The Waste Isolation Pilot Plant: A Potential Solution for the Disposal of Transuranic Waste (1996)

To dispose of transuranic (TRU) wastes in the Waste Isolation Pilot Plant (WIPP), the Department of Energy (DOE) must demonstrate that the repository complies with applicable regulatory standards. These standards address

• transport of TRU wastes to WIPP,
• protection of workers from hazardous materials and mining hazards during operations, and
• the long-term performance of WIPP after wastes have been emplaced and the facility has been closed.

This report focuses on the third issue.

Two general standards of the Environmental Protection Agency (EPA), the first of which is defined in two parts, concern long-term repository performance:

1.  
1. EPA's Standard for the Management and Disposal of Spent Nuclear Fuel, High-Level and Transuranic Wastes, 40 CFR Part 191 (or "Part 191"): 40 CFR 191 specifies requirements of geologic disposal systems in the United States.
2. EPA's Criteria for the Certification and Determination of the Waste Isolation Pilot Plant's Compliance with Environmental Standards for the Management and Disposal of Spent Nuclear Fuel, High-Level and Transuranic Radioactive Wastes, 40 CFR Part 194 (or "Part 194"): 40 CFR 194 addresses the specific application of 40 CFR 191 to WIPP.

2. The Resource Conservation and Recovery Act (RCRA) of 1976, 40 CFR Part 268: This law establishes a procedure to track and control hazardous wastes from the time of generation to the time of disposal.

This chapter is devoted mainly to a discussion of what is and what is not known by the committee about the long-term performance of WIPP in terms of the EPA standards. Compliance issues relating to RCRA are addressed in less detail. Issues associated with operations of the repository and transportation of waste to the repository are not considered. The most likely pathways, or "scenarios," through which radioactive materials could be released to the accessible environment are described, together with the performance assessment (PA) approach used to evaluate those scenarios against the EPA standards and some of the results to date. The chapter concludes with comments on the quality of the performance assessment work for WIPP, a brief discussion of other long-term radiological compliance issues, and some overall conclusions.

The EPA standard for long-term isolation of the types of radioactive wastes intended for emplacement in WIPP consists of both (1) the quantitative containment requirements and individual and ground-water protection requirements, and (2) the more qualitative assurance requirements, which are intended to provide confidence that the desired level of protection is achieved. The assurance requirements for waste characterization and monitoring are discussed later in this chapter.

A key aspect of the quantitative standards of 40 CFR Part 191 is the distinction between disturbed and undisturbed performance of the repository.

Undisturbed performance refers to the case in which any releases of radionuclides to the accessible environment occur as the result of reasonably foreseeable natural processes. Releases due to human intrusion or from unlikely natural events—defined in

Part 191 as events having less than a 1 in 10,000 chance of occurring in 10,000 years—are excluded from the undisturbed case.

Undisturbed performance of the repository is governed by the individual and ground-water protection requirements of 40 CFR Part 191. The individual protection requirements limit the committed effective dose to any individual resulting from releases from the undisturbed repository to 0.15 milliSievert (15 mrem) per year. The ground-water protection requirements set concentration limits for potable sources of ground-water that might come in contact with the wastes.

Disturbed performance includes consideration of inadvertent human intrusion into the repository. The EPA standard, Part 191, and supporting guidance require that releases of radionuclides resulting from drilling into the WIPP repository be considered, along with other specified disruptive events, named below, that could affect the performance of the repository. The containment requirements, reproduced below, apply to the disturbed performance of the repository.

Part 194 directs DOE to assume that the frequency of boreholes drilled into the WIPP site be based on the rate of drilling observed in the Delaware Basin during the 100 years prior to the time of the compliance application, by taking into account both deep drilling (i.e., boreholes that would reach the depth of the WIPP repository) and shallow drilling (i.e., boreholes that would not reach the depth of the repository). EPA specifies that the assumptions about future drilling practices with regard to borehole diameter and plugging practices should be consistent with practices in the Delaware Basin at the time of the compliance application.

Part 194 also requires that DOE consider the effect that mining of any potash of economic grade (in today's market) would have on the hydraulic conductivity of formations above the repository, and the effect that fluid injection for enhanced oil recovery in nearby oil and gas wells would have on the repository.

40 CFR 191.13 requires that

The key words in this regulation are "reasonable expectation," in recognition of the uncertainties inherent in making assessments over a 10,000-year time frame.

In paraphrasing 40 CFR 191.12, a performance assessment is defined as an analysis that identifies all significant processes and events that could affect the repository and evaluates the likelihood of each process or event and the effects of each on the release of radionuclides to the environment. To the extent practicable, these estimates are combined into an overall probability distribution displaying the likelihood that the amount of radioactive material released to the environment will exceed the values specified in Table 1, Appendix A to 40 CFR 191 (40 CFR 191, 1995, p. 11).

In comparison to the radioactive waste standards adopted by most countries, Part 191 is unique in that, in addition to regulation based on radiation dose, repository compliance also is based on calculations of release fractions of selected radionuclides (Table 2.1). The containment requirement addresses the ability of a

repository to isolate waste from the environment, without distinguishing releases that would lead to significant doses from those that would not. One advantage of this requirement is that, unlike a dose limit, it does not require assumptions about where people will live, how and where they will grow food, and what water sources will be used. A disadvantage is that the Part 191 containment requirements do not differentiate between transport of radionuclides off-site in a pristine aquifer that is a likely exposure pathway and—in the case of WIPP—transport via the Culebra Dolomite, in which the high content of dissolved solids makes the water unfit for human consumption.

Dealing quantitatively with human intrusion is a difficult issue. Any assessment of releases or risks from human intrusion into a repository is inherently arbitrary because the nature and rate of future intrusion over a 10,000-year time frame cannot be known. One advantage of basing the future human intrusion rate on the historical rate, as EPA does, is that the effect of known resources on past exploration and production activities is considered. However, whether or not this knowledge will predict accurately what may happen in the future—especially when institutional controls preventing intrusion have been lost—is uncertain. The difficulty does not concern specific details of EPA's guidance regarding how to treat intrusion in performance assessment, but rather that such guidance is arbitrary and of unknown validity with respect to how future events actually will occur. The same issue is of concern in Europe and other countries dealing with the regulation of geological repositories. In general, radionuclide releases resulting from human intrusion tend to be treated qualitatively in these countries, for example, as a consideration in site selection.

