Appendix G An Overview of WIPP Compliance Issues
This appendix is a nontechnical synopsis of the major themes and findings of this report. The primary goal of this report is to present a scientific and technical appraisal of DOE activities in connection with the Waste Isolation Pilot Plant (WIPP) radioactive waste isolation program. This entails detailed discussions of complex issues related to engineering measures, biochemical processes, and hydrogeological aspects of the site. Radioactive waste isolation is also a topic of considerable interest and concern to the general public, and it is important that the public understand the basic concepts underlying the design and operation of the WIPP facility.
This appendix aims to describe the WIPP project clearly and concisely by using a "nonspecialist" approach—that is, one that avoids specialized technical terminology as much as possible—to discuss salient issues pertaining to the project. It traces the history of the facility and summarizes the committee's main concerns, conclusions, and recommendations regarding Department of Energy (DOE) activities at WIPP.
Deep Geologic Disposal
The objective of the nation's nuclear waste disposal program is to place solid waste in a location where it cannot return to the biosphere by any foreseeable natural process. A generic method has been selected: bury the material deep within the earth's rocky crust. Salt beds are particularly suitable because they are very stable over geological time scales and have the ability to flow and permanently seal the excavations in which the waste is placed. WIPP is located in such a formation.
The plan for waste disposal at WIPP is to excavate a series of underground rooms in the salt deep below the earth's surface. Radioactive waste will be emplaced in the rooms, and then the shafts to the surface will be filled with salt and other materials to close the facility. The weight of the overlying rocks will compress the disturbed salt and reduce its ability to transmit fluids and gases to the very low values of the undisturbed salt, thereby "sealing" the repository.
The worldwide use of salt cavities to store compressed gases attests to the remarkably low permeability of deeply buried salt (Berest and Brouard, 1996). Salt also has a well-known ability to flow under applied stresses. At the depth selected for the repository, crushed salt will reconsolidate to the same low permeability as that of the original formation, and even large cavities eventually will disappear.
Numerical modeling can predict the flow behavior of salt far into the future for any defined situation. The flow is slow enough to allow ample time for underground operations, yet fast enough to be essentially complete in a hundred years or so.
Project Administration And Regulation
The WIPP project was begun in the mid-1970s. It has been the responsibility of DOE and its predecessor agencies, but many years passed before it was determined that the U.S. Environmental Protection Agency (EPA) would be responsible for certification. Disposal regulations then had to be developed in parallel with field activities. The "Final Rule" for the Criteria for Certification was published on February 9, 1996, as this report was being prepared.
The long timeline associated with transuranic (TRU) waste poses problems for the regulation of WIPP, just as it does for design and construction. It seems obvious that protection of human health should be fundamental, yet it is impossible to predict what humanity will be doing in the vicinity of WIPP during the next ten
millennia. Given the nature of the EPA standard and the apparent lack of release scenarios for WIPP under undisturbed conditions, it is not necessary to calculate individual doses as part of a compliance analysis. The part of the EPA standard that is relevant to WIPP is a release limit that avoids the necessity of making arbitrary assumptions about population distributions and human activities in the far future. Unfortunately, this release limit fails to account for one of the advantages of the WIPP location—namely, that most of the ground-water in the area is so salty (i.e., brine) that it is undrinkable.
The release limit is specified in a rather surprising way: not as an absolute amount, but as a fraction of the waste inventory. The amount is measured conventionally in curies (a measure of radioactivity) rather than mass. This specification can be translated approximately for nonspecialists as follows: the expected waste inventory is of the order of 6 million curies, and the permitted release fraction of plutonium-239, the dominant radioactive isotope of concern, is 0.0001 of that, or 600 curies. This is the same as a mass of Pu-239 of about 20 pounds.
Regulation in terms of release did not entirely relieve EPA from the necessity of making arbitrary decisions. How long a period is implied by the word "permanent"? EPA settled on 10,000 years, although the half-life of plutonium is 24,000 years. How long can one assume "administrative control" of the WIPP area after closure? The EPA chose 100 years. What kind and frequency of human intrusion should be considered? The EPA chose, reconsidered, then chose again. DOE is now required to assume that drilling will continue in the same manner and intensity as today for the next 10,000 years. What credit should be given to the present generation for assuming extra risks that reduce risks to future generations? The EPA is silent on this point.
