So I was recently inspired by the answers that I saw on this stack exchange to the following question: Feasibility of H.G. Wells' Continuously-Exploding Atomic Bombs

And it occurred to me that you can make the core as hot as you want, as long as it doesn't touch anything. So, envisioning ping pong balls hovering on hair dryers (via Bernoulli's principle), I was wondering if you could build a reactor where a mass of uranium is left hovering in a (fast-moving) column of coolant inside a larger tank. The motion of the liquid serves to cool the core, as well as keep it in place in the vessel. Incidentally, if you use many smaller uranium pellets in the core, and they are held together in this column by Bernoulli's principle, you might be able to stop the reaction by slowing down the coolant flow and letting them all fall down and separate.

So, is it possible to build a reactor based on these ideas?

  • 6
    $\begingroup$ People are trying to make better fuel pellets that don't crack easily yet you... $\endgroup$ – user6760 Jan 14 at 3:06
  • $\begingroup$ Nuclear reactors do use coolant already - water, mostly. $\endgroup$ – Halfthawed Jan 14 at 3:10
  • 1
    $\begingroup$ And remember, if you are using uranium as fuel, you have to wait 3 days to turn it back on if you turn it off for ANY reason. Or you might get another Chernobyl $\endgroup$ – Mindwin Jan 14 at 14:25
  • 3
    $\begingroup$ Another approach is a homogeneous reactor, which uses uranium suspended in water or some other liquid. The reactor is a closed pipe, with a pump that circulates the fuel. There's a bulge in the pipe, and the bulge is sized so that the combination of liquid and fuel inside it reaches critical mass. Everywhere else in the pipe the material is hot, but not critical. Just keep pumping! $\endgroup$ – Pete Becker Jan 14 at 14:52

Of course you could, but it's a VERY bad idea...

With enough time and money, such an engineering feat as that you have just described is surmountable. You could certainly design and build a nuclear power plant that does this, but no sane person who has the expertise to design a nuclear reactor would ever do this. It is the equivalent of spending a fortune to have all the seatbelts, airbags and crumple zones taken out of your modern car.

Modern nuclear reactors work on a fail-safe design. They actually need power to keep the reactor operational. There's more to it than this of course, but basically the control rods are set so that without power, they fall into a position that controls and limits the reaction of the nuclear fuel. They require power to be raised out of the way so as to let the reaction build up. That means that if the power to the reactor fails for any reason, the reaction stops rather than building up to a critical mass and melting down the reactor.

Your design does the exact opposite. It basically requires power to slow the reaction, not the other way around. In short, while you have power you can control your reaction quite well and in a visually spectacular manner, but if your power cuts out for any reason, or you get a mechanical failure to your coolant fans, or any other problem, your reaction rate and temperature actually starts to increase. You'd end up with a meltdown.

In short, you want your reactor to fail safely, or in other words fail in a way that the default position of the reactor without power or during a mechanical breakdown of some sort means that the reaction stops. This design only allows the reaction to be controlled IF the reactor is operating correctly. At your first failure in the plant, you are going to have a really bad day.

  • 1
    $\begingroup$ 👍Agreed faster flow means low pressure and temperature reducing the amount of steam to drive turbine, also moving target means reduced or extinguished neutron capture rate. $\endgroup$ – user6760 Jan 14 at 4:48
  • 3
    $\begingroup$ A set of moderator rods can be fixed below the levitating fuel pellets. When the flow stops they drop into an array to shutdown any chain reaction. A set of moderator rods can be held in place, as per real world reactors, which will drop when the power fails or someone presses "reactor shutdown". $\endgroup$ – a4android Jan 14 at 7:12
  • 1
    $\begingroup$ The fuel can be held in place by the flow but fall apart when the flow stops, spreading the fuel in separate tubes. Since the fuel is very dense it cannot be an issue. $\endgroup$ – jean Jan 14 at 14:11
  • 5
    $\begingroup$ Note that a fail-catastrophically (vs fail-safe) design might make sense in some worldbuilding contexts. Imagine if the reactor is powering containment of something that would be extremely dangerous if it escaped - in that case, turning into a bomb could be preferable over powering down if something goes wrong (or if someone tampers with it). $\endgroup$ – R.. GitHub STOP HELPING ICE Jan 14 at 15:34

Your design seems like an absolute safety nightmare. I'll also note that your safety idea of the little fuel pellets separating when the air flow is removed rather depends on them not being so hot that they start melting, which rather defeats the whole point of the exercise.

The underlying idea, though, that if you remove the tiresome need to avoid melting your reactor down into deadly slag you can run it really hot and so gain all sorts of efficiency gains and various other benefits is a) quite true and b) something that people have looked into, though not necessarily for the purposes you were thinking of.

