The Fukushima nuclear event of March 2011 dramatically revealed the potential risks of holding significant quantities of spent nuclear fuel at wet pools requiring continuous water circulation to maintain safe cooling. The housings for four spent fuel pools were badly damaged, and all pools lost cooling and nearly suffered fuel exposure. These conditions had the potential to result in catastrophic radiation release, rivalling or exceeding safety concerns over the nuclear reactors themselves. In contrast, the nine casks of spent fuel in dry storage at the Fukushima site hit by the same earthquake and tsunami experienced no material damage and posed no safety concerns.

It is unlikely that any US reactors face a comparable environmental threat, but due to the inability to implement a timely spent fuel disposal programme at Yucca Mountain, all of the commercial nuclear plants in the US have spent fuel pools that are filled with roughly five reactor cores of spent fuel, and most have also had to build on-site dry storage facilities (Independent Spent Fuel Storage Installations or ISFSIs) for handling fuel discharges in excess of pool capacities.

At the same time, the US government is financially-liable for paying nuclear plant owners for their costs of expanded on-site fuel storage facilities, to the extent that the maintenance or expansion of those facilities could have been avoided had a permanent repository for spent nuclear fuel, for example, the Yucca Mountain Nuclear Waste Repository, been built and in-service in 1998. These liability costs are accumulating on the order of $250 to $350 million per year (based on annual costs of $4 to $9 million for operation and maintenance of each facility only), as the US Department of Energy (DOE) continues to default on its obligations to remove spent fuel from reactor sites. In addition, litigating those liabilities have added substantially to that expense.

A better means of handling this spent fuel, with regard to both costs and safety, would be for the federal government to restart a spent fuel handling programme at one or a few centralized, interim dry storage facilities. This idea was recently endorsed in a January 2012 report by the Blue Ribbon Commission on America’s Nuclear Future, but no studies to date have assessed what size and pace of programme might address today’s needs.

A system of dry storage for spent fuel at one or a few large, federal facilities would offer a number of engineering and economic advantages over the current practice of holding spent fuel at individual reactor sites. The lack of a repository like Yucca Mountain has forced the industry to develop considerable expertise in dry storage cask design, fuel handling, and site monitoring. Building on this operational experience and strong safety record, a federal programme could circumvent some of the political and engineering obstacles that paralyzed the Yucca Mountain project. A DOE-run programme to transfer title of the spent fuel to the federal government would also address the government’s ongoing liability problems. Finally, centralized dry storage preserves longer-term policy and engineering optionality, by serving as interim storage for some future permanent repository or acting as recycling locations for a closed fuel-cycle industry of the future.

Reliance on centralized interim dry storage facilities would require a fuel removal programme to pick up and transport fuel from individual plant sites, similar to what was originally envisioned for deliveries to Yucca Mountain. Had Yucca Mountain been built on schedule, the DOE’s original spent fuel removal programme would have been timed and sized to pre-empt the need for at-reactor storage expansions. Since this did not occur, there is now a much different (and greater) backlog of spent fuel requiring a storage solution. The federal fuel removal programme needs to be redesigned in light of today’s waste inventories as well as heightened concerns about safety and the debate over renewed nuclear development. The appropriate design of such a programme depends on (at least) two related issues:

(1) Determining what size and pace the overall programme should have, and

(2) Deciding what fuel to pick up first.

Again, lessons from the failed Yucca Mountain programme can be brought to bear on these issues. The new programme would be best designed by first laying out specific goals—for instance, prioritizing full decommissioning of shut-down sites and/or setting fuel density targets for at-reactor storage pools—and then allowing nuclear owners to negotiate and exchange fuel pickup rights based on site-specific needs for fuel removal.

