Tuesday, January 15, 2013

To reprocess or dispose? A look at fuel cycle triage

A recent study by former colleagues of mine from Oak Ridge National Laboratory raises some interesting questions about the future direction of U.S. nuclear fuel cycle. My colleagues have been presently engaged in a scientific triage study for used nuclear fuel disposition options. One of the largest parts of their work has simply been in collecting the massive amount of data on the 67,600 metric tons (1 MT = 1000 kg) of commercial used nuclear fuel in the U.S., including issues such as how long it was burned in the reactor, the fuel type, and the initial enrichment, with an objective of being able to accurately characterize the composition and location of every used nuclear fuel assembly presently in the U.S. (I also am tangentially involved in this work, funding an undergraduate for data collection and am hoping to expand my role into doing modeling work in support of this effort).

The overall goal of this work is to support a more informed decision framework to specifically look at how we deal with spent fuel inventories in the U.S. - in other words, performing a triage analysis on what fuel would be the best candidates for various fuel cycle options (including direct disposal versus recycling). Given that some fuel is inherently going to be less suitable (read: more expensive) for recovering actinides as future fuel material, the goal is to sort out what can be disposed of immediately and what might be preserved for future fuel cycles.

Their (surprising) finding was that of the present inventory, 98% of the current used fuel inventory (by mass) could be disposed of without leaving open the option of future retrieval while still allowing for the ability to facilitate a future closed fuel cycle in the U.S. This conclusion was based upon the assumption that the U.S. would eventually open a fuel reprocessing facility; even under this assumption most of the present inventory of used nuclear fuel is not needed to support such a cycle. Some of this is simply due to the large inventory of used nuclear fuel in the U.S. - at nearly 68,000 metric tons of heavy metal with the largest fuel reprocessing centers having a throughput on the order of 1,000-1,500 MTHM per year, there is simply more "legacy" fuel out there than a typical facility would ever usefully process.

Their decision analysis was based on several factors, including the value of material which would be recovered (older fuel tends to have less plutonium available for recovery, and the plutonium is of lower quality); complexity (older fuel has other complicating factors such as different types of cladding material - like stainless steel - which can complicate potential recovery and thus make it less preferable to newer fuel), and simply the amount of material needed to sustain a closed fuel cycle (given the time before such a facility would come online, it is anticipated more than sufficient inventories would be present to sustain a closed fuel cycle without drawing into older fuel). Likewise, they considered what fuel assemblies might be useful to future reprocessing research efforts by DOE (such as used, highly-enriched fuel from naval and research programs).

To many who advocate exploiting the resource potential of used nuclear fuel (myself included), this is a jarring conclusion. There has always been a tacit assumption in mind that domestic reprocessing would not only include future inventories of used nuclear fuel, but help to alleviate the pressure on current demand for geologic repository space by making use of the readily available inventories out there. Yet beyond looking at what is economically practical (i.e., prioritizing the most valuable fuel for recovery), the report brings in an eye-opening reality - given the fact that the U.S. has spent the last thirty years committed to a once-through fuel cycle track, there is simply more used fuel than a single modern reprocessing facility would have capacity to handle, especially given the stable influx of fuel coming out of future reactors which would form the foundation for a future closed fuel cycle. As a result, much of this "legacy" fuel becomes unnecessary to support such future fuel cycles.

A more important implication relates to geologic disposal itself. The plans for the (now likely former) Yucca Mountain site called for a 50-75 year "retrievability" window; in other words, the repository was to be operated for an extended period which would allow for retrieval of used fuel out of the repository for other uses. (After the retrieval period, it was generally assumed if no use case had emerged by this point, permanently closing the repository was the most reasonable option).

Designing a repository with future retrievability in mind doesn't come for free; it essentially adds another engineering constraint (read: cost) to the problem and ultimately requires further analysis of how the repository will perform in containing waste in addition to the "post-closure" period. (It also tends to bias one's choice of geology - a feature of salt-based repositories like WIPP is that they are explicitly not designed to be retrievable - the heat from nuclear waste packages generally causes salt to plastically deform around waste packages, effectively "sealing them in.")