The way in which human intrusion is handled in regulations 40 CFR 191 and 194 implies that WIPP, or any potential repository, could "pass or fail" on the basis of a human intrusion scenario that is, by necessity, arbitrary, especially in its assumptions concerning future technology. Indeed, human intrusion rates or scenarios could be devised that would cause any geological repository to comply or fail to comply with EPA standards. This issue of human intrusion has been considered at length in the NRC report Technical Bases for Yucca Mountain Standards (NRC, 1995).

Until recently, federal law dictated that WIPP must comply with both the radioactive waste standard (40 CFR 191) and the RCRA standard (40 CFR 268), the latter of which establishes requirements for the disposal of chemically hazardous wastes (see Box 2.1). The 40 CFR 268 standard prohibits the land-based disposal of untreated hazardous wastes under most circumstances. This "land ban" has a "no-migration" provision by which a variance can be granted to permit the disposal

of untreated waste if the applicant can demonstrate that the waste will not move after disposal. In 1989, DOE submitted a No-Migration Variance Petition for the planned underground test program at WIPP and received approval for a 10-year period. In 1995, a Draft No-Migration Variance Petition was submitted by DOE for full-scale operations at WIPP.

During the 104th Congress, bills (U.S. Congress, House of Representatives, 1995, 1996; U.S. Congress, Senate, 1996) were introduced to amend the WIPP Land Withdrawal Act. Among other features, these bills would exempt WIPP from compliance with long-term RCRA requirements. In hearings on a House bill, EPA indicated that it was not opposed to such a change. An attachment to a letter from Mary Nichols, EPA Assistant Administrator for Air and Radiation, and Elliot Laws, EPA Assistant Administrator for Solid Waste and Emergency Response, to Senators Craig and Kempthorne, stated that

In this letter, EPA indicated that, under current law, it was obligated to continue to enforce RCRA provisions at WIPP. The letter illustrates the degree to which DOE and EPA have worked in a cooperative, constructive manner to address significant issues regarding compliance with RCRA regulations. On September 23, 1996, the President signed into law (P.L. 104-201) an amendment to the WIPP LWA that exempts WIPP from the federal RCRA requirements.

Performance assessment (PA) encompasses the overall process of assessing whether or not a waste disposal system meets a set of performance criteria. For the WIPP PA, the system is a deep geologic repository disposal system in bedded salt for DOE TRU waste, and the performance criteria are various

long-term environmental metrics in U.S. government regulations (Rechard, 1995, p. Glos-14).

Iterative PAs of WIPP are being performed for DOE by Sandia National Laboratories. The performance assessments are intended to provide interim guidance while final compliance evaluations are being prepared.

The methodology has been developed to calculate a performance measure in the form of a complementary cumulative distribution function (CCDF) that permits comparison with the EPA release limits for radioactive waste disposal. The CCDF, a generally accepted form for depicting risk by a specific performance measure, was popularized by the Reactor Safety Study (U.S. Nuclear Regulatory Commission, 1975) for point estimates and by the Zion/Indian Point risk studies performed by industry (Pickard et al., 1981) for estimates with uncertainty. For WIPP performance assessments, the performance measure takes the form of normalized releases to the accessible environment of the radionuclides listed in Table 2.1. Under a normalized release limit, the estimated release is reported as the fraction of the release limit allowed under Table 1, Appendix A of Part 191. A release estimate equal to the allowable limit would be a normalized release of 1. Figure 2.1 is a hypothetical CCDF illustrating compliance with the 40 CFR 191.13 containment requirements.

The basic framework of the WIPP performance assessment methodology is the Kaplan and Garrick (1981) "triplet" definition of risk. This definition of risk is founded on the principle that to determine risk, the following three basic questions must be answered:

1. What can go wrong?
2. How likely is it to go wrong?
3. What are the consequences?

"What can go wrong" is examined in the form of scenarios that can lead to releases of specific radionuclides to the accessible environment; "how likely," is examined by calculating the probabilities of those scenarios; and the "consequences" are examined in terms of the overall likelihood of various levels of release. The results are then cast into CCDFs of the form illustrated in Figure 2.1. The procedure for developing the CCDFs results in a structured set of scenarios leading to the specific consequences of interest; summing the probabilities of the scenarios for each release set; and plotting the complementary cumulative probabilities as a function of normalized release rates, usually on log-log coordinates (see Appendix B for more details).

A key part of the performance assessment model involves deciding on an appropriate set of scenarios, that is, identifying what can go wrong. The WIPP performance assessment team used the following five-step selection process for human-initiated scenarios (Cranwell et al., 1990):

1. compiling features, events, and processes (FEPs) that could affect the disposal system;
2. classifying events and processes to enhance consistency and completeness;
3. screening individual events and processes;
4. combining events and processes into specific scenarios; and
5. screening scenarios to identify and eliminate those that have little or no effect on the performance assessment.

Final selection of the scenarios in Figure 2.2 was based on screening events and processes according to probability, consequence, and physical reasonableness. The process resulted in the set of eight scenarios illustrated, although calculations were made on the basis of the first four scenarios only. A more detailed discussion of scenarios is presented later in this chapter.

The complexity in the performance assessment results from two main factors:

1. The numerous variables (approximately 50 in all) that represent the hydrological, chemical, mechanical, thermal, and transport processes involved in the transport and physical process models.
2. The effort required to perform sensitivity and uncertainty analyses on the models.

Documentation detailing the WIPP performance assessment effort includes a wide range of source materials documented in various reports and studies—including a 1992 Performance Assessment (the 1992 PA; Sandia National Laboratories, 1992, Vol. 3); the various iterations of the systems prioritization method (SPM; Sandia National Laboratories, 1995); and the Draft Compliance Certification Application (DCCA; DOE, 1995a).

Uncertainties in the knowledge base were propagated through the scenarios, and sensitivity studies were performed so that results of the analysis include a quantification of the uncertainties associated with the calculated release rates.

FIGURE 2.1 Hypothetical CCDF illustrating compliance with the containment requirements.

Source: Sandia National Laboratories (1992, Vol. 1, Figure 3.3).

FIGURE 2.2 Potential scenarios for the WIPP disposal system. Source: Sandia National Laboratories (1992, Vol. 1, Figure 4.1). NOTE: TS refers to an event in which subsidence results from potash mining; the yes/no decision answers the question of whether potash mining is considered in the scenario. The base case depicted here represents the undisturbed repository. The E1 and E2 scenarios are discussed later in this chapter (see Figure 2.4 and 2.5). For further discussion of potash mining, see Chapter 4.

