In This Issue
Spring Issue of The Bridge on Emerging Issues in Earth Resources Engineering
April 14, 2014 Volume 44 Issue 1

Geologic Disposal of Spent Nuclear Fuel

Friday, April 18, 2014

Author: John A. Cherry, William M. Alley, and Beth L. Parker

An Earth Science Perspective

Climate change in the form of global warming is a widely accepted threat to humanity on a global scale and attributed at least in part to anthropogenic greenhouse gas (GHG) emissions. Nuclear power is a well-established source of electrical energy that produces minimal GHG emissions (Kharecha and Hansen 2013). Yet expansion of nuclear power worldwide has stalled except in Asia and production is poised to shrink in Europe and North America as old reactors are decommissioned and few new ones built. Public support for nuclear power is entangled with the unresolved issue of final disposal of the used (spent) fuel. One solution for used fuel and other high-level radioactive waste is entombment in deep, low-permeability geological repositories (DGRs; 300–600 m below ground surface) that are separated from the biosphere. The US effort to find a deep geological repository began in 1957 with a report by a National Academy of Sciences committee (NRC 1957). An international consensus favoring DGRs also began around this time.

The benefits of nuclear power to combat climate change are global. Therefore, there are ethical and economic arguments for cooperation toward international repositories using geologies that are exceptionally favorable for long-term radionuclide containment. We briefly review progress in the United States, Canada, and other major nuclear power–producing countries.

Deep Geological Repositories

The concept of deep geological repositories for the long-term storage of nuclear waste relies on the following assumptions: (1) rock mass—and the associated stability of the physical, hydrological, and chemical environment—will constitute the primary physical barrier and part of the chemical and transport barriers for radionuclide isolation from the biosphere; (2) the rock is suitable for mining, and the open cavity will remain stable during the time necessary for emplacement and retrievability; (3) the rock mass around the cavity can withstand long periods of high temperature and dissipate the heat from the waste; and (4) the cost of creating the DGR is affordable and justifiable.

The main evaluation stages in the DGR concept are detailed site characterization, system design and evaluation, vault creation by mining, loading with spent fuel, a period of performance monitoring and assessment of retrievability, and final closure by infilling and sealing accompanied by further performance monitoring. These stages may span 100 years or more depending on technological advances for spent fuel reuse or reprocessing.

Although the quest to create DGRs began nearly 50 years ago and many billions of dollars have been spent, none yet exists and all nuclear power–producing countries currently house their used fuel in temporary storage facilities near the ground surface. Sweden, Finland, and France are getting close to constructing DGRs, but Canada and the United States have fallen far behind with no sites selected for detailed assessment (beyond Yucca Mountain, Nevada); in Canada a number of sites are in the preliminary (desktop) evaluation stage of assessment.1

While there are many reasons for the failure to develop a used-fuel DGR (Alley and Alley 2013), this paper focuses on a critique of key issues from the geoscience perspective:

  • the exceedingly long time frame and issues concerning the predictive capabilities generated by field-derived, site-specific information most relevant to radionuclide migration over geologic time;
  • the advantages and disadvantages of different rock types for mining, radionuclide containment, and retrievability; and
  • the relative decision weight placed on complex mathematical models versus geoscience-based analogues for predicting radionuclide behavior in deep groundwater systems.

Our emphasis is on the used-fuel disposal problem in the United States and Canada (which together account for almost half of the global inventory; IAEA 2008), although much of our argument also applies to other countries seeking similar solutions.

The Promise of Radionuclide Containment over Geologic Time

The nuclear industry2 is unique in setting a management goal that extends over future geologic time to isolate its wastes from the biosphere. This is an exemplary moral position.

In the 1970s the time scale of this promise in Canada and the United States was more than 10,000 years, a duration likely selected because it represents time since the last continental glaciation and thus the longest time scale over which geosphere isolation predictions could be reasonably envisioned. The time scale of the promise has since grown: the nuclear industry in nearly all countries now proposes that isolation should exceed 100,000 years and preferably approach 1,000,000 years or more, equivalent to the time needed to reduce by decay the radioactivity in the used fuel to nonhazardous levels (Figure 1). This promise is based on the expectations that the used fuel will be contained for some time in metal and clay packing in the DGR and that the site-specific geology, hydrology, and geochemistry will ultimately provide containment over future geologic time.

