Hacker News Reader

Comment by robonerd | Hacker News Reader

The PDFs here have a lot more information: https://sam.gov/opp/af490b568d2a438498afa1e80bce63e5/view

A few takeaways; they intend for such an engine to eventually support long duration human spaceflight (going to Mars.) The propellant for the NTR engine to be liquid hydrogen. One of the problems DARPA anticipates with using such an engine for such a mission is needing to store liquid hydrogen longer than the present state of the art.

The PDF doesn't seem to mention it, but I think the Advanced Cryogenic Evolved Stage (ACES) is probably relevant to this project. Does anybody know what kind of duration they expect to get from ACES? I'm not sure but I think it's weeks, not months.

View on HN

ACES is effectively dead, although a lot of its ideas ended up getting into Vulcan's upper stage. Long duration in its context means days. It was intended to do something similar to what SpaceX is doing with Starship, using the boiloff gas to pressurize the tanks.

The problem with longer duration storage of hydrogen is that there really isn't any option besides going with a denser or thicker material, while modern rocket wall thicknesses are measured in millimeters of lightweight metals or composites.

However, the convenient thing about NTR is it should be a lot easier to switch to something less prone to seeping through everything. It would be a matter of weighing the losses from needing a heavier tank against the losses from using heavier propellant.

Long term hydrogen storage isn't that bad with the proper architecture. You need a cryocooler which can be powered by the nuke, and thermal shielding for the tank which in vacuum is thin film and of minuscule weight.

Hydrogen leakage and structural embrittlement are overblown, i.e. the Space Shuttle tank is one of the most mass efficient architectures in history and it was full of liquid hydrogen. Terrestrially, you can buy a Toyota hydrogen car today. Materials matter, but people act like the thing needs to be made of 4" plate and will fall apart if you look at it. Scaling helps here too, as volume increases to the third power while wall area increases to the second.

The thing will, if there is any sense in the architecture, be assembled in orbit so gossamer heat shields and the like won't be a problem, nor will an extended assembly program that makes with a separately launched nuclear reactor.

For ISRU Mars return, water is incredibly abundant and there's no concern with "wasting" residual oxygen. For lunar applications, water may be scarce but oxygen is abundant in regolith.

You can't beat hydrogen as a fuel. As the lightest molecule, you get the highest exhaust velocity for the least energy input.

In vacuum, the required insulation is cheap and easy: more layers of the famous metallized crinkled plastic foil. The stuff that (with gold-colored metallization) is an iconic part of "the" satellite/space probe design.

The hard part about that insulation is that on earth, you need to sustain a vacuum in the annular space while overall being light due to the LH2 itself being light. Ideas would be to get tension fibers bridging that annular space, the inner tank with the LH2 being slightly pressurized, and thus the outer wall being kept from large-scale buckling (and small-scale buckling is cheap to reinforce for with an isogrid (triangle honeycomb) or other similar reinforcement structure on the outside of it). But in space, the outer wall isn't needed, because space is already a vacuum.

Would that suggest a staged approach where the long-range vehicle is fueled up in orbit?

A pure vacuum shuttle should be fine with getting launched and then wrapped with the insulation afterwards. Launching it empty is likely not really easier than launching it full, due to LH2's low density. But that's not important.

That seems to be why GP suggested wrapping it after reaching orbit.

> pace Shuttle tank is one of the most mass efficient architectures in history and it was full of liquid hydrogen.

That being said, it didn't have to last very long while filled with LH2/LOX; a few hours at most prior to launch, and a few minutes during launch. They were never reused, unlike the orbiter and SRB segments.

More's the pity. They actually used extra fuel to keep it from joining the shuttle in orbit. That much raw aluminum pressure vessel in orbit could have been so useful!

> You can't beat hydrogen as a fuel. As the lightest molecule, you get the highest exhaust velocity for the least energy input.

You can't beat it in terms of exhaust velocity, but you can often definitely beat it in terms of whole-system performance.

