Single NTR vs. NTR cluster

Quaoar · 2018-11-08 08:52:12

Two ore more NTR engines are safer than one, because if one goes out the other(s) can continue the mission.
But there are many troubles with multiple nuclear rocket engines:

1) they are all off-centered, so in case of one engine out, the others need to be gimbaled more than what a single engine needs for TVC.
2) they are mounted on a bigger platform so they need a bigger and heavier shadow shield
3) during start-up, the thrust build-up is not perfectly symmetric in the engines, generating a torque that must be compensated by gimbaling or using RCS thruster (the same may happen during cool-down).
4) multiple engines irradiate each other, needing extra shielding material between them to avoid neutron interference and resulting a reduced engine life

On the contrary a single NTR spaceship has a cleaner, lighter and simpler design, but in case of engine out is doomed.

What I ask to the experts of this forum is if it is possible to build a single NTR engine with redundancy of turbopumps, control drums and other vital parts, and with independent elements, where a single element failure only results in a lower thrust without compromising the mission.

Thanks to all

Last edited by Quaoar (2018-11-08 09:01:08)

elderflower · 2018-11-08 09:59:04

A pebble bed reactor, perhaps?
Idle spare parts are parasitic mass so reducing the capacity of each of multiple units such as pumps would be the way to introduce a degree of functionality with one of them out of order. This will still carry a mass penalty: two lower capacity units will be more massive than a single one with the same total capacity. Also small machines are less efficient than bigger ones.

Quaoar · 2018-11-08 10:59:04

elderflower wrote:

A pebble bed reactor, perhaps?

Pebble bed are very interesting: 1000 s of Isp and T/W ratio up to 30, but I was speaking in general about any type of NTR, designed with independent element, that can be excluded in case of failure without compromising the whole rocket.

GW Johnson · 2018-11-08 11:35:14

A cluster of engines gives you engine-out capability. This is true for any kind of engine, not just nuclear thermal, of course. How many engines to cluster is driven by throttle-down capability of the engines, the effects of losing one out of however many there are, and the varying thrust requirements imposed by the mission design and propellant burn-off upon the propulsion.

During an engine-out, the asymmetric thrust geometry will act to throw the vehicle out of control. Fast-reacting gimbal capability is required to counter this in most designs, a capability far more stringent than in normal operation. This is because the engine thrust centerlines are mounted parallel at zero gimbal. This makes the nominal mounting cosine factor on thrust to be unity, for highest-possible thrust and impulse delivery. But that may be a false economy.

I would suggest a variation where thrust centerlines are canted to point through, or a tiny bit ahead, of vehicle center of gravity. There is no asymmetric thrust geometry problem at loss of one engine, greatly reducing the gimballing design requirements, and lightening-up that subsystem. The cosine factor is not unity, but generally within less than 1% of unity. At engine loss, only a tiny vehicle attitude correction (not a major thrust gimbal effort) realigns thrust vector with intended path.

I honestly don't know why this hasn't been done all along with multi-engine stages. It was done with shuttle, and successfully, until Challenger's unsurvivable event. And since, to the end of the program.

This approach makes really good sense for a multi-engine lander. You don't have much time to recover attitude and control during an engine out, in a landing scenario.

The proven NTR design is the one last tested during the Rover program as NERVA. Those folks had an improved design intended to be tested next, but never got the chance. That would be the best design to recreate and verify in test, then use. Isp ~ 900-1000 sec vs ~800-900 with NERVA, T/Weng ~ 5-6 vs 3-4 with NERVA. All still straight LH2 propellant.

None of these had shielding, and none were ever tested as a cluster. The stage itself was a shadow shield for whatever it pushed, taking advantage of 1/r-squared spreading for fast intensity reduction. The engines had a moderator about the reactor. The effects of adjacent particle sources upon moderator and reaction is a very good question for a nuclear engineer, which I am not.

GW

Last edited by GW Johnson (2018-11-08 11:58:39)

kbd512 · 2018-11-08 12:59:27

Hydrogen-impregnated stainless steel foams or simply foams with voids can dramatically lower the mass of shielding and research has shown them to be substantially more mass-efficient (more effective at blocking Gamma and Neutron radiation at the same time with a single material) than the prototypical High-Z (Gamma) / Low-Z (Neutron) material combinations, like DU (Depleted Uranium) or W and LiH, for example.

The use of particle beds, liquid fuels, or vapors would also boost Isp, as a function of the heat such materials can withstand before core material damage occurs. The particle and molten salt with suspended fuel particle (liquid) technologies are the most well-developed. Newer refractory ceramic metals like Hafnium-Carbide provide better containment options for higher heat levels required for higher specific impulse operation.

The Tri-Modal designs could provide primary electrical power, lower-thrust / higher-Isp operation, and higher-thrust / lower-Isp operation for breaking orbit by using supplementary LOX to create an "afterburner" effect where very hot H2 is mixed with O2 to provide additional thrust.

GW Johnson · 2018-11-08 13:09:19

My point was that none of those really great ideas has been tested in an actual engine firing. The only thing that ever was, was the now-ancient NERVA core design. You cannot rely on the untested idea, no matter how attractive. That great potential demands testing. None as occurred since the last firing at Jackass Flats, NV, in 1974. The problem is not the technology, it is the politics.

GW

kbd512 · 2018-11-08 13:57:47

GW,

Since there's been no testing of NTR's since the 1970's, by that logic this engine will simply never exist. NASA's not really interested because it costs real money and takes a real commitment to development of the technology. And yes, I'd agree about the politics being the primary problem.

