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.
2 ADEQUACY OF TREATMENT OF GEOLOGY AND HYDROGEOCHEMISTRY The disposal system proposed in the KBS reports depends for its effectiveness primarily on the long-term integrity of copper canisters surrounded by bentonite and placed deep in crystalline bedrock. Canister integrity is a function of both the nature of the canister material and the environment in which it is placed. To demonstrate that a satisfactory environment for the canisters will exist over the long term requires showing that (1) masses of sound bedrock can be found of sufficient size and sufficiently low permeability to serve as repository sites; (2) the characteristics of groundwater under natural conditions (quantity, rate and direction of movement, and chemical composition) can be predicted and will not affect a copper canister adversely; (3) ground- water movement will not be adversely influenced by con- struction of a repository or by the heat generated from the radioactive waste; (4) tectonic movement in bedrock will not seriously damage the repository; and (5) pos- sible renewed glaciation will not adversely affect the repository. Evidence in support of these five conclu- sions, as marshaled in the KBS-2 report, was critically examined in the NRC review of that document and found to be generally convincing, although some gaps and weak- nesses were noted. The purpose of this chapter is to take a fresh look at the KBS-2 report conclusions in the light of research over the past five years. REPOSITORY SITES Can bodies of crystalline bedrock be found in Sweden of sufficient size and sufficiently low permeability to serve as repository sites? 14
l5 KBS-2 described exploration at four possible sites (Karlshamn, Krakemala, Finnsjon, and Stripa), including details of surface geology, study of cores from deep boreholes, geophysical measurements both on the surface and in boreholes, and measurements of hydraulic conduc- tivity at many places in the boreholes. Bedrock at all the sites consists of granite and metamorphic rock cut by many fractures and zones of crushing, with relatively high permeability in some of the fractures and very low permeability in the solid rock between. The problem at each site was to determine whether a body of rock existed at a depth of roughly 500 m sufficiently free of fractures and large enough (covering an area of at least 1 km2) to permit excavation of a repository. To make such a judgment requires geologic inference from the surface studies and borehole data, and with current technology the data so obtained can never be sufficient to make the judgment with complete confidence. The KBS authors con- cluded that the number and permeability of fractures at three of the areas made them questionable as repository sites, but the Karlshamn area appeared to have a volume of sound rock large enough, although barely so, to accom- modate a repository. The NRC subcommittee concurred in this evaluation of the geologic data. The authors of KBS-3 retain the Karlshamn site (also referred to as the Starno site) as a good possibility (TR 83-56), and they describe exploration in four additional areas (TR 83-40, TR 83-52, TR 83-53, TR 83-54, TR 83-55). One of the new areas was eliminated quickly on the grounds that it had too many water-bearing fractures, but the other three looked promising enough to warrant a more detailed study than had been attempted at Karlshamn. In particular, more long boreholes were drilled, most of them to depths greater than 500 m, and at various angles so that they would intersect differently oriented frac- ture zones. This provided a partial answer to an objec- tion that had been raised regarding the earlier work, that the boreholes at Karlshamn were too few to justify the attempted geologic extrapolation. Data from the three new sites (Fjallveden, Gidea, and Kamlunge) indicated rocks and structures much like those at Karlshamn; again at these sites the volumes of relatively unfractured rock estimated to be present at appropriate depths are barely large enough for repository construction. The new studies, like the older ones, start with geologic mapping from surface exposures, aided by tracing of lineaments on air photos; they emphasize appropriately
16 the nature and spacing of fracture zones. Based on the surface maps, boreholes can be located and oriented so as to intersect fracture zones at various depths, making it possible to construct a reasonably complete three- dimensional model of the fracturing. Geophysical tech- niques are used both at the surface and in boreholes to test the properties of individual fractures and crush zones and to explore for others that may have been missed in the mapping and drilling. Cores from the longer boreholes permit detailed study of rock compositions and textures to depths of 500 m and more. Hydraulic conductivitiesâboth specific values for individual fracture zones and average values for large volumes of rockâare measured by standard techniques on sections of the boreholes isolated by packers. Application of these various techniques by Swedish geologists and geophysi- cists, as described in KBS-3 and its supporting technical documents, is fully up-to-date and consistent with modern practice in other countries. Thus the KBS-3 authors have apparently accomplished their purpose, which is simply to show that places can be found in Sweden where the bedrock has the necessary properties for repository construction. None of the places examined so far is ideal, and active investigation is under way to find better ones. The investigation is handicapped by the reluctance of many property owners to permit detailed study and drilling. (Although this is not clear from the written report, in conversation the KBS authors say that often an area that looks especially promising must for this reason be passed over for a second-best site.) But the four sites that have been studied in detail seem sufficiently attractive to ensure that, when the time comes to actually select a repository location (at least a decade hence), a good one will be found in Sweden. Estimating geologic conditions deep underground from surface observations and a few boreholes, even when aided by geophysical data, is always somewhat speculative. Details of underground conditions can be ascertained only by sinking a shaft and exploring laterally from its bottom, and surprises are to be expected in such explora- tion. A skeptic can always find grounds for criticizing geologic extrapolation; but where sinking a shaft is not practical, there is no alternative to dependence on surface and borehole data. In the panel's opinion the KBS authors have acquired about as much information from such data as it is practical to obtain and have made a good case for their conclusions.