> You can't beat hydrogen as a fuel. As the lightest molecule, you get the highest exhaust velocity for the least energy input.

This is probably right, but the way you said it made me wonder. Would it be possible to strip electrons from atoms, then use just the electrons as propellant? Or would the ensuing static charge of the spaceship render this infeasible? I imagine it'd pull in electrons from all around itself, but I don't know how the numbers come out.

- "Or would the ensuing static charge of the spaceship render this infeasible?"

Back-of-the-envelope math says a large spaceship will reach 100 kV potential at a charge imbalance of around 1e15 electrons (total mass: 1e-15 kg). So yeah, completely unfeasible.

(It's asking the wrong question though. Electric thrusters aren't thermal systems, and aren't limited by molecular weight as severely as thermal engines are. You can get stupidly high Isp (>200 km/s) out of heavy ions, just by raising the voltage).

https://en.wikipedia.org/wiki/Ion_thruster#Comparisons

And, in fact, you WANT heavy ions, as the thrust/area of an ion engine at a given exhaust velocity scales as the square of the ion mass/charge ratio. The thrust in an ion engine is limited by space charge (where the charge of the ions between the accelerating grids becomes similar to the charge on the grids). Using heavier ions also reduces the ionization energy/mass. There's been work on using molecules or small droplets ("colloidal thrusters") to get even higher mass/charge, but you need to totally avoid generation of fragments with low mass/charge as they will dominate the current.

Funny that you mention this, Ion thrusters do exist. They are a thing but with very limited uses cases. They still need a kind of propellant gas like Xenon or Krypton that gets used.

https://en.wikipedia.org/wiki/Ion_thruster

If you strip the electrons off some atoms and use just the electrons[0] as reaction mass, you will eventually get a large enough electric charge you can no longer throw the electrons away from you. Electric forces behave similarity to gravity, so while it wouldn’t normally be phrased like this, you could say your engine exhaust will eventually no longer have escape velocity from your ship.

For this reason, ion drives do things to neutralise the net charge.

(If you meant using them as a power source rather than reaction mass, it’s technically possible but that’s called a capacitor and they have very low energy density).

[0] or, by symmetry, just the nucleus.

You didn't address this part of the parent comment:

> I imagine it'd pull in electrons from all around itself, but I don't know how the numbers come out.

I never thought of it before but it seems like that should work. "Space" is actually a neutral plasma, right, so it should be full of free electrons. Those should neutralize the ship before any significant charge builds up. It seems like you should be able to use space itself (or more accurately the interplanetary medium) as a massive ground plane to complete the circuit for the charged exhaust beam.

Space is pretty empty. From what I can find, the interplanetary medium is around 5 particles/cm^3 compared to the exhaust from the ion thruster which seems to be around 10^6 particles/cm^3 and disperses to 10^4 particles/cm^3 further away.

Source: https://trs.jpl.nasa.gov/handle/2014/15643

"As the charge-exchange plasma density near spacecraft is at least about three orders of magnitude larger than the solar wind plasma density, the plasma environment of DS1 spacecraft is completely dominated by the charge-exchange plasma in the plume."

I'm assuming you're asking about ejecting the electrons and remaining positively-charge hydrogen ions separately, since keeping the hydrogen around would be a waste of mass.

I'm no expert at space propulsion, but I think this would have a few issues:

- Hydrogen has a pretty high ionization energy, even higher than Xenon

- As you said, static charge buildup

- Momentum = mass * velocity. Electrons have 1/1836 the mass of a proton, so for the same momentum you need a much higher velocity

- Imparting velocity is harder for low-mass particles because they tend to zoom off very quickly if not contained

Or just photons. That's what the genie gives you when you ask for a torch drive—should've been more specific.

>Scaling helps here too, as volume increases to the third power while wall area increases to the second.

Not really. As surface increases the wall tearing force at the given pressure is increases too, so you have to increase the wall thickness, and thus the mass of the tank also grows close to the third power.

