Salado Hydrogeology, Gas Pressure, and Room Closure
Studies of physical, chemical, and biological processes involving the rock salt environment provide important conclusions about the behavior of waste-filled excavations in the Salado Formation. This information is reviewed briefly in the following four sections. The first discusses the hydrogeological properties of WIPP salt (see also Appendix C). The second draws general conclusions about gas generation. The third section discusses the influence of room closure, with local stratigraphy (Appendix A) and the creep behavior of salt (Appendix D) used as additional inputs. The final section discusses the combined effects of brine inflow, gas generation, and room closure.
In assessing deep geological disposal as a potential means of isolating radioactive waste from the biosphere and worthy of further research, the National Research Council suggested rock salt as a particularly suitable geological material (NRC, 1957). The fact that water-soluble formations such as the Salado Formation in New Mexico have remained in place for hundreds of millions of years indicates that, except for slow dissolution at the margins, they are hydrologically inactive. Another desirable characteristic of salt is that over time, it will flow around the waste and encapsulate it completely.
The Salado is approximately 600 m thick, extending about 400 m above and 200 m below the horizon selected for the WIPP repository. The Salado contains bedded, continuous layers of relatively pure halite, impure halite with clay and polyhalite constituents, and interbeds of anhydrite, accompanied by underlying clay seams (Freeze et al., 1995a, pp. 1-8).
Some of the interbeds are clearly evident and continuous over many kilometers and have been designated as "marker" beds. They serve as a guide in keeping the excavations within given bedded layers and maintaining a fixed separation from the nearest anhydrite interbeds above and below.
Excavations at WIPP have not revealed any feature that would cause DOE to question the assumption of the very low permeability of the Salado. However, concerns have been raised (Bredehoeft, 1988) that although the Salado appears to be dry, it may still be saturated with brine, that is, the microscopic voids between the crystals that make up the salt beds could be filled with brine and the voids interconnected (see Box 3.1).
In this case, interstitial brine would tend to flow down the pressure gradient and into the excavated repository. Although the porosity of the halite is not high,1 such a permeable-medium model considers the entire Salado Formation as a brine reservoir, able to flow into and fill the excavations. The brine would corrode the metallic waste containers, generating an underground "slurry" of contaminated brine. Through chemical and bacteriological processes, this brine might generate enough gas to raise the pressure above lithostatic (i.e., the pressure generated by the weight of overlying rock).
Several underground investigations and analyses of the interconnection model have been conducted, and a number of inconsistencies have been noted (Freeze and Christian-Frear, in preparation; Beauheim et al., 1991; McTigue, 1993). These include the following:
- Brine seepage into excavations in the Salado halite occurs in variable small quantities in unpredictable locations.
- Brine composition is quite variable, even among samples separated by distances of only tens of centimeters. Over a period of 230 million years (the
The porosity of the Salado is generally about 1 percent by volume, and the interstitial moisture content is typically less than 0.5 percent by weight.
BOX 3.1 Permeability and Fluid Inclusions of the Salado Formation
The permeability of the Salado Formation salt is an important parameter in evaluating the capability of the WIPP repository to withstand hypothesized future flooding by an external source of water that would flow through any interconnected pores in the salt. Valuable insight into permeability is achieved by examining the composition of fluid inclusions. Fluid inclusions are tiny quantities of liquid or liquid plus gas enclosed in salt crystals of the Salado, which represent samples of the fluid from which the salt crystals formed or were subsequently recrystallized. Compositions of the inclusions show substantial variety, which has led some researchers (e.g., Roedder et al., 1987) to speculate that the salt may have recrystallized in the presence of different fluids that have flooded the salt beds at various times since their formation. However, more recent work (Bein et al., 1991) suggests that variations in chemistry of fluid inclusions is attributable to depositional and recrystallization processes that occurred millions of years ago and indicates that these variations do not require fluid flow through interconnected pores. The most recent work by Jones and Anderholm (1993, 1996) also suggests that the observed variations in fluid inclusion composition can be explained by natural and expected variations in the marine and marine-marginal environmental conditions during Salado deposition in the Permian. This means that the inclusion fluids probably have been isolated since the salt was formed. Thus, the Salado salt probably has remained essentially impermeable since its formation (see also Appendix C).
