Are nuclear engines really low thrust at sea level? Why?

[farmerben]

The subject of the thread is the question.  In KSP the NERV engines have terrible thrust at sea level.  Is this necessarily the case?  Why?

[magnemoe]

 Not really, that depend only on the nozzle, however I guess they have pretty low TWR and they need to use bulky hydrogen for best performance. You also has radiation issues, reactor is just a bit radioactive before started, afterward its get dirty. 
So its mostly about game balance. 

[Shpaget]

Wikipedia has some data on actual NERVA engine that was built and tested

• Diameter: 10.55 meters (34.6 ft)
• Length: 43.69 meters (143.3 ft)
• Mass empty: 34,019 kilograms (74,999 lb)
• Mass full: 178,321 kilograms (393,131 lb)
• Thrust (vacuum): 333.6 kN (75,000 lbf)
• ISP (vacuum): 850 seconds (8.3 km/s)
• ISP (sea level): 380 seconds (3.7 km/s)

ISP is not bad at all, but maybe a bit heavy.

[KSK]

Throwing some numbers out here for comparison.

The Timberwind 45 design was specced as:

Diameter: 13.94 ft (4.25 m)
Vacuum thrust: 99208 lbf (441.3 kN)
Sea level thrust: 88305 lbf (392.8 kN)
Vacuum specific impulse: 1000 s
Sea level specific impulse: 890 s
Engine mass: 3300 lb (1500 kg)
Thrust to Weight Ratio: 30
Burn time: 449 s
Propellants: Nuclear/LH2

That's apparently substantially higher performance than any NERVA NTR, although I'm not sure if it was ever anything more than a CAD file. The specs for the Timberwind 75 and 250 (which is a real beast) have better sea level thrust but both still have that T/W of 30. 

For comparison, the Merlin 1c has a sea level thrust of 420kN but a T/W of 96. Admittedly, that's a generous comparison, since the Merlin is supposed to be about the highest T/W kerolox engine going at the moment but even so: 30 vs 96 is quite a difference. Plus the Merlin isn't throwing out mildly radioactive exhaust.

As to why NTRs have relatively low thrust to weight, I'm guessing (and it really is a guess) that you can't have too high of a mass flow rate through them. Couple of reasons I can think of: 

1.  With an NTR your propellant is also your coolant and (for hydrogen at least) a moderator for the reactor. Throw too much propellant out the back and you cool your reactor down far enough that you start losing that lovely ISP that was the point of you using an NTR in the first place.

2. Physically throwing a lot of propellant through a solid reactor core isn't easy.

3. Related to 1, throwing too much coolant through your reactor gives you some really nasty (well nastier) thermal gradients to have to deal with inside your reactor.

If any actual engineers want to chip in, I'd be really interested to hear what you've got to say!

[Gargamel]

Sort of OT, as we're talking liquid fueled engines designed for space flight, but I watched this last night on the Nuclear air breathing engine:

 

[IncongruousGoat]

To address the question specifically, no, nuclear engines don't necessarily have to have vacuum-optimized nozzles. However, nuclear engines' low TWR makes them impractical for use at sea level, where TWR is far more important than specific impulse, and so were all built with large, high expansion ratio vacuum optimized nozzles.

[Kerbart]

5 hours ago, farmerben said:

The subject of the thread is the question.  In KSP the NERV engines have terrible thrust at sea level.  Is this necessarily the case?  Why?

A regular nozzle like the Nerva uses is either optimized for vacuum or for (some) atmospheric pressure.

Given that you're basically having an open-ended nuclear reactor that blows out propellant after being fed through the core one can guess the lack of enthusiasm for an atmospheric version. In addition, while having excellent I_sp, the thrust-to-weight ratio isn't that great, and in general, when at sea level, you want lots and lots of thrust. Even with a sea-level optimized nozzle it's still not going to be a very good engine for lift-off boosters.

[farmerben]

 

Here is an old school video about the NERVA engines.  There is a graphite moderator, and a rotating control rod design.   Presumably this means the nuclear reaction can be controlled at various levels of hydrogen throttle.  They say the engine operates at 4000 degrees Fahrenheit.  It can start and stop repeatedly.  I would assume the operating temperature is chosen based on the other materials (besides uranium and hydrogen) that make up the structure of the rocket engine.    

 

 

[farmerben]

At least in KSP, onion staging always beats bamboo staging.  If someone can explain the advantage of lifting an engine that is not firing, that would explain why nuclear sea level engines are not desirable.  