The scaling benefit is that you save insulation.

Apparently Bruno has been talking about two orders of magnitude improvement to Centaur V's duration. Seems far fetched to me, but I think months of duration would be necessary to make this engine worthwhile (the PDF is talking about the value of this engine for getting astronauts home from Mars quickly in emergencies; that would only be possible with months of duration at least I think.)

DARPA says they're expecting designs using liquid hydrogen, and as far as I understand liquid hydrogen would be the most efficient propellant for an NTR. What might the best storable alternative be?

From what I understand, Bruno didn't say that Centaur V has those two orders of magnitude improvements, rather that they're aiming to push improvements of that level over the next few years. That said, I don't think it's too far fetched, assuming that the long duration version is separate from the regular version (ie it can be heavier to support denser tanks).

Liquid Hydrogen would be most efficient in a pure physics sense, but due to the mass tradeoffs with storage tech, there may be other propellants that are comparable in a practical sense. I'm not informed enough on the matter to say exactly which would be better, but for a somewhat comparable point of reference, Hydrogen+Oxygen is the most efficient propellant for chemical rockets but when accounting for the special tanks needed for storing hydrogen, methane can achieve pretty comparable performance due to being perfectly fine in a thin-walled stainless steel tank.

I don't see how an NTR helps you in any way to get to Mars or back. Heavy engine, voluminous tanks (~70 kg/m³), criminally wasted ISRU material (you have to throw away 88.9% of the water that you mine on site, whereas a hydrolox or methalox system uses almost all of it and the methalox system can even mix it with considerable amount of CO₂ for better system performance). The performance figures for such a system will be terrible. At best a LANTR (not just an NTR) might be somewhat useful for cislunar uses. For Mars flights not even LANTR may be useful.

I'm skeptical too, but DARPA is saying the DRACO program is for getting to/from Mars quickly:

> The DRACO program intends to develop novel nuclear thermal propulsion (NTP) technology to enable time-critical missions over vast distances in cislunar space. Unlike propulsion technologies in use today, NTP can achieve high thrust-to-weights similar to chemical propulsion but with two to five times the efficiency. This enables NTP systems to be both faster and smaller than electric and chemical systems, respectively. The propulsive capabilities afforded by NTP will enable the United States to maintain its interests in space, and to expand possibilities for the National Aeronautics and Space Administration (NASA)’s long-duration human spaceflight missions (i.e., to Mars). Because of the ability to transit space faster than other propulsion systems, the NTR engine can return astronauts to Earth much faster in case of an emergency and similarly ensure reduction of overall trip time and exposure to deleterious impacts to astronaut health which come with long-term spaceflight.

> but with two to five times the efficiency

I suspect from the number that they're talking purely about Isp. Once one performs a whole system analysis, it's much less rosy for (non-LA)NTR.

I agree. Furthermore, besides the mention of Mars they're also talking about cislunar space in that same paragraph, but chemical propulsion seems sufficient in cislunar space. It's only takes a days to return from the moon with chemical propulsion, which proved sufficient in the past.

LANTR would improve performance of lunar landers/cislunar shuttles, especially for variable specific impulse which is what LANTR could plausibly do without much trouble -- start with high oxygen flow for high thrust and high propellant mixture density, decrease oxygen flow later in flight for higher terminal Isp. This brings you the performance of a multi-stage vehicle without staging, and LANTR can even with high oxygen flow deliver Isp significantly higher than what hydrolox has, with propellant density several times higher than what pure-hydrogen NTR gives you.

I've thought about trying to optimize the performance of such a variable Isp vehicle, but it requires calculus of variations skills that I'm lacking at the moment. I guess I need to take a look at that. But there's a decent chance that with a such a vehicle, you could move from the "we need to mine ice on the Moon" to the "we just need to extract oxygen from lunar soil; we can bring hydrogen from LEO" territory, which would be a win for lunar flights (for example you wouldn't be limited to polar region bases where you'd need to mine water to get back home).