17 The weakest point of the KBS argument relates to infer- ences it draws about amounts and flow of groundwater over long future times from measurements of hydraulic conduc- tivity in intricately fractured rock. The adequacy of the factual basis for these inferences is examined in the next section. GROUNDWATER IN ITS NATURAL STATE Can the quantity, rate and direction of movement, and chemical composition of groundwater in granitic rock be predicted from measurements made on or near the surface, together with data from a few boreholes? Groundwater is expected to permeate a repository after it has been filled and sealed, and the preconstruction pattern of groundwater flow will reestablish itself within a time that is short in comparison with the very long half-lives of many of the interred radionuclides. If the movement of groundwater is slow, the bentonite around each canister should be stable indefinitely, and the small amount of water that diffuses through the buffer will have little effect on the copper canisters. More rapidly moving water, however, might eventually erode channels in the bentonite and bring corrosive agents into contact with the copper surfaces. To demonstrate that suitable repository sites can be found, therefore, requires that detailed information be acquired about amounts, movement, and chemical composition of groundwater at each candidate site. Efforts to obtain such information have been a major part of the KBS proj- ect, assigned by the KBS staff to geologists, hydrolo- gists, and geophysicists of the Swedish Geological Survey. Groundwater movement in granite and other crystalline rocks is largely confined to fractures, which commonly can be seen to cut the rock in many directions. The rock itself has very low permeability, but parts of some frac- tures may carry a good deal of water. Studying ground- water movement is thus chiefly a matter of determining the fracture pattern over an area and estimating flow rates in individual fractures or groups of fractures. The problem is difficult because actual observations must be limited to rock exposures on the ground surface and to measurements in boreholes, and the number of boreholes must be kept small to avoid perforating a potential repository site with too many openings. Observations may be supplemented by geophysical measurements, both on the
II surface and in boreholes; but even with a combination of the most modern techniques, discerning complex fracture patterns at depths of a few hundred meters is subject to much uncertainty. Eventually, when an actual repository site is to be explored in detail, sinking a shaft and driving tunnels will make possible more accurate mapping of fractures and measurements of groundwater flow; but the preliminary studies of the KBS project must be carried out without this advantage. The authors of KBS-2 were well aware of the difficul- ties they faced in predicting groundwater conditions at depth from surface and borehole observations, and critics of the KBS-2 plan identified this as one of its most vulnerable parts. The NRC subcommittee, after extensive review of the technical documents and conversations with the KBS investigators, decided that the geological and hydrological work was well done, that enough conservative assumptions had been made to compensate for the uncer- tainties, and that the measurements were "adequate to ensure that the quantity of water reaching the copper canisters in a well constructed repository will not be damaging." The additional work reported in KBS-3 is not qualitatively different from that done earlier, but the methods of measurement have been refined and abundant new data have been gathered. Groundwater conditions at the four new sites discussed in KBS-3 (Fjallveden, Gidea, Svartboberget, and Kamlunge) and at the experimental site at FinnsJbn have been exten- sively studied (TR 83-43, TR 83-45, TR 83-52, TR 83-53, TR 83-54, TR 83-55). A critique of the treatment of groundwater flow in KBS-3 involves three questions: (1) Are the reported measurements of hydraulic conductivities accurate? (2) Are the interpretations of these measure- ments valid? (3) Are the numerical predictions of flow rates likely to be reasonably accurate? 1. Measurement of hydraulic conductivity (K). The bulk of the measurements used to estimate values of K were made, as in KBS-2, by injecting water into sections of boreholes sealed off with inflatable rubber packers. The test sections ranged in length from 2 to 25 m (TR 83-45) . The value of K for each isolated section, expressed in terms of that for an equivalent homogeneous porous medium, is inferred by analysis of recorded values of the amount of water injected and the time variation of pressure in the borehole section. This is a standard procedure for such measurements. There are potential