Salt mines elsewhere in the world have indeed flooded on occasion, but this is typically due to a combination of excessive extraction of salt during mining, presence of fractures in the salt caused by the disturbance of the mining, and close proximity to aquifers or surface water bodies. However, these conditions do not exist for the WIPP repository. The overburden pressure and deformation properties of salt (Appendix D) will close and heal any fractures that might develop in the disturbed rock zone and seal and isolate any residual gas-filled or brine-filled void spaces in the repository horizon. No permeable formations in the vicinity of the repository would serve as a suitable "source" or "sink" of ground-water.
- age of the Salado), these compositional differences would have been eliminated by diffusion had the pores of the formation been interconnected.
- Rock salt behaves essentially as a viscous fluid, with a zero yield (shear) stress, when the load is applied over a long time (see Appendix D). Such flow behavior would eliminate the possibility of connected pathways between the salt pores. Certainly, rock salt cannot sustain stress concentrations around microfissures over geological time. The inability of salt to sustain a shear stress over a long time is also indicated by the fact that the in-situ state of stress in the salt is isotropic (i.e., equal in all directions).
- Brine pore pressure in both halite and anhydrite marker bed units at WIPP has been found to be approximately 12 MegaPascals (MPa), which is between hydrostatic (6 MPa) and lithostatic (15 MPa). These measurements are consistent with isolated brine.
- In-situ testing (see Appendix C, and Beauheim et al., 1993b) indicates a large variability in intrinsic permeability, ranging from less than 10-21 m2 for pure halite to as high as 10-18 m2 (Freeze et al., 1995a) for anhydrite interbeds. When anhydrite is subjected to internal pressures (e.g., by high gas or brine pressures in the waste rooms) approaching lithostatic, its permeability may increase significantly due to the generation of incipient fractures and eventually open macroscopic fractures. Median values of 10-21 m2 and 2.5 × 10-19 m2 are used in computations by the Department of Energy (DOE, 1995a, pp. 6-73, 6-75) for halite and anhydrite permeabilities, respectively.
- Stress concentrations around underground boreholes and excavations in the Salado generate a disturbed rock zone (DRZ). The DRZ is a local region of enhanced permeability consisting of microfractures that spread at a slow and exponentially decreasing rate into the salt from any newly generated surface (created, for example, by borehole drilling).
All of these observations support an alternative to the interconnection (Darcy flow) model. In the alternative model, the undisturbed salt is essentially impermeable. Brine seepage comes from isolated domains of porosity containing brine that, in the undisturbed state, will be at lithostatic pressure. These domains become interconnected if they are penetrated by the micro and
macrofissures of the DRZ. If the DRZ should extend to the marker beds, Darcy flow from these beds into the excavations could occur.
Appendix C addresses brine flow into excavations in the Salado and considers the following two possibilities: (1) impermeable salt containing permeable marker beds, from which brine flows into the excavations; and (2) a permeable Salado Formation with Darcy flow into the excavations. In both cases, by using the permeability values considered to be most appropriate on the basis of current data, the total inflow appears to the committee to be too small to result in significant brine accumulation in the rooms. With appropriate design width and sealing of the entrance drifts to the rooms, it should be possible to minimize extension of the DRZ to the marker beds and avoid connection between the beds and the excavation (see discussion of room seals in Chapter 4).
In conclusion, brine inflow to excavations in the Salado is likely to be significantly less than was thought to be the case several years ago–and will occur primarily via anhydrite marker beds. Careful attention to room filling and backfilling could aid in further limiting brine inflow.
Because the current design of WIPP calls for sealing all shaft communication between the repository horizon and the surface, gas generation could conceivably create a pressurized repository. Three sources of gas generation in the sealed repository have been identified (Lappin and Hunter, 1989):
- anoxic corrosion, caused by any inflow of brine that might react chemically with the carbon steel waste drums, generating gas (predominantly hydrogen); and
- bacterial reactions with some of the organic constituents (e.g., cellulose).
