Dr. David Snow
PhD, Engineering Science
University of California-Berkeley

In several places in his paper, Dr. Snow states that a near-surface or above-ground, centralized, monitored retrievable storage facility, possibly at WIPP, is now the only option to disposal at WIPP. However, it is CARD's position that the WIPP waste should remain in monitored retrievable storage at the generator sites while research is pursued to find a truly safe method for final disposition of this waste. The Department of Energy has stated that the WIPP waste could stay safely at the generator sites for at least another 50 years. CARD believes that the risks from transporting 35,000 shipments of waste around the country to WIPP or another centralized facility are great and outweigh the risks from properly storing the waste where it is generated or currently stored.

The (Short Version) of this paper is also available in the Non-Technical Section.

At WIPP, radioactive waste is being disposed of permanently in drums and boxes placed in rooms excavated in the Salado salt beds. Like all other excavations below the water table, the repository will saturate, and dissolved radioactivity can ultimately escape via boreholes, shaft.s or fractures to the overlying Rustler evaporites. The most evident aquifer in the Rustler, the Culebra dolomite, is claimed by DOE to provide such slow transport that the Rustler can be considered an adequate barrier to waste migration. But performance assessment modeling, based on insufficient exploration data, unsupportable deductions and faulty assumptions led to that claim. This paper asserts that the Rustler formation overlying and down-gradient of the WIPP repository will not provide the claimed geologic containment because karst conduits are present that will facilitate rapid, ephemeral flow. If disposal is not halted and timely rectified, escaping radioactivity may reach Nash Draw within a thousand years, contaminating the Pecos River and Rio Grande. Until a suitable disposal site or method is engineered, a monitored retrievable storage facility may offer the only alternative.

In 1998, the Department of Energy (DOE) became certified to dispose of transuranic (TRU) waste transported from military generation sites around the country, placing it irretrievably in rooms excavated in deep salt beds beneath the Waste Isolation Pilot Plant (WIPP), Carlsbad, New Mexico, and is presently expanding the facility. The urgency to eliminate surface stockpiles and to decontaminate bomb- making plants and test facilities at Rocky Flats, Argonne, Arco, Mound, Oak Ridge, Hanford, Savannah River, Nevada Test Site and the Lawrence Livermore and Los Alamos Laboratories gave stimulus to the conclusion that WIPP would provide safe, permanent isolation, that excessive plutonium and other radionuclides will not be conveyed in solution by groundwater moving from the repository through covering rocks to the accessible environment (surface or 5 km distant) in less than the 10,000 years mandated by 40CFR191.

Deep geologic disposal has seemed to be the answer. Natural radioactivity in the biosphere results in mild carcinogenesis that humans accept, while uranium ore bodies that could not be tolerated are shielded by intervening thicknesses of rock. Low solubility of ore minerals, and dilution and slow transport as solutes in groundwater have protected us. But in the last half-century, mining and processing of uranium ore, fission, bomb manufacture, and waste reprocessing have concentrated uranium and transuranic elements at numerous accessible places. It is logical to reverse the destination of the radionuclides, such as plutonium and americium resulting from bomb making, by burying these wastes deep in the earth. But such geologic disposal methods have proven so difficult to validate that every nuclear nation remains burdened with dangerous stockpiles of spent fuel, unneeded bomb triggers and contaminated materials. WIPP is the first and only permitted disposal facility in the world for TRU wastes. But because adequate containment conditions were not proven at the WIPP site, the facility does not make the case for safe geologic disposal. A single monitored retrievable surface or near-surface facility may be the most prudent U.S. repository until a better technology develops.