19 difficulties with measurements of K made in this fashion, but, unless there is a "borehole skin" effect, errors should not be unreasonably large (TR 82-06). The packer tests yield values of K that are representa- tive of the rock in the immediate vicinity of the bore- hole. In relatively unfractured rock (K values of 10~9 to IfTll meters/second (m/s)), the effective radius of influence in the tests ranges from 0.15 to 1.5 m (TR 83-43). To estimate the hydraulic conductivity of larger rock masses, an alternative method, referred to as an "interference test," was used. In interference tests, pressure changes in boreholes are recorded as a result of injection or pumping carried out in a nearby borehole. Such tests were conducted at all four sites but were limited to the upper part of the bedrock (0 to 150 m) because the necessary equipment required boreholes of large diameter that could be driven only to this depth, at the time when the testing was being performed (Leif Carlsson, Swedish Geological Survey, personal com- munication, 1983; TR 83-43, TR 83-45). The K value measurements reported in KBS-3 are more numerous than those in KBS-2 because more deep boreholes (to depths of 600 m and more, well below the chosen repository level of 500 m) were drilled at each site. In the opinion of the panel, the measurements are consistent with the current state-of-the-art. Within the limits of current understanding of flow in fractured media, the measurements themselves can be considered accurate. 2. Interpretation of measured values of K. The values of K are highly variable, as might be expected in rock cut by fractures of varying width, varying orienta- tions, and varying amounts of secondary filling material (chiefly clay minerals, chlorite, and calcite). For sections of a borehole cutting rock in which fractures are absent or tightly sealed, K values are at or near the limit of measurement, 10~11 m/s; for sections with more open fractures, values range up to 10~6 m/s or higher. The higher values in all of the studied areas are largely limited to the upper 300 m, below which level values greater than 10~Â° m/s are very few. This distribution has led the Swedish scientists to fit exponentially decreasing curves to the data on K versus depth. The curves permit estimates of average hydraulic conduc- tivities for different depth ranges in the rock, but many points deviated considerably from the averages indicated by the curves.
20 One can question the meaning of such averages, on the grounds that all or nearly all of the flow represented by an "average" for a given rock thickness might be concen- trated in a few highly conductive fractures. If a repository were intersected by such fractures, it might conceivably be damaged by the concentrated flow even though the predicted "average" flow was suitably small. Damage from such flow seems unlikely, however, because measured hydraulic conductivities in the depth range 400 to 600 m, even for the most conductive fracture zones, are nearly all less than 10~8 m/s. On the assumption of a hydraulic gradient of 0.1 percent for these depths (estimated from the leveling-out, with depth, of gradients observed near the surface), the conductivity would give a flow rate of only 0.3 liters/square meter/year (1/mvyr) . A more conductive fracture or fracture zone would pre- sumably either be sealed or avoided in selecting the location for the repository. The exponential curves and the estimated average hydraulic conductivities seem a reasonable interpretation from the measured K valuesâ provided, of course, that the uncertainties are recog- nized and conservative values are used. One criticism of the treatment of K values in KBS-3 relates to the interference tests. Although these tests were evidently performed at each site, the data are not presented or discussed. If the "average" values of K indicated by the exponential curves are a valid represen- tation of the large-scale, three-dimensional flow in granite, then the values of K derived from interference tests should be close to those predicted by the regres- sions. In the only published comparison of values derived from the two methods of testing, at the Studsvik experimental site, hydraulic conductivities given by interference tests were considerably larger than those given by packer tests (TR 82-10). All more recent work, however, indicates excellent agreement between results of the two methods. This includes unpublished analyses of measurements at the four recently studied sites (Hans Carlsson, KBS, personal communication, 1984) and an additional unpublished study at the Stripa experimental facility (Leif Carlsson, Swedish Geological Survey, personal communication, 1983). Thus in five of six comparisons between the two methods, agreement is reported to be satisfactory. On the basis of these reports, but without the opportunity of checking the actual data, the panel considers that the KBS interpretation of the measured hydraulic conductivi- ties is reasonable.
