The limits of solid-core nuclear thermal rocket technology

[DLR]

I am wondering how far out nuclear-thermal rocket technology may carry human explorers. If we set L2 as our point of departure, would it be possible to fly human missions out to Jupiter and return them in a reasonable amount of time, assuming in-situ propellant production in the Jovian system?

How would a relatively near-term Nerva-type reactor compare with more advanced designs, such as the Dumbo or the Timberwind reactor rocket?

[tnphysics]

DUMBO >> NERVA in every respect, except that some development is needed. However, NERVA had safety issues IIRC (loss of fission products)

[DLR]

What is the functional difference between Dumbo and Nerva? You mean Nerva loses fission products, which Dumbo does not? What were the specific impulse and thrust to weight specifications of proposed Dumbo engine designs?

[DarkenedOne]

DUMBO >> NERVA in every respect, except that some development is needed. However, NERVA had safety issues IIRC (loss of fission products)

How is NERVA unsafe from the loss of fission products? 

[tnphysics]

GH2 attack on the C fuel cladding

I think that ammonia has been insufficiently considered as a prop in designs hot enough to dissociate most of the hydrogen to atomic H. The Isp could still be very large, with much higher density.

[DLR]

Loss of fission products should be a non-issue in deep space.

[tnphysics]

It was regarding launch from Earth-a major use for a sufficiently powerful NTR, as SSTO RLV can be easy in some cases (e.g. LOX augmentation, or a Isp in the high 1000s, the latter possible if the H2 dissociates into H atoms).

[DarkenedOne]

It was regarding launch from Earth-a major use for a sufficiently powerful NTR, as SSTO RLV can be easy in some cases (e.g. LOX augmentation, or a Isp in the high 1000s, the latter possible if the H2 dissociates into H atoms).

Exactly.  In order to use nuclear rocket launch from Earth the reactor power has to be greater than that of even commercial reactors that power cities.  Getting those to be accepted would be a political nightmare.

However the high ISP does the greatest good for latter parts of the trip.  That is largely why LH2/LO2 is used a great deal for upper stages while kerosene is used for lower ones.

The best application for nuclear rocket would be in space.  There high ISP being in the 900s would allow us to go further will less fuel.  Also a very important feature of them is that they can use many different gases for propellant, which makes ISRU using them pretty easy.

[93143]

Dumbo had motor T/W values as high as 129 (for molybdenum elements) or 79 (for tungsten elements) in the original 1957 report.  A complete engine would require additional systems (turbopumps, nozzle, and so on), but my BotE calculations indicate that it could be feasible to reach or exceed an engine T/W of 30 or so, possibly as high as 60, even with the tungsten elements.

Isp for a tungsten Dumbo was between 845 and 965 s, according to my after-analysis of the report using CEA.  The range appears to be a margin for manufacturing irregularities.  Modern manufacturing techniques, and possibly materials science, should result in numbers tending towards the high end of that range.

I like the tungsten design because of the higher Isp, and because apparently a tungsten-uranium-oxide cermet with tungsten (alloy?) cladding supposedly solves the erosion issue in NERVA-types.

The post with my Isp and T/W number-jumbling in it is here:

http://forum.nasaspaceflight.com/index.php?topic=23228.msg665977#msg665977



For a Hohmann transfer to Jupiter you need about 8.8 km/s to leave and about 5.6 km/s to arrive.  Let's use that as a baseline.

...well, I threw together a simplified interplanetary trajectory analyzer this afternoon, but it apparently needs a ridiculous number of timesteps to achieve the accuracy I wanted, so...

Wikipedia says that LEO to L2 is 3.43 km/s, and the reverse is 0.33 km/s.  This implies that if you start at L2, using impulsive burns, you should be able to get 8800 m/s departure velocity with 330+3210=3540 m/s, if you do the second burn really close to Earth.  I figure about 75 m/s gravity loss for an initial vehicle T/W of 0.2, or 250 m/s for 0.1.  With Isp over 900 s, the 330 m/s to get from L2 to LEO is pretty small, but it does lighten the vehicle; let's say 70 m/s for 0.2 and 240 m/s for 0.1.  Total 3610 or 3780 m/s.