Quaoar · 2018-11-08 14:49:35

GW Johnson wrote:

I would suggest a variation where thrust centerlines are canted to point through, or a tiny bit ahead, of vehicle center of gravity. There is no asymmetric thrust geometry problem at loss of one engine, greatly reducing the gimballing design requirements, and lightening-up that subsystem. The cosine factor is not unity, but generally within less than 1% of unity. At engine loss, only a tiny vehicle attitude correction (not a major thrust gimbal effort) realigns thrust vector with intended path.

That's nice, but as the propellant tanks become empty the center of mass move backward, so we have to increase the canting angle: I refer to a modular "star-train" like yours, that will be the protagonist of my next novel. She will mount gas-core NTR because mission deltaV is about 20-22 km/s.
So what do you suggest: one or two engines?

GW Johnson wrote:

I honestly don't know why this hasn't been done all along with multi-engine stages. It was done with shuttle, and successfully, until Challenger's unsurvivable event. And since, to the end of the program.
This approach makes really good sense for a multi-engine lander. You don't have much time to recover attitude and control during an engine out, in a landing scenario.
The proven NTR design is the one last tested during the Rover program as NERVA. Those folks had an improved design intended to be tested next, but never got the chance. That would be the best design to recreate and verify in test, then use. Isp ~ 900-1000 sec vs ~800-900 with NERVA, T/Weng ~ 5-6 vs 3-4 with NERVA. All still straight LH2 propellant.
None of these had shielding, and none were ever tested as a cluster. The stage itself was a shadow shield for whatever it pushed, taking advantage of 1/r-squared spreading for fast intensity reduction. The engines had a moderator about the reactor. The effects of adjacent particle sources upon moderator and reaction is a very good question for a nuclear engineer, which I am not.
GW

In this picture of the NTR-Copernicus the extra shields between the engines are clearly visible.

Quaoar · 2018-11-08 15:02:10

kbd512 wrote:

Hydrogen-impregnated stainless steel foams or simply foams with voids can dramatically lower the mass of shielding and research has shown them to be substantially more mass-efficient (more effective at blocking Gamma and Neutron radiation at the same time with a single material) than the prototypical High-Z (Gamma) / Low-Z (Neutron) material combinations, like DU (Depleted Uranium) or W and LiH, for example.

That's very interesting: I found this article about the topic:

https://core.ac.uk/download/pdf/82628810.pdf

Thanks

SpaceNut · 2018-11-08 20:38:14

I am wonder what the issues are when using multiple engines as to why we should not right size individual ractors to power each engine rather than one brute nuclear power source?

kbd512 · 2018-11-08 21:44:10

SpaceNut,

It comes down to reactor design. You're talking about designing an entirely new reactor and rocket engine. In our discussion of how reactor thermal power output scales with core volume, the relationship is not at all linear and the mechanisms required to control reactivity do not behave in the same way. The fundamental principles of operation don't change, but there aren't sufficiently detailed models available. Nowhere else are GW-class reactors the size of a small industrial air compressor tank used. The turbo machinery required to push Hydrogen through the core would have to change as well. We're talking about designing and testing an entirely new rocket engine at that point. The NERVA program engines produced thrust outputs that roughly corresponded with the thrust outputs of the LOX/LH2 powered RL-10 (25,000lbs) and J-2 (250,000lbs) engines. That made sense, given the notion that the NTR engines were intended to be as close to higher-Isp like-kind replacements as available technology would permit.

elderflower · 2018-11-15 03:47:06

It occurs to me that any engine involving a nuclear reactor is going to have to deal with post operation heating due to short half life isotopes continuing to break down after the reactor is shut down. This will probably require a small flow of hydrogen through the core. The thrust when it is dumped can be balanced by splitting the flow into two opposite streams, but there will still be a negative effect on overall Isp. Reactors don't go from high power to no power like chemical reactions.

Quaoar · 2018-11-15 03:58:28

elderflower wrote:

It occurs to me that any engine involving a nuclear reactor is going to have to deal with post operation heating due to short half life isotopes continuing to break down after the reactor is shut down. This will probably require a small flow of hydrogen through the core. The thrust when it is dumped can be balanced by splitting the flow into two opposite streams, but there will still be a negative effect on overall Isp. Reactors don't go from high power to no power like chemical reactions.

Correct. Every NTR needs a cool-down hydrogen flow before the shut-down: during this operation the Isp gradually decrees from 900 to 300 seconds, so there is a slight reduction of the whole system Isp. The burn time is calculated in a way to include the cool-down thrust.
After the cool-down, the residual decay heat is dissipated by the little radiator of the electric generator (the black truncated cone in the tail of the spaceship posted above), if the rocket is bimodal.

New Mars Forums

Announcement

#1 2018-11-08 08:52:12

Single NTR vs. NTR cluster

#2 2018-11-08 09:59:04

Re: Single NTR vs. NTR cluster

#3 2018-11-08 10:59:04

Re: Single NTR vs. NTR cluster

#4 2018-11-08 11:35:14

Re: Single NTR vs. NTR cluster

#5 2018-11-08 12:59:27

Re: Single NTR vs. NTR cluster

#6 2018-11-08 13:09:19

Re: Single NTR vs. NTR cluster

#7 2018-11-08 13:57:47

Re: Single NTR vs. NTR cluster

#8 2018-11-08 14:49:35

Re: Single NTR vs. NTR cluster

#9 2018-11-08 15:02:10

Re: Single NTR vs. NTR cluster

#10 2018-11-08 20:38:14

Re: Single NTR vs. NTR cluster

#11 2018-11-08 21:44:10

Re: Single NTR vs. NTR cluster

#12 2018-11-15 03:47:06

Re: Single NTR vs. NTR cluster

#13 2018-11-15 03:58:28

Re: Single NTR vs. NTR cluster

Board footer