Safety in the (Very) Long Run
Assessing long-term performance of a proposed geologic nuclear waste disposal site
The monumental pyramids of Egypt, built around 5,000 years ago, are among the oldest man-made structures on earth. But a modern-day project has been proposed by the U.S. Department of Energy (DOE) that would create a repository so safe and strong, yet so inconspicuous and remote, that it would isolate humankind from some of the world's highly radioactive nuclear waste for twice as long as the pyramids have stood.
The proposed geologic repository at Yucca Mountain, in Nevada, would be designed such that its subterranean waste storage packages will be long-lived and that should any releases of radionuclides occur in the future, they would be small and spread slowly.
In 2002, the White House recommended the site and then Congress approved Yucca Mountain as the site of a geologic repository for spent nuclear fuel and high-level nuclear waste and directed the DOE to seek a construction authorization from the U.S. Nuclear Regulatory Commission (NRC) for the repository. The recommendation followed two decades of congressionally mandated research involving an array of scientific and technical disciplines to ensure the long-term safety and integrity of the site.
The NRC, which is responsible for developing regulations and setting forth technical requirements and criteria consistent with the Environmental Protection Agency standards for licensing the repository, established the federally funded Center for Nuclear Waste Regulatory Analyses (CNWRA) in 1987 at Southwest Research Institute (SwRI) to provide technical assistance and research. The NRC has taken no position on the suitability of the Yucca Mountain site.
High-level Radioactive Waste
DOE is proposing that seventy thousand metric tons of radioactive waste go to the proposed Yucca Mountain repository1. Although most would be spent nuclear fuel from civilian electricity-generating reactors, numerous other types of nuclear waste are to be disposed, including the byproducts of reprocessing spent nuclear fuel, high-level waste from defense activities and numerous waste forms from research and other activities. Prior to disposal, some high-level waste will be vitrified, or heated and mixed with other materials to form glass, then poured into stainless steel canisters. The spent nuclear fuel will be disposed in the form it was used in the power plant pellets stacked and sealed in metal alloy tubes called cladding.
If improperly handled, the high-level waste can be deadly. While a fatal whole-body dose for humans is about 5 sievert (Sv) of acute exposure, the surface dose rate for a typical spent fuel assembly exceeds 100 Sv per hour, 10 years after removal from a reactor2. Radioactive isotopes will eventually decay to harmless materials. Some isotopes decay in minutes, hours or days while others decay very slowly. For example, thorium-234 has a half-life of 24.1 days, while uranium-238 has a half-life of 4.5 billion years. Therefore, the wastes must be disposed at a site that adequately isolates them from human beings for a very long time.
DOE's proposed design for Yucca Mountain would provide twofold protection through a system of natural and engineered barriers. The natural system would capitalize on the site's geology as well as its desert climate and sparse vegetation. Furthermore, the area surrounding Yucca Mountain has a relatively low population density. The main attributes for the natural system are the surrounding rock's ability to slow down the movement of water and chemically remove radioactive particles from contaminated water and hold them in place, and the long distance that radioactive particles would have to travel through the rock before reaching an area where water is likely to be used by anyone.
The engineered, man-made system is proposed to be embedded in the geologic medium, nearly halfway between the land surface and the water table approximately 600 meters below the surface, to add a second barrier against release and migration of radioactive materials. The engineered system consists primarily of a corrosion-resistant waste package covered with a metal shield to prevent water and chemicals from dripping on and entering the waste package. The waste packages would be stored on stands so that any water pooling in the disposal tunnels would not contact them. Between 8,800 and 12,000 waste packages are expected to be used by the DOE for disposal.
The CNWRA, under contract from the NRC, has developed a system of tools, referred to as Total-system Performance Assessment (TPA) tools, for independent analysis of the repository performance3,4. The NRC will use these to review a potential DOE license application for the proposed repository. The main component, referred to as the TPA code3, provides a system-level framework for quantitative understanding of the key factors controlling the degradation of the engineered system, the release of radionuclides from the repository, the transport of radionuclides through various environmental pathways and possible human exposures. The TPA code was developed by a team of CNWRA and NRC scientists and engineers with combined expertise in risk assessment, hydrology, materials science, mechanical engineering, rock mechanics, geology, volcanology, geochemistry, health physics and computer science.
Estimating the performance of any underground repository is a complex exercise with substantial uncertainties and an extremely long period to be analyzed (the regulatory compliance period is 10,000 years). These uncertainties are attributable to incomplete knowledge of the characteristics of the natural system and to the uncertain performance of man-made structures over a very long period.
Any detailed computer modeling must consider credible scenarios such as earthquakes and volcanism; coupling among thermal, hydrological, chemical and mechanical processes; spatial variations in the emplacement of the waste form; rock physico-chemical properties; uncertainty in the process; various conceptual models; and the likelihood that possible future events will actually occur. A key strategy is to simplify the problem such that it is tractable, yet realistic enough to produce a credible analysis.
