Thursday, April 5, 2012

Bow Before The Mighty Thor!... ium.

Recently, a friend of mine posted a video which featured a series of clips of Kirk Sorenson selling his LFTR (Liquid Fluoride Thorium Reactor) concept to the masses.  I had made some fraction of a response based on what I'd seen in the video at the time.  Nonetheless, I felt it merited further study, so I looked a little deeper into the literature and now have the need to write a little more on the subject.

I do have to admit that a lot of what I hear Sorenson say sounded more like shpiel appealing to the masses rather than hard science, but that's not to say he doesn't know the science.  In fact, I'm fairly sure he does.  But since the video clips came from public speeches, he's playing to a crowd who really don't know much about the subject.  He focuses a great deal on the shortcomings of the current technology and a lot on the strengths of the LFTR design in terms of efficiency and safety and so on.  The numbers he quotes sound incredibly exaggerated on the face of it, but in fact, they are mostly accurate, if a little misleading.

In reference to this subject, though, I suggest that people watch both Kirk Sorenson's TEDxYYC talk on the LFTR and Bill Gates' TED talk titled "Innovating to Zero," during which he evangelizes the TWR (Traveling Wave Reactor) design.  I don't particularly favor one idea over the other, but I think having the combo of those two videos at the very least helps put some things into better perspective.

One of the figures that sounds really insane from the Sorenson video is where he says that current nuclear reactor tech depending on whether it uses light or heavy water coolant,  uses anywhere from 0.5% to 0.7% of the available energy of the fuel.  Well, this is technically true, but it is misleading.  What he leaves out is that even using 0.7% of the available energy of the fuel means that the top-end heavy water reactor technology exhibits over 90% thermodynamic efficiency, which is not "terrible" as he puts it.  Even America's "light" water reactors apparently reach over 65% efficiency.  No matter how you want to talk about it, that is significantly better than the very best that can be offered by coal and petroleum burning generation systems.  There are further developments in other countries which really raise the levels of efficiency and safety, but when talking to an American crowd, it's actually fair to confine the stats to the comparatively inferior technology that operating U.S. reactors have simply because of the fact that the U.S. hasn't invested in nuclear energy  development since 3-Mile Island.  I'd be shocked if even 5% of Americans realize today that 3-Mile Island was actually a successful execution of safety systems that prevented any danger.

The real problem isn't thermodynamic efficiency, and there's no way that switching to thorium or any other fuel will change that by itself.  The issue is how much of the fuel is actually being used.  This is something that Bill Gates touched on in his mention of the TWR.  It's not that we're getting poor efficiency -- it's that we're only using less than 1% of the actual fuel.  Out of all the uranium mined, somewhat less than 0.8% of it is U-235, and the rest is fertile, but not fissile U-238.  Even after using the enriched uranium which has a much higher percentage of U-235, we're left with tons of depleted uranium which mostly never gets used in actual fuel.  Most of it ends up in storage, and that which doesn't typically becomes ammunition.  Any type of breeder reactor, whether it's running on thorium or uranium is going to be able to address that side of the issue, so in that sense, thorium isn't all that special.

Thorium is indeed more abundant, and that also means it should be cheaper.  Currently, it's not because it's a specialty metal and demand is low, so there isn't a lot produced in any usable form.  Enriched uranium oxide runs somewhere around $1600 per kg, while usable thorium fuel runs about $5000 per kg.  That's right now.  In a hypothetical thorium economy, it's quite reasonable to presume that thorium will come down very very far in price as the demand goes up and countries that hold hefty thorium reserves will actually start to open up their wares.  Uranium is anyway about 600x more plentiful than gold, and thorium is about 4x more abundant than that.  Nonetheless, I would point out that cost of fuel is not a huge factor here.  Currently, a 1 GW-throughput nuclear plant would be expected to spend about $30-$35 million per year on fuel, which is not bad considering that we're not even burning 1% of the uranium that's actually mined.  However, if you move to a breeder reactor type that actually creates and burns fissile material out of the U-238, that brings their annual fuel costs down below $1 million.  And that's assuming 1 cycle.  It goes down further still when you get into recycling and re-burning the waste.  Yeah, thorium will theoretically bring this even further down as well, but now you're dealing with small enough sums of money that the point becomes more or less moot.