21 3. Numerical prediction of flow rates. Groundwater flow rates under natural conditions are predicted in KBS-3 by using a three-dimensional finite-element approximation to solve the governing equation (KBS-3, p. 6:1/21; TR 83-45). This equation (i.e., the theory behind the calculations) assumes that Darcy's law provides a valid description of flow in fractured granite. The assumption is that a large mass of granite with an interconnected network of fractures will, in fact, behave like a porous medium if the spatial scale is large enough. The panel agrees that this assumption is reasonable for a rough calculation of discharge per unit area ("Darcian veloc- ity") through a region. However, the application of such calculations to actual velocities in fractures is prob- lematic because of the difficulty of accurately estimating the "effective porosity" associated with the hydraulically connected, conducting fractures. The use of finite-element models in groundwater hydrol- ogy is by now almost routine. They can be used for accu- rate calculations of groundwater flow if (1) the under- lying theory (Darcian flow) is valid; (2) the hydraulic properties of the medium (most importantly, conductivity) are specified correctly; (3) the boundary conditions (impervious basements, flow divides, and so on), including the geometry of the boundaries, are known; and (4) the elements are of an appropriate size and shape to yield an accurate numerical solution. The first two items have been discussed above. Geological boundaries can almost never be specified with great precision; and so additional uncertainty, over and above that caused by difficulties in measurement and interpretation of K values, is intro- duced by this incomplete knowledge. Finally, because of computer limitations, the element grid used in KBS-3 was rather coarse. In some cases, this resulted in a physi- cally implausible solution (i.e., one in which conserva- tion of mass of groundwater was violated) and/or in failure of computed groundwater trajectories to be consistent with the specified boundary conditions (TR 83-45). Similar problems beset all efforts to model ground- water flow. To calibrate and validate numerical models, extensive field data are needed, especially data on groundwater heads over time. Such data were not available prior to preparation of the KBS plan. Thus the calculated results must be regarded as no more than semiquantitative. The lack of precise numbers, however, is not serious for purposes of safety evaluation. This can be illustrated
22 by relating flow rates to hydraulic conductivities and hydraulic gradients in a simple application of Darcy's law. Average hydraulic conductivities (from the packer tests) lie between 10~8 and 10~10 m/s; hydraulic gradients at repository depths are lower by 1 or 2 orders of magnitude than surface slopes, which in the subdued topography of eastern Sweden are unlikely to be greater than 10 percent. For various combinations of these two variables, Darcy's law gives the flow rates shown in Table 2-1. Now the flow rates calculated by the model for the different areas lie between 0.01 and 0.06 l/mVyr. As Table 2-1 shows, to get flow rates substantially higher than these values would require unrealistic combinations of gradient and conductivity. Thus, in the Panel's opinion, the flow rates cal- culated in KBS-3, although subject to a good deal of uncertainty, are derived by reasonable methods of calculation using the available data. Studies of groundwater chemistry, like studies of groundwater motion, have added little that is new between KBS-2 and KBS-3, but they have introduced significant refinements of method and have added greatly to the data base (KBS-3, p. 7:1/10; TR 83-34). Especially notable have been efforts to improve and check measurements of Eh (redox potential), a notoriously difficult and often unreliable part of water analysis (TR 83-40). By com- paring readings obtained with two kinds of electrodes, platinum and amorphous carbon (gold was also tried, but found to be subject to sulfide poisoning), and by check- ing the readings against Eh values calculated from overall groundwater composition, the KBS authors have verified that their listed Eh values are meaningful. TABLE 2-1 Plow Rates or Flux (l/m2/yr) as a Function of Gradient and Hydraulic Conductivity Hydraulic Conductivity Gradient 10~10 m/s 10~9 m/s 10~Bm/s 1.0% 0.1% 0.01% 3xl0~2 3xl0-3 3xl0~4 3X10~1 3xl0~2 3xl0~3 3 3xl0~ 1