Situation today

Today, 55 out of 75 commercial reactor sites in the United States have licensed and built on-site dry storage facilities, or ISFSIs, to house a portion of their spent nuclear fuel from prior years of operation. Of the remaining 20 sites, nine are currently pursuing ISFSI licensing (so presumably have plans to build ISFSIs in the near future), while 11 have not yet announced plans to pursue ISFSI licenses. Figure 1 shows the locations of ISFSIs currently licensed. Note that four site-specific licences are for non-commercial reactor sites (DOE TMI Storage, DOE Idaho Spent Fuel Facility, Private Fuel Storage, and GE Morris) and four site-specific licences are at sites that also carry general licenses (Surry, Robinson, Oconee, and North Anna).

Figure 2 shows an illustration of the DOE’s original fuel programme’s likely fuel handling capacity in comparison to needs over the period from 1998 to 2042. The graph depicts the cumulative industry-wide spent fuel discharges in the colour-shaded areas, and the total removal capability of the intended DOE programme as the red line sloping steadily towards the upper right. At the planned start date of the programme in 1998, there would have already been about 40,000 metric tons of uranium (MTU), with about 4400 MTU of this stored at shut-down sites and a small amount (1600 MTU) in excess of existing pool storage capacities at the time (that is, already moved to plant-specific ISFSIs). The majority of the remaining spent fuel would have been (and was, in 1998) in the at-reactor fuel pools. Annual discharges from reactors to storage pools were increasing at around 2000 MTU per year (the slope of the top of the shaded area in the figure), but only a small portion of this (the blue area) would have been occurring at reactors facing pool capacity constraints. That is, most plants would have still had unused pool capacity, as shown in the grey shaded area.

ISFSIs typically cost about $40 million to build, around $0.8 to $1.0 million per cask, and then a few million dollars per year to monitor and maintain. Since 1998, nuclear owners have collectively spent about $3 billion building ISFSIs and cask systems at nearly every reactor site. Going forward, owners and utility ratepayers will continue to spend roughly $200 million per year on ISFSI operations and maintenance (O&M). As explained further in this paper, some of these ongoing costs could be avoided with a few centralized interim facilities if a new federal programme begins soon.

If there is a silver lining to these on-site solutions and increased costs borne by most plants, it is that the experience gained in developing private ISFSIs will help improve the efficiency and effectiveness of a new federal spent fuel programme. Several storage canister types and cask designs have been vetted and approved by the NRC. ISFSI development has also resolved a number of technical obstacles to spent fuel management, including accommodating non-standardized fuel such as failed fuel and Greater-Than-Class-C waste. So far, nuclear owners have transported more than 1400 casks from at-reactor pools to ISFSIs, each usually holding about 10 to 12 MTUs of spent fuel. In aggregate, this is equivalent to about 25% of the total US commercial spent nuclear fuel discharged to January 2011, according to the Nuclear Energy Institute. Since the mid-1990s (after it became clear that nuclear owners could not expect timely fuel removal under the DOE programme), ISFSIs have been built at a rate averaging three new facilities per year, as shown in Figure 3. This experience in the planning, engineering, and operational aspects of dry storage gives the industry a stepping-stone to larger-scale dry storage.

New programme: size and pace

Although Yucca Mountain was not developed, the funding for the project did occur and continues today, at roughly $800 million per year from US commercial nuclear plant operators, based on the Nuclear Waste Fund fee of 0.1¢ per kWh of generation. Access to these funds for a new federal fuel removal programme using centralized dry storage facilities would require legislative changes (which are addressed in detail in the 2012 Blue Ribbon Commission report), but with those changes the current funding mechanism would likely be sufficient to cover most or all of the costs of building and managing large-scale federal interim dry storage facilities. A 2009 study by the Electric Power Research Institute (EPRI–1018722) estimated the capital cost for a 60,000 MTU site (about 6000 casks) to be $757 million. A site this size would be able to handle all the spent fuel currently in at-reactor ISFSIs and all the additional spent fuel discharges through 2030. Two of these facilities would be large enough to hold the entire industry’s discharges from existing plants, including all fuel currently stored in wet pools and all future discharges through 2050. The capital cost for one of these sites could be covered with a single year’s collections under the current funding mechanism — meaning the new programme could be pursued without putting any new strain on federal budgets (no new taxes or borrowing).