Thus, figuring out what spent fuel has little potential prospect for future recovery represents an technical triage which can help simplify a future repository design (as well as open up options for where such repositories might be located). In essence, separating out the "wheat" (fuel more useful for recovery) from the "chaff" (fuel which has limited potential for recovery) allows for a more intelligent approach to used fuel disposition which can ultimately make constructing a future permanent geologic repository cheaper and easier.

Of course, the standard caveat applies: the hardest part of opening any geologic repository has never been technical so much as it has political. Nonetheless, the ORNL report offers a rather bracing conclusion as to what a future U.S. fuel cycle may look like, even if the decision is made to restart reprocessing in the U.S. Ultimately, the vast majority of the current inventory of used nuclear fuel may yet still be destined for direct disposal, simply due to the realities of waiting over three decades before finally deciding to reconsider our rather ill-fated national decision to abandon a closed nuclear fuel cycle.


  1. This is fascinating. Thanks for sharing.

  2. But this is based on MOX and current thermal reactors right? It's always seemed strange to me that people would classify MOX as recycling at all, since it doesn't really do anything meaningful about the waste and muddies the water for real recycling where you would use up everything but the fission products.

    1. The assumptions made here aren't necessarily confined to a thermal-MOX scenario; this is simply an examination of fuel assemblies needed to sustain separations research and how much fuel would be required as initial feedstock to supply an industrial-scale separations facility. I agree that there's the issue of how much thermal MOX really "buys" one in terms of destroying long-lived actinides or in reducing repository burden; this is why I think the DOE term "modified open cycle" is probably more appropriate to thermal MOX, in that it represents auxiliary energy recovery more than it does a full recycle / closure of the cycle.

    2. This research seems to have been conducted only from the point of view of research, not a commercial operation. Why would anyone assume the start up of an entirely new industrial process that limited itself to a single production facility? A more important question in my mind is how valuable could the material be if used to its highest and best use, instead of assuming it is all waste to be disposed of as cheaply and as quickly as possible.

      I hope that your colleagues release details of the material composition so that people like my friend NNadir can perform analysis using a completely different paradigm.

    3. @Rod: Let's be honest here - at ~$20 billion a pop, chances are pretty good we're going to only get one reprocessing facility to start with. Unless capital costs for reprocessing come down significantly (say, with pyroprocessing or another variant), it doesn't seem likely we'd start off with more than one industrial-scale facility, especially if we were to go the conventional PUREX route. So, that assumption seems reasonable to me.

      I will say that having followed this research for about the last year or so from its start, I was a bit disappointed to see the rather sweeping conclusion based mostly upon an input/output ratio, when there was considerable work being done (and still is being done) on isotopic characterization.

      To answer your question, have a look at Figures 16 (Pu-239 equivalence for fuel assuming thermal recycle), Figure 17 (Pu-239 equivalence, fast recycle). Now advance the clock 20 years.

      A relevant consideration in my mind would be, "What fuel will be *worth* going after? As you see from the graphs, used fuel is a lot like uranium deposits - not all of it is of high enough grade to be worth initially going after - we'd want to hit the high-grade stuff first. Which in this case appears to be high-burnup PWR fuel. Considering now we'd have 20 years of the highest-grade material, the assumption that much of this (although I'm not sure I agree that it would be almost all of it) becomes a little more plausible.

      In any case, much of the data is calculated from the RW-859 used fuel database; a lot of the work going forward has been in collecting irradiation history data from the utilities to fill in the gaps.

  3. This study doesn't make much sense to me. Sure, if you assume we will continue to generate >2000 MT of used fuel/year, and we only build a single recycling plant that has a capacity of <2000 MT / year, of course you don't need any starting inventory to keep the recycling plant going. But if we were going to recycle, why would we stop with one plant? Presumably we would be recycling for a reason, probably because the price of mining new uranium was higher than the cost of recyling old uranium. In that case, we would want enough plants to recycle as much uranium as we need to fuel our reactors each year.

    I can sum up the report as follows: if we keep mostly doing the same thing we've been doing, then we will keep doing the same thing we've been doing. However, we can guarantee that life in 2033 will not be same as life in 2013. That probably means that the nuclear power industry will be substantially different (SMRs, fast reactors, recycling plants, etc.), so these simplified assumptions will not apply.