23 Measurements of dissolved sulfide, made rapidly with an Ag/Ag2S electrode, have likewise been compared to results from titration and found to be valid and repli- cable. The extent to which samples may have been contaminated by drilling fluid, even after lengthy flushing of the water sources, was checked by analysis for iodide added as a tracer to the water used for drilling. The many new analyses for granitic bedrock in the four recently studied areas (TR 83-17, TR 83-19, TR 83-4l, TR 83-59, TR 83-70), carried out with these several improvements, corroborate earlier work. Except for a very few aberrant figures, values of pH lie between 7.5 and 9.5, of Eh between 0.0 and -0.3 volt, of HS~ below 0.1 mg/l, of Fe2+ between 0.1 and 7 mg/l, and of total organic carbon between 2 and 7 mg/1. The uniformly slightly alkaline and slightly reducing character of groundwater from depths of a few hundred meters in Swedish bedrock seems abundantly confirmed. The water is in equilibrium with its bedrock environment, in the sense that the concentrations of silica and the major cations (Na+ , K+, Ca++, and Mg++) are within the fields of stability representing equilibrium with the common altera- tion minerals found in the fractures and crush zones (kaolinite, chlorite, smectite, laumontite, and calcite) (TR 83-59). In summary, work on groundwater during the past five years has added substantially to support for the con- clusion that reasonable estimates for the quantity, rate of movement, and chemical character of groundwater at depths of a few hundred meters in crystalline rock can be made from careful geologic mapping at the surface, logging of boreholes, examination of drill cores, a full panoply of geophysical tests, measurements of hydraulic conduc- tivity, and chemical analyses of many samples. Uncer- tainties still remain about interpretation of the widely variable values for hydraulic conductivity and about the validity of using Darcy's law for calculating flow in fractured rock; but the uncertainties have been narrowed by comparing results of different measurement methods, and they can be partially compensated for by use of con- servative values. The panel agrees with the KBS scien- tists that the accumulated evidence at most of the sites selected for detailed study points strongly to groundwater behavior and groundwater composition favorable to the location of waste repositories of the kind envisioned in the KBS-3 plan.
24 EFFECTS OF REPOSITORY CONSTRUCTION ON GROUNDWATER Can adequate predictions be made of the effect on groundwater movement of disturbances caused by a repository, particularly the fractures due to blasting during excavation of shafts and tunnels and the heating caused by the buried waste? Construction of a repository will necessarily disturb the natural flow of groundwater, and the question arises as to whether the disturbance will persist, after the repository is closed, in a form serious enough to affect the buried canisters. Part of the KBS design for a repository is the use of a bentonite-sand mixture for backfill, which is intended to have a hydraulic conduc- tivity comparable with that of the original rock. If the backfill behaves as expected, the long-term effect of the repository on groundwater flow should be small. (The effectiveness of bentonite as a barrier in backfill and seals is considered more fully in the section in Chapter 4, "Retardation in the Near-Field.") Questions are still pertinent, however, about possible increased flow of groundwater through the rock material adjacent to shafts and tunnels that was fractured by blasting during construction, and about possible thermal convection cells set up by heat from radioactive material in the canisters. Both questions were treated at length in KBS-2, and little new has been added in the work for KBS-3. Oral statements by KBS personnel (September 1983) indicated that recent experimental work at Stripa has verified earlier predictions that the very localized fracture enhancement near shafts and tunnels would not greatly compromise the low hydraulic conductivities of the rock matrix and thereby disrupt the existing regional flow. A recent study of thermal circulation by means of a mathematical model (TR 80-19) supports KBS-2 estimates that the low planned temperature of a KBS repository (maximum designed temperature of 80Â°C) would not be sufficient to cause appreciable convection. The calcu- lations indicate, in fact, that movement of groundwater in a repository located beneath a hill would actually be slowed by a slight increase in temperature. The panel agrees with the earlier NRC subcommittee that a good case has been made for lack of adverse effects on the natural movement of groundwater caused by the presence of a repository.
