There are a huge array of design possibilities for a Thorium molten salt reactor / liquid fluoride thorium reactor, but this post takes a very simplified approach to map a small part of the fascinating and diverse landscape with a little rough drafting. We concentrate on thermal spectrum designs. I’m strictly an amateur in these matters so everything I write should be taken with a grain of salt and checked more carefully with people who actually know what they are talking about. 🙂
My references are Kirk Sorensen’s energyfromthorium.com and David LeBlanc’s excellent slide set. Highly recommended!
One or Two Fluids?
One choice is having two fluids (core and blanket) or just a single one (Thorium-232 and Uranium-233 salts mixed in the same vessel).
The single fluid reactor seems like a much simpler alternative. There are problems though that when Thorium-232 absorbs a neutron, it becomes Protactinium-233 (Pa) before becoming Uranium-233. The half-life of Pa is about 30 days, and if it catches a neutron, it becomes U-234, which is not a fissile material. So the neutron was wasted. This is pretty bad since Pa’s neutron cross section is over twice that of Th’s so it wants to catch those neutrons. But there are ways around this.
Protactinium Extraction?
For one, you can perform protactinium removal from the single fluid core. This involves liquid bismuth and I’m not very familiar with the process, but it is more complex than U or FP separation (see below). The separated Pa-233 is put into tanks to wait for its decay into U-233 before it is put back into the core. This also enables pure U-233 extraction which is somewhat problematic in a proliferation sense (I’m not very knowledgeable in this direction).
In a two fluid reactor, you can probably get by without Pa removal by just making the blanket bigger, so there is much more Thorium than Pa at any given time, giving a better statistical chance for a given neutron to hit a Thorium and not a Protactinium atom. This can also involve fancy geometries, since the core stays the same size. For example there can be a long empty gap between the core and the blanket through which the neutrons fly.
Uranium Extraction?
In a single fluid design the new Uranium is already in the core after it is produced from the Thorium, so this step is not needed. But in a two fluid design it is practically mandatory. The ThF4 and UF4 mix is bubbled with Fluorine gas, and the UF4 turns into UF6, which is a gas. The gas is separated from the liquid. UF6 handling is already a known 60 year old industry from nuclear fuel processing. The UF6 is changed back to UF4, which is then inserted into the core. The ThF4 is inserted back into the blanket of course.
Moderation, How?
The most traditional way would be to moderate the reactor with solid graphite rods. There are thermal spectrum spreading and since the fuel is in liquid form, expansion and convection effects that mostly help with control here, but single fluid reactors still can have situations of positive power coefficients. Graphite has issues of swelling and shrinking under radiation and a limited lifetime. Regularly changed moderator pieces would be radioactive waste as well. Loose graphite rods or even pebbles could be ways to prevent stress cracking.
There have been designs without any solid moderators as well. The salts moderate some anyway, and some reactors run simply at faster spectrums, the latter again bringing control problems.
And lastly, there is the heavy water moderated MSR, of course loved by many Canadians because of their CANDU expertise, the current heavy water moderated solid fuel reactors. The heavy water would not act as a coolant at the same time, but would be separated in thermos style vacuum jacketed pipes. Any breach would boil the moderator away, reducing reactivity. This jacketing would enable easy low pressure vessel construction, like for the rest of the reactor.
Geometry?
The first MSR designs were naturally small and spherical. A number of geometry proposals have surfaced: pipes coming from many directions into a bundle of criticality or a cylindrical core with a cylindrical blanket around it. There are many conflicting requirements like maintaining criticality, preventing overcriticality, minimizing neutron waste, minimizing inventory size, coping with thermal expansion, passive safety, ease of construction and transportation, scalability…
Inventory Size?
A small fissile inventory has a number of benefits. The reactor is simply smaller for a given power ratio, making it cheaper. Also less U-233 for the starting load needs to be produced elsewhere. If the breeding ratio is low and the fissile inventory large, it takes long (perhaps twenty years) for an MSR to produce enough U-233 to start another new MSR, probably requiring big production of U-233 elsewhere, changing overarching infrastructure plans significantly.
Fission Product Removal?
