What is the Ideal System for Running a Nuclear-Powered Heat Engine in Space?

[vemvare]

(I'm not sure I put this thread in the right sub-forum)

If we need to operate a nuclear-powered heat engine in space to run a generator, meaning it can only be cooled by radiation, what kind of system should we go for, if we want the power per unit mass ratio to be as high as possible?

I've "researched" this on my own, and found a few things:

#1 Due to the Stefan-Boltzmann law being a fourth grade equation radiator temperature being high is very important.
#2 The reactor experiments that were part of the NERVA-program were by far the "hottest" reactor cores ever operated, at an exhaust temperature of 2370K.
#3 The mechanical strength of all known metals and alloys seems to drop so much as to become almost useless at 1500K or so.
#4 Graphite is apparently strong even at extremely high temperatures. But graphite being graphite, I can't imagine it being very shock-resistant or able to resist "grinding".

Could a ceramic/graphite stirling engine be built? How would heat be carried from the "cold sink" to the radiator? What would the ratio of the cold to hot sink temperatures ideally be?

 

[Danger]

Hi, Vemvare. For now, I would forget about NERVA completely since it's a rocket engine. Instead, try researching SNAP reactors. (I can't give you much info off the top of my head, because it's almost 40 years since I read up on them for a project of my own.) NASA has been using them for longer than that. The most effective one, if I recall, used twin Brayton cycle turbines. I think that the thermal transfer to the radiators was through "heat pipes", but I'm not sure. (The acronym stands for "System for Nuclear Auxiliary Power" and was used for things like Keyhole satellites.)
I think that there was a competing system that used the Rankin cycle, but I can't remember what it was called or how successful it was.

 

[vemvare]

Ahh, thanks, great! I had never heard of that particular program. Found http://www.gnnallc.com/pdfs/NPP 01_SNAP_Reactor_Overview_Voss_US_AFWL_AFWL-TN-84-14.pdf
I've much info to dig through...

The point of mentioning NERVA was to showcase that nuclear reactors can be run at very high temperatures, naturally a closed system would be very different from a rocket engine.

 

[mfb]

#1 Due to the Stefan-Boltzmann law being a fourth grade equation radiator temperature being high is very important.

And for efficiency of the heat cycle, you want the temperature to be low. There will be some trade-off between radiator size, heat transport, reactor efficiency, overall mass and so on, without easy solutions for parts of it.

Hot graphite is quite reactive and its mechanical properties are problematic even at room temperature.

 

[Danger]

Another consideration, in your favour, (which mfb's post brings to mind) is that structural integrity won't be as much of an issue for you. A well-behaved little generator humming away to itself won't experience anything like the sort of punishment that a rocket engine goes through during launch (especially at max-Q), and even the subsidiary systems will have it easier (such as not undergoing the "thermal shock" of cryogenic fuel entering the combustion chamber).*

*NERVA, though, circulated the fuel in a coil wrapped around the nozzle to both pre-heat it and act as a cooling system for the engine.

 

[mfb]

*NERVA, though, circulated the fuel in a coil wrapped around the nozzle to both pre-heat it and act as a cooling system for the engine.

I think most liquid fuel rocket systems do that.

 

[Danger]

I think most liquid fuel rocket systems do that.

Could very well be. I haven't investigated any of them, aside from linear aerospikes which are too weird to consider in this case. I doubt that any of them had the same differential between liquid H₂ and O₂ and the operating temperature, though, which would make pre-heating less proportionally effective in a NERVA. The latter ran at 2727°, which was all that the structural integrity of the machine would allow.

 

[vemvare]

So the reactor system could potentially operate at an even higher temperature than NERVA? I don't know much about nuclear reactors, but I'm thinking that if a failure scenario is that they are prone to "heat excursions", parts of the core could destroy itself if the operating temperature is high enough, while the immense flow of coolant in a NERVA-type engine would make this less likely.

That document is a rich source of information. Apologies if this post seems a bit disjointed. The SNAP 50, which was conceptually verified but not built, would use an UC (uranium carbide) or UN (uranium nitride) fast reactor and a potassium Rankine cycle.

Li would be used used as primary coolant, I guess due to its neutronic properties, Li-7 wouldn't become radioactive while K would.
(kelvinized and rounded numbers for convenience)

Specs:
Reactor output temp: 2000+ F, a.k.a 1366K
Turbine inlet temperatue: 1339K
Condensing temperature: 977-1033K
Radiator temperature: 977-1033K

Weird stuff is going on in that document, the scematic on page 90 in the pdf (page 83 in the scanned in text) seems to show a tertiary cooling system with NaK and lower radiator temperature. Different stages of the development of the concept? It is also later stated that both UC and UN was investigated and UN was found to be preferable.

I can't find any information on the LCRE, (lithium cooled reactor experiment), said to have operated at a temperature of 1903 C, or 2176 K (p 11/ 18 in the linked document).

They also mention the Brayton cycle, using an inert gas as secondary coolant, and the boiling-condensing of something in a "heat pipe" as tertiary coolant (the radiator).

As far as I've understood so far, the stress on the moving parts at high temperature does seem to be the "bottleneck", otherwise they'd just boil something hotter, like Li. Is it possible that a MHD-system could be used?

An interesting side note, perhaps a space based version of something akin to UHTREX could separate out the fission products from the primary coolant loop and simply send them off into space.

 

[Danger]

So the reactor system could potentially operate at an even higher temperature than NERVA?