I'm also skeptical how radioactive the normal exhaust would be.  There are a few radical neutrons in the exhaust, but these are mostly thermal neutrons (having been moderated by graphite).According to wikipedia: neutrons from U235 fission have energy 2 MeV.  Moderators bring them to the thermal range of 1 eV.   By comparison a neutron bomb released neutrons at 14 MeV.  I suppose all the thermal neutrons released on Earth would be absorbed by something... making normal matter slightly more radioactive than normal.  I'm not suggesting this is environmentally benign, but probably way to low to be considered a weapon.  

 

[IncongruousGoat]

1 minute ago, farmerben said:

At least in KSP, onion staging always beats bamboo staging.  If someone can explain the advantage of lifting an engine that is not firing, that would explain why nuclear sea level engines are not desirable. 

There are three main reasons why rockets in real life use linear staging. The first is that, with linear staging, you can get high specific impulse in vacuum by using vacuum-optimized nozzles. As an example of this, consider Falcon 9. The Merlin 1D engines used on the first stage of Falcon 9 have a sea-level Isp of 282s, and a vacuum Isp of 311s. The vacuum-optimized Merlin 1D used on the second stage, on the other hand, has a vacuum specific impulse of 345s, which is much more efficient. In KSP, where a good half of your burn to orbit is done in the atmosphere, it's reasonable to use sea-level optimized engines for the whole flight. On Earth, however, only perhaps a sixth of the burn to orbit is done in the atmosphere, making vacuum efficiency important. The second reason is that rockets in real life tend to have small upper stages, which aren't really practical in an onion configuration for structural reasons. The third reason is that onion staging (a.k.a. asparagus staging) in KSP usually involves fuel crossfeed, which we haven't really worked out how to do in real life. It's much easier said than done.

Basically, onion staging is impractical for a bunch of reasons that don't matter too much or at all in KSP but are very important in real life.

[Mad Rocket Scientist]

Nuclear engines in KSP are weaker than those in real life for balance reasons. Any NTR beyond the earliest prototypes are spec'ed at overall better stats than KSP, although their higher mass/kn will probably mean that they will never be used as a lifter engine.

[AeroGav]

If you build one into a space plane however you don't need a thrust to weight ratio greater than one.     This massively changes the game,  in KSP at least.     I won a stock payload mass to orbit fraction challenge with a design that used no oxidizer     Here;s a slightly more usable airplane based off that entry - orange tank to orbit capable

 

https://kerbalx.com/AeroGav/Andromeda

[wumpus]

On 10/23/2018 at 9:19 PM, IncongruousGoat said:

Basically, onion staging is impractical for a bunch of reasons that don't matter too much or at all in KSP but are very important in real life.

The previous poster's description of onion staging "firing all engines during lift off" was how early rockets were staged at all and still used in Soyuz.  Of course, nobody has yet built a kerbal "magic fuel line", so no fuel is transferred from drop tanks to the sustainer engine, but real life fuel tanks are far more mass efficient than anything designed by kerbals.  Note that if you can get bamboo staging to work (in orbit or otherwise outside an atmosphere), then that is going to be optimal in KSP (you might have to use "Pe kicking" to get anywhere,  because it works best with low-TWR engines).  Again most of the reason it makes so much sense is thanks to the inefficiency of kerbal fuel tanks so don't expect to see it much IRL (the lack of "magic fuel lines" is a bigger problem).

On 10/23/2018 at 8:42 PM, farmerben said:

They say the engine operates at 4000 degrees Fahrenheit.  It can start and stop repeatedly.  I would assume the operating temperature is chosen based on the other materials (besides uranium and hydrogen) that make up the structure of the rocket engine.    

If you want the engine to start and stop repeatedly (which of course you will for any mission that can justify nuclear propulsion), then lowering the thrust makes the whole start/stop cycle wildly easier.  Nuclear engines have a relatively long cool down between adding control rods and the (secondary) nuclear reactions stopping.  Presumably some of this would be included in the delta-v of the burn, but you will still need to feed cooling hydrogen through the reactor while it cools down, leaving the end of the burn rather inefficient.

[Racescort666]

1 hour ago, wumpus said:

real life fuel tanks are far more mass efficient than anything designed by kerbals.

It's worth noting that the fuel tanks in KSP all have the same wet/dry ratio regardless of size. This isn't the case in real life as you get the square-cube law. Quick calculation says that the FL-T800 has an areal weight of 35 kg/m2 which is super high considering that the Space Shuttle ET is less than 20.

[IncongruousGoat]

3 hours ago, wumpus said:

The previous poster's description of onion staging "firing all engines during lift off" was how early rockets were staged at all and still used in Soyuz.