Figure 1

We use the term promise rather than goal because of the way the industry presents the DGR concept to the public. No formal examination of its scientific foundations was conducted when the promise was first envisioned in the 1960s and, in our view, prediction of radionuclide transport and fate over geologic time was unrealistic based on what science then had to offer. More reliable, longer-term predictions are now possible but their period of high confidence is much shorter than the promise. This creates an impasse between government policies and science. The uncertainty inherent in longer-term (i.e., >10,000 years) predictions should not be used to discredit the role of science in today’s decisions. Rather, the main tradeoff under consideration by the nuclear industry is between (1) continuing to store used fuel near the surface, leaving future generations to deal with the associated risks and DGR costs, and (2) proceeding with monitored DGR storage and leaving future options (e.g., retrieval, reuse) open (Alley and Alley 2014).

Advantages and Disadvantages of Different Rock Types

The 1957 NRC report considered granite and basalt but ultimately determined that disposal in salt beds or domes was the most promising solution, because of their presumed dryness and lack of transport pathways, among other factors. Salt was the cornerstone of US waste disposal policy for two decades, and a salt DGR in New Mexico for intermediate military radioactive waste has operated successfully since 1999. But salt creep makes this mineral poorly suited for keeping the cavity open for monitoring and retrievability. Moreover, prospective salt areas may be jeopardized by unsealed holes from previous drilling, and geological sequences with salt may have resource value such as oil or gas and be subject to future drilling.

In the 1980s Congress limited the focus to tuff (a form of volcanic rock) at Yucca Mountain, Nevada, where the thick unsaturated zone was initially selected based on the advantages listed in Table 1. But after much study, the future of the Yucca Mountain site is unclear3 and the US quest for a used-fuel DGR has begun again. The following five geologies have been identified as offering favorable characteristics (e.g., Farvolden et al. 1985): granite, salt, shale/clay, shale cap, and tuff; their characteristics are summarized in Table 1. Besides salt, granite and shale have been explored most extensively. 

Table 1

Granite and Shale

As indicated in Table 1, there is no ideal geology, but some offer better prospects than others. Granite is excellent for mining and accommodating heat but, because it has uncertainties due to fractures (known or suspected) and weak attenuation mechanisms, it does not allow credible transport and fate predictions. It has therefore been abandoned by most countries that have a politically/socially acceptable alternative.

Shale beds, where fractures are closed or extremely small, offer the best possibilities for containment predictions over future geologic time and can withstand heat effects potentially better than salt, but shale exhibits more difficult properties for mine stability. Figure 2 shows a conceptual example of a DGR in low-permeability strata where groundwater flow has no influence on radionuclide migration.

Figure 2

Neuzil (2013) notes that, although other countries have selected shale, this option has been absent from the US repository program. He concludes that research in other countries has yielded a much better understanding of the isolation afforded by shales and that they may offer potential to host most of the world’s radioactive waste.

There are also possibilities for making use of more than one rock type in DGR design. Bredehoeft and Maini (1981) proposed a repository in granite (for shaft stability) situated beneath shale. Building on this, Russell and Gale (1982) drew attention to the favorable sedimentary geology beneath Canada’s main nuclear power production center, where there is much used fuel in dry surface storage at the Bruce Nuclear Power Development in Ontario, 200 km west of Toronto and near Lake Huron. Here, there is the potential for shale layers for containment and deeper layers for the repository vault. Although Canada did not begin to assess this option until 2005, it has advanced toward a decision to create a DGR for low- and intermediate-level radioactive waste (but not a used-fuel repository) at this site.

Canada initiated a selection process in 2010 to find a willing community for a used-fuel DGR; 21 communities situated on both granite and sedimentary rock, including some in the Bruce region, are undergoing the initial evaluation process.

In our view, any one of multiple geologic settings in good circumstances combined with appropriate engineered barriers may offer sufficient containment for the establishment of adequately secure DGRs. Even with the best engineered barriers, however, DGRs in each geology type face substantial uncertainty when the time frame extends too far into the future (e.g., approaches geologic time such as 10,000 years or more).

International Efforts

The Canadian effort to find a DGR site began in 1974 when Atomic Energy of Canada selected crystalline rock, represented primarily by granite, as its only geologic option, a choice that subsequently became government policy (Aikin et al. 1977). Although the primary reason appears to have been remoteness from population, this was not acknowledged as the main selection criterion and no science-based argument was given as to why granite should offer long-term radionuclide containment. Canada is now committed to granite and shale as equal options at the social acceptability stage of site evaluation, without distinguishing between reliance on the engineered barriers versus the rock in terms of DGR design concept.

About the time that Canada began its granite-focused DGR quest, Sweden, Finland, and the United Kingdom also selected granite. France identified it as the favored option, with sedimentary rock (shale) as a secondary possibility. Belgium, lacking crystalline rock options within its borders, pursued clay as its only option. Germany has had a long-standing focus on salt. After consideration of a wide range of geological options, China selected granite as its preferred option in the 1980s (Wang 2010).