Various geologic media have been studied as potential hosts for the permanent disposal of U.S. radwastes. Crystalline rocks, including granite, basalt and metamorphics, though possessing low permeabilities that assure slow transport rates, were disqualified because all sites are water-saturated and connected to potable water sources. Though pervasive but tight fractures are the typical conduits for leakage, it has never been possible to prove the absence of preferred pathways, such as fault zones, that would shorten travel-times to streams or aquifers. Since 1957, rock salt has been considered a favorable medium because of the perception of a nearly impermeable, self-sealing nature. However, abandoned salt mines near Lyons, Kansas were disqualified in 1972 because the overburden rocks are perforated by many old drill holes. Salt domes of the Gulf Coast region were considered but disqualified due to the probability that fluids in a sealed cavern subject to creep closure would be ejected, ultimately reaching overlying aquifers. WIPP is similar in these regards. Bedded salt near Hereford, Texas failed as a candidate by decree of an economy-minded Congress, together with the candidate site in basalt at Hanford, Washington. That left. for active investigation only a site in welded tuff at Yucca Mountain, Nevada for civilian power-plant radwastes, and bedded salt at WIPP for military radwastes. Though high permeability and deep drainage of the fractured tuff at Yucca Mountain provides an unsaturated environment, remaining uncertainties of future climate conditions, faulting, volcanism and seismically-driven hydrothermal upwelling (Hill, et. al. , 1995; Hill and Dublyansky, 1999) cast doubts on the adequacy of waste isolation there. The WIPP site, selected in 1975 (U.S. DOE, 1997, p. 2-11), survived the purging of salt sites while the Atomic Energy Commission clothed the early investigations in military secrecy. At the onset of controversy concerning karst features over the WIPP site, the consulting role of the U.S. Geological Survey was discontinued, making Sandia National Laboratories the sole consultant to DOE. Technical oversight improved when the 1987 No-Migration Petition to the new EPA disclosed project weaknesses. Since then, DOE has maintained a mission-oriented relationship with its consultant, Sandia National Laboratories. Consequently, serious faults in the investigations have been perpetuated from the early 1980's through the 1996-1998 certification period (U.S. DOE, 1996).

The underground facility lies 2,150 ft. below the surface and is ultimately to measure 4350 by 5050 ft. in dimensions, a significant part of the Land Withdrawal Area (LWA, Figure 1), often referred to as the site. The geology has been summarized by Powers, et. al., (1978) and Brinster (1989). As Figure 2 shows, there are about 1300 ft. of salt and anhydrite beds of the Salado Formation above the repository. Above the salt is the Rustler Formation (Figure 3), comprised of thick beds of anhydrite interspersed with salt and dolomite, dipping about 1⁰ ENE. The Rustler is 400 ft. thick at P-10 beneath the east boundary of the LWA, thinning to 302 ft. at P-6 beneath the west boundary. Above the Rustler lies the Dewey Lake Formation, 199 to 534 ft. of redbeds of fine-grained, uncemented muddy sandstones and siltstones. Overlying the Dewey Lake is the Santa Rosa formation, 0 to 217 ft. of coarse sands and conglomerates interbedded with shales, mainly beneath the eastern half of the LWA. Windblown sand masks most of the semi-arid surface of the Mescalero Plain, which slopes gently west to Livingston Ridge, the edge of Nash Draw (Figures 1 and 5). It is Doe's position that any contaminated brines that escape the repository will be conveyed upwards via exploration boreholes through the Salado but not to the surface. Rather, contaminants are predicted to move laterally via the thin Culebra dolomite aquifer in the Rustler (Figure 3). On the site, an ill-defined water table is in the Dewey Lake redbeds, sloping westerly. But in the Culebra, the piezometric surface slopes westerly, southwesterly or southerly, depending upon place and time of observation. Because bedding dips northeasterly, nearly in the opposite direction, the water table offsite to the west of the LWA is in the Rustler, leading to outcrop areas in Nash Draw, 1.1 to 4 miles beyond (Figure 1), or to Malaga Bend on the Pecos River, 13 to 15 miles southwest of the LWA. Nash Draw is a subsidence trough formed by dissolution of the underlying Salado salt, which caused fracturing, brecciation and subsidence of the overlying Rustler (Kelly, 2000). Uncertain regional boundary conditions and a great range of transmissibilities across the site leave doubts about the paths and destinations of contaminants that may escape the repository, thus uncertainties of the time to reach outlets or the nearer accessible environment, nominally the subsurface limits of the LWA.