On March 31, 1995, DOE submitted to EPA a draft analysis of WIPP compliance with 40 CFR 191 for the undisturbed case; this multivolume submission was supplemented in July 1995 with a performance analysis for the disturbed case (i.e., human intrusion; DOE, 1995a). DOE's current schedule calls for new performance assessment calculations for the compliance certification application to be completed and documented by September 1996, for subsequent submission to EPA a month later. Earlier performance results have been published in the 1992 PA (Sandia National Laboratories, 1992) and in the SPM analyses (Sandia National Laboratories, 1995). Compared to these earlier analyses, the DCCA included updates to

the analytical codes and revisions of specific analytical assumptions, where dictated by EPA guidance (although this DOE submission predated the release of the final version of Part 194). The analytical assumptions of the DCCA appear to be largely unchanged from those of the 1992 PA and the SPM.

Although the DCCA indicates compliance with Part 191 containment requirements, the results of this analysis are too aggregated to allow any conclusions to be drawn about which scenarios and pathways contributed to the calculated releases. However, careful study of the 1992 PA does provide some insight into the relative magnitude of release scenarios.

CCDF curves summarizing radionuclide releases to the accessible environment resulting from cuttings removal and ground-water transport fall substantially below release limits promulgated by EPA. Although this is an important result, it must be noted that the rate of intrusion assumed in the 1992 analysis was less than is now required in the final Part 194, issued on February 9, 1996. The two main contributors to radionuclide releases to the accessible environment identified in the 1992 PA were (1) drill cuttings brought to the surface from exploratory drilling and (2) radionuclides transported by ground-water flowing through the Culebra following human intrusion.

Some of the results from the 1992 PA with respect to the 10,000-year containment requirements of the EPA standard are as follows:

• Where the intrusion probabilities are based on expert judgment and dual-porosity transport with chemical retardation, the mean CCDF is more than an order of magnitude below EPA limits.²
• Where an intrusion rate of 30 boreholes per square kilometer per 10,000 years and dual-porosity transport without chemical retardation are assumed, the mean CCDF is approximately one order of magnitude below EPA limits.
• Where an intrusion rate of 30 boreholes per square kilometer per 10,000 years and single-porosity, fracture-only transport (with little retardation) are assumed, the mean CCDF is less than an order of magnitude below EPA limits.

These three 1992 PA results represent a sample of all the different modeling assumptions and parameter values that were examined to assess their impact on the mean CCDF. The compliance outcome is a common result, with variations on the degree of sensitivity. Of course, individual scenarios can be found to produce CCDF curves that exceed the EPA limits, but for them the important issue is their likelihood, or frequency of occurrence (see Appendix B for more details). This information is used to determine the mean CCDF curve, the one used to assess compliance.

The containment requirement for WIPP (defined by Table 1 of Part 191, shown here as Table 2.1) specifies a release limit of 100 curies of plutonium per million curies of TRU waste, equivalent to a release fraction of about 10^-4, at least initially. Although release limits are specified in Part 191 for various isotopes, plutonium appears to be the most significant, in terms of both the inventory and compliance. Based on inventory estimates (DOE, 1995c) adjusted for the estimated date at which the repository would be closed, the WIPP inventory of thorium, uranium, neptunium, plutonium, and americium isotopes with half-lives greater than 20 years (which determine the release limit), is around 6 million curies. The corresponding release limit for plutonium-239 (Pu-239) would be 600 curies, or about 10 kg (approximately 20 pounds).

Figure 2.3, taken from the 1991 PA (Sandia National Laboratories, 1991), describes the radionuclide inventory as a function of time. The inventory is expressed in EPA units used for a single waste panel. An EPA unit is a convenient way to express the inventory, based on release limit definitions in Appendix A of 40 CFR 191. As can be seen, after about 500 years, Pu-239 is the dominant radionuclide of concern at WIPP.

As noted earlier, the EPA standard sets forth two other quantitative requirements in addition to the containment requirements: (1) individual protection requirements; and (2) ground-water protection requirements. The individual protection requirement considers the radiation dose to humans in the accessible environment for 10,000 years of undisturbed performance. Because performance assessment results indicate that for the undisturbed case, brine (the medium for radionuclide transport) in contact with waste will not migrate more than a few ''tens of meters" from the waste emplacement panels in 10,000 years, the 1992 PA did not include performance estimates for individual protection requirements. Subsequent assessments (e.g., SPM, DCCA) have also concluded that, under undisturbed conditions, waste would not migrate. The PA work (DOE, 1995a, Section 8.1) uses a bounding calculation to show doses to humans resulting from brine transport to be orders of magnitude below natural background; hence, compliance with the individual protection requirement can be demonstrated easily for the undisturbed scenario at WIPP.

The PA work has noted that the ground-water protection requirement does not apply to WIPP because there is no source of potable ground-water as defined in the EPA standard (Sandia National Laboratories, 1992, Vol. 1).

FIGURE 2.3 Time-dependent inventory expressed in EPA units for a single waste panel.

Source: Sandia National Laboratories (1991, Vol. 4, Figure 2.4.2). More recent estimates (DOE, 1995c) of the WIPP inventory differ from those used in this figure.

In summary, the sole long-term radiological compliance issue is whether or not the containment requirements are met. Performance calculations indicate that if the shaft seals are effective, there will be little, if any, release or exposure from the undisturbed repository. For the disturbed case, compliance depends directly on the assumptions and analyses made for the frequency and consequences of human intrusion into the repository.

Assessment of the likely performance of WIPP begins with the identification of scenarios by which wastes could be released from the repository. The scenarios that have been evaluated in greatest detail are those considered by DOE to represent the major mechanisms and routes for release of radioactive materials.

Scenarios applicable to the undisturbed case considered in WIPP PA work include the following:

1. leakage of brines containing radioactive materials up the shaft and/or through a disturbed rock zone around the shaft up to the Culebra Dolomite and flow to the defined accessible environment via the Culebra; and
2. flow of WIPP brines directly along anhydrite marker beds in the Salado Formation-salt.

These scenarios are mentioned briefly here and represent mechanisms for potential release of radionuclides that underlie the considerations in Chapters 3 and 4.