Operating such a thing inside a gravity well, such as on the surface of Earth, is a pretty terrible idea, because any sort of failure will ruin your reactor as a load of very hot, intensely radioactive gloop sloshes onto the floor of the containment structure and melts and burns away til it cools off enough, or alternatively intense radioactive fissioning gas will rise up and plate itself all over the ceiling, burning and melting its way through as it does. The former would be indistinguishable from a meltdown, and a lot harder to avoid, and the latter is a whole new and exciting kind of catastrophe.

In space, however, you don't have to worry about gravity or even convection, if you don't want to. Hell, in an emergency, there's the option of blowing out the reactor core into space where it won't do anyone any harm (not recommended in orbit around a planet. or possibly even in a planetary magnetosphere. or near other ships. etc).

Behold, the Open-cycle Gas Core Nuclear Thermal Rocket!


(image from Project Rho, more information on this and many other nuclear rocket designs at the link)

You can run your reactor at a fairly toasty 55000K, and the only compromise you have to make is that some of the uranium plasma will leak out in with the hydrogen, but you can't make an omelette without breaking a few eggs, eh?

  • 3
    $\begingroup$ The minimum density of fissile material that can possibly go critical is 9 gm per liter. That's very difficult at 55,000K unless it's at very high pressure. Also, a big uniform mass is not very efficient for a reactor. Even if it's a mix of uranium and moderator, it's going to be very difficult to get that to go critical and also not be prompt. That is, the margin between critical and BOOM! is very small. I'd have to spend several hours, maybe days, with my nuclear software to determine if this could work. But I would be surprised if it did. $\endgroup$ – puppetsock Jan 14 at 14:31
  • $\begingroup$ @puppetsock I suspect that the people who drew up the original research papers didn't simply pluck those figures out of their nether regions. I don't have any PDF copies of them to hand though, so I'm afraid you'll have to turf them up yourself. I'd also like to point out that even if it was plausible, it seems like a pretty terrible idea for so many other reasons that I don't see anyone actually bothering to make such a thing. The closed-cycle "nuclear lightbulb" designs seem to have had more research, but I suspect they'd be eclipsed by fusion designs before they ever materialisd... $\endgroup$ – Starfish Prime Jan 14 at 14:36
  • 1
    $\begingroup$ I have the criticality safety standard document here on my hard drive. That's where the 9 gm per liter comes from. And, at 55,000K, assuming an ideal gas of U235, you absolutely cannot get a reactor to go critical at less than 50,000 atmospheres. $\endgroup$ – puppetsock Jan 14 at 14:46
  • 2
    $\begingroup$ @puppetsock here's some: High Specific Impulse Gas Core Nuclear Reactors, 1971, propellant injection at 1000atm, Reactor moderator, pressure vessel, and heat rejection system of an open-cycle gas-core nuclear rocket concept, 1973, propellant at 500-2000atm. $\endgroup$ – Starfish Prime Jan 14 at 15:12
  • 2
    $\begingroup$ @puppetsock Why would there be a minimum density for criticality? The way I understood the question up to now, it's a simple matter of what a neutron encounters on average before leaving the reactor. And, given any non-zero density, I can simply make the reactor big enough so that the average neutron will encounter enough fissile nuclei along its way to the surface. Of course, there is a limit if you also have a minimum cooling medium density in the core, but with a pure uranium plasma in vacuum, I don't see why there should be a minimum density for criticality. $\endgroup$ – cmaster - reinstate monica Jan 14 at 20:57

Possibly Possible but Probably Impractical

Keeping an object supported via a column of moving fluid requires that the fluid is moving upwards at the terminal speed of the object. Uranium and plutonium are crazy dense, and spherical pellets are not going to offer much aerodynamic resistance, so you'd either need to use molten salts or some other very dense fluid as your coolant or very small fuel pellets that can be kept aloft by water in order to keep the required jet speed manageable.

The resulting design would not be very fault-tolerant, since you'd now have a critical mass of fuel which must be kept suspended continuously by a precisely-controlled jet of fluid at all times in order to function correctly. And yes, you could design the bottom of the fluid container so that a failure of the jet results in the fuel dispersing across the bottom of the container, but you'd need some way to get from that state back into a functional one with minimal outside intervention.

Either way, though, if you're going working with spherical pellets of fuel, then you might save yourself a lot of fiddling with high-powered jet pumps by building a Pebble Bed reactor instead. These use the same sort of spherical fuel pellets piled in a static bed, with an inert gas coolant flowed up through them. Used pellets are removed from the bottom, new ones are dropped in on top, and otherwise the pellets just sit there happily reacting away in their pile.


Something along those lines may be incorperated into a fusion reactor. One of the main problems with fusion is that the plasma needs to be really hot, 100,000,000 K. This will melt anything it touches so it needs to be kept away from the walls. Magnetism has been the way up to now but perhaps adding a fast flowing gas might be able to assist the magnets.

New contributor
Paul McCarthy is a new contributor to this site. Take care in asking for clarification, commenting, and answering. Check out our Code of Conduct.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.