Not really accurate. The only early rockets to use a stage-and-a-half design were the original R-7 and the original SM-65 Atlas. Even then, both of those were rather quickly augmented by the addition of a third stage when they were put to use as orbital launch vehicles. Plenty of other early rockets, especially on the American side, used traditional staging. Thor-Able, Juno I/II, Vanguard, Titan I/II, and Scout to mention a few. The Shuttle also lit all its engines on the pad, but I think we can all agree that the Shuttle should not be taken as an example of how to do literally anything right.

Those two early ICBMs were only built the way they were because air-ignition of a liquid-fuel engine was unproven technology at the time. Later ICBMs and orbital vehicles would, almost without exception, use serial staging, or a mix of serial staging and radial boosters.

[farmerben]

8 hours ago, wumpus said:

 

If you want the engine to start and stop repeatedly (which of course you will for any mission that can justify nuclear propulsion), then lowering the thrust makes the whole start/stop cycle wildly easier.  Nuclear engines have a relatively long cool down between adding control rods and the (secondary) nuclear reactions stopping.  Presumably some of this would be included in the delta-v of the burn, but you will still need to feed cooling hydrogen through the reactor while it cools down, leaving the end of the burn rather inefficient.

Hmmm thats interesting.  I wonder if would need a long cool down, or if the shutdown would last on the order of 1 second?  

So the reactor is turned sub-critical with hydrogen flowing and the core temperature begins to drop.  This lowers the ISP.  The question is how much temperature change is needed for safe operation, and how much time this takes.  I'm guessing it only takes on the order of a 1% temp drop which happens in less than 1 second, but I could be wrong.  

[wumpus]

14 hours ago, farmerben said:

Hmmm thats interesting.  I wonder if would need a long cool down, or if the shutdown would last on the order of 1 second?  

So the reactor is turned sub-critical with hydrogen flowing and the core temperature begins to drop.  This lowers the ISP.  The question is how much temperature change is needed for safe operation, and how much time this takes.  I'm guessing it only takes on the order of a 1% temp drop which happens in less than 1 second, but I could be wrong.  

If you want to wait for 1% of normal operation (which seems wildly excessive, I'm sure that during the burn it radiates at least 1% into the rest of the spacecraft) you are likely waiting (and dumping 1% of your hydrogen from a normal burn) for an entire hour.  Cut it to 2% and you are dealing with dumping hydrogen (at twice the rate?) for 10 minutes.  You need to eat 4% if you want the shutoff to be in (10-60) seconds.

4% (or even 2%) of a slow burn for a Hohmann transfer is one thing, but it is another thing when you need sufficient thrust to leave a launch pad.  It even is a wildly huge problem for a Hohmann transfer.

From a little googling: https://www.quora.com/How-long-does-it-take-to-shut-down-a-nuclear-reactor

According to wiki:

Quote

Building on the KIWI series, the Phoebus series were much larger reactors. The first 1A test in June 1965 ran for over 10 minutes at 1090 MW, with an exhaust temperature of 2370 K. The B run in February 1967 improved this to 1500 MW for 30 minutes. The final 2A test in June 1968 ran for over 12 minutes at 4,000 MW, at the time the most powerful nuclear reactor ever built.

NERVA NRX (Nuclear Rocket Experimental), started testing in September 1964. The final engine in this series was the XE, designed with flight design hardware and fired in a downward position into a low-pressure chamber to simulate a vacuum. SNPO fired NERVA NRX/XE twenty-eight times in March 1968. The series all generated 1100 MW, and many of the tests concluded only when the test-stand ran out of hydrogen propellant. NERVA NRX/XE produced the baseline 75,000 lbf (334 kN) thrust that Marshall required in Mars mission plans.

The ISS has heatsinks that can handle 70kW of power, so they can handle %4 of 1.75MW reactor.  While I'd expect the reactor cooling fluid to be hotter than the ISS's (and thus more efficient over mass) you can begin to see the problem.  Either you build a ship with orders of magnitude larger heatsinks than the ISS, or you have to deal with cooldowns in terms of hours (leaking fuel the whole time), probably both.  For a simple comparison, the ISS heatsinks are about 1/10th the size (don't know the mass off the top of my head) the size of the solar panels.

There's a reason I suggest slagging and ejecting the entire reactor core after each burn, although I don't expect there is any way to make it "the cheap part" of a nuclear rocket engine.

[farmerben]

The reactor design of the NERVA is different in several important respects from a conventional power plant.  It uses a different control rod mechanism and higher enrichment fuel.  Still, lets consider the decay heat problem is roughly similar to what the people on quora are describing.  This implies that a nuclear engine requires a cooling system that can dissipate secondary decay heat about 5%-7%  of critical heat.  If we have that type of cooling system, then we can shut off the hydrogen flow within about one second, before the reactor temperature has dropped significantly.