Repository performance is judged according the radioactivity accumulated over 10,000 years outside the accessible environment boundary. Limits in Curies based on health effects at a surface body of water are prescribed by the EPA in 40 CFR 191.13 for all radionuclides in the expected inventory of materials disposed. Performance Assessment (PA) is a modeling exercise of prodigious complexity (Helton, et. al., 2000), designed to take into consideration all physical properties of the wastes, the repository and its geologic environs, including effects of climate and man- made perturbations. It is a collection of models, each of which embodies a time-dependent mathematical representation, a method of numerical evaluation and a computer code. There is a model to represent the two-phase inflow and outflow of brine and gas between a sealed repository and anhydrite beds in the salt a few feet above and below the repository horizon. There is a model to calculate the hydrogen that will evolve from the reaction of brine with disintegrating steel drums and cellulose in the waste. There is a model to calculate the changing radionuclide inventory as it decays, dissolves in the brine and is absorbed by engineered backfill materials. There is a model to describe the groundwater flow in the Culebra dolomite aquifer (assumed to be the sole conduit) above the repository, and a model to calculate transport of each radionuclide species through time and distance along Culebra pathways. PA incorporates several different scenarios, such as no human disturbance during the 10,000 years, or mining of potash in the overlying McNutt Member of the Salado, or oil-well interceptions of the repository and/or pressurized brine reservoirs in the underlying Castile Formation, plus implications of borehole drilling, plugging and blowout prevention. Because a single PA requires many man- years of effort and millions of dollars worth of Cray CPU time, few have been done. Scores of physical properties are factors in the computations, many of which are ill-defined, so are best described as distributions. The methods of Monte-Carlo sampling are used, several hundred iterations providing a range of answers, each a distribution function of the cumulative releases over the 10,000 years. The proportion of iterations that satisfy the EPA criteria measures the probability of acceptable results. Any scientist has to admire the elegance of the computation procedure even if his expertise covers only a part, while the manager or politician reveres that which is understandably baffling. But scrutiny reveals that underlying the mathematical elegance and the comprehensive array of data manipulated are numerous assumptions, many of which have over-riding significance to the results. Some of the hydrogeological assumptions violate the perceptions of qualified critics, such as Anderson, 1978, Ferrall and Gibbons, 1980, Barrows, 1982, Snyder, 1985, Phillips, R. H., 1987, Snow, 1998 and Hill, 1999, and arguably differ from actual conditions. Management has exerted its will to succeed in the licensing process, influencing the scientific staff to adopt models and select studies favorable to Doe's objectives, biasing the results of PA. A critic must use Doe's own data and draw inferences from his own observations to show where investigations have gone astray. There are areas of geology, rock mechanics and hydrology that deserve re-assessment in that light.

The starting point for all PA model realizations is the calculation of brine inflows aft.er the repository is filled and sealed. As noted by Brinster (1989, p. II-19), the Salado is not homogeneous salt, but rather, “ . . . consists of salt rhythmically interbedded with anhydrite, polyhalite, some glauberite, and some thin mudstones.” The salt was formerly believed to be so impermeable that the rooms would remain dry, but the appearance of small brine seeps soon aft.er opening the first research rooms showed that DOE must contend with a wet waste environment. Excepting direct recovery of solid waste carried to the surface along with cuttings from inadvertent oil wells, all other scenarios entail radionuclide transport via flowing groundwater. The project might have been aborted if DOE had been more respectful of the historic problems of water in salt and potash mining. At WIPP, brine that accumulates will eventually saturate downdip openings, corrode containers and packaging and dissolve radionuclides. Generated gas will collect updip. The computed brine inflows depend on the measured permeabilities of fractured anhydrite beds above and below the repository horizon, but only the 3.0-ft. thick Marker Bed #139, 9.5-ft. below the repository floor, Marker Beds A and B, totaling 0.7 ft. thick and lying 4.3 ft. above the roof and the 0.6-ft. Marker Bed #138, 39.2 ft. above the roof have been modeled as inflow contributors. That limitation was due to the assumed extent of the DRZ, the disturbed rock zone (fractured salt) expected to form around the rooms as they close. The consequences of gas generation, cavity pressurization and two-phase outflows of brine and gas through those four anhydrite beds indicated (by PA) that the undisturbed scenario poses no hazard of a significant breach or accumulation beyond the accessible environment.