It's not for getting to/from Mars quickly. It's for giving money to congressional districts quickly.

This is more the sort of engine you develop if you're going for an Apollo style mission where there's a mother craft that goes into orbit and a separate lander goes down to the surface. A NTR's poor TWR compared to conventional combustion rockets means it would be a bad ascent stage.

I wouldn't assume the plan relies on ISRU at all but if it do having to carry the resulting hydrogen up to orbit on the ascent stage will be a big limiting factor so not keeping the oxygen isn't so large a flaw. And if you're carrying the fuel to orbit on another rocket you want to get as high an ISP as you can manage with what you bring up.

All of which isn't to say this would be a good plan. I've drunk the SpaceX koolaid on the topic. But if it's a bad plan at least it isn't a stupid one and there are reasons behind things.

Yeah, I did notice that the original NTR plans arose from the wish to upgrade Saturn V with its limited "throw weight" at third stage separation (http://www.astronautix.com/s/saturnc-5n.html). It doesn't seem to make a lot of sense to design a propulsion unit for a sixty year old mission architecture today, though.

Yeah, the SpaceX Kool-aid, at least as regards Mars colonization, is about as lethal as the Jonestown variety.

There will be no Mars colonization on Starships, whatever Elon says. Starship is just wholly inadequate to the task. Neither will there be any sub-orbital passenger or freight service on it.

Starship should be adequate for lofting lots of Starlinks, for getting to the moon, and for boosting just amazingly well-equipped 100t outer solar system probes and telescopes. It might suffice for a quick visit to Mars with a half-dozen crew. (BTW, 9, not 6 months, each way.)

Probably the only way to make even that work would be to send two ships strung on a cable, nose to nose, spun for centrifugal gee force, so they could still walk when they got there. Maybe the second ship carries hydrogen (as ammonia?) to bond to ISR carbon to come home on. And solar panels, to crack the carbon.

But the first thing any attempted colonist would transmit back is "Can I please come home?"

A Mars colony will involve living in small metal boxes, surrounded by danger, with ruinous prices on things, if they are available at all.

We can get the same experience today, without going to Mars, in places called "prisons".

I think, in the medium term, all Musk is actually aiming at, is a crewed research station on Mars, with a few dozen people living in it. Rather similar to what we already have in Antarctica.

People will sign up to go. First person to step foot on Mars gets their name in the history books, next to Neil Armstrong. The rest get to join a very elite club. I suppose prison is kind of a club too, but nothing elite about it.

And Musk will call it a "colony"–aspirational naming. And maybe, one day, in centuries to come, it will actually evolve into one. I don't think Musk has really thought a lot about how to get from the "colony-in-name-only crewed research station" to a genuine colony – that's too many steps ahead. He just trusts he'll work it out when he gets there, or if he doesn't live that long, somebody else will.

"Aspirational naming" is quite a curious synonym for lying.

And to say that Musk has not thought much about X, for any X, is quite an understatement. The closer you look at anything he says, the less evidence of thought you can find. Today is a golden age for glib grifters.

Well, look at SpaceX: he founded it, he remains its CEO&CTO, and in 10 years it has gone from 0% market share to over 50% global market share–which was achieved, not through anticompetitive subterfuge, but simply by building a substantially better product at a substantially lower cost (and whatever government subsidies were involved, were made available in even greater amounts to competing companies which failed to leverage them into the same market success). Obviously he must have some capacity for intelligent thought to be able to pull that off. Of course, he employs many brilliant engineers, without whom none of that would have been possible–but, as founder/CEO, he created and sustained the corporate environment which made it possible for them to achieve that.

Yet, he does literally none of the work, and every unscripted public statement shows he understands nothing of the technical details beyond what he has learned to parrot.

He did not found PayPal, Tesla, or Neuralink, although he has often claimed to. Hyperloop is 100% grift.

