Toward On-Site Closed Nuclear Cycles Not Requiring Deep Burial of Waste
- First Online:
- Cite this article as:
- Fowler, T.K. & Ahn, J. J Fusion Energ (2010) 29: 188. doi:10.1007/s10894-009-9256-4
We discuss a non-chemical means for onsite reprocessing of spent fuel from hybrid reactors such as LIFE and also deep burn fission reactors. Using a plasma-based Archimedes Filter of standard design, actinides could be removed in a few passes through the Filter to qualify as TRU waste that could be disposed of in a site like WIPP. An improved Filter is discussed that could reduce waste to 1 cubic meter per year, suitable for shallow burial.
KeywordsArchimedes filter Deep-burn reactors Nuclear fuel cycle Geologic disposal High-level wastes
This paper addresses the role of the nuclear option in meeting the California Energy Future goal of essentially zero carbon emission by 2050. A 1-GWe reactor consumes about 1 metric ton of fissile fuel per year, turning it into fission product waste. There is also nuclear waste when a plant is decommissioned, and secondary waste incidental to plant operations. Here we address what has been the main focus of public attention, the steady stream of high-level waste (HLW) represented today by about 20 metric tons/GWe/year of light-water reactor (LWR) spent fuel consisting of the fission products, Pu and minor actinides (Np, Am, and Cm), all mixed with a large quantity of unburned uranium. We will focus on advanced “deep burn” reactors, such as the traveling wave reactor (TWR) being developed by TeraPower or the far-reaching LIFE fusion/fission hybrid concept at LLNL . A neutron spectrum favorable to breeding in deep burn reactors allows a burnup of uranium much higher than the 5% typical of LWR’s. However, our waste disposal concept may also have near-term applications to existing LWR’s.
Suppose that a device co-located with a deep burn reactor could separate from the waste stream all actinides and selected fission products to be recycled and burned up inside the reactor, leaving about 1 ton/GWe/year of waste that would qualify as TRU waste, or even as Class C waste suitable for shallow burial. Just such a device is the Archimedes Filter initially aimed at removing actinides from waste at Hanford with considerable cost savings due to the reduced volume of HLW requiring deep geologic disposal. The Archimedes Demonstration Facility proved the concept at a throughput rate of 90 tons/year, more than needed, with 95% separation of heavy atoms.
Obstacles to similar ideas in the past, all involving chemical processing, have been cost, secondary waste, nuclear weapons proliferation concerns , and debates about what technical improvements would make any difference in repository regulations. The Archimedes Filter is a non-chemical approach with the potential for alleviating costs and reducing secondary waste, and as a “batch filter” that cannot separate pure plutonium it largely eliminates proliferation concerns. The desirable features of an Archimedes Filter are inherent in the concept, consisting of a Waste Injector and RF power to turn waste into a plasma flowing along magnetic field lines, followed by extraction of heavy ions by an electric field perpendicular to the magnetic field. More development work is needed on the Waste Injector, discussed below.
The regulatory issues require a sufficient change in technology to warrant a new appraisal of risks. While this is not entirely predictable in the context of the California Energy Future for 2050, we will use current NRC regulations as a guide to technological requirements, taking NRC Class C as a guide for shallow burial, and the TRU waste category that could be disposed of in a repository such as waste isolation pilot plant (WIPP). Our goal is to demonstrate the versatility of the Archimedes Filter to adapt to future guidelines (for example, stricter standards to take better account of environmental transport paths for isotopes such as 129I or 135Cs).
On-site processing by an Archimedes Filter may have particular advantages for deep burn reactors, as an alternative to the idea that a deep burn reactor core like that for the TWR would be sealed and shipped to a customer who after decades of use ships it back full of waste. Deep burn reactors are breeders with a fast neutron component—from fission in TWR, from fusion in LIFE—that can burn up residual actinides and selected fission products that the Archimedes Filter would recycle through the reactor. Both TWR and LIFE can burn many fuel forms: LWR spent fuel, weapons Pu, natural U, etc. They thrive on depleted uranium as fuel, thus enabling profitable use of accumulated junk that will itself pose environmental hazards over time.
Example waste stream (excluding unburned U)
Recyclable Pu (4% 241Pu)
Minor actinides (1.8 × 10−4% 242Cm)
Total Pu and minor actinides
Fission products, long-lived
Activated 14C, 59Ni
Fission products etc. 1/2 life >30 years
Short half-life, 90Sr(Y), 137Cs(137mBa)
3.4 × 106 Ci
All fission products, activated material
3.4 × 106 Ci
Class C accountable
0.06 Ci/gm of waste
6 × 105
1.0 × 10−4 Ci/gm
Also shown in Table 1 is the current NRC standard for actinides in Class-C low level waste intended for shallow burial, and the Dilution factor required to meet the standards. While the waste volume is only of order 0.2 m3, we see that dilution alone is not an option. However, at the demonstrated collection efficiency of 95% for the original Standard Archimedes Filter, a mere fivefold dilution to ≈1 m3 together with only four passes through the Filter—4 tons/year rate versus 90 tons/year demonstrated—would meet current NRC standards for TRU waste. Thus, recycling and burning actinides removed by the Archimedes Filter would yield waste in which only fission products would determine the waste category. Deposition codes developed by Archimedes Technology could confirm these conclusions.