In all of these countries the repositories are positioned below the water table and there is thus a risk that radionuclides could escape to the biosphere if there is active groundwater flow. The exception to groundwater-zone DGRs is the abandoned Yucca Mountain option, although this site also does not avoid potentially significant water movement over geologic time scales.

Most major nuclear power–producing countries (France, Switzerland, the United Kingdom) have selected argillaceous sedimentary rock (e.g., extremely low permeability shale) as their priority, provided it is within their borders and politically/socially acceptable.

China has found that the granite at its DGR focus area in the Gobi desert has fractures with water flow; this was not expected and may cause reconsideration of other options in China such as shale. Sweden and Finland concluded that containment predictions in granite cannot be made over long periods of time with sufficient confidence (Mazurek 2010); but because these countries have no geologic alternative they are constructing granite DGRs that rely on very expensive enhanced engineered barriers for the primary containment.

Siting and Transport Considerations

An important consideration is that a granite repository for Canada would require both long-distance transportation of waste from southern Ontario and enhanced containment of the transported waste by engineered barriers to achieve the same isolation prospects as sedimentary rock. These arrangements are expected to result in substantial additional DGR costs. This brings to the forefront the tradeoffs between (1) the political advantages of granite site remoteness and relative ease of public acceptance and (2) the long-standing promise for strong scientific evidence of containment over geologic time while communicating with transparency about uncertainties around the science and cost.

A further consideration is that used-fuel disposal efforts are based on the premise that each country that produces nuclear power will establish its own DGR. This is reinforced by legislation mandating no used-fuel transport across borders, despite much international cooperation in scientific and engineering aspects of nuclear power development. For countries with minimal nuclear power production (e.g., Mexico, Brazil, Czech Republic, South Africa), creation of a national repository does not make economic sense. Moreover, the best geological conditions for a DGR may be near an international border, as would be the case if the Canadian used-fuel DGR were proposed for the Bruce region near Lake Huron.

The Challenge of Predicting Long-Term Behavior

Oreskes (2004, p. 369) argues that “In all but the most trivial cases, science does not produce logically indisputable proof about the natural world. At best it produces a robust consensus based on the process of inquiry that allows for continued scrutiny, re-examination and revision.” We agree that claims to predict radionuclide containment over long times must be supported by strong scientific consensus, with ever growing evidence, if they are to be highly credible to the public. For these reasons, the selection of an approach to establish consensus is important (Saltelli and Funtowicz 2014). There also must be a commitment to long-term monitoring and reevaluation of DGR design and operation in recognition of the inherent uncertainties.

To fulfill the promise initiated decades ago, radionuclide migration to the human environment (i.e., the biosphere) must be predicted over geologic time. Several general approaches are used to predict such migration, of which we highlight two: performance assessment and the safety case.

Performance Assessment

Performance assessment (PA; or total system PA) is a complex structured mathematical approach with general features developed from risk assessments of nuclear power plants. It is based on systematic identification of all natural and anthropogenic features, events, and processes believed to have potential to contribute significantly to the failure of the repository (Long and Ewing 2004). A single analysis may thus require hundreds of component models with thousands of input parameters.

The mathematical models are founded on conceptual models, and when these are substantially wrong in a nonconservative direction all validity of the calculations is lost. For the geosphere, many model input parameters cannot be measured reliably and are therefore represented by uncertainty. However, much uncertainty around too many parameters results in loss of credibility. In short, the PA approach has maximum complexity but minimal transparency.

The Safety Case

The safety case relies on conceptual models to explain why a repository is thought to be safe (Long and Ewing 2004). It should be transparent and include multiple independent lines of reasoning based on evidence rather than relying on a multitude of calculations.

A groundwater system allows for the development of simple transparent models (along with more complex and less transparent models) and features natural analogues for system evaluation. Site-specific chemicals or isotopes in the pore water serve as analogues for radionuclide behavior and can provide strong lines of independent evidence of slow chemical migration over geologic time, where migration is diffusion controlled and free of the influence of active groundwater flow (e.g., Clark et al. 2013). Thus, based on safety case results, countries that initially selected granite switched to extremely low permeability shale primarily because active radionuclide migration due to groundwater flow in fractures is much less likely in shale (as mentioned above).

Mathematical modeling is still needed to interpret field data, but the models representing sedimentary rocks such as shales are relatively simple, based on a small number of significant processes, and documented migration of nonradionuclides in the geologic past is used as an analogue for radionuclide behavior near the repository over future geologic time. The approach involves quantitative, field-derived data, is mathematically elegant, and can be clearly explained thanks to the simplicity and small number of control parameters for which quantification is needed.