The fallacy of that conclusion stems from a misconception of the behavior of the Salado overburden. The 13-ft. high by 33-ft. wide rooms will be short-lived. Large open fractures appear in the ceilings of all rooms within months of mining. Several roof-falls and floor heaves have already occurred, so an extensive array of roof bolts has been installed to delay the failure of the remaining experimental rooms long enough to fill them with drums. These, and all future rooms will suffer collapse of major roof slabs bounded above by weak clay-bed partings. Such falls will crush the drums, and liberated waste will penetrate the fractures. DOE has assumed roof fractures extending upwards only to Marker Bed #138, but as creep subsidence incorporates whole panels and then the repository width, horizontal slip and openings will occur on successive higher clay seams, most bounding stiff anhydrites. Horizontal slickensides observed in Rustler clays at the Exhaust Shaft. (Holt and Powers, 1986) manifest the shear failure to be expected in Salado clay seams, a consequence of local subsidence. Inclined fractures laterally limiting roof slabs will interconnect the rooms and panels via the slip surfaces and fractured anhydrite beds situated farther above the repository, each of which will contribute to increasing inflows of brine. Experience at potash mines in similar salt sequences (notably at K-2 Mine in Saskatchewan) indicate that such roof behavior is typical. At the Canadian mines, the fractures sometimes breach the top of salt into an aquifer, causing inflows that flood the mine (Tofani, R., 1983, Van Sambeek, 1993). Aft.er shaft. leakage, such roof breaching is the next most common cause of flooding of salt and potash mines, all of which ultimately flood because they lie below the water table and have inhomogeneous, deforming roof-rocks. Such subsidence experience invalidates the assumed limited height of the disturbed rock zone around WIPP rooms and the continuous plastic creep and room closure envisioned by the designers. Fundamental to subsidence prediction at WIPP is that ultimately, inclined fractures, as shown by MacIntosh (1990) will bound a de-stressed region extending over entire panels of rooms. Because each anhydrite bed (numbering about 40 above the repository) has a thin, weak clay parting at each of its faces, the bedded, subsiding roof will be split into many independent beams, like a glu-lam beam that has come unglued. The deflections and accompanying inclined fractures will drain not just Marker Beds #138 and #139, plus A and B, but many more anhydrite beds above them, increasing the inflow rates accordingly. Some fractures extending up from the repository will not anneal, but because of movement and flow, will remain open for gas and brine leakage during the years of pressurization by closure and gas generation.

Far-field pore pressures approaching lithostatic in salt not only drive the brine inflows along the anhydrites and clays, but also ensure that little subsequent outflow follows the bedding during pressurization. In PA, the Salado above anhydrite marker #138 up to the Rustler is assumed to be salt with very low permeability distributed uniformly around a median of 6.1 X 10^-20 ft..². Utilizing Darcy's Law (Helton, 1991, p. 83 and Table 4), flow upward across the 1300 ft. of salt (about 85%) and fractured anhydrite beds (about 15%) up to the Culebra dolomite aquifer was computed as though the interval is a continuous porous medium. What is significantly wrong with the model is that it assumes no fracture conduits reaching high above the panels. Rather, as pressures in the sealed repository rise, gasses will cause the subsidence fractures to propagate unstably to higher levels where smaller rock stresses prevail, facilitating subsequent brine leakage to the Culebra and other aquifers much sooner and at higher rates than the PA model predicts. Because of the non-conservative assumption that the Salado is structureless, and consequently because the rock mechanics model is unrealistic of long-term subsidence, the conclusion from PA calculations that the undisturbed scenario is innocuous has to be wrong. Histories and subsidence behavior of the analogous Salado interval above the McNutt horizon at nearby potash mines of Eddy County, NM could have been studied, reported and modeled, to derive more realistic WIPP- site subsidence predictions. Sandia rock mechanics wanted to do that at the Horizon (Amax) Mine (Crosser, 1998), but funding was denied them. It is common geotechnical experience that in a significant proportion of dams, tunnels, aqueducts or deep mines, if there are potential but unknown geologic defects or mechanical inhomogeneities, failure will occur by reason of unexpected hydrologic effects. Therefore, in sensitive and doubtful situations, especially conservative assumptions are appropriate. Instead, WIPP modeling employed idealistic assumptions of continuous media, when discontinuous (fracture) properties would have been appropriate. The conservative expectation is that subsidence fractures at WIPP will propagate first by gravity, then unstably upwards to the Rustler due to the gas pressure generated in the repository, followed by contaminated brines aft.er the gas has dissipated and rooms become saturated. Sealed shaft.s and boreholes nearby will probably retain their integrity unchallenged, because fractures will provide easier egress for fluids.

There is currently a concentrated leakage occurring into the Construction and Salt Handling Shaft. from the top of Dewey Lake redbeds at 20 to 70 ft. in depth, thence into the repository, believed to arise from runoff at the parking lot. In European potash mining experience, such incipient karstic shaft. leakage has been found to be irreparable. The first drop of water signals the eventual flooding of the mine.