Under the first scenario, human exposures could occur due to water extraction from the Culebra to add water to stock ponds,³ and subsequent consumption of beef from cattle that have consumed this water. This scenario, involving flow out of the Culebra, is evaluated against both the containment requirements and the individual protection requirements.

The scenario involving flow along the anhydrite marker beds is evaluated against the containment requirements only, because the marker bed brines are not potable (Davies, 1989; Brinster, 1991; Yaron and Frenkel, 1994; Sexton, 1996), nor do they reach the surface.

Scenarios for the disturbed case involve releases resulting from boreholes drilled inadvertently into the waste. One scenario considered is the direct release of waste through drill cuttings brought to the surface. Other disturbed case scenarios include releases that occur through flow up a borehole into the Culebra dolomite, with subsequent transport in the Culebra; the effect of potash mining above the repository; and waterflooding from nearby water injection for enhanced oil recovery.

The analysis of releases of radionuclides through cuttings brought to the surface in drilling mud includes several factors. In addition to the direct volume of waste displaced by the borehole (i.e., the area of the borehole times the repository height, with the assumption that all such wastes reach the surface), Sandia has made calculations of additional releases due to the erosion of wastes by drilling mud and the movement of wastes in the vicinity of the borehole because of internal gas pressure in the waste.

The quantity of waste intersected directly by a borehole can be determined simply from the frequency and diameter of boreholes assumed to be drilled for the period beginning 100 years after repository closure and continuing until 10,000 years after closure. The previous analyses based on the 1992 PA assumed 30 boreholes per square kilometer per 10,000 years; the current (1996) 40 CFR 194 requirement specifies that the historic record over the past 100 years should be the basis for the borehole rate, which, based on presentations to the committee by EPA and DOE staff,

leads to an intrusion rate closer to 45-50 boreholes per square kilometer during a 10,000-year period.

For 50 boreholes per square kilometer per 10,000 years with a 0.073-m ² cross-sectional area (i.e., 12-inch diameter) borehole, the fraction F of waste released from cuttings (see Box 2.2) is 3.6 × 10^-6 (this includes an adjustment to account for no intrusion during the first 100 years after closure). As noted above, the allowable fractional release is about 10^-4, initially, and increases with time. Thus, for 12-inch diameter boreholes drilled at a rate of 50 boreholes per square kilometer per 10,000 years, direct cutting releases apparently are less than 4 percent of the allowable limits. Even when allowing for increased release due to waste erosion by drilling mud and spallation, extraction of the waste as cuttings is unlikely to approach the release limits.

This calculation illustrates that the fraction of the Table 2.1 limit released as cuttings is independent of

1. the volume of waste disposed of in WIPP,
2. the geometry of the waste rooms (i.e., the ratio of height to width and length), and
3. the plutonium content of the waste.

To the extent that erosion and spallation are low in comparison to the volume of waste that would be released directly by displacement, the release fraction also does not depend on any waste treatment technology that may be employed. The cuttings release fraction does depend on (1) the assumed intrusion frequency, and (2) the assumed diameter of the intrusion borehole.

Therefore, the cuttings calculation is largely independent of the characteristics of the waste and the repository design. Furthermore, it is related to the location of the facility only to the extent that the location affects the assumptions made regarding the frequency and nature of intrusions over the next 10,000 years.

In the "E1" and "E1E2" scenarios, DOE postulates that a borehole through the repository may encounter a pressurized brine pocket in formations below the Salado Formation in which WIPP is located, resulting in the release of radioactive materials. The two specific scenarios analyzed for this type of release are illustrated in Figures 2.4 and 2.5. Figure 2.4 shows the E1 scenario, in which a single borehole penetrates the facility and continues downward into a pressurized brine pocket; Figure 2.5 shows the E1E2 scenario, in which two boreholes penetrate the repository. Because brine is assumed to migrate through the waste in the E1E2 scenario, this scenario is the more significant contributor to total calculated release.

Such brine flow, presumably, would augment the salinity of existing water in the surface and shallow subsurface regions. As noted previously, ground-water from the Culebra Dolomite at the WIPP site and in adjacent areas downstream of a hypothetical release already is too high in concentrations of dissolved solids to be potable to livestock or for humans. Regions of potable water from the Culebra are southwest of the site and are not in the direct path of a hypothetical release.

The rate and quantity of plutonium released to the accessible environment through the E1 and E1E2 scenarios depend on a number of factors. In PA sensitivity analyses, the three factors that were found to have a significant effect on calculated releases are

1. the solubility of plutonium in WIPP brines,
2. the potential for retardation of radionuclides in the Culebra, and
3. the potential for movement of actinides in colloids.

As a result of these PA results, expanded work to determine the effects of solubility, colloid formation and retardation on releases has been undertaken (see Chapter 5).

The committee believes that the PA model assessments of the E1, E1E2, and other scenarios involve technically unrealistic assumptions that have not been probed in past sensitivity analyses. For these scenarios, these technical assumptions also appear to be unreasonably conservative, specifically in the lack of credit given for compartmentation of the waste panels and rooms. These issues are discussed in more detail below.

Construction of the repository produces a network of stress-induced cracks that form a disturbed rock zone (DRZ) in the Salado salt bordering the excavated areas. The cracks are expected to close and heal in time due to salt creep. The permeability of the DRZ region, initially higher than that of the undisturbed halite because of these cracks, is restored to a value close to that of the intact salt over time.

The July 1995 update of the DCCA (DOE, 1995a) assumes that the salt around the repository between marker beds 138 and 139 (the anhydrite beds above and below the repository) remains an unconsolidated disturbed rock zone (DRZ) for the full 10,000-year regulatory period, with a permeability of 10^-15 m². For comparison, the undisturbed halite is assumed to have a permeability that ranges between 10^-24 m² (i.e., essentially impermeable) and 10^-20 m². The draft compliance document does note that "although the DRZ is modeled conservatively in this assessment, it is the subject of a modeling study, and assumptions and treatment of this region may be different in the final Compliance Certification Application" (DOE, 1995a, p. 6-76).

The unconsolidated DRZ assumption appears to be inconsistent with the well-established properties of salt. A major feature of salt is that it behaves as a viscous liquid and creeps. The salt around waste rooms will reconsolidate as the pressure on the DRZ exerted by the backfilled rooms rises back to the lithostatic level. Such reconsolidation is estimated to take on the order of a few hundred years at most after repository closure.