Kirk Sorenson's primary criticism of the TWR design that Terrapower has proposed is that it's just really difficult and expensive to build and demands a huge array of backup safety systems.  There are, however, people who level the same criticism back at the LFTR design.  Much of the design layout and safety systems he speaks of for a LFTR are actually quite true, and in fact, they are demonstrated to work.  However, they've only been shown to work in very small-scale test reactors (ones that operate in the 5 MW range).  To go up by a factor of 200-300x the output (yes, we'll need that), you've got to scale up the design, and passive safety systems like a salt plug DO NOT scale up so easily.  Sure, he talks about a modular design, but that's really not going to be cost-effective in the very large scales of output.  This is probably why Sorenson's company Flibe Energy (Flibe, I'm guessing, comes from Fluoride-Lithium-Beyllium coolant) is only looking at military contracts where the goal is to build small localized reactors very cheaply.  This is fine for powering a military base or something, but to provide power for the nationwide grid, you've got to scale up the design to high output.  TWRs, at least, have very detailed simulations to show that the design works at the very large scale -- around the 2 GW range.  The level of these simulations also elucidate all the engineering challenges that they'll have to overcome and so on.  However, all they have is simulations...  but since the TWR is a particular variety of FBR (Fast Breeder Reactor), it's not as if they're working from a starting point that hasn't been tried and tested in large scale applications.

The very idea of a breeder reactor is not really new.  It was at least theorized back when Fermi managed the very first sustained nuclear chain reaction at CP-1 at the University of Chicago.  Thorium itself necessarily must operate in a transmuting environment because thorium is not a fissile material.  Th-232, however, is fertile for the creation of U-233, which is fissionable.  What you're really burning here is U-233, not Th-232.

Sorenson makes a big to-do about the very low output of wastes in comparison to a U-235 cycle.  Well, how correct that is really depends on your point of view.  Burning U-233 produces just as much aggregate waste as U-235.  What is different is that U-233 fission produces a lot of the familiar products like Sr-90 and so on, but very little of any of the transuranic (i.e. heavier than uranium) waste that people are so terrified of.  That's the stuff that people talk about as being so dangerous it has to be buried for tens of  thousands of years.  Of course, fast reactors also solve this issue as well and also leave behind wastes that are hazardous for comparatively short times, so thorium isn't really special in this regard.  And when Sorenson talks about the waste being a tiny fraction of that compared to PWR/PHWR, he's also counting the wasteful nature of the overall pipeline what with all the leftover depleted uranium that never even enters the reactor in the first place.  It's also counting the fact that the enriched uranium that does enter the reactor is hardly pure U-235, so you end up with some non-burned material coming out, compared to a more "pure" Thorium fuel input -- so it's really more like a "waste" of the overall process from mining to reaction.  Since a LFTR necessarily transmutes and burns all the fuel with which it is supplied, you end up with only waste on the outer end.  Watt for watt, fissioned atom for fissioned atom, you don't actually have a waste advantage.  You just have a cleaner pipeline than PWR/PHWR/BWR etc.  Of course, it is fair to point out that the low transuranic wastes is a valuable shift.  It means that nothing is left behind that will remain radioactive for several millennia.

Something that Sorenson also points out if you listen to any of his full-length talks is that regardless of whether you're talking about transuranics or not, the radiation concerns are severely overblown.  The risks and dangers are a pitiful little joke compared to what people believe them to be.  Your exposure due to other sources of radiation make nuclear waste pale in comparison.  The only real thing you have is that because you can't create weapons-grade fuel from a LFTR, nuclear proliferation isn't a serious concern, and that's a real solid positive.  However, this is also true for any fast breeder reactor design as well, because it would actually use all those transuranics and weapons-grade fuel and power the reactor all at the same time.