In short, the safety case for a DGR geosphere that can provide robust analogue information allows a substantial degree of validation of the model concept over past geologic time as opposed to untested or untestable complex calculations. Although use of the term validation is controversial among groundwater modellers, data validation is a well-established component of the scientific method. Model-derived data (output) can be evaluated with respect to precision and accuracy (uncertainty) in a similar manner to field and laboratory-derived data-sets, yielding a quantitative assessment of the numerical model performance and the conceptual site model it represents. Thus validation can demonstrate understanding of the hydrogeologic system and indicate the relative favorability of geosphere conditions with allowances for uncertainty.

However, predictions over millions of years of geologic time cannot realistically account for all catastrophic disruptions (e.g., glaciation, volcanism, earthquakes) or other major influences that are unprecedented and beyond reliable quantification. Furthermore, the repository could produce heat or gas generation effects that change the hydrogeologic system and render invalid long-term predictions based on analogues. This uncertainty can best be accommodated by PA monitoring and adaptive management strategies and designs that can accommodate new understanding and new technologies. 

Promises and Processes

In its DGR quest, the nuclear industry strategy was aimed at overcoming public fear and distrust concerning radioactivity. But three critical mistakes were made decades ago.

The first was to promise the public that radioactivity would be isolated over future geologic time. This promise lacked credibility then, when DGR geosphere science was in its infancy, and still defies common sense. A reasonable goal is for a repository to have sufficient containment capability to isolate radionuclides from the biosphere for several thousand years, and to achieve this capability through engineered barriers and the geosphere in whatever combination can be shown with strong scientific evidence to be reliable.

The second mistake was to frame the prediction problem around complex mathematical models founded more on theory than on empirical evidence. Although common in PAs used for nuclear reactor performance and safety, this approach is contrary to what is accepted in the geoscience disciplines, where nature has provided hundreds of millions of years of performance data to use in the form of analogues. The nuclear industry failed to recognize that the main scientific challenge in repository prediction required a geoscience approach based on site characterization with multiple lines of evidence founded on geology, hydrogeology, and geochemistry, with the interpretations of the geoscience evidence providing insights to the future. This requires application of the scientific method well established in these geoscience disciplines.

The third mistake was pursuit of DGRs without transparent decision making. Minimal transparency was an established (and necessary) part of nuclear science and engineering during the Cold War and this created a paternalistic approach, but even decades later the industry has yet to embrace the scrutiny of either the scientific community or the public.

In terms of decision-making processes, the DGR task has long been the domain of the US Department of Energy and, until 1998, Atomic Energy of Canada (in 2002 a new agency, the Nuclear Waste Management Organization, was made responsible for the Canadian DGR effort); the core business of both organizations in this context is development of nuclear power, not assessment of geoscience systems for waste disposal. A direct result of the isolation of decision making from mainstream science has been delays in recognizing both the limitations of granite and the favorable containment capabilities common to many types of sedimentary rock. For example, the value of site-specific analogues with migration controlled by diffusion rather than groundwater flow to evaluate near-surface entombment of low- and intermediate-level radioactive waste in low-permeability clayey deposits was well known by the mid-1980s (e.g., Gillham and Cherry 1983), but did not become part of the used-fuel DGR thinking until the late 1990s.

Concluding Remarks

It is essential to build a robust consensus in the scientific community about selection and assessment of repositories. But although such consensus is a prerequisite to public trust, it has been extremely limited to date. The public must trust that the process is reasonable and transparent and that the nuclear industry is committed to long-term responsible guardianship of its wastes. Moreover, the geographic location of a repository must satisfy public opinion and political considerations in addition to scientific criteria, and cost and science uncertainty issues should be vetted with transparency. To achieve the broadest credibility, the time frame promised for radionuclide containment must be shortened and the process intuitively reasonable, open, and adaptive to monitoring results subject to rigorous independent peer review.

If it is generally accepted that global climate change due to fossil fuel use is worth combating and that nuclear power is a proven large-scale alternative for baseload electricity, then there is a moral argument in favor of proceeding with DGRs as a lower-risk alternative to continued surface storage. And if the human species can persist for the next several thousand years without the used fuel causing harm, our generation and the next few will have discharged their responsibilities well.


We thank Richard Jackson, Mark Logsdon, and Don Lush for helpful comments on manuscript drafts, Gillian Binsted for editorial assistance, and Kristina Small for graphic design. The views expressed in this article do not represent the official policy or position of the organizations that the authors represent.