A consequence of this conservative assumption is that boreholes that penetrate the salt between waste rooms are assumed to communicate with the waste and serve as release pathways, no matter when the boreholes are drilled. In addition, boreholes that penetrate different panels are analyzed as an E1E2 pathway, because under the conditions assumed throughout the entire repository, flow is calculated to occur via the DRZ. Sealing of rooms and panels so that this communication is prevented seems to the committee to be an entirely practical, cost-effective way to reduce the significance of the two-borehole (E1E2) scenario (see Chapter 4).

The more realistic treatment advocated here, of modeling time-dependent DRZ closure, adds complexity to the PA model. This complexity is warranted because it removes the non-physical conservative assumption that the DRZ does not heal. Removing an unrealistic assumption from the PA model has the added benefit of making the sensitivity analyses

derived from the PA more accurate representations of the true dependencies on parameters in the model.

It is conservatively assumed in the performance assessment that boreholes retain their initial diameter and have a permeability equivalent to that of a silty sand (10^-14 m²) for the entire 10,000-year regulatory period, based on present information of known and defensible natural processes that would fill the hole in time. No credit is given to mechanisms that would cause the boreholes to plug or shrink or that would interrupt the hypothetical flow of radioactively contaminated brine along the greater than 650-m borehole length. However, if further study should identify a natural mechanism that would cause the flow in the boreholes to be restricted, this could result in a significant reduction in calculated releases. Flow through an E1E2 scenario can occur only if both boreholes are open. With boreholes that stay open for times that are short in comparison to the 10,000-year regulatory period (e.g., 1,000 years), some borehole pairs from drilling events at random times will not result in a release because the first borehole will have closed before the second has been drilled. Further study to develop a better understanding of potential borehole closure mechanisms is recommended by the committee to examine an important component of the model calculation of releases.

As noted previously, calculations of releases over 10,000 years based on a DRZ that does not reconsolidate close to the value for intact salt are unrealistic (see also Box 4.1 in Chapter 4). This overly conservative assumption prevents the creep and healing behavior of the Salado halite from being properly included in past sensitivity analyses of WIPP performance. With effects such as DRZ reconsolidation excluded from a sensitivity analysis, an opportunity is lost to evaluate the full potential for cost-effective measures to demonstrate compliance. Stated another way, realistic DRZ modeling may lead to a demonstration of compliance using a cost-effective engineering feature such as backfill, which would lessen the dependence of the compliance demonstration on the work addressing each of the three key factors enumerated above. These factors have emerged from the existing PA work and associated sensitivity analysis, and they involve technically complex experimental work. For example, addressing the first factor, that of plutonium solubility, involves completion of ongoing plutonium solubility tests that could be the time-limiting step in determining WIPP compliance, if no more straightforward and cost-effective way to demonstrate compliance were identified. If the DRZ healing behavior were included in the PA sensitivity analysis, confidence in compliance might be increased due to a lesser dependence on results from work in these three more complex areas.

Overly conservative (to the point of being physically unrealistic) analytical assumptions, such as those described above, not only lead to a pessimistic assessment of potential repository releases, but also prevent the identification of design factors that are important to performance. If one assumes, for example, that the salt reconsolidates in times much less than 10,000 years, then compartmentation of WIPP has enormous benefits regarding calculated releases through the E1E2 scenario. The probability of one borehole being drilled into any given area of the repository decreases in direct proportion to the area involved. The probability of two or more such boreholes penetrating the same waste area decreases even more rapidly as the area is reduced. When one also considers that, for flow to occur, boreholes must overlap both in location and in a time window determined by their duration, the great potential value of compartmentation is clear.

A high degree of compartmentation can be obtained at relatively low cost. The current plan calls for waste panels consisting of seven rooms plus access drifts along each end of the rooms in a "ladder" arrangement (Figure ES.1). If, after the rooms are filled with waste, the access drifts are backfilled with compacted salt, then a high degree of isolation would be obtained. Changes in the design layout to ensure improved compartmentation seem entirely feasible, especially if, as is currently estimated (DOE, 1995c), the TRU waste inventory stored in WIPP is significantly less than the original design inventory. This issue is discussed in more detail in Chapter 4.

The main mechanism through which individuals could be exposed to radioactive materials from WIPP is the migration (through mechanisms discussed below) of material carried by water that is to be used directly or indirectly by humans. Some ground-water sources at WIPP are too saline for consumption by humans or livestock or for irrigation, so therefore, these do not appear to pose a significant risk to humans, even if WIPP radionuclides were to migrate into them. The ground-water pathway that could most directly result in human exposure at WIPP is movement of water containing radionuclides into the Dewey Lake Red Beds (see Appendix A), in which existing ground-water is known to be potable, with exposure resulting from direct or indirect human use of this water.

Although many analyses of radionuclide releases from WIPP have been made by Sandia National Laboratory and other DOE contractors, the radiation doses that could result have not been assessed for some reasonably identifiable scenarios and pathways. Most analyses of WIPP performance have focused on regulatory compliance with the EPA containment requirements, that is, calculations to estimate the quantity of radioactive materials that could be released across an arbitrary compliance boundary within 10,000 years. There has been less analysis of individual doses because the individual dose requirements of the EPA standard apply only to an undisturbed repository. For example, because of the high salinity of water in the Culebra, the analyses of releases to the Culebra do not directly translate into an assessment of doses to individuals. However, these analyses do contribute to understanding of how WIPP would perform because they include extensive consideration of processes, mechanisms, and parameters associated with the site and with the transport of radioactive materials from the repository.

The individual protection requirements (i.e., EPA's 40 CFR 191.15(a)) require that the committed effective dose for the undisturbed repository not exceed 15 mrem/yr for 10,000 years. Because limitation of individual doses under the undisturbed case is a regulatory requirement, DOE has conducted analyses (DOE, 1995a; DOE, 1990; Lappin et al., 1989) of this issue. The process addressed by these analyses is the leakage of radionuclide-containing brines through shaft seals into the Culebra. Human doses are calculated based on the assumption that water from the Culebra dolomite is used in stock ponds for cattle, and human exposure occurs through beef consumption (see Box 2.3). The individual doses for the undisturbed repository are estimated to have a peak value of about 10^-8 mrem/yr, well below the EPA standard of 15 mrem/yr (and below the natural background level of roughly 300 mrem/yr).