Once built, there should be considerable operational cost savings from centralization compared to the ISFSI O&M costs being incurred at the numerous private sites. The 2009 EPRI assessment, as well as studies of possible large, private ISFSI sites, estimated the cost of annual steady-state operations to be $3.7 to $8.8 million per year. These annual operating costs are quite close to what are currently being incurred (on average) at each of the 55 private ISFSIs, so switching to a few federal facilities could save more than $200 million per year in at-reactor operating costs. The variable cost of each storage cask, including the canister and concrete overpack, would be about $1 million per cask, totaling about $600 million per year for a programme transporting 6,000 MTU per year. Variable transportation costs would be on the order of $28 million per year, using the EPRI assumption of $280,000 per rail shipment and 100 shipments per year (which would imply in this example six casks per shipment). Again, all of these annual operating costs could be funded with the assessments being collected already.

In terms of land use, dry storage poses no material or novel problems. For instance, the ISFSI at the shut-down 619 MW Connecticut Yankee plant carries the entire 28-year output of spent fuel discharges from that plant in 43 casks on a pad approximately the size of a hockey rink (about 200 feet by 100 feet, or about 465 square feet per cask). Scaling this up to about 10,000 casks needed to accommodate the entire industry’s total spent fuel discharges through 2030 would require 107 acres, equivalent to about 97 football fields — not a large facility or land requirement. Nuclear waste, though heavy, is surprisingly compact compared to the vastly larger waste streams of other energy sources. A single 1000 MW modern supercritical coal plant, for example, will burn approximately 2.5 to 3 million tons of bituminous coal per year, leaving about 10%, or 300,000 tons, of that as fly ash and bottom ash.

The framework for transportation from reactor sites to centralized dry storage would not be materially different from what had been envisioned for Yucca Mountain. Fuel would be loaded into transportation casks and delivered by rail and truck to the storage sites. If all the waste was moved by rail in 10 MTU casks (weighing about 60 tons in their overpack transportation casks), then moving 3000 MTU per year would entail moving just one 60 MTU trainload (a bit less than the size of two discharges from a typical operating pressurized water reactor or PWR) per week for 50 weeks per year. This would require about a six-car train, so it would impose a very minor logistical and scheduling burden on the United States’ rail infrastructure. Even if moving slowly, at 15 mph for an average trip distance of 1500 miles, transportation would take 100 hours, so only a few such trains would be needed to service the entire industry. Likewise, the fuel handling at each end (loading and unloading) could be done at a rate of a few days per cask, based on experience at decentralized ISFSIs.

While siting one or more federal dry storage facilities would no doubt entail some of the same public debate and protest as affected Yucca Mountain, there are a few locations in the United States that are already experienced in dealing with nuclear waste and might be compatible with expansion to handle spent fuel. One such location is the US DOE Waste Isolation Pilot Plant near Carlsbad, New Mexico. This site has been in use since 1999 for interring transuranic (TRU) wastes from nuclear research facilities in salt caverns. Handling of above-ground spent fuel canisters requires far less engineering complexity than is being applied to these TRU wastes. The surrounding region is a sparsely populated desert and little or no agricultural or alternative use, and the facility already has sophisticated security, waste handling, and monitoring capabilities in place (as well as a large, adjacent uranium enrichment plant).

The government’s failure to build a permanent repository ready for service by 1998 means that we now need a faster, larger programme in order to ‘catch up’ to the original timeline for fuel removal. As we have demonstrated, however, neither the physical size of the sites nor the logistical burden of transporting the waste should interfere with developing a new and larger programme. Indeed, the relevant scale (as shown below) would be about twice the size of the programme planned for Yucca Mountain—a size and pace that had already been considered to be achievable based on prior programme analyses. In some of its mid-1980s planning documents, the DOE considered a 6000 MTU per year acceptance rate, assuming deliveries to two repositories instead of one.