Maybe...
The reason that NERVA was restricted to 2,727° was because that was the temperature at which one or more of the components either melted or, as you said, suffered structural failure. I don't know, right off how they got that high, because the melting point of uranium is around 1,300°. For some reason (possibly too many drugs as a kid back then), I thought that they had alloyed it with zirconium. I'm really going to have to check that out. Also, I might be confusing it with the KIWI series of engine that either predated NERVA or ran parallel to it. In any event, the core seemed to hold up better than other bits such as the nozzle throat and chamber various attachment thingies. Again, this was a very long time ago. I was back in something like grade 11 or 12 when I was looking into it. (I remember because my grade 12 chemistry teacher marked me "wrong" on a test question regarding uses of hydrogen and I put down "nuclear rocket fuel". He wanted things like "peanut butter". If it wasn't in the textbook, it didn't exist. So I brought in my notes on NERVA, plopped them down on his desk, and got the grade corrected when he managed to understand the articles 2 days later.) I could have just said "rocket fuel", but I had the NERVA research in my briefcase and didn't feel like copying a bunch of articles about the Apollo program or likewise. This was back when I had to drive 5 miles to the library and copy everything, including diagrams, by hand on notepaper.
But, I digress...
I've been a bit too busy with other stuff (and still am) to check out the link that you found to SNAP, but it seems from what you posted that it's far more advanced than it was back when I looked at it (same time as I did NERVA). I think that SNAP-10 and -20 were the only ones (at least declassified ones) at the time. I'll certainly try to read it soon, or at the very least download it so I can check it at my leisure in case the site folds or something. Thanks for pointing it out.
Actually, it couldn't hurt to try sending a message to Astronuc and ask his advice. He's a nuclear engineer, and actually quite friendly once you get past that beard.

 

[Astronuc]

So the reactor system could potentially operate at an even higher temperature than NERVA? I don't know much about nuclear reactors, but I'm thinking that if a failure scenario is that they are prone to "heat excursions", parts of the core could destroy itself if the operating temperature is high enough, while the immense flow of coolant in a NERVA-type engine would make this less likely.

That document is a rich source of information. Apologies if this post seems a bit disjointed. The SNAP 50, which was conceptually verified but not built, would use an UC (uranium carbide) or UN (uranium nitride) fast reactor and a potassium Rankine cycle.

Li would be used used as primary coolant, I guess due to its neutronic properties, Li-7 wouldn't become radioactive while K would.
(kelvinized and rounded numbers for convenience)

Specs:
Reactor output temp: 2000+ F, a.k.a 1366K
Turbine inlet temperatue: 1339K
Condensing temperature: 977-1033K
Radiator temperature: 977-1033K

Weird stuff is going on in that document, the scematic on page 90 in the pdf (page 83 in the scanned in text) seems to show a tertiary cooling system with NaK and lower radiator temperature. Different stages of the development of the concept? It is also later stated that both UC and UN was investigated and UN was found to be preferable.

I can't find any information on the LCRE, (lithium cooled reactor experiment), said to have operated at a temperature of 1903 C, or 2176 K (p 11/ 18 in the linked document).

They also mention the Brayton cycle, using an inert gas as secondary coolant, and the boiling-condensing of something in a "heat pipe" as tertiary coolant (the radiator).

As far as I've understood so far, the stress on the moving parts at high temperature does seem to be the "bottleneck", otherwise they'd just boil something hotter, like Li. Is it possible that a MHD-system could be used?

An interesting side note, perhaps a space based version of something akin to UHTREX could separate out the fission products from the primary coolant loop and simply send them off into space.

There are a number of considerations in developing a power system, especially one that pushes materials to their technical limits, i.e., temperature and stress. One has to consider material compatibility and strength, which usually diminish as temperatures increase.

Here is a site that discusses high temperature ceramics.
http://www.hexoloy.com/high-temperature-ceramics/creep-resistant-material

Some of the issues with NERVA included the high temperatures and high mass flow rate. The high coolant (hydrogen) flow rate meant substantial buffeting of the fuel elements. That would be less of a concern in a liquid metal core at lower flow rates.

However, in a reactor, one must consider the consequences of fission products (2 atoms per 1 U fission), the effects of neutron irradiation on structural materials, including atomic displacements and transmutation, and fuel-cladding chemical interaction.

More later.

 

[vemvare]

I'm currently reading up on materials physics, in order to better understand this. I haven't abandoned the thread...

 

[vemvare]

http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20110009914.pdf
Yet another paper.

This one is interesting, it has a rather low radiator temperature, and this is justified in that apparently the specific power (W/kg) of any heat engine, or in this case MHD system increases rather rapidly with a smaller temperature differential. Radiator mass quickly becomes the least of the problems, as generators, turbines and so on goes bulkier.

How can a plant with 12,89MWt and 2,76MWe be said to have a "total plant efficiency" of 55,2%?

If 10,13MW doesn't end up as electric power, then that must be radiated off, which would be 22056,7m^2 of radiator, yet apparently that doesn't blow the concept out of the water.

I'm missing something here... Also, in the diagrams it looks like 600K would have the lowest alpha, which if I understand it correctly in this context means weight-to-power ratio.

 

[Astronuc]

How can a plant with 12,89MWt and 2,76MWe be said to have a "total plant efficiency" of 55,2%?

If 10,13MW doesn't end up as electric power, then that must be radiated off, which would be 22056,7m^2 of radiator, yet apparently that doesn't blow the concept out of the water.

I'm missing something here... Also, in the diagrams it looks like 600K would have the lowest alpha, which if I understand it correctly in this context means weight-to-power ratio.

The efficiency doesn't look right given the net output and thermal input, but I'll have to read the paper to understand what they are talking about. Also, they are short on details of the NFR, and 1800 K is rather hot. I'd be curious about the reactor and fuel design. Fission products are rather mobile a those temperatures. I wonder if they plan on venting the Xe and Kr from the fuel.