Three important features of early analyses are that water in the Culebra appears to be too saline for consumption by cattle, transport by colloids was not considered, and pathways to the Dewey Lake were not included. In addition, the analytical treatment of the behavior of the shaft seals and repository in this analysis appears to be conservative, that is, likely to over-estimate the releases (see Chapter 4).⁴ Taking all these considerations together, the committee concludes that the net effect will probably be to lower the already very low concentrations and doses indicated by DOE's analysis of radionuclide releases through shaft seals.

The ground-water protection requirements (i.e., EPA's 40 CFR 191.24(a)) require a calculation of levels of radioactivity in ground-water in the accessible environment from an undisturbed repository for 10,000 years. These requirements were not evaluated in the 1992 PA work under the assumption that no relevant potable ground-water sources exist. Recent estimates of radioactivity released from an undisturbed WIPP repository to ground-water are 10^-3 pCi/1, well below not only the EPA standards (ranging from 5 to 15 pCi/1), but also the natural background level of roughly tens of pCi/1 (DOE, 1995a).

Thus, for the effectively sealed and undisturbed WIPP repository, the committee has identified no

credible, probable mechanism for release of, or exposure to, radionuclides and concludes that DOE will be able to demonstrate compliance with the EPA standard by a wide margin.

As discussed earlier, DOE's PA showed that cuttings produced by drilling through the repository do not appear likely to cause the release limits to be exceeded. That is, although inhalation exposure from dust containing radionuclides that are brought to the surface as drill cuttings represented the dominant release scenario in the 1992 PA, the releases were within regulatory limits. If this is correct, compliance will depend on the results of other scenarios, such as the E1 and E1E2 scenarios, in which pressurized brine from below the repository horizon flows through the

repository, into the Culebra Dolomite, and then into the defined accessible environment. Releases associated with these scenarios appear to have been calculated by using highly conservative assumptions about the disturbed rock zone around the repository.

Individual exposures that might result from WIPP in the event of human intrusion have not been analyzed extensively because such exposures are not covered by the standard. The scenario thought to represent the dominant exposure pathway is an ingestion exposure assumed to result from the following sequence of events. Two vertical boreholes are inadvertently drilled through the repository in the future, one of which punctures a hypothesized pocket of pressurized brine in the Castile Formation. As a result, brine from the Castile Formation flows through the repository, dissolving and entraining radionuclides that flow up the second drill-hole into the Culebra, which is above the repository but below the surface. These radionuclides would then sorb onto the mineral surfaces of the Culebra (see Appendix A for a more complete discussion of the geologic formations at the WIPP site). ground-water from this formation would then be withdrawn and fed to cattle⁵, which subsequently, would be eaten by humans.

This Culebra-to-beef scenario is not the only one that is relevant to an assessment of all possible exposure pathways to humans. That is, using PA calculations of the Culebra pathway as a basis for inferring that minimal exposure would result from a WIPP repository depends on the assumption that there are no other pathways or scenarios worth considering. This is not yet certain. For example, in addition to the postulated main pathway through the Culebra, an alternative (or secondary) pathway exists through the shallower Dewey Lake Red Beds. The committee believes that the potential for individual doses received by drinking water drawn from this formation, located closer to the surface than the Culebra, should be analyzed and documented. That is, the first element of the triplet definition of risk, "what can go wrong," should include the possibility of a borehole release from the repository to the Dewey Lake Red Beds. This is a concern because the Dewey Lake is known to contain potable ground-water, and because it is the shallowest water-bearing unit in the area, it is also the easiest subsurface unit to be tapped for water supply by a future society. Thus, the potential for individual doses to be received by individuals (as a consequence of contamination from a breached repository) may be greater than those arising from an equivalent release into the Culebra. Because the Dewey Lake is less transmissive than the Culebra, the relative amount of plutonium reaching the Dewey Lake and the rate of lateral spreading may be less than for the Culebra. A quantitative analysis of this issue has not yet been presented to the committee in documented form for review.

Assessment of the isolation performance of the repository when disturbed by human intrusion is more complicated than in the undisturbed case. Because of gaps in the data available to date and the preliminary nature of the PA work reviewed by the committee to date (i.e., prior to the 1996 PA), it is not possible to make a compelling case that the radionuclide releases will comply with the EPA standard for the disturbed case. However, for the WIPP repository disturbed by future human activity, the committee has noted three ways in which on-going or additional work may lead to a demonstration of compliance. These are:

1. Re-evaluation of the probability and/or consequences assigned to highly speculative scenarios of future human activities may reduce the estimated risk of radionuclide release.
2. Experimental and field programs in progress or planned may show that key parameters (e.g., actinide solubility, sorption, radionuclide travel times, colloidal transport) are well within the range required to reduce the impacts of human activities such that releases fall within the acceptable range.
3. The implementation of available engineering options (e.g., compartmentation, treated backfill), which have not been considered in published DOE PA analyses, can substantially reduce the consequences of human intrusion.

Based on the PA analyses published by DOE to date, which the committee believes include some very conservative assumptions (i.e., tending to over-estimate releases), the committee concludes that incorporating the above three considerations very probably will allow DOE to demonstrate that a WIPP repository can comply with the EPA standard for the disturbed case. These topics are treated in the remaining chapters of the report.

Development of the WIPP PA has been a pioneering effort: it was the first detailed assessment of an actual radioactive waste repository to be published and has been the model for many subsequent analyses worldwide. The linkage of many computer codes and the capacity to use probability distributions for parameters and to propagate uncertainty through the analysis are major technical achievements. Nonetheless, in several important aspects, the full potential of the PA has not been realized in the WIPP project.

The committee noted several concerns above regarding the analytical treatment of some aspects of WIPP performance, for example, in the failure to consider salt reconsolidation and compartmentation of the waste rooms in the assessment. A more general problem is that the parameter values used in the PA reflect an uneven mixture of conservatism and realism. This is not unusual in risk assessment in which it is accepted practice to use a series of iterative assessments, based initially on conservative bounding values, followed by substitution of more realistic analyses that make a significant difference. A potential problem with this approach in the case of a complex analysis is that sensitivity analyses will fail to identify parameters that are most important to performance when those parameters are represented by a bounding value. A concern also is raised in Chapter 6 that some parameters assumed to be independent are, in fact, interrelated.