The worst case scenario would involve total dissolution of all steel drums in brine, with generation of the equivalent amount of hydrogen, plus bacteriological conversion of all cellulosics, plastics, and rubbers to a mixture of methane, CO, and CO2. Based on this scenario, by considering gas compressibility but not solubility, reactivity, or the likelihood of interbed fracturing at lithostatic pressure, an early calculation (see Molecke, 1979a, b) indicated that the resulting pressure in the sealed repository could exceed lithostatic.
The generation of such quantities of gas cannot occur under undisturbed conditions because the quantity of brine required in this calculation is not available inside a sealed repository, and an external brine reservoir is sealed off from contact with the waste. However, the E1 and E1E2 scenarios do postulate an unlikely but possible mechanism for the introduction of very large quantities of pressurized brine. Accordingly, theoretical and experimental studies of gas generation have been undertaken (Freeze et al., 1995a) to provide estimates of the amount of gas generation to be expected under any credible set of conditions.
Early in the gas generation studies, DOE-sponsored researchers concluded that the amount of ionizing radiation in the expected WIPP inventory would not be enough to generate significant quantities of gas by radiolysis (see Molecke, 1979a, b). The remaining two sources of gas have been studied in more detail.
Early laboratory experiments at Sandia National Laboratories (Molecke, 1979a, b) designed to permit estimates of chemical and biological gas generation rates were inconclusive because of the small and sometimes inconsistent effects observed. After a hiatus of several years, a larger program (Brush, 1990) was undertaken. This program, performed by recognized specialists, is now approaching completion (Brush, 1994; Beauheim et al., 1995).
The following conclusions about the major corrosion process have been reported:
Anoxic corrosion of steels … will produce significant quantities of H2 and consume significant quantities of H2O if
(1) … sufficient brine enters the repository after filling and sealing;
(2) significant microbial activity … does not occur (microbial activity will produce CO2 or CO2 and H2S, which will passivate steels …) (Brush, 1994, p. 6),
Under humid conditions (gaseous, but not aqueous, H2O present), anoxic corrosion of steels and other Fe-base alloys will not occur … (Brush, 1994, p. 7).
With respect to microbial activity, the conclusion announced in the same document is less definite: "… although significant microbial gas production is possible, it is by no means certain" (Brush, 1994, p. 7).
Results of experimental studies (Brush, 1994) show that anaerobic gas generation rates in brine are likely to be quite low—approximately one micron per year rate of dissolution of the steel surface (Brush, 1994, p. E-9) and to be negligible if the waste is exposed only to gaseous phase "brine humid" conditions rather than being immersed in liquid brine.
Time Dependence of Bacterial Activity
Even if the experimental program eventually succeeds in measuring bacterial gas generation rates, a question will remain about extrapolating short-term, laboratory-derived rates to repository compliance time scales. Linear extrapolation may be appropriate for certain physical and chemical processes, but living organisms are more complex. Can bacteria continue to be active metabolically over extremely long periods in a closed system?
Because biologists ordinarily do not deal with long times and closed systems, there seems to be no literature on this subject, especially for the halophilic environment. However, literature does exist on life in extreme environments, and there are basic principles that are true for all forms of life under all conditions.
In the closed WIPP facility, organic and inorganic nutrients must be supplied from the emplaced waste and the enclosing salt. Neither of these can be regarded as a rich source of the complete range of compounds and elements required by the organisms likely to be present. Because extremely halophilic bacteria tend to have unusually complex nutritional requirements, they are not likely to thrive in an environment with a restricted range of substrates (Kushner, 1978). Nevertheless, it is probable that limited amounts of suitable nutrients are present and therefore, that the potential exists for metabolism and cell reproduction.
Unlike the culture medium in a laboratory flask, WIPP waste would not be homogeneous. Bacteria require an aqueous medium to dissolve and transport nutrients and to disperse metabolic products. Unless the repository floods, there is no continuous aqueous medium to give full access to the scattered supply of nutrients.
If a repository room does become flooded (by human intrusion, for example), biological activity will bloom for a time. The activity cannot continue indefinitely because of the bacterial generation of waste products that inhibit further metabolism and the conversion of limiting nutrients to forms that are not available to succeeding generations of bacteria.