[wumpus]

The issue is that you are taking the problem that doomed Fukushima (cooling water sufficient for normal operation stopped flowing at all) and putting it a vacuum.  While I couldn't find the mass of the ISS's heatsink, it does have about 2 tons of ammonia just to pump around the heat.  And 5-7% of NASA's NERVA would take about 3 orders of magnitude more cooling capacity of the ISS.  You'd need something like heat pumps to drive the radiators to their melting point, and then it would still be massive.

I still think you would "just" stage your reactor rods.  I'd expect that what's left would be pretty radioactive, but not bubbling away at anything like 5-7%.

[winged]

On 10/23/2018 at 10:40 PM, KSK said:

Throwing some numbers out here for comparison.

The Timberwind 45 design was specced as:

Diameter: 13.94 ft (4.25 m)
Vacuum thrust: 99208 lbf (441.3 kN)
Sea level thrust: 88305 lbf (392.8 kN)
Vacuum specific impulse: 1000 s
Sea level specific impulse: 890 s
Engine mass: 3300 lb (1500 kg)
Thrust to Weight Ratio: 30
Burn time: 449 s
Propellants: Nuclear/LH2

That TWR wasn't realistic if I understand correctly:

https://forum.nasaspaceflight.com/index.php?topic=39436.0

Quote

* Supposedly Timberwind was posited to be uniquely high-TWR as a result of Silicon Carbide construction materials.  NASA has done work on Tungsten Cermet materials that have apparently surpassed this, but I'm unclear exactly how realistic Timberwind was.

Quote

FYI Timberwind wasn't a normal reactor but a "pellet-bed" reactor and the supposed thrust-to-weight was calculated on a very low internal drag which never appeared in what testing was done.

Secondly ALL tested NTRs were re-start-able and this was demonstrated on multiple runs of the NERVA/KIWI designs. Specifically multiple re-starts, deep-throttling, and long run times were part of the design criteria. Timberwind was different kettle of fish with a set of different design criteria, pretty much none of which were met in reality.

 

[Brotoro]

On 11/1/2018 at 6:08 PM, farmerben said:

The reactor design of the NERVA is different in several important respects from a conventional power plant.  It uses a different control rod mechanism and higher enrichment fuel.  Still, lets consider the decay heat problem is roughly similar to what the people on quora are describing.  This implies that a nuclear engine requires a cooling system that can dissipate secondary decay heat about 5%-7%  of critical heat.  If we have that type of cooling system, then we can shut off the hydrogen flow within about one second, before the reactor temperature has dropped significantly.

The NERVA shutdown was supposed to take about 45 seconds. Afterward there would be a cooldown phase that would span many hours (or a few days) where there would be thrust pulses (something like a dozen or a hundred-something, depending on how long the main burn was) as liquid hydrogen was run through the engine to carry off heat. The thrust of these pulses would be accounted for when planning the orbital maneuver. The specific impulse during these pulses would be a measly 400 seconds.

 

Edited November 3, 2018 by Brotoro

[Hypercosmic]

is it possible to make a separate coolant loop that takes the heat from the core away to the radiator when there's no propellant flowing through?

I imagine that the coolant loop, other than permitting quicker engine shutdown, could also be used to heat the propellant itself removing the need for the propellant to touch the reactor, with some, probably significant loss in exhaust velocity as a trade-off.

Edited November 3, 2018 by Hypercosmic

[Ho Lam Kerman]

On 11/1/2018 at 1:09 AM, IncongruousGoat said:

the original SM-65 Atlas. Even then, both of those were rather quickly augmented by the addition of a third stage when they were put to use as orbital launch vehicles.

Wait- am I missing something out, or did Mercury Atlas use an upper stage?

[IncongruousGoat]

3 hours ago, Ho Lam Kerman said:

Wait- am I missing something out, or did Mercury Atlas use an upper stage?

No, you're not. However, Atlases flying unmanned payloads at the time of Project Mercury were augmented by one of several upper stages. Unless I'm mistaken, the only payloads to be launched into orbit by an unagumented Atlas were the SCORE satellites in 1958 and, of course, Mercury capsules.

[DDE]

On ‎11‎/‎3‎/‎2018 at 1:28 PM, Hypercosmic said:

is it possible to make a separate coolant loop that takes the heat from the core away to the radiator when there's no propellant flowing through?

Sure, but at that point system complexity and dead weight start to go out of control. A few minor additions and you can break out the nuclear-electric drive, with, what, 15 times the Isp?