Aikin AM, Harrison JM, Hare FK. 1977. The Management of Canada’s Nuclear Wastes. Canadian Ministry of Energy, Mines and Resources, Report EP 77-6, August 31.

Alley WM, Alley R. 2013. Too Hot to Touch: The Problem of High-Level Nuclear Waste. New York: Cambridge University Press.

Alley WM, Alley R. 2014. The growing problem of stranded used nuclear fuel. Environmental Science and Technology, doi: 10.1021/es405114h.

Birkholzer J, Houseworth J, Tsang C-F. 2012. Geologic disposal of high-level radioactive waste: Status, key issues, and trends. Annual Review of Environment and Resources 37:79–106.

Bredehoeft JD, Maini T. 1981. Strategy for radioactive waste disposal in crystalline rocks. Science 213(4505):293–296.

Clark ID, Ai T, Jensen M, Kennell L, Mazurek M, Mohapatra R, Raven KG. 2013. Paleozoic-aged brine and authigenic helium preserved in an Ordovician shale aquiclude. Geology 41(9):951–954.

Farvolden RN, Pearson R, Davison CC. 1985. Hydrogeology in radioactive waste disposal. Hydrogeology in Service of Man: Memoirs of the 18th Congress of the International Association of Hydrogeologists, Cambridge.

Gillham RW, Cherry JA. 1983. Predictability of solute transport in diffusion-controlled hydrogeologic regimes. Proceedings of the Nuclear Regulatory Commission Symposium on Low-Level Waste Disposal: Facility Design, Construction, and Operating Practices, September 28–29, Washington.

Hedin A. 1997. Spent Nuclear Fuel: How Dangerous Is It? Svensk Kärnbränslehantering AB (SKB) TR-97-13. Stockholm: Swedish Nuclear Fuel and Waste Management.

IAEA [International Atomic Energy Agency]. 2008. Estimation of global inventories of radioactive waste and other radioactive materials, IAEA-TECDOC-1591. Vienna.

Kharecha PA, Hansen JE. 2013. Prevented mortality and greenhouse gas emissions from historical and projected nuclear power. Environmental Science and Technology 47:4889–4895.

Long JCS, Ewing RC. 2004. Yucca Mountain: Earth-sciences issues at a geologic repository for high-level nuclear waste. Annual Review of Earth and Planetary Sciences 32:363–401.

Mazurek M. 2010. Far-field process analysis and radionuclide transport modelling in geological repository systems. In: Geological Repository Systems for Safe Disposal of Spent Nuclear Fuels and Radioactive Waste. Ahn J, Apted MJ, eds. Woodhead Publishing Series in Energy 9. Cambridge, UK: Woodhead Publishing Ltd. pp. 222–257.

MIT [Massachusetts Institute of Technology]. 2003. The Future of Nuclear Power: An Interdisciplinary MIT Study. Cambridge, MA.

Neuzil CE. 2013. Can shale safely host US nuclear waste? Eos 94(30):261–268

NRC [National Research Council]. 1957. Report on Disposal of Radioactive Waste on Land. Washington: National Academy Press.

Oreskes N. 2004. Science and public policy: What’s proof got to do with it? Environmental Science and Policy 7:369–383.

Russell DJ, Gale JE. 1982. Radioactive waste disposal in the sedimentary rocks of southern Ontario. Geoscience Canada 9(4):200–207.

Saltelli A, Funtowicz S. 2014. When all models are wrong. Issues in Science and Technology (Winter):79–85.

Wang J. 2010. High-level radioactive waste disposal in China: Update 2010. Journal of Rock Mechanics and Geotechnical Engineering 2(1):1–11.


1 Disposal in deep (>2 km) boreholes is also being considered in the United States but is not addressed here.

2 We use the term nuclear industry here to refer collectively to the combination of government regulatory agencies, national laboratories (US Department of Energy and Atomic Energy of Canada Limited), and nuclear power producers, acknowledging that the federal governments control the direction and most of the public communications.

3 The Obama administration discontinued funding in 2011, but debates continue in Congress and the courts.


About the Author:John A. Cherry (NAE) is a distinguished emeritus professor, University of Waterloo, and director of the University Consortium for Field Focused Groundwater Research based at the University of Guelph, Canada. William M. Alley is director of science and technology with the National Ground Water Association and former chief, Office of Groundwater, US Geological Survey. Beth L. Parker is a professor in the School of Engineering at the University of Guelph, director of the G360 Institute for Applied Groundwater Research, and Canadian Natural Sciences and Engineering Research Council Industrial Research Chair in Fractured Rock Contaminant Hydrogeology.