> Yet, he does literally none of the work

This sounds like labor-theory-of-value BS. Management and leadership is work, and it's essential.

Compare SpaceX to Blue Origin - while SpaceX has succeeded in conquering over 50% of the orbital launch market, Blue Origin still hasn’t made it to orbit - nor has ULA’s new rocket using Blue’s engines. Is that the fault of Blue’s engineers? I don’t think that’s fair - some of them are just as brilliant as SpaceX’s. The real blame, I think, is at the executive level. Bezos has never invested anywhere near as much of his time and energy and personal wealth into Blue as Musk has invested into SpaceX. And Musk has made much better choices of executive leadership (Gwynne Shotwell vs Bob Smith). That’s just one example of the massive difference the ability and commitment of a founder can make to the success of a business. (You’d think on a forum owned by a Silicon Valley VC firm, that point would be uncontested and accepted as obviously true.)

Shotwell anyway has to know that the whole Mars colonization shtick is total BS.

He has been lucky in some of his hires.

I have perhaps naive expectations about somebody styling himself Chief Technical Officer. I guess Chief Grift Officer would be too revealing.

CTOs aren’t always super-technical-and even those who are, while the CTO of a small startup might realistically have an expert-level understanding of all the business’s core technologies, that is no longer a realistic standard when talking about a multi-billion dollar firm with a highly complex or diverse tech stack. Arguably, one of the most important tasks for a CTO, is to be able to tell the difference between good engineering executives and bad ones. And, judged by that standard, Musk actually has done a very good job as SpaceX CTO, much better than many of its major competitors. Doing that requires understanding the technology well-enough to distinguish engineers and engineering leaders who really understand it from those who are just pretending to do so-and I think it is obvious Musk does understand the technologies at SpaceX (and Tesla too) well-enough to successfully make that distinction. People seem to be holding him to an unrealistic standard, which I doubt they’d actually apply to a CTO who wasn’t named Elon Musk.

I would say he gets a free pass nobody else does. If the CTO of Intel spouted things as idiotic as he does routinely, they would have to resign.

What about Larry Ellison, CTO of Oracle? Frankly I think Larry Ellison could say any crazy thing he liked, and no one would really care, and he'd stay CTO and chair of Oracle's board. Because C-suite executives get a "free pass" all the time–especially when they combine their C-suite role with a substantial ownership interest in the company (true of both Ellison and Musk). But most C-suite execs, the average person has never heard of them, and so they don't care what they say. Whereas, Musk is a controversial celebrity, so people judge him by rather different standards than the thousands of other near-anonymous CEOs, CTOs and billionaires in the world.

> you have to throw away 88.9% of the water that you mine on site

Nobody cares about that. People care only about the end results, not about the efficiency (or inefficiency) of the intermediate steps. As long as you can travel to Mars and back in half the time it takes with a chemical rocket, exactly zero people will shed tears for all the oxygen wasted after splitting water on Mars.

Besides that, chances are you will be able to find uses for the oxygen you produce. I don't need to remind you that people breathe oxygen, astronauts included.

> People care only about the end results, not about the efficiency (or inefficiency) of the intermediate steps.

Look up for the guy claiming hydrogen is best and most efficient, even though performance per dollar is poor.

Speaking of performance per dollar, what does the nuclear fuel for an NTR cost?

The NERVA engine [1], [2] used 60kg of 92.5% enriched Uranium.

The online nuclear fuel cost calculator [3] shows that at current Uranium market prices, the cost for 1kg of such highly enriched Uraniums is about $50k, so the whole engine core would come at about $3 MM.

This engine had a weight of about 2.5 metric tons, a thrust of 75 kN and a specific impulse of 860s.

For comparison, the weight of a SpaceX Raptor engine is about 1.5 tons, it has a thrust of 1.8 MN (24 times higher than Nerva) and a specific impulse of 360 s (2.4 times lower than Nerva's).