The PA identification of actinide solubility in brine and retardation as being among the most important parameters of WIPP performance is probably correct, but to a lesser degree than indicated by the sensitivity analysis. This is because other performance factors, such as DRZ permeability, were not sampled as analytical variables and, thus, their importance was not identified by the sensitivity analysis. As a result, an opportunity may have been missed to demonstrate compliance more quickly and economically than could be done through work on solubility and retardation.

In several iterations (the 1992 PA and SPM efforts), DOE has conducted sensitivity analyses of analytical parameters and used them to identify program research priorities. However, the PA modeling capability does not appear to have been used in a similar way to evaluate design alternatives, despite opportunities to do so. In 1991, DOE published the report of an Engineering Alternatives Task Force (EATF; DOE, 1991), a study conducted at least in part to respond to recommendations from the WIPP Committee (NRC, 1989). A similar study was published in September 1995 (DOE, 1995b) in response to EPA requirements. As described in Chapter 4, both of these reports included very positive comments about the value of sealing waste rooms; however, the current design (DOE, 1995b, p. ii) calls for panel seals but not room seals. Despite these two engineering studies, the PA model has not been used to evaluate what may well be the major potential benefit of such a design modification.

Waste characterization—the process of identifying and classifying the chemical, physical, and radiological constituents of each drum of waste—is a critical aspect of every waste management project. For a project such as WIPP, many of the waste characterization needs are set by agencies such as the EPA, the New Mexico Environment Department (NMED), the U.S. Nuclear Regulatory Commission, and DOE itself. It is obviously impractical, if not impossible, simply to overlay the requirements of each of these individual agencies. Attempting to do so quickly degenerates into an exercise in a complete analysis of every possible component. Such analysis is costly and time-consuming and would result in the highest potential

health hazard to the employees involved, with no clear need for all of the information collected.

DOE has in view the implementation of ''performance-based" waste acceptance criteria for WIPP and extension of that concept to define waste characterization requirements (DOE, 1995a, Section 4.3.2, pp. 4-7). The concept is straightforward and reasonable: the waste characteristics that should receive the most attention in characterization are those that are most important from the viewpoint of the performance measures on which the isolation capability of WIPP is evaluated.

With respect to long-term radiological protection, the requirements proposed by EPA in 40 CFR 194.24(b) are consistent with a performance-based approach. For each performance measure important to isolation capability (e.g., activity, permeability, and porosity), EPA proposes that DOE use performance assessment to establish a range of waste characteristics within which the facility will comply with regulatory standards. For those waste characteristics that are comparatively unimportant to isolation, the ranges will be wide. This approach should help avoid unnecessarily detailed characterization efforts.

The extent to which existing waste drums will have to be opened and characterized is not yet clear, because specific details of the characterization required have not yet been established. In particular, the method by and extent to which a knowledge of historical processes can be used to characterize existing wastes have not been established. Similarly, the requirements for using statistical methods to characterize waste based on measured values from a subset of that waste are not yet defined.

In the committee's view, it is essential that any waste characterization program be established with a clear understanding of potential uses of the characterization information. Measurements of TRU waste characteristics can be expensive and may lead to occupational exposures. The value of information gained by such analysis must exceed the cost and risk of obtaining the information, but these trade-offs can be considered only if the sensitivity of repository performance to different waste characteristics is understood. Where costly waste characterization requirements appear to offer little value in terms of ensuring health protection, DOE should seek to have the requirements reinterpreted.

Existing waste characterization regulations have been developed for surface waste management facilities rather than for a deep repository. These regulations reflect a general bias that more waste characterization is better than less and that the applicant has the burden of explaining why particular requirements should be waived in a particular case. 40 CFR 194 states that "any compliance application shall describe the chemical, radiological and physical composition of all existing waste proposed for disposal …" (40 CFR 194.24(a), Federal Register, p. 5240).

This is consistent with a performance-based waste characterization because a description of composition is not the same as a full characterization. The need for descriptive material is recognized. In the committee's opinion, there is no scientific basis for requiring a full characterization.

It appears that the project is concentrating too much on the problem of how to characterize waste and insufficiently on the question of whether to characterize it. The value of extensive characterization of WIPP wastes is questionable on several counts:

1. Whatever the current physical properties of wastes, they are likely to change significantly over the 10,000-year regulatory period of the facility. As the repository consolidates due to salt creep, the waste will be compressed and compacted.
2. The containment requirement of 40 CFR 191 limits the fraction of WIPP wastes that can be released over 10,000 years in terms of a percentage of the initial inventory; the absolute quantities that can be released are not defined. The total amount of radioactivity in WIPP will determine the release limits. For releases as cuttings, inventory information is largely irrelevant (see discussion of human intrusion earlier in this chapter). For releases due to solubility-limited flow, such releases may be largely independent of inventory, but release limits will depend on the total radiological inventory—that is, for the solubility-limited case, the more waste that is placed at WIPP, the easier it will be to comply with 40 CFR 191.
3. Performance assessments to date indicate that little, if any, waste migrates under undisturbed

• conditions and that releases from human intrusion—assessed for various stylized scenarios, which include some apparently very conservative assumptions—appear to be small compared to the release limits. If the waste is not going to migrate to locations where humans may come into contact with it later, then the qualitative, common-sense answer to characterization is that it does not matter what the waste contains. A quantitative answer to the degree of characterization required would emerge rigorously from PA calculations, using the release limits specified by EPA standards, which apply irrespective of whether the releases pose a risk to humans.

DOE is required to perform a monitoring study as part of the information to be submitted in the compliance certification documentation. DOE is required to assess the feasibility of monitoring in both preand post-closure phases. A key element is that the monitoring techniques cannot impair or degrade the containment of waste.

In the committee's opinion, this DOE monitoring study should distinguish carefully between current technologies and expectations for future technologies. The committee believes that this study should also assess carefully those technologies that must penetrate the integrity of the closed repository to be effective. Simply stated, it may be easy to monitor parameters that will not provide any useful information about the facility and very difficult to monitor parameters that could be informative but almost assuredly require breaching the repository so that probes, cables, and other devices, can be routed to the surface. DOE and EPA must recognize that a monitoring program typical for RCRA facilities may be impossible to implement for the WIPP facility. Although the committee believes that the requirement for DOE to perform this monitoring study is appropriate, both EPA and DOE must be prepared to face a situation in which long-term, post-closure monitoring within the repository may be impractical. EPA, importantly, recognizes that such a monitoring program may be impossible to achieve without impairing the integrity of the repository. However, monitoring of the subsurface environment above the Salado (in particular, with sensors in the Culebra) could provide warning of a breach, probably without compromising the repository.