Consider a specific example. Phosphorus typically makes up 2-3 percent of the dry weight of a microorganism. Phosphorus is indispensable to life; it is a component of the genetic material (DNA) of every living cell. Some of this phosphorus may be recycled promptly after the death and lysis of a particular cell. However, in a closed environment such as WIPP, a small remainder may be retained in a biologically inactive form for long periods. DNA samples at least 50,000 to 100,000 years old have been isolated from several sources (Williams, 1995). The phosphate deposits of Florida and Tennessee are a striking example of the way a life process can steadily lose a critical material in a "side stream."
Similar arguments can be made concerning other elements and compounds essential for the metabolic activities of bacteria in WIPP. In a closed system, it can be anticipated that at least one of these materials eventually will become limiting, thereby halting reproduction and metabolism.
"Real Waste" Tests
The DOE WIPP program plan of 1990 called for an engineering-scale program of "real" gas evolution tests to be run in parallel with laboratory studies. The plan involved emplacing a large number of drums of actual waste in one or more underground rooms, which would then be sealed and monitored. When this program encountered both practical and political objections, it was redesigned to use specially constructed metal "bins" to be filled with actual waste, placed underground, and monitored. This program also was canceled.
What remains of the real waste concept is the Source Term Test Program set up in 1994 at Los Alamos National Laboratory (LANL; see Chapter 5). As part of this program, gas evolution from selected samples of actual waste under various conditions is to be monitored at both the drum and the liter scale. As discussed in Chapter 5, no results from this program have been reported yet.
Conclusions on Gas Generation and Gas Pressure
The committee thinks that pressurization by gas generated in an undisturbed repository is not a serious concern for three reasons:
- Laboratory experiments have recently established that hydrogen generation by anaerobic chemical corrosion of steel can occur in WIPP, but only when liquid brine and metal come into physical contact. Gas evolved from such brine-induced corrosion is limited by the contact area of brine with the steel drums. The brine-humid conditions anticipated in the repository restrict the expected gas-generating mechanisms to fewer than the set originally considered.
- Biological gas generation will be no more than a brief transient because it will be limited by brine availability as well as by nutrient depletion.
- Although gas sealed in waste-filled rooms can eventually approach lithostatic pressure due to creep closure of the rooms, it will represent a small total volume and a small gas pressure energy.
As a result of an extensive rock mechanics research program, prediction of creep closure of repository excavations at WIPP is relatively straightforward. The time-dependent stress-strain behavior of salt is understood sufficiently well to predict closure for most design situations (Freeze et al., 1995b; Freeze and Christian-Frear, in preparation; Munson, in press). Studies to date show agreement between predicted and observed room closure rates within less than 10 percent. Further details are presented in Appendixes C and D.
Marker Bed Considerations
The main source of brine inflow to the repository rooms would be from three adjacent anhydrite layers, two above and one below the rooms. Marker bed (MB) 139, located about 1.5 m below the 4-m-high disposal room, is about 0.9 m thick. Anhydrite "a" and anhydrite "b" (each 0.1-0.2 m thick) are some 2 to 4 m above the 4-m-high disposal room, and MB 138 (approximately 0.2 m thick) is about 11 m above the ceiling (see Figures 3.1, 3.2; and Beauheim et al., 1993a).
The rooms are wide enough (10 m) that the normal rock fracturing and collapse processes around the rooms (i.e., formation of the DRZ) may cause these layers to be brought into hydrological connection with the waste rooms. Over a period of several years from the time of excavation, the DRZ will extend progressively to a distance above and below the repository comparable to the half-width of the room (i.e., about 5 m above and 5 m below the room).
The waste emplacement room dimensions (4 m high × 10 m wide × 91 m long) and the proximity of marker beds suggest that the DRZ formed around each room (including roof sag, floor uplift, and wall fractures) could, after some time, provide hydrologic contact between the waste-filled rooms and one or more marker beds.
Hydrologic contact is a matter of concern because it could create a path for fluid transfer between adjacent rooms. Even when undisturbed, the marker beds are more permeable than the neighboring halite. The
"best-estimate" anhydrite permeability is of the order of 10-19 m2 , which is two orders of magnitude greater than the corresponding best estimate permeability for undisturbed halite (Freeze et al., 1995a, b). (See the discussion below on Backfilling and Compartmentation).
