Travelling wave reactor anyone ?

  1. vanesch

    vanesch 6,236
    Staff Emeritus
    Science Advisor
    Gold Member

    Several times now, I've been in contact with people who talked about the "travelling wave reactor", but I can't find any "official" information about it.
    The only thing I could find that sounded somewhat serious is in MIT's "Technology Review":

    Is this a joke, or is it more serious and in that case, has anybody any extra information about it ?

    It sounds simple: a small layer of highly enriched U-235 "feeds" a natural uranium block that is transformed slowly in Pu, and a "wave" of burning fuel travels through the block. But I see tons of problems with that design, the most obvious one to me being that you will never ever get enough "useful neutrons" that way to keep the chain reaction going and the breeding going.
  2. jcsd
  3. Astronuc

    Staff: Mentor

    I've not heard of this particular concept, but I've heard of coupled cores.

    I'm skeptical about the highly enriched material to get it going. I'd be interested in how they do reactivity control. I don't see this going on for decades.

    That's an amusing comment about licensing discussions with reactor vendors. There aren't that many.
  4. I am pretty skeptical of how they would keep it going too. neutrons dont just travel to the next layer of atoms, so there must be some fissions happening at some larger delta x, which means there are fission products there. I expect these f.p.s will build up enough to halt the 'forward progress' of the wave...

    How is it controlled??
  5. maybe i am viewing this diagram wrong... hopefully I am viewing this diagram wrong, if they intend to use such a device. It looks to me as if the coolant never surrounds these tubes where this wave is travelling back and forth. How would this "reactor ever be cooled? And there is an expanse for gasses to build up? How dumb is that? These bundles of pillars would reach extremely high temps rather quickly. And would be prone to blistering and galling. I see many flaws to this design from here. I have to be viewing this wrong
  6. Astronuc

    Staff: Mentor

    I suspect the intent is to have a highly enriched zone that propagates like a wave. The zone starts at one end/face/side and propagates through the breeding material in which U-238 is converted to Pu-239, or Th-232 is converted to U-233. The geometry essentially looks like a flat plate.

    Let's say the highly enriched (~90%?) starts on the left side of the breeding zone, so the left side of the HE zone has a reflector, e.g. Be, and the right size is the breeding zone. The reaction starts, and the fast flux penetrates the breeding zone. Fission products accumulate in the highly enriched zone in proportion to the ∑fΦ(xi).

    I wonder about the neutron spectrum, and whether the reactivity control is accomplished by a combination of Doppler and density (as a function of temperature). This seems to be an interesting exercise on paper, but I'm highly skeptical with respect to practicality.

    I'm curious about the transport of fission gases (Xe, Kr) and volatiles to the plenum volume, and the conduction of heat to the liquid metal coolant.
  7. vanesch

    vanesch 6,236
    Staff Emeritus
    Science Advisor
    Gold Member

    In fact, there are 2 reasons why I have difficulties with the principle on which this scheme is based, apart from a lot of practical issues such as the fuel design, burnup, cooling etc...

    The first one is that the momentarily critical mass is a plate, which is a strange geometry for a critical structure. It's not easy to see how the chain reaction can go on in a plate, unless it is thick enough, but if it is thick enough so that the border losses are low, I don't see how you can breed enough "with the neighbours".
    The second problem is that the breeding is supposed to go in one direction (the direction of the "travelling wave"), but neutrons go both ways. Exactly the same amount of neutrons that go "left" and breed, will go right and get lost in burned-up fission products. And that neutron loss is too large, because it means you need a k_inf of over 3
    (1 for the chain, 1 for the breeding, and 1 "lost"). Unless there's a way to have a reflector move with the burning plate, so that these neutrons get back to do something useful.

    With all that, I don't see any particular advantage over more standard designs of breeders. I only see extra difficulties.
  8. Astronuc

    Staff: Mentor

    Initially one could place a Be reflector on the face of the highly enriched zone (plate), but then the reflector will not travel with the fission zone. Once the wave starts to move, I wonder if the fission zone expands, i.e. the front face moves into the breeding zone at a greater rate than the following side moves.

    I agree that there seems to more problems/disadvantages than advantages.