The 40 CFR 191 regulation require "long-term monitoring" of the closed facility, but without indicating what this implies. How long is long-term? What is to be monitored? The committee believes that long-term monitoring must be done from the surface. Observations that would require instrumentation inside the seals pose a risk of adding potential leakage pathways. The risk from monitoring systems that fail and become part of a leakage circuit could be higher than the risk due to leakage from a reliable unmonitored facility. Considerable advances are being made in remote investigation and monitoring by geophysical methods, but these have not yet matured sufficiently to provide monitoring of a repository. In summary, the committee believes that the EPA, in 40 CFR 194, has proposed monitoring requirements that do not appear feasible today.

The need for a monitoring study nonetheless provides an excellent example of how information from the operational phase at WIPP can be used to deal with compliance issues. It may be feasible, for example, to install probes and conduits into a room or panel that is then sealed. This would provide a direct measure of the ability to perform the monitoring anticipated in the regulations and to assess the reliability and integrity of the monitoring system. Should one of the seals fail because of the monitoring system, the consequences will be relatively slight because the failure could be recognized and repaired easily. A pilot study during the operating period should be informative regarding any early failures of the monitoring systems.

Two of the most common practices for enhancing the quality of technical work are a process of competent peer review and an appropriate quality assurance program. The WIPP performance assessment activity has been under substantial review, by peer review groups and others, since 1989. With WIPP, the question is not a lack of review, but rather the quality of the review and whether or not the reviews being performed are the ones needed. The reviews fall into three main categories: (1) formal technical reviews, (2)

continuing review and feedback, and (3) miscellaneous review and advisory groups.

The first group of reviews is most likely to have continuity of impact. Reviews that appear to be particularly important are those from the PA Peer Review Panel, the State of New Mexico (both the Environmental Evaluation Group [EEG] and the Attorney General's Office), stakeholder reviews, and the EPA.

Despite the many peer review efforts and the superior technical ability of those performing them, the WIPP PA effort is centered on complex models, making meaningful probes difficult by outside review groups interested in understanding the effects and parameters most important to the outcome and interested in developing a simple, effective way in which these dependencies can be properly viewed. Given the earlier comments about the apparent conservatism in the analysis and the resultant misidentification of sensitivities, it appears that the peer reviews have overlooked these important shortcomings of the WIPP PA and have not resulted in a balanced, realistic assessment of WIPP performance.

One reason for this may be that the technical experts engaged in performance assessment have valued incremental technical innovations over transparency of the process. Understanding of the key aspects of WIPP radionuclide isolation performance need not be buried in a complex computer model. The descriptions of WIPP release scenarios described above are an attempt to indicate that the identified scenarios through which wastes could be released from WIPP are easy to visualize and that the factors contributing to releases can be identified readily. As the WIPP project moves forward in the compliance process, the hope of the committee is that the value of technical transparency and accessibility will be embraced by the project.

The quality assurance (QA) philosophy and procedures covering performance assessment are still under development. This documentation (see Sandia National Laboratories, 1991; 1992) covers QA procedures for (1) parameter selection and use of expert judgment, (2) analyses and report reviews, and (3) computer software supporting PA.

The development and implementation of QA procedures have progressed most visibly in the first of these three activities. Ideally, the QA program is to be implemented fully before final PA results are considered suitable for comparison with 40 CFR 191. However, full implementation alone is not sufficient to guarantee a high-quality result, which can be achieved only through the combination of a competent project team, peer review, and regulatory compliance. The final results have yet to be seen.

Performance assessment has a major role to play in demonstrating compliance with EPA standards and providing assurance to the public that transuranic waste can be stored without posing a significant health risk to the public in the Waste Isolation Pilot Plant. Although performance assessment did not originate with WIPP, the project has done much to advance the development and applicability of PA to radioactive waste disposal. As with any new development, be it a machine or a thought process such as performance assessment, initial expectations typically are much higher than actually can be achieved. Part of any criticism of the WIPP PA program may reflect these initial expectations.

Federal regulations relating to waste disposal define performance assessment as an assessment of the likelihood that the amount of radioactive material released to the environment will exceed a set of specified values. The analyses required to make such an assessment are capable of much more than merely demonstrating compliance with specific radionuclide release limits; they can be used on a much broader scale to enhance decision making and the overall efficiency of the WIPP project. The committee believes that these opportunities have not been exploited fully by DOE.

The general conclusion is that the WIPP performance assessment program is providing valuable insights on the performance capability of the repository, including the integration of features, events, and processes associated with total repository performance. The program is essential to success in opening the repository and can serve as the primary knowledge base for a meaningful risk management program during the operational phase. The extent to which performance assessment can become a primary monitor of repository

performance clearly depends on the scope and resource base of the PA effort. The committee believes that scope must include performance measures that go beyond radionuclide release rates to the accessible environment. In particular, the WIPP PA, although quite responsive to the letter of the law, is too narrow in scope to serve as a robust basis for risk management—a goal beyond that explicitly required by federal regulations but clearly in the interest of public health and efficient management of public resources.

An example of an opportunity for continued use of PA in the manner recommended here is in waste characterization. PA can provide guidance to waste characterization requirements. The impacts, both in terms of economics and of human exposure to the radioactive materials, can be significant. However, both DOE and EPA seem committed to an extensive waste characterization program that is only minimally defined by the need for information as input to the PA process. EPA is apparently requiring waste characterization efforts well beyond that necessary for assessing performance of the closed repository. The DOE seems intent on fulfilling these requirements with no clear scientific basis for the need for this information

Remaining issues of concern relate to the timeliness of PA results, the transparency of the PA model, and its limited scope in terms of aiding WIPP program decision making. The complexity of the model and its limited set of performance measures handicap its usefulness in project decision making, because the mechanisms, processes, parameters, and events that contribute to risk are difficult to understand. If a simpler model could be developed that would represent the full model reasonably and adequately, such a simpler model—with expanded performance measures—would benefit the project. Similarly, the results and insights of the performance analysis, as reported in the DCCA, are largely inscrutable. A simplified description of the scenarios and associated processes under which waste from WIPP could be released and people could be exposed to radionuclides would be a valuable complement to the WIPP performance assessment report.