The simultaneous presence of radioactive and chemically toxic wastes in WIPP presents a "mixed-waste" conundrum. Must EPA enforce Resource Conservation and Recovery Act (RCRA) provisions (relating to disposal of hazardous waste) at WIPP? A recent law (P.L. 104-201) removes this requirement at the federal level, providing DOE with an exemption from the need to demonstrate that volatile organic compounds (VOCs) would not escape from the repository. The justification for this exemption is based on the assessment that the toxicity of the radionuclides in WIPP is far more important than that of the chemicals.
Certification Of Wipp
In October 1996, DOE intends to submit to EPA an application for a certificate of compliance. A critical item in the documentation package will be the results of the performance assessment (PA) group at Sandia National Laboratories (SNL). PA is a set of methodologies used to model the long-term behavior of radioactive waste disposal systems.
The purpose of a performance assessment is to answer three questions:
- What can go wrong?
- How likely is it to go wrong?
- What are the consequences?
The first question was answered years ago by inviting interested parties to submit "scenarios" in which human or natural events lead to the release of radionuclides from the repository. After extensive review of a large number of proposed scenarios, eight were selected on the basis of their credibility and significance.
The PA group at SNL is charged with answering the other two questions, using an approach mandated by EPA. The group is developing a large computer model of WIPP. The model interconnects many large special-purpose programs, addressing, for example, site hydrology, geology, and design features. Into this model go descriptions of the eight scenarios, along with data on the many physical and chemical parameters and rate constants required for the computations.
Providing data for this model has been and continues to be a major activity of SNL and its subcontractors. Laboratory and field work play a large part; other approaches include the systematic "elicitation" of expert opinion for matters that are difficult to measure (e.g., the solubility of certain actinides in saturated brine) or impossible to measure (e.g., the best way to communicate to future generations the danger of digging within the WIPP reservation).
In one of the eight scenarios, the repository remains undisturbed by humans after it is closed. For this "undisturbed" case, neither the committee nor DOE's contractors could think of any way in which there is a credible possibility for release of radionuclides.
The other seven scenarios involve human intrusions (HI) of three kinds, alone and in combination, and with various frequencies. A variety of possible circumstances, such as the presence or absence of brine, accompanies each of these "disturbed" scenarios. The existence of many possible scenarios and many kinds of data makes the computations very complex. The computer can handle this complexity, but the computations are nearly inscrutable to all but specialists.
A problem arises for certain data used in the computations: what numerical value should be used when only a broad range of possible values is available? Choosing the "worst case" value (i.e., the least desirable value) of every range maximizes conservatism but introduces an unnatural bias. When faced with such uncertainty, statistical theory provides a means to make predictions on the basis of a defined range in values. This is accomplished by making many simulations for the same scenario, but in each simulation a different value for each uncertain parameter is selected randomly. If implemented carefully and if a large number of such model simulations are made, the average of the numerous predictions can provide a reasonable estimate of the likely outcome for comparison with the compliance criteria.
The output of such simulations is a calculation of hypothetical releases of radionuclides from the repository. Associated with each calculated release is a calculated likelihood (probability) that such a release will occur. Each of these calculations of release and probability of release is called a "realization." The realizations are thus a series of related probabilities and releases that can be plotted on orthogonal axes (e.g., horizontal and vertical, with releases on the horizontal axis and probability on the vertical axis). The releases appear as points, and the curve connecting them would look like a stepped line or a smooth curve, depending on the number of realizations. The analysis of each scenario includes many realizations.
Finally (see Figure 2.1), scenario probabilities are combined for graphical display as complementary cumulative distribution functions (CCDFs). Plotted on the same scale is a single, rather simple graph representing in probability form the EPA maximum-release specifications.
Chemistry, Biology, And Geotechnology
It has been the practice of the PA group to be conservative in modeling. This is sensible, except that two or three excessively cautious estimates may combine to produce an unrealistic model result, or a conservative assumption may come to be regarded as a "given." Therefore, it is important that conservative assumptions be confronted constantly with new information as it is developed. This confrontation has not always occurred in the WIPP PA.