Information gained from detailed process models at the subsystem level, natural analog studies, and laboratory and field work are vital to carrying out the performance assessment. To account for the uncertainty inherent in characterizing a large and complex system for such long periods, probabilistic simulations are used.
The following is a description of a hypothetical scenario where release of high-level waste is assumed to produce a radiological dose to a hypothetical human receptor (assumed to be a member of a farming community near Yucca Mountain). For this scenario it is assumed that the engineered system degrades, allowing water infiltrating through the mountain to enter waste packages and dissolve radionuclides. As water flows through the unsaturated and saturated geologic media beneath the repository, it transports radionuclides to distant irrigation and drinking water wells.
Many factors affect the magnitude of these processes. The climate is likely to change with time and will likely increase the present-day infiltration rate over a 40,000-year period before it plateaus and the decreasing trend begins. Topography of the mountain, vegetation, soil cover thickness, presence of fractures, faults, lithophysal cavities (voids that occur naturally within rocks) and heterogeneities in the hydrogeology of the unsaturated zone will determine how much rainwater will infiltrate and move through the rocks. The water infiltrating into the mountain will get diverted around the drift, then around the waste package before entering small cracks in a compromised waste package and dissolving the spent nuclear fuel.
Heat generated from the decaying radioactive waste could lead to thermal-hydrological-chemical coupled processes affecting the environment for a period of time. During this time, the drip shield and waste packages are susceptible to accelerated corrosion and the waste form is susceptible to accelerated dissolution.
Conduction, convection and radiation heat transfer will govern temperature and relative humidity variations at the repository near field, the drift wall, the waste package and the waste form. Water vapor created by above-boiling temperatures could likely move away from the waste package and perch above the near-field area after condensation, but may move back toward the repository in the fractures (known as refluxing) even when the waste package is still above the boiling point of water.
Seismicity in the region may induce rock falls in the drifts, or storage areas, which could cause a failure of the drip shield and waste package in the unbackfilled repository.
For release to occur, the waste package would have to fail and water would have to contact the radioactive waste. Waste packages and drip shields could fail because of hidden defects. Based on a limited literature survey, 0.1 to 1 percent of waste packages are assumed to undergo premature or initial failures because of manufacturing defects and emplacement accidents.
The chemical composition of the water will change under elevated temperatures and affect degradation of the engineered barrier, and thus the rate of radionuclide release. However, most waste packages are anticipated to eventually corrode, resulting in a breach of the waste package. Corrosion could occur in the presence of very little water (humid-air corrosion) or in an aqueous environment (under a film of water). The size and shape of a breach caused by corrosion, the conditions in the near-field environment and rate of water percolation into the drift, waste form (leaching) dissolution rate, and radionuclide solubility will determine the rate of aqueous releases of radionuclides from the waste package.
The progress of the release of radionuclides depends on the location of the breaches in a waste package. If breaches occur primarily on the upper half, the waste package could form a vessel similar to a bathtub in which the water must fill to the height of the breach before radionuclides spill out and are released. If breaches occur at the top and the bottom of the waste package, the water dripping on the waste form would release radionuclides without any accumulation. Radionuclides leaving the waste package could diffuse slowly through the medium surrounding the waste package or flow more rapidly through interconnected fractures.
There are other scenarios for the failure of the engineered system that are less likely but may have greater consequences. Those scenarios include:
Displacement of a yet-unknown fault intersecting the repository, causing shearing of waste package and exposing radionuclides for release into the groundwater;
Intrusion of magma that damages waste packages by a combination of high temperatures, corrosive conditions and large mechanical forces, exposing radionuclides for release to groundwater; and
Extrusive volcanism that damages waste packages, entrains waste and ejects radioactive material with airborne volcanic ash. This phenomenon could allow the ashspent fuel plume to move to great distances in the direction the wind is blowing.
The receptor could be exposed to radionuclides transported either through groundwater or through the atmosphere as a result of volcanism. The farm community could drink contaminated water or use contaminated water for other residential purposes such as gardening; ingest contaminated, locally produced food products; be exposed to surface contamination, such as inhaling radioactively contaminated dust; or get direct exposure to radioactivity. Calculations have been done for a hypothetical receptor at 20 kilometers from the repository where irrigation wells may be located.
Dealing with Uncertainty
Because of the long time scales and inherent process uncertainties, it is not possible to have perfect knowledge or understanding of repository performance. However, decision-makers can still make well-reasoned, risk-informed decisions regarding the safety of the proposed repository if the effects of uncertainties are evaluated.
Parameter and model uncertainties, spatial-temporal variabilities and design alternatives, if available, are considered so that a large number of potential scenarios can be considered.
Uncertainty and variability over a range of potential scenarios are propagated in the TPA tools by Latin Hypercube Sampling (LHS), a stratified Monte Carlo sampling approach. The uncertainties of key model input parameters are quantified by assigning probability distributions to them. Events that originate external to the repository such as volcanism and faulting could have high consequences but very low probability of occurrence. Such small probabilities often challenge probabilistic analyses. The TPA tools handle such shortcomings through specialized techniques.