Simply reducing actinides to the Class C standards required for TRU waste suitable for disposal at WIPP is probably a transformative change in waste management . We already have real experience with such disposal options, unlike deep geologic disposal for HLW. On the technology side, producing only TRU waste does not require further development of the Standard Archimedes Filter except to adapt the Waste Injector to the reactor waste composition. While injection of a surrogate material to simulate Hanford waste was remarkably successful in the Archimedes Demonstration Facility, at the time Archimedes Technology ceased operations in 2006 problems had appeared for certain applications so that the Waste Injector remains a development item.
Ref. 4 inventory (per bin)
Deep burn reactor inventory (kg)*
5 × 10−3
All activated mat’l
Fission products, short-lived
275 Ci/gm avg.
Fission products, long-lived
5 × 10−4 Ci/gm
1.8 × 10−4 Ci/gm
All Mat’l except actinides
As is typical , initially most of the fission product radioactivity in Table 2 is in the short-lived isotopes 90Sr (and very short-lived 90Y) and 137Cs (and very short-lived 137mBa). Strontium and cesium could perhaps be separated by a Modified Archimedes Filter adapted to extract several batches of masses around selected atomic masses. One might then be able to recycle Sr and Cs to be transmuted inside the reactor. Or we could use the Modified Archimedes Filter only to reduce the mass of material requiring a cool off time on site. The total mass involved is about 60 kg/year including non-radioactive isotopes of Sr and Cs that the Filter could not separate from 90Sr and 137Cs (or residual long-lived 135Cs).
Class C standards for fission products, etc
3 passes (m3)
The volume in Table 3 is the Inventory divided by the Standard, giving the dilution volume per year required to meet the Standard. We see that diluting the waste to 1 m3, found sufficient to handle residual actinides after recycling, would also handle all fission products and activated material, if we also recycle 94Nb, 99Tc and 126Sn to be burned up in a TWR or LIFE (and perhaps 129I and 135Cs which would not require further treatment under present regulations but might in the future: see “Appendix”). Per the table this requires three passes through a Modified Archimedes Filter with 95% collection efficiency, versus four passes for the actinides. Just three or four passes to handle all actinides and relevant fission products is an order of magnitude fewer passes than is required for competitive schemes . The actual efficiency may be greater than 95%, due to a lower plasma density, yielding even fewer passes.
We tentatively conclude that, for a Deep Burn Reactor with a modified Archimedes Filter, dilution of the waste in ≈1 m3 of packaging could reduce all waste to Class C. Just 1 m3 of Class C waste per GWe per year would be much cheaper to dispose of than the equivalent HLW. The activity disposed of, equaling 216 Ci/year/GWe, is no more than the Class C waste already being disposed of by California utilities (for 2004–2008, average Class C shipped to depositories per year by all California utilities combined = 350 cu.ft. = 10 m3 containing on average 700 Ci total ). Reducing the waste volume to 1 m3/year also might even open the option of on-site disposal of all of the waste, for some sites. With the possible exception of waste from on-site refabrication of recycled actinides and fission products, other waste disposal mainly concerns decommissioning. The non-chemical Archimedes Filter would be contaminated at decommissioning but should not produce a significant waste stream aside from secondary waste due to maintenance etc.
The required Modified Archimedes Filter might divide the 1 ton/year throughput into four or five mass batches: all actinides in one batch (U, Pu and minor actinides); Sr and Cs in other batches; Nb, Tc and Sn in others; and the bulk containing everything else. For example, an optimized batch containing the 29.4 kg/year of 99Tc from the Inventory of Table 2 might total 100 kg/year including non-radioactive isotopes for a filter capable of extracting a mass batch over the range 95 < A < 104. Preliminary neutronics runs for the LIFE neutron spectrum applied to this batch indicate an 84% burnup of 99Tc in 1 year despite a chain of transmutations that regenerate 99Tc . However, recycling 100 kg/year of fission products might consume too many neutrons, so that isotopic separation of 99Tc is probably required. The LIFE spectrum would burn up >99% of pure 99Tc in 1 year .