29 TECTONIC STABILITY Can areas of bedrock be found in Sweden where the possibility of tectonic movement that might damage a repository is negligible? The crystalline bedrock of Sweden is part of the Baltic Shield, a large area of Precambrian rock that is generally recognized as one of the most stable parts of the earth's continental crust. Large faults with spacings of kilometers criss-cross the shield, cutting it into blocks within which fracturing is minor. Displacement on the faults, both major and minor, records movement that was largely completed in Precambrian time, more than 600 million years ago. More recent movement in bedrock has been restricted to small displacements along some of the old faults: in effect, the blocks show some evidence of jostling about over geologic eons, but interiors of the blocks have been largely unaffected. Present-day earth- quakes in Sweden are few and small (Richter magnitudes seldom greater than 4); and where resulting fault dis- placement from individual quakes can be found, it is limited to a few centimeters, again on a few of the ancient fault lines. From such evidence, the KBS-2 authors concluded that repositories placed well within the stable blocks are in little danger of disruption by tectonic movementâeither slow deformation or the sudden displacements accompanying earthquakesâand the earlier NRC subcommittee agreed that evidence for the conclusion was convincing. Research in the last five years has only further con- firmed the essential stability of Swedish bedrock. Much attention has been given to movement of some of the blocks over the past few hundred thousand years resulting from the melting of successive Pleistocene ice sheets and the consequent periodic release of bedrock from the load of 3-km-thick masses of ice (KBS-3, p. 8:14/19; TR 83-57, TR 83-58). Interest has also focused on attempts to relate fault movements and measured bedrock stresses in Sweden to large-scale movements of crustal plates, par- ticularly to possible compression of the Scandinavian peninsula against the Baltic Shield by plate movement away from the Mid-Atlantic Ridge (Bath, 1983). Bath also reviewed recent work on seismic risk: he estimated the magnitudes of possible future earthquakes for many areas covering 1Â° latitude by 2Â° longitude and found for the most active of such areas maximum expectable magnitudes of 5.1 over a period of 100 years and 6.4 over 500 years.
26 For roost of the areas, the estimated extreme magnitudes are much smaller. Bath recommended that waste reposi- tories should be located away from active fracture zones, and he suggested a depth of 1400 m rather than the 500 m planned by KBS, on the grounds that seismic waves show a marked increase in velocity at about this depth and thus imply a decrease in rock fracturing. Bath noted also that seismic activity is concentrated in a zone running north-northeast through central and eastern Sweden, close to the Baltic shore north of Stockholm. It is doubtful that earthquakes would appreciably damage a repository built within a stable block, even if it were located in this seismic zone (as it would be at Gidea or Kamlunge), but the very slight risk could be avoided by placing the repository near the less seismic southeastern coast (for example, at Karlshamn or Fjallveden). A principal critic of the KBS-2 report was N.-A. Mbrner, who claimed that evidence for the tectonic stability of Sweden was far from convincing (KBS TR 18). He based his skepticism chiefly on many years of study of the levels of old marine terraces. The unloading of Scandinavia by the melting of the last Pleistocene ice cap has resulted in a slow elevation of the land with respect to sea level, an elevation recorded in terraces marking old stands of the seashore. MSrner's study of the terraces enabled him to follow details of the uplift. He concluded that the uplift could not be entirely explained by glacial retreat and that it must also involve a tectonic component of unknown origin that is still active. This led him to look for evidence of recent tectonic activity, and he thought he could see such evidence in small faults cutting glaciated surfaces and huge blocks of bedrock in some glacial moraines. In Morner's view, Sweden has been active tectonically since the end of glaciation, and such activity will be heightened in the next glacial period, so that locating waste repositories in Swedish bedrock would be foolhardy. In the intervening five years, Morner has given no further expression to his skeptical opinions. The panel had no opportunity to interview him, and inquiries of other Swedish scientists elicited only the speculation that he had either changed his views or tired of urging them in the face of steadily mounting contrary evidence. It may also be, however, that Mbrner feels partly jus- tified by a change in the climate of opinion in Sweden. Bath (1983) noted, for example, that most Swedish