Subsystem and System-Level Performance Estimates
The current climate at Yucca Mountain is arid; however, based on historical analysis, the climate is predicted to become moister and cooler over the next 40,000 to 60,000 years before returning to drier conditions around 90,000 years. Under the current climate, infiltration through the mountain may be as much as 31 millimeters per year during the regulatory period, but could increase during the wetter period that follows. Higher flow rates could potentially release a larger mass of radionuclides from the engineered system to the natural system.
The peak repository temperature could reach an average of 165 degrees Celsius. Elevated temperatures might increase the spent nuclear fuel dissolution rate, form aggressive chemistry, alter water flow rates and increase the rate of corrosion of the waste package and the drip-shield.
If other mechanisms have not already breached the waste packages, corrosion processes are estimated to breach them over 38,000 to 131,000 years, but that is significantly beyond the regulatory compliance period of 10,000 years.
A small fraction of the annual precipitation would infiltrate below the ground surface to the repository horizon and contact the waste package. A smaller fraction is estimated to contact the waste form, and more than 99 percent of the water precipitating on Yucca Mountain corresponding to the repository area or "footprint" is estimated to not enter the waste packages.
While the impact of rockfall is still under investigation, current calculations suggest that seismically induced rockfall or drift collapse would not fail waste packages over the entire regulatory compliance period of 10,000 years. A small percentage of waste packages could fail from fault displacement or from intrusive igneous activity (magma flow in the drifts), but the probability of these conditional failures is low and the consequence is not high.
Groundwater release of radionuclides is therefore dominated by the initially defective failure of a small percentage of the waste packages during the 10,000-year period. A very small percentage of waste packages could fail from extrusive igneous activity (volcanism), but even at this low level of failure and a very small probability of occurrence the risk (calculated as consequence times probability) may not be negligible because the consequences of such a failure may be high. Accounting for these disruptive events, nearly 98 percent of the waste packages are estimated to remain intact for at least 10,000 years.
TPA analyses project that radionuclide groundwater concentration arriving at the receptor location from initially defective waste package failure will peak any time between 3,700 to 32,000 years, while the peak concentration from corrosion will occur between 41,000 to 100,000 years. Radionuclides with a combination of high solubility, long half-life and large initial inventory exhibit the largest release rates. Although the unretarded (chemically not sorbed in the rock) radionuclides take an average of 640 years to travel in the groundwater aquifer, compared to 280 years in the unsaturated zone below the repository, the longer flow path, combined with greater retardation in alluvium, substantially delays transport of radionuclides to the pumping well. System-level performance estimates are not presented here, but such estimates will be presented by the DOE in its license application. NRC's performance assessment calculations are primarily for checking DOE's calculations.
Sensitivity Analysis Results
In addition to explaining how the number of waste packages failed or the amount of radionuclides released from the engineered and natural systems with time affects the dose risk to the receptor, the TPA tools also facilitate the conduct of sensitivity (or uncertainty importance) analyses with parameters, repository system components and alternative conceptual models to determine the importance of uncertainties to repository performance.
The importance of such analyses cannot be over-emphasized because of the relatively limited data for developing process models such as distribution functions to represent uncertainty in the system model). Calculations show that a 10 percent change to the mean of the distribution function of some parameters could result in as much as a 150-percent change in the dose risk estimate.
Similarly, an example conceptual-model sensitivity analysis (using an alternative spent nuclear fuel dissolution model) decreased dose risk by nearly 100 percent.
Repository component-level sensitivity analysis showed that repository performance is relatively sensitive to the performance of the waste package, the unsaturated zone and the saturated zone.
The results of sensitivity analyses are used to help determine how NRC should focus its high-level waste program and how pre-licensing interactions with the Department of Energy should be conducted.
The suite of probabilistic TPA tools developed at the CNWRA under contract to the NRC has helped the CNWRA and the NRC to independently evaluate the performance of the proposed repository at Yucca Mountain. These tools have been central to identifying influential repository components, models and parameters. These, in turn, have helped the NRC to identify the areas where additional focus may be needed to reduce uncertainties. The TPA tools have helped NRC in developing a risk-informed, performance-based regulation4 and have aided the CNWRA and the NRC in refining their understanding of how various key features, processes and events in an integrated system model influence risk estimates. This knowledge will be important to prepare for, and actually review, the license application expected to be submitted by the Department of Energy in the near future.
As inherent to any performance assessment model, the results are based on numerous simplifying assumptions and the model uses currently available site-specific data while site investigation, design and testing continue. But the flexibility to conduct system-level analysis with different conceptual models, using data inside and outside the range of uncertainty, has been effective.
Comments about this article? Contact Mohanty at (210) 522-5185, or firstname.lastname@example.org.
Published in the Summer 2003 issue of Technology Today®, published by Southwest Research Institute. For more information, contact Joe Fohn.