Engineering to Improve Predicted Repository Performance
As discussed in Chapter 2, the committee judges the probability of radionuclide release from a well-sealed Waste Isolation Pilot Plant (WIPP) repository under undisturbed conditions to be very small. Concern has been raised that corrosion of the steel waste containers exposed to brine could generate gas at substantial pressure in the repository. Although the gas itself would be essentially nonradioactive, it could entrain radioactive material in the process of being released if a drill hole penetrated the repository at some time in the future.
In examining such possibilities, it is useful to consider ways in which the likelihood of potential releases could be reduced by changing the design of the repository or the form of the waste before it is placed underground. It is technically possible, for example, to incinerate the waste at the surface, thereby eliminating the possibility of gas generation underground. However, incineration also poses some risk of radionuclide exposure and would be very costly.
The Department of Energy (DOE) has conducted a number of studies (Lappin and Hunter, 1989; Marietta et al., 1989; Butcher, 1990; DOE, 1991, 1995b)1 to examine a variety of such options to reduce risks. A short list of engineered alternatives is discussed below.
Repository Design And Excavation Alternatives
Contact of drums with brine could be substantially lower than suggested in Chapter 3 because of both the probable behavior of the excavations (see Box 4.1) and the use of relatively simple engineering design changes, as noted below.
Natural and Engineered Sumps
Closure of the waste storage rooms, which are designed to be 4 m high × 10 m wide, will result from roof sag and floor uplift. Development of the disturbed rock zone (DRZ) and the associated floor uplift will create a "sink" for any brine flowing into the room, while simultaneously raising the waste drums above the brine, thus minimizing any immersion of the drums were liquid brine to be present. Also, because the rooms will follow the 1° inclination (i.e., a dip of 1.8 percent) of the marker beds (MBs), brine will tend to accumulate at the lower end of each storage room, creating the possibility of a drainage sump at this end of the rooms to remove the brine from contact with waste.
Design Considerations to Minimize DRZ Effects
If entrances to each room were to consist of two excavations, each 3 m wide, 3 m high, and about 25 m long, separated by a wider pillar, and if the center height of the excavations was located midway between the anhydrite layers, it should be possible to reduce the DRZ for each 3-m-wide excavation to no more than about 1.5 m above and 1.5 m below the roof and floor horizons, respectively (probably considerably less if the rooms are excavated and filled rapidly, within several weeks). In this way, it should be possible to avoid disturbing the anhydrite (brine-bearing) layers or connecting them with the DRZ, thus avoiding any inflow of brine into this section of the excavations. The 25-m section could be sealed with crushed salt or prefabricated compacted salt blocks, thereby isolating each room or panel from its neighbors; these seals (free
BOX 4.1 Creep Closure and the Disturbed Rock Zone
Excavations will creep closed by movement inwards from the entire volume of salt, and eventually by a lowering (subsidence) of the surface, and/or gaps between the salt and overlying more elastic formations. A crude analogy to the viscous behavior of salt is the filling of a cavity of air made in a volume of water. The air cavity will fill by flow of water towards the cavity. The water occupying the volume of the cavity will result in a (slight) lowering of the water surface level.
Measurements at WIPP show that open rooms tend to close at a constant rate of about 1 percent per year. After about a hundred to a thousand years, then, it will be difficult to identify where the original rooms were excavated, because these rooms would be filled in by flow of the entire salt mass.
The Disturbed Rock Zone (DRZ) refers to a region, usually within one diameter or less of the walls of an excavation, in which the difference between the rock stresses prior to and after excavation is sufficient to produce microcracks in the salt. In the immediate vicinity of the walls the microcracks may coalesce to form open fractures. It has been shown, both by laboratory experiment and field observations at WIPP, that the microcracks and fractures can be healed when the deformation of the salt towards the excavation is resisted by a "back-pressure." This back-pressure can be generated by placement of seals or by back-filling the excavations with crushed salt. Calculations (Callahan et al., 1996) indicate that the permeability of a shaft seal of crushed salt that has been dynamically compacted to 90 percent of the density of intact salt (Hansen and Ahrens, 1996; Brodsky et al., 1996) will achieve a value close to that of the intact salt within 50-100 years. The DRZ will disappear over the same time period. Back-filled rooms will require longer to come to equilibrium at lithostatic pressure, but should certainly reach a condition similar to the shaft within a few hundred years.
of brine flow into the DRZ) should close as rapidly and tightly as the shaft seals.