    1. @Damon: I think you're missing a key part of the issue here however, which the inherent delay going forward before any reprocessing facility would come online. The ORNL study assumes a latency of 20 years - which means about another 40,000 MTHM of feedstock to draw upon. Second, reactive worth is going to decrease with time, due to the decay of Pu-241 (i.e., fissile worth is determined by Pu-239 + Pu-241). Your most "valuable" fuel from a reactivity standpoint will be the newest ejected fuel.

      Basically, given the assumptions, you're going to have a large feedstock to draw upon of fuel which overall preferable and reach steady-state relatively quickly in terms of facility input.

      The question then comes down to whether drawing "deeper" into the past is worthwhile. You raise an interesting point about economics, but there are potentially other non-economic reasons why reprocessing may be considered as well (such as reducing repository burden - which is somewhat debatable under a thermal MOX scenario, but it is another relevant consideration). But I think one of the main take-away points to consider is that given the likely reprocessing scenarios envisioned (i.e., a 2000 MTHM/year facility would still be the largest in the world), much of the current SNF inventory is simply not needed for a future program, particularly given that future fuel discharges will serve this need.

      I'll admit that I was a little disappointed to see that there wasn't a very deep consideration of looking at broader categorization of fuel for potential recycle; essentially the report separates out a small minority of fuel for future study and, based upon the assumptions of a future reprocessing timeframe, consigns the rest to the repository, rather than looking at the the inventories to establish a "cut" point based on the projected fissile worth. However, arguably given the data such an analysis could be done, particularly in light of say, an alternative fuel cycle scenario (i.e., more facilities, or fuel cycle projections based on future uranium prices). In this sense, the work represents a potential starting point for other questions.

    2. Interesting that elimination of Pu and its radiotoxicity isn't listed as a consideration for reprocessing.

    3. While it's true that Pu-241 will decay and reduce the recycling value of older used fuel, it's also true that older used fuel has a much lower burnup. This means it has less Pu, but more of the Pu is the best kind - Pu-239. Existing reactors were built in the days when we thought reprocessing/MOX would happen (like it did overseas), so enrichments and burnups were low. With recycling, there is little incentive to achieve high burnups. Some of the oldest fuel probably also has the highest remaining energy.

      However, it is also likely that we would not be able to use LWRs to effectively burn the recycled fuel. I agree it is likely we will not be recycling on a large scale within 20 years. We probably won't have to do large scale recycling for 50-100 years, given the large amounts of uranium that have been discovered in recent years. I believe centralized dry storage will be effective for this time period, so we have time to do research on advanced recycling technologies.

      I agree the intent of the ORNL report was to justify use of a place like WIPP as a repository. However, I would argue that even in a salt deposit, the canisters will still be retrievable. If we were technologically able to dig the hole in the first place, we should still be able to retrieve the canister later. It may not be easy or cheap, but it is still retrievable.

      Ultimately, I believe the nuclear industry wants the used fuel question to be "solved". Up until now, we have been working towards a repository as the "solution". However, used fuel is very stable and easily managed. As we all know, the problems are mainly political in nature. Politics are really only effective at dealing with imminent problems, not long-term issues. Used nuclear fuel will never be an imminent issue, so I don't expect political leaders to ever "solve" it. The root cause of our problems is the agreement in 1982 for utilities to hand the used fuel over to the federal government for disposal (in return for the $1/MW-hr high level waste tax). The utilities generate the used fuel, and we should be dealing with it ourselves.

    4. @Damon: Have a look at Figures 16 & 17, which gives Pu-239 equivalence for thermal and fast recycle, respecively. While you're correct that the Pu-239 fraction of low-burnup fuel is higher, this mostly is true for BWR fuel. For PWR fuel, the total Pu content ends up outweighing the lower Pu "grade", especially for higher-burnup, more recently discharged fuel. The report also looks at the neutronics of fast-spectrum recycle (Figure 17) - the late-discharged, high-burnup PWR fuel is going to be *much* more useful than low-burnup BWR fuel, or even some of the higher-burnup fuel.

      I will put the caveat that they did this in kg Pu-239 equivalent per assembly, which has the obvious problem that PWR assemblies tend to be bigger than BWR assemblies. A good question then would be to normalize Pu equivalent per unit MTHM, which would more closely tie to the fuel "worth" versus amount processed.