Whether a new DOE programme could work off the backlog of spent fuel stored on at-reactor ISFSIs in a reasonable timeframe depends primarily on the actual programme start date for any reasonable transportation rate. It is more important to start as soon as possible than to wait to conduct a larger programme. However, the goal of ‘avoiding additional at-reactor storage costs’—which shaped much of the original programme’s design—can no longer be a primary programme objective, because those costs, to a large degree, have already been sunk. Today’s programme requires re-defined objectives that recognize the industry’s current situation and concerns. Once these objectives are defined, the programme’s required transportation rate can then be re-evaluated and sized for compatibility.

Any new spent fuel removal programme should be sized and paced to support goals that:

  • Remove all fuel at decommissioned or decommissioning plant sites to allow full completion of decommissioning and restoration at those sites
  • Avoid major additional capital investments in at-reactor storage, particularly new ISFSI builds or ISFSI expansions
  • Reduce government liability for operating and maintenance costs at existing ISFSIs.

Priority for shut-down reactors

Prioritizing fuel removal at decommissioning plant sites is a relatively easy solution for solving a number of issues relatively quickly. There are currently 14 shutdown commercial sites holding about 3500 MTUs in total. This fuel could be transported to centralized dry storage in the first two to three years of the programme, assuming it quickly ramped up to a 3000 MTU per year capacity like the original programme plans often assumed. Thus, only the first few years of the new programme would have to be dedicated to this goal. Once that fuel was removed, these shut-down sites could be fully decommissioned back to greenfield conditions, thus addressing the broader policy question of how commercial reactors complete their service and retire. The United States litigation at these sites would also have a foreseeable end, once the DOE took title to the fuel. Thereafter, the programme capacity would be available to meet other programme goals until about 2035, when the next major wave of decommissioning is likely to occur.

Avoid ISFSI expansion

Depending on when the new programme starts, there could still be opportunities to avoid some at-reactor capital expenses associated with building new ISFSIs (for the 11 plants that have not yet announced plans to build an ISFSI) or to avoid expanding ISFSIs at existing facilities. A programme to start removing fuel by 2020 would be timely enough to avoid these new ISFSIs, but it would need to be in place soon, before plant owners must begin planning and building their own dry storage.

Because there are only a few potential new ISFSIs, it may also be possible to accommodate both this goal and shut-down priority with a programme barely faster than the original programme’s 3000-MTU-per-year pace.

Figure 4 shows fuel for the 14 currently decommissioning reactors, as well as future decommissioning reactors (shown in green areas). The small amount of fuel volume shown in dark blue shows additional fuel storage needs at sites currently without an ISFSI, representing the need for a few new ISFSI builds after 2020 absent a fuel pickup programme. The fuel volume shown in medium blue labeled ‘ISFSI expansion’ reflects a more extreme version of the ‘avoid ISFSI expansions’ policy since it shows all fuel that will need to be packed and loaded onto existing ISFSIs in the future from 2020 and beyond, including loadings of additional casks onto ISFSIs that may not need platform expansion. (For comparison, the blue area in Figure 2 corresponds to the three blue areas combined in Figure 4.) With a start date of 2020 and the DOE’s original transportation rate, both programme goals of ‘priority for shut-down facilities’ and ‘avoid new ISFSI builds and expansions’ would be mostly achievable, as seen by the fact that the dark, upward sloping line for cumulative programme removals (beginning in 2020), stays roughly on pace with the sum of spent fuel at shut-down facilities plus incremental discharges needing ISFSI storage.

Reduce O&M costs

There are significant economies of scale to be enjoyed in dry storage O&M expenses, and the DOE could reduce total industry-wide O&M by an order of magnitude or more—eventually saving around $250 to $350 million per year—with centralized facilities. However, to fully achieve this goal the DOE would have to both remove all fuel at existing ISFSIs and avoid all new ISFSI builds and expansions. This goal is not achievable without a transportation rate higher than the original programme planned. Figure 5 shows the total volume of fuel requiring dry storage in the blue areas: dark blue is fuel already stored on ISFSIs, while light blue is the incremental quantity added between 2012 and 2020. A new programme starting in 2020 with an annual pick-up rate of 6000 MTU per year for the first ten years (through 2029) could allow the programme to prevent ISFSI expansions and clear fuel from existing ISFSIs, albeit with some delay as the programme also works off a 2020 backlog of 36,000 MTU of on-site stored fuel in ISFSIs. After the first ten years of removing spent fuel at 6000 MTU per year, the programme could be scaled back to 3000 MTU per year and still keep up with new discharges and fuel removal at future decommissioning sites.