Some redesign of the room and pillar layout may be needed if room compartmentation is introduced. The general principle of designing operations to minimize disturbance of the marker beds prior to room sealing is a useful one to follow. Such seals also should reduce potential impacts of the E1E2 human intrusion scenario substantially (see Chapter 2). Because the projected inventory of transuranic (TRU) waste destined for WIPP may be less than the design capacity of the repository (DOE, 1995c), the separation of individual rooms could be increased and the room seal designs improved.
Room Seals, Panel Seals, and Backfill
Once filled with waste and backfilled, the rooms should close within about 100 years, given the absence of brine in the DRZ (see Appendixes C and D). Although a central "core" of crushed waste materials in the rooms will probably retain significant local permeability, these materials can be isolated from each other by effective room and panel seals. There seems to be no reason why such seals cannot be designed to be as effective as shaft seals. Thus, the room and panel seal permeabilities would decrease progressively (e.g., to 10-16 m2 or lower within 100 years), as assumed for the shaft seals (see discussion below), and, given the essentially viscous long-term behavior of salt, would approach the undisturbed value for salt (i.e., essentially zero) after 500 to 1,000 years. Thus, for at least 9,000 of the 10,000 years for which E1E2 intrusions must be considered, individual waste-filled rooms would be isolated from each other.
The effectiveness of room and panel seals in reducing radionuclide releases has been noted in DOE reports:
The concept of sealing individual rooms or portions of rooms, using thick salt 'dikes' which isolate smaller volumes of waste from each other, was considered the most feasible facility design alternative. … (DOE, 1991, p. A-39)
This alternative was considered for mitigating the effects of the two-borehole [i.e.,
the E1E2] scenario, and to a limited extent, the single borehole drilled into the Castile brine. The EAMP [Engineered Alternatives Multidisciplinary Panel] modified this alternative by suggesting that floor to ceiling salt seals could be installed at each end of the waste disposal rooms, as well as at appropriate locations within the rooms. This would decrease the effective permeability of each waste disposal panel, and prevent hydraulic communication between the two boreholes. If this alternative is implemented, it would appear to effectively eliminate the effects of the two-borehole scenario …. (DOE, 1991, p. A-30)
A final report of the Engineered Alternatives Cost/Benefit Study (EACBS) (DOE, 1995b) considers a wide variety of engineered alternatives, ranging from the use of backfill around the waste drums to reduce the void volume (and, hence, the time required to compact and seal in the waste) to the sealing of individual rooms. The prudence of considering relatively simple measures such as room seals has been recognized in concluding statements such as the following: "Communication between the rooms during an intrusion scenario is significantly reduced (gas, brine, and radionuclides)" (DOE, 1995b, p. B-7).
The committee is disappointed that, as of the time of preparation of this report, these engineered alternatives, despite their appearance in the Engineered Alternatives Task Force (EATF) and EACBS reports and their benefits being noted in Systems Prioritization Method (SPM) efforts (Sandia National Laboratories, 1995), have been neither evaluated by the WIPP program in past compliance documents nor incorporated into the 1995 performance assessment process (DOE, 1995a). The 1995 baseline model used (DOE, 1995b, Introduction) for compliance purposes is a connected repository, a suboptimal design (see Chapter 2). The benefits of using backfill are also supported by recent SPM results (Prindle et al., 1996).
Pre-Emptive Mining of Potash
The McNutt Member of the Salado Formation above the repository horizon contains potash that could be attractive for mining in the future. The field extraction methods currently used in New Mexico potash mines result in substantial ground deformation, both above and below the mined-out horizon, and surface subsidence. The ground deformation and subsidence over the mined-out area involve fracturing of the overlying strata, such as the Culebra Dolomite and the Dewey Lake Red Beds. This, in turn, may increase the hydraulic transmissivity of these layers significantly.
It is not yet clear how important this increased transmissivity might be in compromising the ability of the Culebra and associated strata to delay release of radionuclides from the repository. Studies by DOE are in progress. Should it be determined that the effects are sufficiently adverse to warrant corrective action, it is technically possible, although expensive, to extract the potash "pre-emptively" (i.e., as part of the engineering design of the repository) in such a way as to avoid subsidence and associated damage to the overlying strata.