This is a very common feature of MSR designs (some once-through no-refuel systems for ships and spacecraft can take this out), but how often it is done, varies. Vacuum distillation means reducing the pressure until the most volatile stuff starts boiling. Only UF4 fuel salt should be left after this. With a single fluid salt there is a problem that the Thorium Fluoride salt boils as easily as some of the fission product fluoride salts and hence something else would need to be done, or just some Thorium be “thrown away” together with the fission products. On the other hand, some fission products are gases like Xenon, which boil already in the core. Many of the fission products are also valuable rare elements by themselves and can be sold. The MSR naturally separates fission products from fuel, unlike current solid fueled reactors, making this line of technology look unexplored and interesting. Since there is less waste sucking up neutrons in the reactor, the neutron efficiency of an MSR can be high and it can get by with little fuel and a small physical size.
Power Extraction, How?
This is more of an “auxiliary” function. The most common design is connection to a helium heat exchanger probably through another salt loop to reduce radiation loads, and the helium would drive a heat engine. High temperatures (compared to solid fueled traditional nuclear reactors) are achievable and all kinds of thermal processes become attractive as well. Since the molten salt solidifies at a certain temperature, the heat exchanger can not go below that. Coupled with the high efficiency and high top temperature, you get small radiators and cooling towers and small water usage compared to traditional reactors.
Control, How?
An ideal MSR is extremely easy to control because of the natural negative feedback of the reactor – increase power usage by increasing coolant flow, the reactor cools and increases reactivity and gets to a steady level. As far as I know, the original aircraft reactor experiment was not throttled in any other way. There simply are no control rods. An MSR would automatically operate in a designed, relatively narrow temperature region of a few tens of Kelvins. In the event of a significant overheating accident, a freeze plug at the bottom of the reactor would melt and the salts would drain into tanks below for slowly cooling down.
Waste Handling
First the most valuable materials like Xenon, Tritium and various other substances could be sold. Since the waste would by majority be fission products and barely no heavy transuranics, they could be stored on-site until they quickly lost enough radioactivity and cooled for final storage of a few hundred years inside rock.
So, What Is The Best Configuration?
As a personal preference, my best guess for the best of all worlds is a two fluid design, with a cylindrical “LeBlanc” geometry. Adding power would simply involve making a longer cylinder. No Pa removal, but the blanket could be oversized. Remember that all the vessels would be at low pressure, and the blanket isn’t even that hot and is only filled with Thorium, which is cheap. Initial core load U-233 amount would still be low.
Next to the reactor would be the fluoridation tower for taking U from the blanket and putting it to the core. Hydrogen would be needed to change the UF6 back to UF4. (Generating chemically nasty HF at the same time.) A vacuum still would remove the fission products from the core, and this would be located close by too.
Since there would be no high pressure vessels, the reactor core and blanket could be quite light and probably road transportable as a whole. The material of choice for the high temperature salt sections would be Hastelloy.
Moderation could either be nonexistent, by loose graphite profiles inside the core, or Hastelloy pipes containing a vacuum walled heavy water pipe. This is an area that needs lots of innovation to minimize maintenance and waste.
According to calculations made by David Leblanc, a 1 m diameter 6 m long “pipe style” reactor (easily truck transportable for construction!) could produce 400 Megawatts of electricity (900 MW thermal) with only a few hundred kilos of fissile inventory! Most of the salt mass would be “filler” (Fluorine, Lithium, Beryllium) and Thorium.
This is a very stark contrast to the current nuclear plants with their huge high pressure vessels, currently probably only manufacturable in one steel plant in Japan, requiring many years of ahead time for ordering, and lots of welding on the build site as well.
End Note
There is real small MSR hardware that has been demonstrated to work, and there are good conceptual designs for real powerplants that could change the world energy generation picture completely. I am not aware of any other energy production technology which is so based on reality and still potentially revolutionary at the same time. This thing has a very good chance of working well. It does not require significant new scientific developments or a fifty year timetable, like fusion. It is not as limited in availability and dependability like wind or solar power. It is good that those alternatives are pursued too, but the MSR is a bigger answer to a bigger problem.
-the Molten Salt Reactor, Real and Revolutionary!
Gravity Loss (now moved to gravityloss.com) 