      I think the reason the industry wants the used fuel issue "solved" related largely to public perception; as much as you or I might view medium-term dry storage as a workable technical solution until that fuel could be processed, most of the public does not. Even then though, I think we'd still bump up against a need for triage - some fuel is simply not worth recovering and would be simply easier to dispose of.

      As far as WIPP goes, I agree that it's technically retrievable in the strictest sense, but that's not really how I'd say the definition has been used with respect to geologic repositories. The point of retrievability is being able to directly pull the package back out with minimal effort - which again is a bit of a foolish notion if we apply it universally to *every* assembly, regardless of which ones may have little future economic value.

    5. I did look at the figures you mentioned. You also have to look at the Pu-238 and Pu-240, which are undesirable isotopes and make the reprocessing facility more difficult to build (due to the high decay heat load and dose rate). Pu-241 also puts out a lot of dose, though it fissions nicely. MOX fuel assemblies built using high burnup fuel would have higher dose rates due to these isotopes, making it more difficult to ship and receive them. These problems can be solved with time, distance, and shielding, of course. But this is why reactors that use MOX recycling do not burn their fuel to high burnups like we do in the US. Existing reactor designs work best with low-burnup MOX. High burnup MOX (as well as 2nd-time recycled MOX) is more challenging and would probably require more design modifications to existing reactors. I understand the AP1000 and ABWR are designed to be able to handle more MOX fuel.

      As to retrievability, even the Yucca Mountain was not very retrievable. The packages were to have titanium drip shields and be buried in bentonite clay. Each storage drift would only have one cask that was easily retrievable (the one in front). The criteria even say that retrievability does not require it to be easy or cheap. I suppose the concern at WIPP is that we probably have not put any large, hot canisters in the salt mine yet, so we might not be sure exactly what will happen if we put a 30-40 kW canister in there.

    6. @Damon: Valid points you bring up re: heat-generating Pu isotopes. This is perhaps a limitation of the ORNL study, in that it's looking mostly at fissile worth and not estimating relative cost of processing. It would be interesting if one could create a cost model to account for these factors and perhaps apply an optimization study based on the collected isotopic inventories within the report.

      As far as Yucca goes, I was under the impression that the bentonite would be put in only following closure of the repository (although I could be mistaken); although the titanium drip shield certainly was a bit of gold-plated silliness brought on in part due to the "retrievability" standard. It seems if one is going to maintain "retrievability," dry cask storage is indisputably the best solution - so why not again figure out what is likely to be useful in the future, store that in casks at the repository site, and bury the rest? It makes for a much cheaper "retrievability" standard.

    7. I'm sure Areva and EdF could provide a good cost model for PUREX reprocessing, which would be a good starting point for DOE.

      The bentonite was planned to cover each canister as it was loaded. The titanium drip shield was obviously ill-advised (it would have required a large fraction of the world's Ti production capacity. Both of these design features were added when EPA/NRC changed from a 10,000 year lifetime to a 1,000,000 year lifetime. DOE didn't think the fancy nickel alloy canister would last a million years if water dripped on it, so they added the drip shield to keep the water off the can. But the drip shield could corrode away too fast, so the bentonite was added. Clay is not very water-permeable, so maybe a little pyramid of clay would keep the water off the can. Engineers can always come up with new ideas when faced with ridiculous design constraints. The used fuel will be less radioactive that uranium ore in 8000 years or so, which was the point of the 10,000 year limit. A million year design life is patently absurd.

    8. @Damon:  The considerations of heat load and dose rate don't apply more than marginally to pyroprocessing, do they?

  4. The recycle decision is based upon the future of nuclear power in USA. With little future, used fuel is only a liability that should be sent to the deep geologic repository. WIPP can accommodate this material without delay, so get started and bury that shit.
    When/if the Global Nuclear Energy Partnership (GNEP) model is needed for delivery of new and recovery of used fuel, the USA can rest assured that the mines, the factories, the supply chains, the technologies, and the workers can be outsourced.
    Food comes from the grocery store, electricity comes from the wall socket and money comes from entitlements. Productivity and technology, why?

  5. Lord, I wish someone would get some form of fusion going, so we wouldn't be building up thousands of tons of radioactive crap to store somewhere every year for hundreds or thousands of years. I think this whole discussion is mildly mad.