Extraction of potash would be carried out in two stages, using a room and pillar method with cemented backfill (Brady and Brown, 1987). In the first stage, approximately half of the potash is mined across the entire minable area. The rooms are then filled with cemented backfill. In the second stage, the ore remaining between the cement-filled areas is extracted, and the voids are again filled with cemented fill. In this way, deformation of the overlying formations would be negligible. If desired, special backfill material could serve as a protective cap over the repository and as a deterrent to future drilling.
Waste Form Modification
Concern about potential gas generation in WIPP (Bredehoeft, 1988) arose from the possibility that large quantities of brine might accumulate in the repository during waste storage (see Appendix C). The 1991 EATF (DOE, 1991), a multi-disciplinary group of over fifty scientists and engineers, considered a wide range of waste form alternatives based on existing technologies or on technologies that could be developed within a few years. Such technologies include methods to shred; compact; cement or grout; incinerate; vitrify; melt metal (producing ingots or slag); add pH buffers, salt, or sorbents; decontaminate metals; and use
noncorrosive containers. Changes in facility design also were considered. These options were screened to fourteen combinations that were chosen for detailed evaluation. Although such technologies either were available or probably could improve the waste form, the 1991 EATF report found that
- costs would be higher than if wastes were untreated,
- worker risk would be higher than for untreated waste handling,
- a waste treatment system would take significant time to implement, and
- treatment options would involve more complicated regulatory requirements.
In general, waste treatment options were viewed as properly held in reserve in case a need to supplement more readily implementable alternatives, such as backfill, arose. The 1995 EACBS (DOE, 1995b) chose to evaluate eighteen alternatives and reached similar conclusions and provided a specific ranking of the various alternatives.
In the committee's view, the conclusions of the EATF and EACBS are appropriate. As discussed in Chapter 3, gas generation is not anticipated to be a significant problem at WIPP; therefore, the use of advanced treatments such as incineration is unwarranted.
Findings, Conclusions, and Recommendations on Engineered Features
The committee believes that the WIPP design should incorporate engineered alternative features, and the performance assessment (PA) should reanalyze the human intrusion scenarios, incorporating (1) the existence of effective room/panel seals, and (2) a DRZ permeability that decreases with time, consistent with the known behavior of salt. Further, it is feasible, probably with no major increase in cost, to design the repository so that waste containers avoid contact with brine, thereby essentially eliminating gas generation, and to ensure that waste panels are isolated from each other effectively by seals that are comparable in permeability (10-16 m2) to shaft seals.
The committee thinks that it is prudent to consider such engineering measures as backfilling of the waste rooms to ensure early reconsolidation of the DRZ and minimization of brine inflow. More advanced options, such as incineration or vitrification of TRU waste, are unwarranted.
Sealing Of Shafts And Boreholes
In the committee's view, the flow of radionuclide-contaminated brine through inadequately sealed vertical shaft excavations at WIPP is the most probable pathway for release of radionuclides from the repository to the accessible environment under undisturbed conditions. Calculations (DOE, 1995d; and Appendix C) indicate that a seal permeability of 10-16 m2 or less, over a minimum seal height of 100 m, is needed over the total (100 m2) cross-sectional area of the four shafts to limit brine flow sufficiently through the seals to provide adequate isolation of the repository for 10,000 years. Figure 4.1 shows the results of PA calculations on the cumulative flow through the seals in 10,000 years.
In the figure, the net total brine flow through a 100-m-high seal of crushed salt (in each shaft) is plotted for various assumed values of the seal permeability. Each of the different flow values shown for a given seal permeability is the result of one "run" (or CCDF "realization"—see Appendix B) of the repository seal model. This model involves a number of parameters such as salt permeability, brine inflow to the rooms, gas generation, room closure, and gas pressure. Values for each parameter were selected at random from a range of values, and the model was run, computing for the particular combination of parameters selected the mean brine pressure at the bottom of the shaft over 10,000 years. In some cases, this mean pressure over 10,000 years was less than the net flow, or "cumulative release" was negative, that is, brine flowed into the repository. In other cases, the reverse was true, that is, the mean pressure below the shaft seal was greater than hydrostatic and so brine flowed out of the repository. Thus, Figure 4.1 shows both positive (outflow) and negative (inflow) values at each permeability tested. For permeability of 10-16m2 or lower, the flows were negligibly small in all cases tested.