If this interconnection should develop while the gas pressure in the WIPP excavation is below the fluid pressure in the adjacent rock, brine will flow in from the marker beds. The amount of brine available is theoretically large because the MB strata are a really extensive. However, the absolute quantity of Salado Formation brine that can enter an excavation in the repository horizon is limited by the remaining void volume of unconsolidated material during creep closure of the salt (see Appendixes C and D).
Should brine come in contact with the steel drums, gas generation would occur. However, several effects, identified and discussed below, suggest that this gas pressure buildup would be limited.
- A pressure rise due to gas generation will reduce the rate of flow into the excavations. If the gas pressure reaches lithostatic, it would open fractures in the marker beds, greatly increasing the marker bed permeability to outward flow from the rooms. Gas would tend to flow out of the rooms.
- Another possible effect of an intersection of the DRZ with the marker beds (particularly those above the repository) is the escape of gas by two-phase flow into the coarser fractures of the marker bed. This becomes appreciable only if the gas-brine pressure approaches lithostatic, when opening of fractures in the anhydrite layers would occur, resulting in a substantial increase in permeability of the marker beds. In this case, as shown by Webb and Larson (1996, Figure 4.9d, p. 37), the 1° inclination of the marker beds results in countercurrent flow (i.e., brine flows into the excavation as gas flows out), increasing the amount of brine inflow. Although variable, the increased inflow is of the same order of magnitude as flow into the unpressurized open excavation.
- At gas pressures below lithostatic (i.e., without interbed fracturing), countercurrent flow is not significant. Furthermore, gas pressure acts to reduce brine inflow.
- During earlier stages of room closure (i.e., 50-100 years), the floor uplift in each waste-filled room would tend to isolate the waste from liquid brine. During this period, the waste drums would be subjected to only a moderately humid environment, in which, as discussed in the previous section, gas generation by drum corrosion is considered negligibly small. However, if the brine remains in communication with the waste-filled drums, the desirability and feasibility of engineered designs to reduce or eliminate return of entrained brine to the waste rooms should be considered.
The Combined Effects of Brine Inflow, Gas Generation, And Room Closure
The preceding discussion indicates that the predicted consequences of the combined effects of creep closure of the waste-filled disposal room, formation (and eventual healing) of the disturbed rock zone (DRZ), brine inflow, and gas generation can vary considerably depending on the particular assumptions made with respect to each of these variables (see also Freeze et al., 1995a). From the arguments outlined earlier in this chapter, the committee concludes that the amounts of brine and pressurized gas are likely to be small.
To assess the overall consequences of these issues on radionuclide isolation, it is necessary to incorporate the ''room effects" into the entire network of excavations, including the shafts and seals. DOE has made the very conservative assumption that all waste-filled rooms remain permanently interconnected via a relatively high permeability DRZ (see Chapter 2) There is strong evidence, however, that placement of crushed salt seals between rooms and back filling of all other underground excavations with salt from mining operations will result in an overall low permeability within the excavation close to that of the intact salt (of the order of 10-20 m2 or lower) within a few hundred years (see Callahan et al.  and Appendix C). Each waste-filled room will then be an isolated compartment within the essentially impermeable Salado Formation. This compartmentation would greatly reduce the probability and consequences of an E1E2 borehole intrusion (see Chapter 2) and also the potential for brine flow up the sealed shafts.
Backfilling and Compartmentation
The development of hydrological contact between the marker beds and the waste-filled rooms could be inhibited somewhat by the use of well-placed (e.g., pneumatically injected) crushed salt backfill in the waste-filled rooms. Although this probably will not reduce the early development of the DRZ significantly because of the difficulty of packing the fill to a high enough density to obtain a rapid build-up of back pressure in the early stages of room closure, sufficient back pressure eventually will develop to reduce the aperture and hence the permeability of the marker beds.
Hydrological communication between the marker beds and the waste-filled rooms is of little significance provided hydrological communication between rooms is prevented. Careful design of entry rooms and early installation of room and panel seals to ensure that waste-filled rooms are isolated effectively from each other, or compartmented—as discussed in Chapter 4—could greatly reduce the vulnerability of a WIPP repository to human intrusion by drilling (see Chapter 2) or by water injection (see Box 3.2).