[1] https://en.wikipedia.org/wiki/NERVA

[2] http://www.astronautix.com/n/nervaalphaengine.html

[3] https://www.uxc.com/p/tools/FuelCalculator.aspx

Seems like you could probably get more efficient by using a nuclear reactor to power an ion drive? Also wouldn't need to cool fuel down to cryogenic temperatures.

At 1 AU from the Sun, and possibly all the way to Mars, advanced photovoltaics may very well be better than a nuclear reactor for powering ion engines: It has very high system-level power/weight ratio (in lab around 300 W/kg, currently in operation around 150-200 W/kg), possibly could even power an ion engine without heavy power conditioning equipment ("direct drive electric thruster"), and also scales down for smaller probes. So for a trip to Pluto, a reactor would be useful, for a trip to Mars, it's hardly necessary.

Ion drives don't scale up in thrust fast enough for it to be worthwhile for manned missions.

I think that's basically what the VASIMIR engine is.

[1] https://en.wikipedia.org/wiki/Variable_Specific_Impulse_Magn...

For a trip to Mars the time it takes an ion drive rocket to reach cruising speed isn't negligable compared to the overall flight time. And missing out on the Oberth effect is fairly significant. If this were a flight to, say, Jupiter though electric drives all the way.

I don’t know that voluminous tanks and heavy engines are necessarily a problem for something that’s designed to permanently live in space - the tanks can essentially just be onion-layered gasbags, and could be km3 in volume if you wanted. As to fuel - don’t get it from heavy bodies. Mine asteroids, minor moons, whatever.

> I don’t know that voluminous tanks and heavy engines are necessarily a problem for something that’s designed to permanently live in space - the tanks can essentially just be onion-layered gasbags, and could be km3 in volume if you wanted.

It's the opportunity cost. At low and moderate speeds (we're talking delta-Vs of 10 km/s and less), the same tankage simply gives you higher performance with chemical propulsion, so for no size of tankage may it actually be advantageous to use an NTR instead of a chemical engine. Only at extreme delta V levels do NTRs actually get better performance, but that's not a mission-to-Mars territory. LANTRs could possibly lower the crossover point, especially with variable Isp, but properly estimating how much requires calculus of variations, as I already said elsewhere.

> As to fuel - don’t get it from heavy bodies. Mine asteroids, minor moons, whatever.

Same issue. Your supply may be limited and/or require effort to extract. NTRs throw oxygen away; hydrolox and methalox engines use it for propulsion. For every tonne of water extracted, you'll go MUCH further if you go chemical, or at least with LANTR instead of NTR.

>Only at extreme delta V levels do NTRs actually get better performance, but that's not a mission-to-Mars territory

Why is that not mission-to-Mars territory? You can shave months off the transit time with >15km/s delta-v.

1) The travel time benefits are degressive owing to increasingly eccentric heliocentric trajectories - the changes in trajectory length get smaller and smaller as your velocity vector stops being colinear with the planet's orbit upon intercept, so you don't really save a lot of additional time. (But you get the most benefits with even small increases above Hohmann transfer speed.)

2) Intercept velocities, on the other hand, are progressive -- pretty much for the same reason, combined with Pythagoras' theorem. At one point you stop being able to aerocapture, even with exerting downward lift in Martian atmosphere to prolong the braking phase.

Owing to these two things, I'm not quite sure that propelling yourself from LEO to Mars at 15 km/s would be a good idea, unless you intend to crash into the planet.

That's not quite right.

Using a Hohmann transfer orbit [1], you get from Earth to Mars in about 9 months.

Using an Aldrin Mars cycler [2], you can get in as little as 75 days. Of course, the Aldrin Mars cycler requires more delta-v, but that's the point, if you have more delta-v you get there sooner.

[1] https://en.wikipedia.org/wiki/Hohmann_transfer_orbit

[2] https://en.wikipedia.org/wiki/Mars_cycler

I am assuming that you can spend the delta-v at both launch and intercept, it sounds like you are assuming it is all used at launch?