Of course, there's one thing that FBRs can do that the LFTR can't, which is to actually make use of the existing stockpile of wastes and actually transmute them into usable fuel which it then burns.  FBRs can also use thorium as well in the long run, but the LFTR design is pretty well made with liquid salts of thorium and its products in mind.  Based on Sorenson's talks, I'm guessing he's using the compact size of the reactor to improve the relative neutron economy, and since Th-232 can apparently absorb even slower neutrons, the temperatures don't need to be that high.  It is a base framework that could be extended to other paradigms, but then you would probably no longer be able to get by on passive safety measures even at a small scale, and you may even need accelerator drive.  Its passive safety measures make it very attractive from a cost and safety standpoint, but then, there are a lot of advancements that move down this line even for conventional reactors (pebble-bed, for instance).

So all that said, I basically have to say that thorium does indeed solve a number of problems, but a few of them are problems that don't yet exist (though they may eventually).  A number of the problems it solves have other solutions that are known and proven technologies (though it's arguable as to what is easier to build and scale up).  And then you have the problems it doesn't solve because it's simply side-stepping them by saying "let's forget about all that and switch to thorium from uranium."  A lot of the valid doubt over the value of thorium stem from the fact that people have only messed with it using relatively old technology.  Currently, India seems to have the only still-operating thorium reactor, and it's actually a converted old PHWR.  Now all the comparisons against existing coal and oil is really not much of a point in my book because saying that your nuclear reactor design is superior to a coal burning plant is like saying that you're morally superior to Ann Coulter.  Not exactly a high bar.  There are some practical engineering challenges with the LFTR which are yet to be explored, and I suspect that as it scales up, you will no longer be able to rely on passive safety measures.  You're going to need lots of active measures and redundancy.  But that's the sort of thing current engineers are well aware of, and I think these aren't insurmountable problems.  Wherever you go with it, nuclear is still the future.

This is especially true if we're to try and power a green energy economy in the future.  Regardless of what you think the future of fuel is for vehicles, somehow or other, an electrically-guided economy means the load on the grid will be massive, and only nuclear has even a dream of a possibility at having a shot at meeting the demands.  The classically "green" methods like wind and solar are still a few orders of magnitude behind, and they probably always will be.  It's just that it's a bit silly to say that there's a single miracle solution that will solve everything, and thorium is certainly not it.  It's just part of a larger long-term plan that will basically be nuclear.

Too bad Mr. Fusion isn't going to happen within my lifetime, though.

3 comments:

  1. You haven't researched your points very thoroughly. Lots of errors, some missing the point.

    See http://liquidfluoridethoriumreactor.glerner.com/ for very clear explanations and links to technical references, but here's some briefly:

    LFTR scales to large sizes just fine, use bigger equipment or duplicate equipment (e.g. removing xenon, transferring uranium from the blanket to the core, extracting fission byproducts, and heat transfer). Did you consider multiple drain pipes to multiple cooling tanks?

    Big advantages to several smaller reactors. For example, 200MW LFTR will fit in a standard truck, assembly in quality controlled factory.

    Yes, nuclear reactors transfer heat to their coolant well. Most engineers talk about thermodynamic efficiency as how much reactor heat makes electricity, megawatt(thermal) vs megawatt(electrical), 35–40% efficiency for standard steam turbines.

    LFTRs absolutely can consume LWR waste. Same fissioning 99%+ of the fuel. The MSRE demonstrated this.

    LFTR has Dramatically less waste than LWR. To really summarize, LFTR 665kg for 10 years plus 135kg for 350 years and 0 for 100,000+ years vs. LWR 35,000kg for 100,000+ years, to make the same 1GW-year electricity.

    Build for the military first because they write their own nuclear regulations, and will get that done Much sooner than the inept US Congress and NRC.

    Most safety in a LFTR isn't engineered or passive, but Inherent safety (based on physics and chemistry): atmospheric pressure, coolant can't boil away (minimum 400 degree C margin to boiling), self-regulating stable fission rate, nothing in reactor to explode incl no hydrogen production, salts chemically bind to most fission byproducts (including the most harmful ones) so they won't release into the atmosphere.

    "Coolant won't boil away" isn't based on some thermostat working; there's no way for coolant to suddenly get 400 degrees C (750 F) hotter.

    We gave up on solid-fueled fast-spectrum reactors before, for control issues; TerraPower will have to deal with those. Plus, somehow guarantee that no pipe ever breaks, since liquid sodium reacts explosively with water or air. A molten-fueled salt-cooled fast-spectrum reactor would have neither of those problems.