We have described above how the programme timing and capacity would need to be sized somewhat differently if various goals and priorities were made pre-eminent. The key finding is that a new programme of 6000 MTU per year for ten years, then 3000 MTU thereafter, could accommodate most economic opportunities to avoid ongoing at-reactor costs and expansions within a decade. Of course, for the first several years of this new programme, not all backlogged needs could be met.

However, it should not be necessary for the government to choose who uses the programme capacity, or to decide for which of those purposes it would be used. Instead, the programme could allocate initial removal rights much like the Standard Contract (now in default) was going to allow, for example, based on the schedule of past discharges from the reactors to the wet pools. Then, the fuel owners should be allowed to exchange their initial queue position rights (each of which is for a share of the programme’s capacity, bestowed in proportion to each participant’s own spent fuel quantities) with each other, via swaps or purchases and sales of rights between themselves and across years when removal is more important to one party than another. Those owners facing a more costly constraint (such as having to build an ISFSI) should be willing to pay more than a party simply interested in drawing down its existing ISFSI, while a shut-down facility capable of decommissioning might be in between. This kind of exchange of economic services and resources is already widely practiced in the industry for a variety of needs, so it would be familiar, easy, and efficient for the participants and the programme as a whole to adopt that practice for spent fuel removal prioritization…

Editor’s note: The paper goes on to discuss the implications of de-densifying spent fuel storage pools, a discussion that has been excerpted because of lack of space. The argument can be read in the full paper, which can be downloaded from

Start date

All of the above assessments of desirable programme size have an assumed start date of 2020. This is arguably ambitious from a political perspective, but it should be quite feasible from an engineering perspective. If the programme were delayed another decade (for example, due to disagreement over new priorities, revised initial allocations, or political agendas), there would be significant adverse consequences for programme economics and for public concerns about safe fuel handling and nuclear viability. As demonstrated in Figure 6, with a 6000 MTU per year programme starting in 2020, the 63 at-reactor ISFSIs then in existence would decline to zero by 2029, and wet pools could be de-densified by about 10%. In sharp contrast, the same programme starting in 2030 would have 66 at-reactor ISFSIs to contend with by 2035, and there would still be 43 of them in operation in 2040. Virtually all wet pools would be full through 2040, unless de-densification was made a priority—in which case, more at-reactor ISFSIs would remain.

The cumulative annual incremental costs of starting later and maintaining more ISFSIs for longer would be about $4 billion through 2050. Even discounting these costs at a rate of 6% and recognizing the benefit of reduced present value programme costs from deferring the expenditures to build the centralized facilities, there would be a net cost of delay of around $1.6 billion to the industry. Perhaps more importantly, this delay in expenditures would further erode public confidence in federal or industry ability to devise durable solutions to foster safer and more economical nuclear power. Opponents of nuclear power may consider this an indirect victory, but they would be compromising their own goals of better waste handling.

This article was published in the October 2012 issue of Nuclear Engineering International

Author Info:

The Brattle Group provides consulting services and expert testimony in economics, finance, and regulation to corporations, law firms, and public agencies.

“Centralized Dry Storage of Nuclear Fuel: Lessons for U.S. Policy from Industry Experience and Fukushima” (with references) is available on


The authors would like to thank Dean Murphy, Brad Fagg, Jerry Stouck, and Robert Shapiro for helpful suggestions in drafting this paper. Any errors are solely the responsibility of the authors. The opinions and conclusions in this report are also solely those of the authors, not of The Brattle Group, Inc. or its employees.