The combination of a Modified Archimedes Filter and Atomic Vapor Laser Isotopic Separation (AVLIS) might serve to separate 94Nb, 99Tc and 126Sn. The Modified Archimedes Filter would separate out perhaps 100 kg/year as feed to a small AVLIS with optics modified to separate isotopes 94Nb, 99Tc and 126Sn . While this AVLIS could perhaps be misused to separate Pu, the throughput capacity of an AVLIS designed for on-site waste processing is small enough not to pose much proliferation threat. Setting aside the difficulty of stealing the highly radioactive core material, the Deep Burn Reactor inventory contains on average <100 kg of Pu at any one time diluted by 900 kg of fission products, of which the AVLIS processor would only be capable of handling 10% (100 kg) in a full year of diverted use, yielding then at most 10% × 100 kg = 10 kg of Pu, comparable to the IAEA Significant Quantity of 8 kg. Thus, an acceptable system might be a modified multi-stage Archimedes Filter that separates actinides down to Class C by four passes through the Filter and which also separates a small quantity of selected fission products to be processed by the small AVLIS, thus allowing recycling of both residual actinide fuel and the worst fission product isotopes. Another isotopic separator developed at LLNL is the Solitron that might be less prone to proliferation risks if its throughput capacity could be increased sufficiently for the waste management application .
As noted above, an important development problem not fully solved when Archimedes Technology ceased operations is the Waste Injector. As in other plasma applications, it is actually the pre-existing plasma that ionizes new material injected into the plasma, the energy being supplied by the RF. This is a feasibility question, not a question of power requirements, since the RF power required to maintain the temperature against radiation (acceptable, at most a few MW) dwarfs the ionization and injection power. Any form of injection that adequately exposes new material to the plasma will do – for example, vapor used in the Archimedes Demo, or tiny “pellets” such as powder produced by pulverization of solids. Material injection into plasmas is a research topic of wide interest for all forms of plasma processing.
To estimate costs, we extrapolate the Archimedes Technology cost estimate for the Hanford application giving about $4 M/ton of throughput, or $0.1 M/ton amortized over 40 years. This includes the cost of the Waste Injector and the cost of RF power, mainly the power to sustain a highly radiative low temperature plasma, based on actual experience in the Archimedes Demo. This cost is to be compared with about $1.5 M/ton to dispose of HLW at a site like Yucca Mountain ($90 B total to store 63,000 tons) and similar costs in Sweden and Japan. The feasibility and cost of on-site fabrication of recycled material, for example as TRISO, requires more study, but we note that ≈100 kg/year throughput is comparable to TRISO facilities at Babcock and Wilcox . A best result might reduce all waste disposal to low-cost shallow burial.
The actual cost of the Archimedes Facility will depend on the number of passes required and the throughput of unburned uranium mixed with the fission products. This is only a reprocessing cost issue, not a disposal issue. The extreme example would be applying the scenario of Fig. 1 to an LWR, with 20 metric tons/year throughput. The actual number for deep burn reactors depends on design. Based on earlier TWR scenarios and LIFE scenarios midway through a 50 year burn, the upper limit is about 1 ton/year of unburned uranium (with recycling giving the opportunity to reclad neutron-damaged fuel assemblies), thus doubling the throughput. Scenarios optimized for waste management could probably reduce this considerably.
Finally, it may indeed make sense to apply the near-term scenario of Fig. 1 to on-site processing of LWR spent fuel. An important difference from deep burn breeders is how to maintain enrichment for the non-breeding LWR. (A TWR also requires enrichment, but only once, to initiate breeding: LIFE does not require enrichment.) One could perhaps recycle the unburned uranium and other actinides in LWR spent fuel on site, augmented by externally enriched fuel; or ship the actinides to an enrichment plant. As noted above, a higher throughput for the LWR increases the cost of the Archimedes Filter and fuel refabrication, though we note that several countries, using chemical methods, have already pursued development to recycle LWR fuel.
More study is needed to assess the technological and regulatory feasibility of on-site waste treatment for deep burn reactors using the Archimedes Filter, and its merits versus similar concepts such as the ALMR with Pyro-processing . If favorable, an on-site closed nuclear cycle could be developed in time to contribute to the goal of virtually eliminating carbon emissions in California by 2050. TeraPower estimates that it could market TWR’s in about 10 years. Early deployment of TWR’s with Archimedes Filters, followed by fusion/fission hybrids, could eliminate all coal plants supplying electricity to California. Indeed, all coal plants in the US could be eliminated by deep burn reactors fueled hundreds of years by existing depleted uranium as fuel, with a progressive improvement in nuclear power from advanced fission reactors to fusion-fission hybrids to pure fusion reactors. A clear path toward an environmentally friendly closed nuclear cycle could help revive work on the Archimedes Filter and motivate an invigorated advanced fission reactor effort, in parallel with impending attempts to achieve fusion ignition in NIF beginning in 2010, and extended fusion burn in ITER before 2020.
We thank John Gilleland at TeraPower, Joe Farmer and Jeff Latkowski at LLNL and our colleague Ehud Greenspan for many helpful discussions, data and ideas that contributed to this paper.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.