    ReplyDelete
  2. "LFTR scales to large sizes just fine, use bigger equipment or duplicate equipment."

    I've seen a lot of conjecture and assertions along these lines, but nothing in the way of experiments or even simulations. Conversely, there are a lot of simulations and much older tests that seem to present issues with scaling up. Particularly the problem of separating Pa-233 out of the blanket and getting it away from the neutron flux so that it can decay into U-233 -- this is really hard at large scales to begin with, and made even harder if you're dealing with a single fluid LFTR, where the fuel and the breeding occurs in the same fluid space.

    "Did you consider multiple drain pipes to multiple cooling tanks?"

    I *ONLY* considered that. How would you scale up without that? I'm pretty you'd never be able to sustain large scale unless you not only had multiple drain tanks, but multiple plugs as well. A single plug for a GW reactor is ridiculous. Overall, I still remain unconvinced. There are too many examples of trying that in experiments where people have calculated that merely keeping the plugs frozen will be a problem.

    Again, it's true that a lot of the working experiments have relied on old technology, like India's reactors, but the end result of that is that it remains unproven in any real practical sense.

    "Big advantages to several smaller reactors. For example, 200MW LFTR will fit in a standard truck, assembly in quality controlled factory."

    Ummm... Whoopee? That's terrific for individual clients who have hefty needs, but not if you're trying to provide power for the whole grid. And that's especially not the case for a hypothetical all-electric future where the load on the grid will basically end up shooting up an order of magnitude or so quite easily.

    I think you're confusing "simple" with "easy." Fast reactors, for instance, are indeed complex, but there has been so much work already done in that field that a lot of techniques and problems have already been solved -- which in turn actually makes them "easier" since there's so much knowledge out there already.

    "LFTRs absolutely can consume LWR waste. Same fissioning 99%+ of the fuel. The MSRE demonstrated this."

    I agree the MSRE demonstrated the possibility, but AFAICT, the advocacy groups don't favor a design that is built with this possibility in mind. More specifically, because the absorption cross-section of U-238 contains a few narrow spikes, your operating temperatures are very different. Fast reactors are made specifically for this, whereas the LFTR that Sorenson sells in his shpiel isn't... moreover, there are talks that Sorenson gives where he specifically talks about the lower-energy operation of the LFTR to be a selling point (particularly on safety). He seems to be centering his talks around a design which is built only to work in relatively low energy neutron economy -- which is generally a good idea if you're trying to sell small modular reactors.

    "LFTR has Dramatically less waste than LWR."

    1 ) I actually mentioned the difference between waste per unit fuel consumed waste of the overall process from mining to burning... the fact that less than 1% of the fuel is actually used in an LWR is the primary difference here because it skews the picture.

    2 ) I would have thought it pretty clear that no one here is talking about LWR or PWR as the end-all solution and Thorium as not valuable. I don't even see why you would think that it's a dichotomy between Thorium and the current status quo. If anything, I'm advocating a move to have breeder reactors (whether Th or U) be the status quo.

    ReplyDelete
  3. "Build for the military first because they write their own nuclear regulations, and will get that done Much sooner than the inept US Congress and NRC."

    That's a fair point. That, and it's much easier to get funding that way. My big question marks don't really apply there, though. I think small modular LFTRs would be perfect for the military and maybe even major corporate/commercial clients. The real difficulties lie beyond that point, and that's the problem when you start taking flights of fancy about anything and speaking as if it will solve every problem.

    "We gave up on solid-fueled fast-spectrum reactors before, for control issues; TerraPower will have to deal with those."

    True, but the only reason I bring them up is because the amount of capital investment already put in on their end means that they've already come a lot farther on this end and come to the point of getting a very "easily" automated regulation that requires no continuous effort to maintain. Admittedly, it's all in the realm of detailed full-scale simulations and not in actual application, but that's still better than rhetoric and conjecture.

    Perhaps Flibe has come a lot farther than I'm giving them credit for, but since they haven't apparently put out any such details in the public domain, I have nothing to go off of.

    ReplyDelete