Conventional nuclear reactors are finicky, complicated and come with a lot of big and expensive secondary cooling systems. What if there was a way to get rid of the cruft?
There are a few fundamentals which must be understood:
A nuclear power plant, like all power plants produces heat which powers
a thermodynamic cycle, some gets turned into electricity, the rest gets dumped into the environment via a cooling tower or similar. Once you get past the nuclear stuff to the steam turbines and whatnot a coal plant and a nuclear plant look quite alike. A nuclear plant is different only in that the heat source is both more finicky and produces dangerous byproducts which must be contained.
Containing the dangerous nasties is generally done by having fuel pellets with a metal shell. The radioactive nasties produced stay trapped inside the shell and so little leaks out that divers can swim in the spent fuel ponds to do repairs. Neutron radiation in the reactor core still manages to make some material in and around the core radioactive but by picking the right materials this problem is minimized.
Chernobyl is a great example of that finickiness in action. One of the reactors was abused a bit, which resulted in an uneven distribution of certain reaction products. When it was turned on later, the reaction was uneven, certain portions reacting more intensely than others. When they turned the power up a bit more, the portions where the reaction was most intense went a bit out of control, there was a small explosion, a fire and we all know how that ended.
Another problem, dropping the control rods into a reactor stops the nuclear fission, but a significant amount of heat continues to be produced. This tapers off over time but that's why fuel bundles have to be stored underwater for a few months after being taken out of the reactor. Fukushima shows what happens when those cooling systems fail. In small reactors, like those found in a nuclear submarine, passive cooling is sufficient to keep everything from melting but in bigger reactors water has to be pumped around or things start melting.
So, aside from Chernobyl style idiocy, a nuclear reactor is a heat source that can't be fully turned off and which, if it overheats and the containment layers are breached releases radioactive materials. Unfortunately, conventional reactors are cooled with water ... at high temperature and high pressure. That's a good way to make steam, but a lousy way to cool things. It limits how hot the reactor can get. A hotter reactor has a bigger temperature difference between it and the environment which makes getting rid of heat easier. In any case there are lots of pumps and valves and backup generators etc. whose job it is to ensure the core stays cool. They have to be redundant and ultra reliable and that costs quite a bit.
So, the obvious implied answer here is to push up reactor core temperatures, or at least increase tolerance to high temperatures. Plenty of proposed designs do that by using a different coolant. A liquid metal or salt which has a high boiling point. They still use a complicated core, fuel rods etc. and if the temperature goes too high it all melts. That's progress but it doesnt go far enough.
This new design has several goals:
--ease of manufacture
--ease of service
--high neutron economy and breeding ratio
--homogenous and easily controllable core
--high operating temperature to provide large delta-t for easier heat exchange
The reactor vessel itself is a re-purposed steel ladle, a steel bucket lined with ceramic bricks, the ladle is partly filled with molten uranium. Above the uranium sits a layer of molten salt. Something with a more reasonable melting point than the uranium itself like FLiBe so external heat exchangers can run at lower temperatures using the salt as a heat transfer fluid. Control of the reaction can be done by adding or removing fuel which changes the size of the uranium blob or by inserting neutron absorbing control rods.
The uranium melt is stirred by convection currents, with heat removed at the top by the molten salt, The molten salt flows into some external apparatus which absorbs that heat (a steam generator, a heat exchanger... whatever) before returning to the ladle.
A word on heat exchangers and other heat exchange devices. They are usually made of metal, in the case of a steam generator, this metal has to withstand the high pressure steam flowing inside it. Steam generators or heat exchangers capable of using the 1000C+ salt directly aren't practical. A workaround solution is to dilute the very hot salt by mixing with cold salt from the output of the heat exchange device.
For ease of maintenance and replacement, as well as to prevent leaks the heat exchange devices rest in a trench which comes off the reactor vessel. Since the trench is meant to contain heat sensitive equipment, it's isolated from the main salt pool by a sort of wall. This allows the trench to be cooled to near the salt's melting point while the main pool remains much hotter. By opening a hole in the wall, some salt can flow between the two sections carrying a controllable amount of heat into the trench, that's the dilution method mentioned earlier, though something more complicated could be done to reduce thermal stress on the refractory nearby. This is to allow ease of equipment replacement. It's like lowering equipment into a pool of water. Components are easily replaceable. If something breaks, pull it out and install a replacement without shutting down. Did I mention that molten FLiBe is transparent? The only thing that can't be replaced is the vessel and it's lining.
Cooling the reactor after a shutdown is a fair bit easier than in competing designs because of the high operating temperature. A portion of the trench walls or lid could double as salt air heat exchanger. If water is removed from the steam generators trench temperature could easily be raised to 700 or 800 C without damage. Compare this to the measly 300-400C in a traditional nuclear reactor. In a worst case scenario, the reactor could be drained into a structure below, or beside it more suited to cooling. The uranium drops into a heat resistant vessel and the salt follows to act as an inert thermal interface material.
The remaining question is the radioactive material. There are two concerns here, firstly, volatile fission products will be coming out of the molten salt. These have to be captured. The molten salt itself is also radioactive, so human workers can't work over-top of the pit or trenches. A containment structure is a good second line of defence but it would be nice to have the interior of the containment structure stay relatively clean so that equipment and robots that go in don't need to be decontaminated extensively. Covering the reactor pit and trenches and generally sealing the gas space above the molten salt seems like a good first line of containment. Keep it at a lower pressure than the containment building and pipe the collected gases to a fission product capture system. Holes for actuators and sensors would be an issue and of course, when replacements have to be done, covers and seals have to be removed. Having mobile extensions to the gas collection system could solve that sort of problem, a second layer of containment could work too. This is really just a matter of convenience, a gas tight containment structure will keep any and all nuclear material contained. Making the space inside convenient to work in is more of an operational concern than a safety thing.
The resulting reactor has several desirable characteristics. The reactor core is liquid uranium (and other stuff), being a liquid, it stays well mixed, being at above 1000C a lot of fission products diffuse out, this includes some interesting things like iodine and xenon, which play a role in making reactors harder to control.
Since there's only fuel in the core, there really isn't anything for the neutrons to hit other than fuel. No casing materials, no support structures, no coolant, sure some neutrons will leak out the sides but that's it and that can be mitigated by making the reactor bigger. The neutron economy, the efficiency with which neutrons are used is really high, lacking a moderator, those will be fast neutrons so fertile material will become fissile fuel, yes this is a breeder reactor. The breeding ratio can be brought very close to the theoretical maximum. This lets us ignore accumulation of poisons since neutron absorbing fission products are offset by increasing quantities of fissile fuel. The end result is a reactor whose fuel never needs to be reprocessed.
Above all this is a big container with molten uranium in it. It's meant to be easy to build. Ideally a coal, plant can switch over to using nuclear heat, a simple conversion. The reverse has been done before.
To keep things reliable, a reactor should have more trenches and heat exchange devices installed than are needed. A single breakdown doesnt require a shutdown. If there is a breakdown chances are it's because of a leak. Fast closing valves are therefore a must and precautions would have to be taken for a steam explosion in the trenches caused by a broken steam generator tube. Since the steam generator is at a higher pressure leakage should be into the containment so contamination isn't really a problem. Close the valves to that steam generator and carry on. The same applies in a shutdown situation with a salt air heat exchanger. Cut off airflow and keep using the others.
Simple, cheap and relatively robust.