Say, those are pretty close to the results from my simulations...

Now, to do the Jupiter approach, I will take the 5.644 km/s Hohmann arrival delta-V, add on the energy from falling at Jupiter from infinity down to Callisto's orbit to obtain 12.9 km/s, subtract Callisto's orbital velocity to obtain 4.7 km/s, add Callisto's escape energy to obtain 5.29 km/s, and subtract the velocity of the spacecraft in LCO to obtain 3.57 km/s or so.  Gravity losses should be on the order of 10 m/s for the higher T/W and 50 m/s for the lower T/W, very roughly.

That gives us a total delta-V of about 7.2 km/s for the higher T/W, or 7.4 km/s for the lower T/W.

...

Using Kirk Sorensen's spreadsheet, with optimistic (for NERVA-type) engine specs of Isp = 925 s and T/W = 10, neglecting boiloff, I get a payload fraction of about 35% for both vehicle T/W cases.  Accepting an engine T/W of 7 results in payload fractions of about 34% for a vehicle T/W of 0.2 and a bit below 35% for 0.1.

Zero-boiloff may or may not be necessary.  If my calculations are correct, at 0.01% per day you'd only lose enough propellant during transit to drop the payload fraction to around 32-33%, which isn't that bad.  But at 0.1% per day, you lose 63% of your remaining propellant during transit...  Lower insolation during the bulk of the cruise phase, due to greater distance from the sun, should make it easier to mitigate boiloff.

For the trimodal TRITON thruster's H2-only specs of 911 s and T/W = 3.6, I get a payload fraction of 31%.  Reducing the vehicle T/W to 0.1 yields almost 33% despite the higher gravity losses, because the engine is so much lighter.

...

Dumbo is not actually much better.  According to my BotE calculations, an engine T/W of 30 is not unreasonable, and the Isp could possibly be higher with development.  Say 950 s.  At a vehicle T/W of 0.2, I get a payload fraction of about 37.5%.  Accepting 925 s Isp reduces this to about 36.5%.

The improvement from T/W isn't as large as one might have hoped, because the LH2 tankage ends up weighing so much...  Dumbo is only vastly superior to NERVA for high-thrust applications.

...

Timberwind is not practical.  The particle-bed concept suffers from local thermal instabilities and would likely melt down in flight.

...

Single-stage chemical does much worse than nuclear, of course, at this level of delta-V.  At a vehicle T/W of 0.2, an RL-10B-2-based stage gets you a payload fraction of just over 17%, meaning that the stage has to be more than twice as heavy to deliver the same payload as a TRITON stage, or almost three times as heavy to match the payload of a Dumbo stage.  Neglecting boiloff, of course.

If you want to do larger delta-V values for a high-energy trajectory, the advantage of NTR over chemical increases dramatically.

...

Please take the above numbers with a grain of salt, as the spreadsheet in question was designed as a simple first-cut tool rather than a detailed design generator.  Also, my delta-V numbers are fairly rough estimates...

Personally I prefer nuclear electric.  Preferably fusion electric, with Polywell or something similar.  Continuous-thrust trajectories with engine Isp of 30,000 s or better.  IIRC, 70,000 s with a sufficiently large Polywell was good for a 76-day transfer to Saturn...

[DLR]

Interesting. What about faster trajectories, say, 1 1/2 years to Callisto ... would Dumbo gain an advantage here because it reduces gravity losses through a faster completion of burns? I remember calculating that for a 450 day trip, the delta v would be around 12km/s at Earth and somewhat less at Jupiter (thanks to Callisto), but I'm not sure, lost the file.