    I wonder if they've done calcs with MCNP?
  9. I dont know, the number of depletable regions, for a trustworthy calculation (and a very fine depletion profile which should be necessary for this type of reactor) would, I imagine, make this incredibly time consuming to run with an MCNP-Monteburns job.
  10. Apart from cooling and control issues, I do not understand why they would want to have the reaction propagating in one direction. Losses could potentially be minimized by having the highly enriched zone in the centre of a "slab" surrounded by breeding material. The reaction could proceed in both directions somewhat reducing losses I would assume. You could even surround it by reflectors on the periphery of the slabs I suppose also. Just seems a weird design trying to get something to proceed in one direction when dealing with an isotropic phenomenon.
  11. vanesch

    vanesch 6,236
    Staff Emeritus
    Science Advisor
    Gold Member

    Indeed. To me the most logical design of a breeder is a rather homogeneous mix of "initial fuel" and of fertile material, such that the breeding ratio is close to 1: the consumed initial fuel is then replaced by newly bred fuel. The problem is of course the build-up of fission products and the diminuation of U-238 concentration. For the neutron balance, the fission products will eat some neutrons, on the other hand the U-238 will eat less of it, so the neutron balance must remain close to the same. Of course you can't burn up everything, as you need a certain amount of U-238 for passive safety (Doppler effect) and in any case your fuel elements will be damaged after a while (first barrier). So after a certain burnup, a reprocessing will be needed. In other words, a "standard" breeding reactor.

    The thing you suggest is the "old" breeder construction, with a fast core, and a "breeding mantle" - that's a perfect plutonium factory then.
  12. I was surprised to see that fast reactor designs were back after I left the field over 35 years ago where I worked at both INEL (EBRII) and HEDL (FFTF). The design may have changed but the physics hasn't.
    First, addressing the statement that the TWR converts depleted or natural U238, implying that you are getting "free" energy, I recall that the "classic" light water reactor generates about a third to a half of its generated power from the U238 in its fuel rods (LWRs used 3% enriched U235 vs. 0.7% natural).
    Second, the "burnt" portion of the TWR will still have a significant portion of Pu238 that was converted from U238 via N absorption & α-decay and the Pu238 can be chemically separated, which is a lot easier than isotopic separation needed for U235 enrichment. I believe that the relative ease of Pu238 separation was the reason why we decided to kill the technology in the 70s as a way to stop proliferation.
    Third, the neutron lifetime in a fast reactor is on the order of milliseconds. Fission product neutrons have energies of over 1Mev and it takes milliseconds to reduce its energy to the 100 - 300 kev range where the fast neutrons react with the Pu238 or U238. This means that you can double the neutron population in several milliseconds before any human can react. In fact, one of the designs was to use annular fuel pellets so that melted fuel could be blown out to the plenum and away from the reaction zone automatically as a way limiting this very short doubling time. This struck me as being a very brittle technology.
    Fourth, fast reactors have to use liquid sodium for a coolant instead of the water used in LWRs because water thermalizes the neutrons. The problem is that sodium reacts violently with water, which usually means that you have to have an intermediate coolant loop to ensure that the sodium does not come into contact with water via a leak in the heat exchanger, for example.
    The nuke accidents at TMI and Fukushima were both loss of coolant accidents where the emergency coolant was not sufficient to remove the decay heat of the core generated by the fission products. The accident took hours or days because it took time to boil off the coolant sitting in the reactor and going to a smaller design may help since you would expect heat generation to scale with volume while heat transfer to scale with surface area but you would get this benefit regardless of reactor type. On the other hand, one of the early deaths occurred as a reactivity insertion accident, which is strongly controlled by the neutron lifetime.
  13. I found another Technology Review that been debunking a rumor that Bill Gates is building the reactors in China:

    Terra Power is talking to India and China about the reactor and with all the NRC regulations in place, I would think they would do R&D and perhaps someday a prototype in those countries than the US.

    Here is Argonne National Lab anaylsis on once-through nuclear reactors, including the TWR:

    It does give some of the description of the core layout as of the reports date. It discusses the wave now being stationary, by shuffling the fuel around. Not all the data is there due to proprietary reasons. Also, it talks about recladding the fuel elements too, which is my main concern. It is a challenge to make materials last 40-60 years in such high temperature and radiation environments.
Know someone interested in this topic? Share a link to this question via email, Google+, Twitter, or Facebook