For example, the pre-1996 PA efforts used release scenarios that assumed the following:
- all storage rooms were interconnected and able to be flooded with brine;
- brine might dissolve all the steel containers and all the TRU elements within them; and
- gas production by chemical and biological processes might raise the pressure within the sealed repository to and beyond the pressure of the overburden (lithostatic pressure).1
These assumptions may be erroneous. The amount of brine that can enter the repository from rock in the immediate vicinity is in fact quite small in the undisturbed case. Brine occurs in the Salado Formation as pockets in "seams" of clay and/or anhydrite (calcium sulfate), and in porous regions that occur irregularly within the otherwise nonporous salt. There appears to have been little or no interconnection within the pockets of brine in the salt for millions of years.
Excavation disturbs the brine pockets in the vicinity of the cavity. During excavations, the rock immediately around the excavation becomes fractured, leading to what is called the "disturbed rock zone" (DRZ). Some of the fractures penetrate the brine pockets and cause the brine to drain toward the excavation. The rate of brine collection has been observed to peak within a few years and then drop off as the DRZ ceases to grow.
Hydrogen generation from reaction between the brine and the waste containers also has been overestimated. Laboratory work has established that brine dissolves steel only when there is actual contact between liquid and metal; where there is no contact, only a thin "tarnish" film forms.
Microbiological experiments have not yet given a definitive answer about biological gas generation, but this too requires liquid water. Moreover, there are fundamental reasons for doubting that organisms can continue to be active metabolically for long periods in an isolated environment containing only limited amounts of the many organic and inorganic factors that are required, even when the food (carbohydrate) supply for microorganisms is unlimited.
The committee uses these considerations to judge that there is only a small likelihood that significant amounts of brine will enter the repository and react with the waste to produce hydrogen gas.
On the other hand, brine flooding does occur in scenarios in which a borehole connects the repository with a pressurized brine reservoir in the Castile Formation below the Salado. The possibility of this type of event is given serious consideration, in part because such a reservoir was encountered in a test boring in the 1970s.
The Actinide Source Term
Scenarios that assume brine flooding followed by leakage of contaminated brine upward into a near-surface aquifer require data on the possible amount and concentration of radionuclides. The "actinide source term" (AST) refers to the kinds and amounts of dissolved or suspended radionuclides that could be released to the biosphere.
The total quantity of radioactive materials expected to be emplaced in WIPP has been estimated from inventory records of the sites where wastes are currently stored. However, estimates of the proportion of waste that actually would be dissolved in the case of brine flooding are very uncertain. This seems like a straightforward chemistry problem, but few laboratories are equipped to handle plutonium (a large component of the radionuclide inventory at WIPP). There is little useful information about this issue in the published literature.
An experimental study of this problem was begun at Los Alamos National Laboratory (LANL) in 1994, but no data have yet become available. The current suite of experiments is scheduled for completion in May 1996. It is possible that useful data will be available in time to be included in the PA in the October 1996 certification package. Evidence from the German waste isolation program, which is also considering a salt repository, suggests that plutonium solubility in brine is likely to be low, but it is important to ensure that the brine at WIPP, which is different chemically in some respects from the German brine, behaves in a similar fashion with respect to solubility.
Hydrology Above The Salado Formation
Some scenarios specify that the radionuclide-contaminated brine will be under sufficient pressure to be transported upward into strata above the repository, where it mixes with water of lower salinity, forming a "plume" of contamination.
Much of the radioactivity in the plume may be trapped in the host rock or at least slowed (retarded) by a combination of known effects including precipitation (formation of a solid chemical compound), sorption (chemical reactions with rock surfaces that inhibit transport), and matrix diffusion (egress into fractured rock pores). However, it is difficult to make reliable predictions of these effects. Laboratory experiments with powdered rock are not necessarily good analogues for the natural environment, and even percolation experiments using columns of intact rock cored from the formation are problematical.
Only extensive field experiments can hope to deal directly with the problem of radionuclide transport in the subsurface around WIPP. Unfortunately, the state
of art is such that even this approach is not necessarily definitive.
As used in this report, the term "non-Salado hydrology" refers to ground-water flow that occurs in aquifers within the rock formations that lie above the Salado Formation. The Culebra Dolomite is the major aquifer, and it is the only one considered in the 1992 PA model. The water in this aquifer, however, is too salty for consumption by either humans or livestock, making the Culebra a poor potential route for radionuclides to reach the food chain in the vicinity of WIPP.