BOX 3.2 Potential Consequences of Brine Injection from Petroleum Recovery Operations on Repository Inflow
Although the permeability of halite in the Salado Formation is probably negligible, there remains some concern about flow through marker beds and other impurities in the formation. In oil-producing regions such as southeastern New Mexico, it is common to inject fluids into the deep subsurface, either to stimulate secondary recovery of oil in partly depleted oil reservoirs (e.g., by waterflooding) or to dispose of the large volumes of brine that typically are produced simultaneously with oil. Both types of operations involve high-pressure injection of fluids into a deep rock formation. If there is a failure in the well casing or in the grout or cement outside the casing, fluid can leak into overlying formations and flow laterally along one of the many anhydrite layers in the Salado.
One such recent occurrence, in the Rhodes Yates Field of Lea County in southeastern New Mexico, apparently caused significant flow through the Salado (reported as an unexpected flow of over 1,000 barrels per hour encountered by a well being drilled in 1991; see Silva, 1994, 1996). The unexpected flow through the Salado was attributed to injection into old (1940s-1950s) production wells in the Yates Formation, some 200 m below and 3 km away from the Salado horizon in the well where the flow was encountered. The flow probably was transmitted through anhydrite marker beds.
Because injection wells operate close to the WIPP site now and similar operations can reasonably be anticipated in the future, accidental leaks from such operations might represent a source of fluid that could migrate into the repository. Brine volumes under such conditions would be much greater than those that would seep into the repository from the Salado under undisturbed conditions. The committee believes that the likelihood and consequences of this type of disturbance occurring at the WIPP site should be evaluated by DOE.
There are differences between the petroleum production operations in the vicinity of WIPP and those in the Rhodes Yates Field (see Fig. A.6, Appendix A). Near WIPP, oil is produced below the 500-m-thick Castile anhydrite and salt formation, which underlies the Salado. The main gas-producing horizons are approximately 600 m deeper. Thus, injection for secondary recovery would be 700 m or more below the repository horizon. At Rhodes Yates, the Salado lies unconformably on the oil-producing Guadalupian series. The Castile Formation is absent.
Given the current improved well completion practices and the presence of the Castile into which fugitive brine may flow before reaching the Salado, the probability that fluid injected for enhanced recovery near WIPP would invade the repository horizon would seem to be lower than the Rhodes Yates incident. The potential of such a leak due to the injection of brine for disposal directly into the Castile, closer to the repository, for the period during which petroleum production at WIPP is likely to continue, is of greater concern. The committee recommends that DOE demonstrate how the performance of the repository will be assured against fluid injections.
Conclusion on the Combined Effects of Brine Inflow, Gas Generation, and Room Closure
The initial brine inflow into rooms is likely to be low enough that rates of gas generation will also be very low. Gas pressure will develop slowly (over hundreds of years) and more rapidly if the void volume of the room is reduced by backfill. Eventually, that is, in some hundreds of years, salt creep will close the rooms around the waste containers, with gas isolated in the voids between the waste in individual rooms and pressurized to the order of the lithostatic pressure. The hazard of such "pockets" of pressurized gas is considered by the committee to be negligible because of its small volume and small gas pressure energy. The committee believes that placement of effective room and panel seals will reduce greatly the potential for hydrological communication between the individual rooms.
Summary And Conclusions
After some hundreds to perhaps a thousand years, the steady-state condition of the undisturbed repository will be one in which the residual void in each room is a small fraction (10-15 percent) of the original room volume. This residual volume will be occupied by a mixture of deoxygenated gases, mostly hydrogen, probably at a pressure approximating lithostatic. These conditions would be established by Darcy flow of brine along the interbeds, controlled by room gas pressure and interbed permeability. In the absence of adventitious brine, only "brine humid" conditions are
expected. Creep closure and gas pressure buildup will serve to limit liquid brine entry.
Waste-filled rooms would probably be isolated from each other, so that the consequences of human intrusion should be considerably less severe than for the "interconnected rooms" scenario considered in the DOE PA. Chapter 4 discusses the engineering feasibility of compartmentation and other, relatively simple, engineering measures that may be taken to further improve the overall performance of WIPP as a TRU waste repository.