I imagine a staged approach. A transfer vehicle, flanked by two (or four) boosters, departs from L2 ... the boosters (similar in size and mass to the nuclear shuttles envisioned by von Braun for his post-Saturn architecture) inject the payload into the Jupiter transfer orbit for a 1 1/2 year cruise and then return to L2 to for reuse. Prior to the departure of the crew on the fast trajectory, several autonomous payloads have been sent out to Callisto on Hohmann or near-Hohmann trajectories, including two (or four) boosters, as well as payload to set up a base on Callisto's surface and several reusable water-collecting landers, these landers supply an orbital electrolysis plant, which produces the propellant for the return journey.

The transfer vehicle then brakes into Callisto orbit, the crew departs for the surface, the vehicle is refueled, docked to the boosters and when a when a return window opens the boosters throw the transfer vehicle towards Earth and return to Callisto orbit for reuse.

I once did back of the envelope calculation showing that all of this may be feasible. But perhaps I had errors in my calculation or my assumptions were too optimistic. I think I used the Timberwind specifications for my engines.

I personally think that the 100MW to 200MW fission reactors you need for quick trips to Jupiter, let alone fusion reactors, are much farther away from maturity than nuclear thermal rocket engines. In addition, nuclear thermal rockets may double as electric power systems during the cruise phase and are capable of working with a number of propellants (depends on corrosion of course), which makes them great for "living off the land" at your destination.

[RanulfC]

Question for 93143;
When calculating T/W for DUMBO did you try applying LOX injection into the engine-bell at any point? (LANTRN-mode IIRC, though the same idea is behind the Thrust Augmentation Nozzle rocket engine)

Randy

[93143]

No.  I played around with LOX injection a bit in Sorensen's spreadsheet, and it looks like it decreases Isp too much to be useful, especially when the basic engine is that good already.

A LANTR Dumbo SSTO is probably a different story, but I haven't looked into that yet.  All I know is that for my all-rocket Polywell-powered SSTO, I needed LOX augmentation to even begin to close the design...

@DLR:  Please excuse my lack of a response.  I'm somewhat embroiled in an attempt to fix a critical design flaw in my Ph.D. code, which doesn't have much to do with orbital dynamics...

[Moe Grills]

It was regarding launch from Earth-a major use for a sufficiently powerful NTR, as SSTO RLV can be easy in some cases (e.g. LOX augmentation, or a Isp in the high 1000s, the latter possible if the H2 dissociates into H atoms).

   Isp in the high 1000's?

  To use a rule-of-thumb, to get a specific impulse of 1000 seconds
for hydrogen propellant in a vacuum, you need to heat the hydrogen to a temperature of around 3,200 K; hot enough to vaporize steel. I think only tungsten and carbon could survive such temperatures structurally, but perhaps I'm wrong and there are alloys and ceramics that can. 

[93143]

According to Figure D-3 in the 1957 Dumbo report, 3200 K gets you about 1000 s at 100 bar chamber pressure.  If you drop the pressure, dissociation allows the hydrogen to hold more energy at the same temperature, so the Isp goes up.  At 1 bar, 3200 K gets you close to 1200 s...

Unfortunately reducing the chamber pressure doesn't seem friendly to engine T/W, since the chamber/throat/nozzle assembly gets huge...  On the other hand, for a high-delta-V deep space stage, T/W is less important than Isp, and a Dumbo could potentially afford to trade off some of the former for more of the latter...

The Russians reportedly achieved 3100 K exhaust temperature running for about an hour on an experimental NTR core rig...

[DLR]

Is something like Dumbo an option again nowadays or are the people at NASA talking about NERVA when they say "we could develop an NTR in the future"?

[TyMoore]

I would just add, that going from a tungsten-uranium oxide cermet fuel elements to an alloy of uranium and tungsten, say 80W20U with 98% U-235 enrichment, clad with tungsten will increase the thermal conductivity of the fuel elements, reduce or completely eliminate hydrogen erosion, and allow operating temperatures in excess of that which would cause UO2 to melt.