It is seen that a lower seal permeability (abscissa) of 10-16 m2 or less is needed to reduce the cumulative brine flow at 10,000 years (ordinate) to negligible levels for all repository parameter combinations. The accuracy of this result is shown by independent calculations performed by the committee and given in Appendix C. Similarly, a permeability of 10-18 m2 or less is needed to limit gas flow through the seal (DOE, 1995d).2
The four shafts at WIPP are each approximately 6.1 m (20 ft) in diameter and approximately 660 m (2,160) ft) deep. The upper 256 m (840 ft) of the shafts passes through relatively permeable water-bearing sediments consisting of evaporites, carbonates, and clastic rocks. The remaining 404 m (1,325 ft) to the bottom of the shaft (and the repository horizon) consists of bedded salt in the Salado Formation. The Salado also extends approximately 200 m below the repository horizon.
Once the repository has been filled with TRU waste, the entire 660-m column of each shaft will be sealed with a variety of seal materials, as shown in Figure 4.2. In designing the seal system, it is necessary to incorporate features that effectively will prevent vertical flow of fluid both within the shaft itself and within an annular region extending about one-half to one shaft radius from the shaft wall into the rock. Within this region, referred to as the DRZ, the initially intact, essentially impermeable salt has been fractured as a consequence of rock stress changes introduced by the process of excavation and by the continuing changes that have occurred over the decade or so since the WIPP shafts were excavated (Figure 4.3). Given the additional 30-40 years needed to fill the repository, these time-dependent effects would be operative for a total of approximately 50 years before shaft sealing could start, creating an annulus of relatively high permeability around the shaft (Figure 4.3). This annular pathway for fluid flow must be eliminated to seal the shafts effectively.
Development and Healing of the Disturbed Rock Zone
The "damage" induced in the DRZ varies progressively from small, isolated microcracks in the regions of low stress change remote from the shaft, through a coalescence of the microcracks, to extensive through-going vertical fractures in the immediate vicinity of the shaft wall. Only in this more extensively fractured region is shaft permeability increased significantly.
Recent gas and brine flow permeability tests (Dale and Hurtado, 1996) around the Air Intake Shaft at the 660-m level and at the top of the Salado (256 m) reveal that there is no measurable change in permeability of the salt at either location, except within an annulus extending approximately 1 m from the shaft wall.3 Thus, the region of increased permeability is likely to be a small part (10 to 20 percent [maximum] of the shaft radius) of the DRZ around each shaft.
Laboratory studies (Brodsky, 1995) clearly indicate that stress-induced fractures in salt specimens can be eliminated when the damaged region is subjected to sustained moderate confining pressure (approximately 5 MPa [MegaPascals]). The permeability of the cracked specimens is found to approach that of intact salt within a few days.
In the full-scale shaft sealing situation, the rate of reclosing or "healing" of induced (permeable) fracture zones around a shaft in the Salado will depend on the rate of buildup of the confining stress. This depends, in turn, on the compressibility of the shaft seal material and the rate of steady-state creep of salt around the shaft (Figure 4.4).
Numerical simulations by DOE contractors indicate that the width of the DRZ annulus at the time the shaft is filled (i.e., 50 years after excavation) will vary from 0.6R (where R [= 6.1 m] is the shaft radius) at a depth of 256 m (i.e., at the top of the Salado, where the lithostatic pressure is 6 MPa) to approximately 0.85R at 660-m depth (i.e., at the bottom of the shaft, where
the lithostatic pressure is 15 MPa). Computations (see Figures 4.3 and 4.4) indicate that with crushed salt that is compacted to 90 percent of the density of intact salt, the DRZ annulus will, 50 years after filling the shaft, have been reduced to 0.05R at the top of the Salado and totally eliminated between 350 m and the bottom of the shaft (660 m), that is, over almost 300 m. After 50 more years, the DRZ is eliminated over the entire 400-m Salado section. Because the region of increased permeability in the DRZ is confined to the portion immediately around the shaft wall, it seems probable that the entire 400-m Salado section in the shaft will achieve a permeability close to that of intact salt within 50 years of filling the shaft.