A considerable amount of field work has been completed to better define the properties of the Culebra, and significant advances have been made. However, the data base and the level of understanding of this flow system still contain notable gaps. A major gap is the lack of attention to a possible radionuclide contamination route through the Dewey Lake Red Beds, in which the water is known to be potable for both humans and livestock. Although a unique definition of the properties of a deep regional system is neither attainable nor necessary, the committee believes uncertainty about some critical issues could and should be reduced further. Field tests in seven new wells drilled at pad H19 at WIPP are scheduled for completion in April 1996, and the results may provide useful new information on contaminant transport and matrix diffusion for the 1996 PA.
Past PA studies have indicated that releases into the Culebra might be serious enough to disqualify WIPP for certification unless it can be shown that source-term concentrations are lower than the high values now conservatively assumed, or that retardation within the Culebra is much greater than zero. In other words, the combination of high plutonium solubility and low retardation under the assumptions of the 1992 PA have the potential to generate noncompliance. A full description of which combinations of parameter values generate noncompliance is an outcome of the PA analysis, and in particular, of the sensitivity analysis derived from the model. Results from past PA studies are illustrative, but use assumptions (such as the number of future boreholes drilled per unit area per unit time) that differ from the parameter assumptions specified by the recent guidance of 40 CFR 194. The future CCDF results of the 1996 PA, especially the derived sensitivity analysis, would contain the quantitative answer to the dependencies on each parameter of the model.
Repository Design Using Compartmentation
Pre-1996 PA calculations were based on a repository design in which separate waste-filled rooms would be connected by the DRZ, which would provide a passage way for gas or liquid. An alternative design would use crushed salt to seal closed each waste-filled room, so that gas or liquid entering one room is prevented from reaching another.
This alternative design provides an approach to demonstrating compliance that the committee expects would lessen the dependencies on (and concerns about) gas generation, the actinide source term, and non-Salado hydrology. Other engineering measures can also be used to improve the repository design. The reconsolidation of salt, discussed earlier, is a powerful design tool, which can be used to create a fully compartmented repository. Access tunnels can be backfilled with some of the salt removed during excavation. Compartmentation and isolation of the waste rooms can be achieved through the use of backfill materials to make effective seals (i.e., barriers to the movement of liquids and gases) to plug all connections made during excavation. Following consolidation of the plugs of backfill material into a tight seal, penetration of any one compartment will have no effect on the rest of the repository.
Such compartmentation would reduce both the probability and the extent of any potential brine flooding associated with human intrusion scenarios. If the rooms were isolated from each other, problems associated with brine entry would be in general reduced, compared with a fully connected repository. By incorporating such a design, the committee believes that the PA model would show that the current state of knowledge about source term and non-Salado hydrology is sufficient to ensure compliance. This expected outcome remains to be demonstrated. However, even should it be possible to certify compliance of WIPP with the available information base, the studies of source term and hydrology should be continued, since these problems are generic to all
concepts of geological waste disposal and could provide considerable additional confidence in the long-term performance of WIPP.
Summary, Conclusions, And Perspective
The readiness of WIPP to open and operate as a radioactive waste repository is the issue at hand in DOE's efforts to demonstrate compliance with EPA regulations. This appendix presents some of the relevant technical issues. The scientific and engineering work of the past several years will be used for the PA computer model calculations by SNL. These calculations compare the projected, estimated performance of WIPP to EPA requirements. The results, a CCDF curve, contain the quantitative assessment of whether the facility can comply with the regulations.
For the undisturbed repository, compliance depends on the ability of the crushed salt seals in the WIPP shafts to develop a sufficiently low permeability to retard the movement of water into and radionuclides out of the repository. This occurs by natural creep processes in a relatively short time (50-100 years), with the salt seals tending to become impermeable eventually, so that releases should be well below the limits of the EPA standard (see Chapter 4).
For a repository disturbed by human intrusion, when evaluated on the basis of reasonable expectation of intrusive activities and their consequences, and using models that would implement available engineering features and do not make overly conservative assumptions, the consensus of the committee is that the WIPP repository could be shown by DOE to comply with the EPA standard.
Compliance with the EPA standard does not address human exposure for the disturbed case. In some release scenarios, such as releases of radionuclides into water at depth in the Salado, the lack of a credible exposure pathway to humans provides an extra margin of reserve in addition to the compliance requirements.