I just haven't figured out what would be the best geometry for a fast spectrum core. Something about this being a Doctoral Thesis level project...

[RanulfC]

I would just add, that going from a tungsten-uranium oxide cermet fuel elements to an alloy of uranium and tungsten, say 80W20U with 98% U-235 enrichment, clad with tungsten will increase the thermal conductivity of the fuel elements, reduce or completely eliminate hydrogen erosion, and allow operating temperatures in excess of that which would cause UO2 to melt.

I just haven't figured out what would be the best geometry for a fast spectrum core. Something about this being a Doctoral Thesis level project...

What? You mean to say that doing a "Doctoral Thesis" just "for-fun" doesn't look good on a resume?

Randy

[Joris]

I would just add, that going from a tungsten-uranium oxide cermet fuel elements to an alloy of uranium and tungsten, say 80W20U with 98% U-235 enrichment, clad with tungsten will increase the thermal conductivity of the fuel elements, reduce or completely eliminate hydrogen erosion, and allow operating temperatures in excess of that which would cause UO2 to melt.

I just haven't figured out what would be the best geometry for a fast spectrum core. Something about this being a Doctoral Thesis level project...

Going with ZrO2+UO2 might be better actually, but tungsten cladding should have a positive effect on the lifespan of a NTR.

I am not sure about the Doctoral Thesis thing,  sounds like an high school project to me. 

[aftercolumbia]

As far as I can tell, hydride intrusion into power reactors is still a bit of a problem, with water and heavy water coolants at temperatures of about 600-650K (CANDU engineers were frustrated by a pressure tube breach in August 1983, the closest a CANDU reactor has ever gotten to a true accident; the hydride (er... "deuteride" if you want to get technical) cracking that caused it was a problem they thought they had licked!)  I can only imagine these would be a heck of a lot worse with NTRs shooting pure hydrogen through at 2500K or hotter.  I'd also recommend the CANDU-like callandria approach to any NTRs meant to operate near Earth or Mars, since it would make the escape of fission products into the exhaust stream almost impossible.  It's heavier and also unfavorable thermodynamics, so your specific thrust (T/W) and Isp would suffer somewhat.

If there is no hazard of having your exhaust hit a radiation-sensitive planet or station, then the escape of fission products is actually a good thing, since it would reduce your stage's residual mass.  The escape of unburnt fuel along with those fission products would be bad, and also, losing fission products takes away some of your delayed neutrons and will make the engine more unstable from a criticality perspective.  I don't think that delayed neutron issue is a big deal, though.  If your maneuvers have some give in the scheduling, and you lose criticality and have to shut down a reactor designed to vent its fission products, it'll blow out its volatile neutron poisons (the most significant being a radioisotope of xenon with a half life of a couple of days) in a few minutes, so you should be able to restart the engine less than an hour after such a scram.  I think the real question would be whether your fuel arrangement can survive the rather extreme thermal cycle.  Another, I believe rather minor, issue with a fission product venting reactor is that the plume expanding behind the ship would be radioactive, and sending some of that radiation back forward at the ship.  This would need to be accounted for in the shielding for the systems and crew, and is likely to add some mass; worth it if that shielding weighs less than the fission products you plan to send out the nozzle.

[tnphysics]

According to Figure D-3 in the 1957 Dumbo report, 3200 K gets you about 1000 s at 100 bar chamber pressure.  If you drop the pressure, dissociation allows the hydrogen to hold more energy at the same temperature, so the Isp goes up.  At 1 bar, 3200 K gets you close to 1200 s...

Unfortunately reducing the chamber pressure doesn't seem friendly to engine T/W, since the chamber/throat/nozzle assembly gets huge...  On the other hand, for a high-delta-V deep space stage, T/W is less important than Isp, and a Dumbo could potentially afford to trade off some of the former for more of the latter...

The Russians reportedly achieved 3100 K exhaust temperature running for about an hour on an experimental NTR core rig...

Oh i was assuming complete dissociation.