The computations above do not allow for the delaying effect of adventitious brine or high-pressure gases that could be introduced into the DRZ fractures. As mentioned earlier, the presence of these fluids would inhibit closure of both the DRZ and the compacted salt. Eventually, the connected, fluid-filled fractures in the DRZ (and regions in the crushed salt) would probably degenerate into isolated spherical inclusions containing gas or brine at lithostatic pressure within the ''healed" salt, but the elimination of permeability may be delayed substantially. It is therefore desirable to ensure that the DRZ fractures and intergranular regions of compacted salt are maintained free of brine or gas during the 50-100 years required for full compaction under dry conditions.
To this end, several strategies have been proposed, including the introduction of supplementary seals above and below the region of the compacted salt seal. Use of the concrete seals, which have a significantly higher initial stiffness than the compacted salt, will result in a
more rapid buildup of back-pressure and correspondingly more rapid elimination of any permeability in the DRZ above and below the compacted salt section, thereby sealing this section early against brine inflow. Brine and gas inflow to the compacted salt column can be eliminated effectively by placing a compacted clay column above and below the compacted salt (Figure 4.2). Although probably unnecessary, it is feasible, at the time of sealing, to "ream out" this inner portion (i.e., about 10-20 percent of the shaft radius, as noted earlier) of the DRZ above and below the compacted salt sections where the concrete is to be placed. The reamed-out sections then would be filled with rapid-curing concrete as soon as possible after reaming, ensuring very early protection of the DRZ in the compacted salt section against inflow of brine.
The benefit of retarding release of radionuclides by placing compacted clay columns at various sections in the shaft (Figure 4.2) should be considered fully in the design and PA analysis of the effectiveness of shaft seals. Present PA sensitivity studies model shaft seal permeabilities but not extensive, detailed shaft designs.
No large-scale demonstrations of the effectiveness of proposed shaft-sealing designs have been carried out. However, the reality of salt creep is indisputable; salt around the shaft definitely will flow to seal in the shaft plug and reheal the disturbed rock zone. Because DOE proposes to use compacted crushed salt as the shaft seal material, this too should reconsolidate eventually to a density and impermeability comparable to that of intact salt. The practical feasibility of using temporary seals to eliminate water from the permanent seal sections during closure has to be examined and demonstrated carefully. Because these seals will be required only at the end of the operational phase of WIPP, there is adequate time to address this design problem.
In the proposed multicomponent shaft seal system (illustrated in Figure 4.2), reconsolidated salt is expected to constitute the major long-term seal component. Ahrens and Hansen (1995) demonstrated the feasibility of large-scale dynamic compaction of crushed natural salt in a shaft configuration to achieve 90 percent of its intact density, providing greater confidence in the efficiency of its reconsolidation to produce an effective long-term seal.
As noted previously, major short-term concrete seal components are expected to provide structural support as well as to retard brine or gas flow substantially during the intervening period before the salt seal components are adequately consolidated. This performance of the concrete component has been documented recently by the Department of Energy (1995d) and, in more detail, in individual reports (Stormont, 1984, 1986, 1988; Van Sambeek et al., 1993). The importance of designing concrete to be chemically compatible and durable in the salt-dominated WIPP horizon has been emphasized by several investigators (e.g., Roy et al., 1983, 1985; Wakeley et al., 1993, 1994, 1995). In-place permeabilities of experimental salt-saturated concrete plugs (1-m diameter) in WIPP showed decreased permeabilities with time (less than 4 × 10-19 m2) after six years (Wakeley et al., 1993). Some remaining concerns exist regarding compatibility of concrete with magnesium-rich brines found at WIPP. Additional confidence in the performance of concrete is provided by field studies of sealing methods in salt or potash mines (Eyermann et al., 1995). The stoppage of major (10 m3/min) flow of pressurized brine in the Rocanville Potash Mine, Saskatchewan, is a good example of the ability to seal excavations effectively in potash (or salt) using current technology.