This starts becoming disadvantageous even faster than just trying to speed up. At that point a much better improvement for you is the development of a magnetoshell decelerator. Mass-wise, there's no situation where propulsive intercept is better since a magnetoshell decelerator will be much more lightweight than even an NTR stage.

Why? Propellant is relatively light in an NTR system as far as I'm aware. I am also assuming that making a nuclear ship aerodynamic would be unfeasible (it will likely be long and only structurally reinforced longitudinally).

Could you make the hydrogen en-route? Solar power to crack water, use the oxygen for breathing? https://www.nasa.gov/content/space-applications-of-hydrogen-... says this is done on the ISS currently.

The mass of the oxygen in water is 8x that of the hydrogen, and you just don't need all the much for humans, and what you do have after respiration (CO2) gets recycled through the Sabatier process (H2O -> O2, H2; CO2 + H2 -> CH4 + H2O)

I.e. water is a quite inefficient storage medium for hydrogen and you're probably better of making heavier containment vessels for liquid hydrogen (of course a calculation could be shown to demonstrate the balance, but a tank weighing 8x the contents is a very long way from the extremely light tanks used in spaceflight)

If you are lugging all of that water mass along you could store it in a jacket around the crew compartment, providing additional radiation protection for most of the trip.

You're going to need humongous solar panels to support this, but since you are in space this isn't an intractable problem. A small but constant acceleration would probably make life better in the spacecraft as well.

Maybe? I believe LH2 has about 35 mols of hydrogen per liter, while water is 55 mols per liter. Storing hydrogen as water seems practical from that perspective, but what of the power needed to split that water? I think you'd need quite a lot of power to split that much water fast (starting a few days before running the engine.) Splitting it slowly over time using solar energy would seem to still leave you with a storage problem, but perhaps a more tractable one.

Maybe instead of electrolysis, they could use heat from the reactor? Thermolysis needs 2500 C though.

NTRs are basically open-cycle gas cooled reactors. The thermal limit on the reactor temp is when does stuff start to melt. Project Rho[0] suggests that's the reactor temp anyways. But you need to be able to separate out the oxygen from the thermolysis stream, rather than just feeding the entire thing into your engine, both because your Isp would go to crap if you tossed the oxygen out too, and you'd have oxidizing your reactor problems. Though, you could just store it all as ammonia, and you get more hydrogen for your buck, and can probably just feed that all through the reactor.

[0]:http://www.projectrho.com/public_html/rocket/enginelist2.php

Right. In a NTR the nuclear fuel has to be hotter than the hydrogen (or ammonia or methane or whatever) propellant so that the heat energy from the first conducts to the second. In a combustion energy the fuel and the propellant are the same substance so you try to limit conduction and can end up with propellant much hotter than the engine.

Wouldn't you just size the engines small enough that they instantly burn off the H2 as you crack it? The solar power should be even and constant so you can size the system to match. It is going to require a very large solar array, especially since your spacecraft is going to be really heavy with all of that water.

You could design a pulse detonation engine for this. Electrolyze water continuously; detonate it in a pipe every now and then. It's a very simple design that gives you quite a bit of performance for hopping in the asteroid belt. Specific impulse similar to a hydrolox engine, or slightly worse than regular hydrolox engines if operating stoichiometrically, although the detonation mode could compensate for that (detonation rocket engines can potentially get ~10% better Isp performance than "classical" rocket engines). However, you get triple the propellant density (water has ~1000 kg/m³; hydrolox is at around 340 kg/m³). This makes it much better compared to a classical hydrolox vehicle wherever gravity is near zero so that you don't need lift-off thrust.

Where are they going to get the water from?

You'd be bringing oxygen and hydrogen along anyways. Why not bring it in water form?

Much easier to store water for long duration than to store hydrogen. It even serves a useful purpose as radiation shielding for some solar events.

Reply on news.ycombinator.com