27 scientists a decade or so ago ascribed all the post- Pleistocene uplift of Scandinavia to glacial unloading, whereas they are now ready to admit a tectonic contribu- tion in some obscure way related to plate movements. Other recent studies (TR 83-57, TR 83-58) have also shown that Mttrner is certainly right in his insistence on some late Quaternary differential movement in bedrock. The best evidence for such movement comes from far northern Sweden, where scarps cutting glacial deposits record ver- tical displacements of up to 30 m in fairly recent geo- logic time. The movement for the most part is along old fracture zones, although some smaller fractures have cut fresh rock. Interpretation of the precise timing and nature of the movement is uncertain. Bath speculated that this is a region where plate motion eastward from the Mid-Atlantic Rift abuts motion southward from rifts in the Arctic Ocean. In any event, these faults are far from any of the studied repository sites except Kamlunge, and no post-Pleistocene displacements nearly this large are known in other parts of Sweden. Despite these exceptional faults in the far north, there seems no longer to be any expressed dissent among Swedish scientists to the prevailing view that the risk of tectonic disturbance to a properly located repository is negligible. The panel regards this view as well founded. More details about post-Pleistocene tectonic movements would certainly be desirable. The probability of future disturbance could be better estimated, for example, if it were known whether the dominant kind of fault movement is normal dip-slip, reverse, or strike-slip. Bath noted that such information is hard to obtain, because Swedish earthquakes are so small and seismographs so widely spaced that good first-motion data are very scarce. He hazarded a guess, based on recent work, that strike-slip movement may be the commonest kind, at least for the larger quakes, but cited other evidence from a few faults for reverse movement on steep eastward-dipping planes. Additional studies of this sort would be of much scientific interest, but it seems doubtful that they would add a great deal to the current assurance that future tectonic activity will be too feeble to harm a repository. The panel concludes that the assurance is adequately supported by currently available evidence.
28 GLACIATION Can the effect on a repository of renewed glaciation be shown to be negligible? Development of another ice sheet in Scandinavia is probable during the next few tens of thousands of years, and its possible effect on a waste repository was con- sidered in great detail by the authors of KBS-2. Not only were the expectable surface erosion and deposition by the ice and its accompanying meltwaters a subject of exhaustive study, but also the effects on groundwater, the effects of the weight of ice on bedrock at depth, the bowing down of the land suface beneath a thick ice cap and the consequent probable incursion of seawater, and the generation of earthquakes by the shifts in mass as the ice waxed and waned. The general conclusion was that a repository, properly located and constructed, would not be adversely influenced by any of the phenomena accompany- ing advance and retreat of an ice sheet, or even by the many advances and retreats that can be anticipated during the next million years. In their review of KBS-2, the NRC subcommittee found that the Swedish scientists had mustered impressive support for this conclusion. Little new has been added in the past five years, and there seems no point in repeating the review of old material here. One possible question was singled out by the subcom- mittee as needing further attention. If, during glacial retreat, a thick front of melting ice stands for a time directly over a repository, might the huge volumes of water, under heads possibly as great as the thickness of the ice itself, alter the groundwater regime sufficiently to change the flow at repository depths? Using data from existing glaciers, one can add details to the model. Presumably, there would be a steep ice front 15 to 30 m high, then an ice surface sloping upward at an angle of no more than 10Â° to the central part of the glacier, where it would probably be 1 to 2 km thick. The hydraulic head in the frontal area of an active glacier may reflect the height to the top of the ice, as demon- strated in some valley glaciers by large quantities of water draining into vertical cylindrical holes through the ice. The glacial front would fluctuate from year to year, but might remain in the general vicinity of the repository for many decades. Now in such circumstances might groundwater flow increase enough to erode bentonite at a depth of 500 m? Could oxygen be introduced from the
29 surface, leading to accelerated corrosion? If some canisters are breached early, could radionuclides be brought to the surface rapidly in large quantity? Such possibilities seem remote, but to prove them completely out of the question is difficult. The KBS scientists further considered the influence of the ice front, in an effort to derive possible conse- quences and to set rough limits, but they concluded that loosely constrained variables are so numerous that the effort is futile (Leif Carlsson, Swedish Geological Survey, personal communication, 1983) . As the Swedish workers point out, since the necessary combination of circumstances is so very special, since the circumstances would not arise for tens or hundreds of thousands of years (after the more active radionuclides would have decayed), since the likelihood of adverse effects is small, and since escaping radionuclides would be enor- mously diluted, the possibility of unacceptable exposure to a remote future generation from this cause seems entirely negligible. The panel agrees with this assessment.