion very slow and prevent movement of possible corrosive agents except by slow diffusion through the clay. To some extent the buffer will also act as a retardant for radionuclide migration should early breaching of a canister occur; this possible function is discussed more completely in the section in Chapter 4, "Retardation in the Near-Field." *Bentonite is a naturally occurring material with considerable variation in composition and properties. The bentonite mentioned in this section and used as a reference material in the Swedish experimental work is a variety from Wyoming with a high ratio of sodium to calcium smectite.
38 In KBS-2 an impressive array of experimental and analytical data was presented to support the thesis that the bentonite would indeed perform up to expectations under the chemical, thermal, and radiation conditions anticipated in a repository. The data were in large part obtained by Pusch and his colleagues at the university of Lulea. In more recent years, Pusch has continued his intensive studies of the properties of bentonite and other clays, both at Lulea and in the underground experimental facility at Stripa. The present review is devoted largely to a survey of the recent findings of Pusch and his co-workers. Some aspects of the emplacement of bentonite were questioned by the NRC subcommittee on KBS-2, particularly the feasibility of placing bentonite blocks in the holes around the canisters without leaving large open spaces, and the feasibility of compacting the backfill in the upper parts of tunnels. Both questions have now been answered by full-scale experiments at Stripa: canisters can be successfully emplaced, either by building the bentonite-block walls around them or by constructing the walls first and then inserting the canisters; and the emplacement of backfill in the upper parts of tunnels, with a final density only slightly less than that of the compacted material below, can be accomplished by a shotcrete process (Stripa 82-06, Stripa 82-07; also, presentation by Pusch to the panel, accompanied by photographs of the operations, September 1983) . The mechanical properties of highly compacted, water-saturated bentonite are of interest in ensuring that canisters do not slowly sink through the bentonite to the bottom of their emplacement holes, and that they will be to some extent protected from shear stresses that may develop in the surrounding rock. A theoretical analysis backed by laboratory experiments has shown that the swelling pressure of the bentonite (10 to 15 MPa under repository conditions) is sufficient to largely restore the original state of stress in the nearby rock. The material is strong enough that the calculated rate of canister subsidence is no more than about 0.01 m in a million years, yet plastic enough to give protection against slight rock displacements (TR 83-04, TR 83-47) . As noted in the section in Chapter 3 entitled "Mechanical Capability," the protection against slow moderate dis- placement is adequate, but a question remains about possible rupture of a canister by rapid horizontal rock movement of more than 0.01 m.
39 Compacted bentonite under its own swelling pressure has been demonstrated to flow into cracks less than 1 mm wide and to seal them. It has been further shown experi- mentally that moving groundwater in such cracks will not, at a later time, dislodge appreciable amounts of the bentonite filling (TR 83-04). The ability of bentonite to control groundwater movement and the movement of dissolved ions is shown by measurements of hydraulic conductivity and diffusivity. For bentonite compressed to a density of 2.0 t/m^ (tonnes per cubic meter, an SI unit numerically equiva- lent to grams per cubic centimeter), Pusch gives a conductivity of 5 x 10~14 m/s (TR 80-16) , a value corroborated, within an order of magnitude, by recent experiments in the United States (Peterson and Kelkar 1983) . For bentonite-sand mixtures like those to be used for backfill, Pusch and Borgesson (Stripa 82-06) give experimental values for hydraulic conductivity: With 10 percent bentonite, at a density of 2.1: 10~9 m/s With 20 percent bentonite, at a density of 2.1: I0-10 m/s In similar experiments, Peterson and Kelkar found With 10 percent bentonite, at a density of 1.98: 6.1 x 10~9 m/s With 30 percent bentonite, at a density of 2.29: 2.2 x 10~12 m/s The agreement is satisfactory, and the range of values is similar to that for the matrix of granitic rock between major fractures. This finding indicates that the backfill in tunnels and shafts will have a permeability comparable to that of its surroundings. Because of the low hydraulic conductivity, motion of ions through the water-saturated bentonite will be dominated by diffusion; for the dif- fusivities of different ions, Pusch gives values in the range 8 x 10~12 to 5 x 10~ll m2/s. The chemical properties of bentonite vary considerably from one sample to another. In KBS-2 the favored ben- tonite was a variety from Wyoming rich in Na-smectite; this material is still used as a reference standard in KBS-3, but a search has been undertaken to find sources that are cheaper and closer to Sweden. Pusch has examined bentonites from many localities, mostly European, and
40 finds much variation in the ratio of Na-smectite to Ca-smectite, the ratio of smectite to other clay min- erals, and the content of such impurities as quartz, carbonate, sulfides, sulfates, and organic matter (TR 83-46, and personal communication, 1983). For most repository purposes, a fairly pure Na-smectite is prefer- able. The composition of bentonite can be somewhat modified, if desirable, by currently available commercial processes: sodium may be substituted for calcium, and much of the sulfide and organic impurity can be removed by several hours of oxidative heating to about 400Â°C. At the time of the KBS-2 plan, Pusch had reservations about the possible effect of heating bentonite to 400Â°, on the grounds that its structure might be irreversibly altered and its expansive properties impaired; but results of recent experiments have convinced him that the change in properties is minor if the heating is not continued too long. No final choice among the varieties of bentonite has yet been made, but Pusch is confident that one or more good sources can be found and that the composition can be altered, if necessary, to meet KBS requirements. Whatever the source of bentonite, the KBS authors think it can be depended on to keep the rate of corrosion low. Presumably, a bentonite with low sulfate content will be selected, and its sulfide and organic content will each be reduced below 200 ppm by oxidative heating. The buffer will limit the supply of water reaching the surface of canisters to a calculated 0.2 to 1.6 1/yr/ canister (KBS-3, p. 20.15), and the amount of oxidants in this quantity of water would be trivial. Bentonite is also known to buffer the pH of contained water to values between 8 and 9.5, thus effectively preventing possible corrosion by hydrogen ion in the presence of chloride even up to seawater concentrations of chloride; if the buffer capacity of the bentonite is ever used up (doubt- ful, because the pH of the groundwater is generally between 7.5 and 9.5, so that little buffering is needed) , the bedrock itself would serve to hold the pH in this range. Possible oxidation of copper by the products of radiolysis is largely prevented by the buffer's ability to slow the escape of hydrogen, and hence to promote the recombination of the radiolysis products. Thus, if the buffer behaves even approximately as expected, the rate of corrosion of canister surfaces will be kept to a very small value. In summary, the panel believes that the experimental data and calculations described in KBS-3 and its tech-
41 nical documents effectively support the KBS claims that, in the system of copper canisters plus bentonite, the canisters are adequately protected from corrosion and stress. The swelling pressure, bearing capacity, and plasticity of the bentonite will provide good mechanical protection; its low hydraulic conductivity will serve as a barrier to active groundwater flow and will maintain a controlled low-corrosion environment immediately adjacent to the canisters.