The seal design incorporates additional components of shaft seals, planned to perform individual functions (DOE, 1995d). These seals include compacted clays (Meyer and Howard, 1983) for use in both non-Salado and Salado members, asphalt to act as a water-stop between concrete members, and possibly other impermeable chemical seal components to act essentially as an O-ring. Issues of concern about the potential use of asphalt as a seal component have not been resolved fully. As with many organic-based materials, there are issues of potential degradation and gas generation, including microbial attack. Addition of lime to the asphalt mix has been suggested as a possible remedial measure (DOE, 1995d). The influence of the different compressibilities of the seal material (here crushed salt and clay) on the rate of buildup of radial stresses in the seal and the shaft wall is illustrated in Figure 4.4.
The sealing of exploratory and other boreholes, as necessary, is expected to draw on extensive experience
with cementing and plugging technology in the oil and gas industry, much of which has involved bedded salt environments. Impressive early tests of the performance of the borehole plug (cementitious materials) under high well pressure conditions in the specific WIPP environment (Gulick et al., 1980) have been followed by other studies, demonstrations, and plug material development (Finley and Tillerson, 1992; Wakeley et al., 1995).
Sealing Of Rooms And Panels
Sealing sections of horizontal drifts presents essentially the same problems as sealing vertical shafts, except that in this case, gravity acts to accentuate development of the DRZ in the roof and to retard it in the floor. One difference is that during placement, it is difficult to compact crushed salt in a horizontal drift to the same density as in a vertical shaft.
Salt creep should compact horizontal seals eventually to the same high level of impermeability (for both the DRZ and the excavation) as in the shafts. If it is assumed, for example, that 100 m of drift can be packed with crushed salt at 70 percent of the intact salt density (i.e., rather than the 90 percent density for the shafts), 200 or 300 years probably will be required to reach 10-16 m2 seal permeability, rather than the 50-100 years needed for the shaft. Technologies of placing drift seals to isolate radioactive waste have been, or are being, examined elsewhere, for example, in Stripa (Sweden) and Pinawa (Atomic Energy of Canada, Ltd., Canada). Salt has the distinct advantage over most other rocks of exhibiting creep, allowing rehealing of the DRZ, and progressive reduction of seal permeability.
Thus, it seems technically feasible to provide, between the waste storage rooms and panels, drift seals that are sufficiently tight, perhaps 300 years after waste emplacement, to eliminate brine communication between the waste-filled rooms and panels at WIPP. Note that any such communication between rooms via anhydrite marker beds (e.g., MB 138, MB 139) would not be affected by the drift seals. These beds have a low permeability, of median value 2.5 × 10-19 m2 (DOE, 1995a, p. 6.75) and therefore would contribute negligible brine, with regions near the excavations becoming drained during the period that the rooms are open before waste emplacement. As discussed at the beginning of this chapter, with careful room and panel seal designs, it should be possible to inhibit room-room hydrologic communication.
Whether or not drift seals contribute to the overall isolation capability of the repository depends on the credence given to the human intrusion scenarios that must be examined according to 40 CFR 194. For the undisturbed case, PA studies by DOE in which no drift seals are assumed show that the shaft seals alone are sufficient to prevent radionuclide release. Thus, for the undisturbed case, drift seals are not necessary to demonstrate compliance (see discussion of Figure 4.1).
Currently it is assumed, for example, as part of the E1E2 scenario (see Chapter 2), that brine from a pressurized reservoir will flow through the repository, carrying radionuclide-contaminated brine to the surface via a second borehole. Provision of drift seals would reduce the possibility of establishing such a flow path and hence would considerably reduce the significance of this "two-hole" type of human intrusion.
The ability of salt to flow under stress is a particularly valuable attribute with respect to sealing excavations in the WIPP repository. While careful attention to good practice during seal construction is very important, it appears to the committee that the design of effective shaft and panel or room seals at WIPP is entirely possible by using currently available technology. The shaft seal design system proposed for WIPP (DOE, 1995d) includes large factors of conservatism and redundancy, which ensure that the requirements of 40 CFR 191 with respect to sealing of excavations at WIPP can be met by a large margin.4 Although room and panel seals are not essential for adequate waste isolation at WIPP given undisturbed conditions, they would reduce the significance of an E1E2 intrusion event considerably.
This is a properly conservative design of the engineered seal. There is no contradiction between the conservatism advocated here and the critical remarks against conservatism in Chapter 2. The latter were levied against unrealistic assumptions of salt performance that could lead to a mischaracterization of how WIPP could perform.