Starship On-orbit refueling - Options and Discussion

[Greg Hullender]

Its like its 1916 and you are building something called an aircraft carrier. It will not be an very good one but it was unlikely to get into carrier vs carriers fights.

I'm eager to hear more about depot-vs-depot fights though! :-)

[meekGee]

I have been following this thread for a while, but have not seen this refueling method discussed. I apologize if I missed the discussion.

Gee, it's only 131 pages over the past five years--read faster! :-)

Seriously, the things that soured me on all rotational approaches in the foreseeable future were:
a) You only need maybe 50 micro-g of acceleration to settle the tanks, so even a very long ullage burn (e.g. hours) won't waste very much fuel.

b) Cryogenic plumbing is the very devil to make work, so anything that needs <i>new</i> plumbing is probably off the table for a long time. Any practical system needs to work with Starships more-or-less as they are today--no new pipes and/or outlets. You <i>definitely</i> want to avoid anything that would add extra mass to each and every tanker Starship--adding mass just to the depot is probably okay.

c) No one has even tried to make a rotating system work in space, much less one that would potentially involve huge changes in moment of inertia from docking, fuel transfer, and sloshing. You could imagine it taking many years to get the kinks worked out.

The ullage-burn approach is so easy and so cheap it's just impossible to beat. Or so it seems, anyway. We'll know more when they start actually trying to do it.

This, and as its just over pressure in tanker pushing liquid gas so gas voids is not an issue, just pump gas back to tanker.
Most of ullage is probably boil off anyway and you can simply burn some methane and oxygen gas for more trust once you nailed it. Not needed for the prototype version however, none has done this before.
Its like its 1916 and you are building something called an aircraft carrier. It will not be an very good one but it was unlikely to get into carrier vs carriers fights.

I'm with you on all this, and I'll reemphasize the topic of docking.

We're not used to docking spacecraft where over half the mass is loose in the tanks.  That's not slosh, that's a couple of independent tsunamis in there.

The only predictable way to do this is get the ships on your fractional-g recipe ahead of time, never going into drift.  Align courses, and dock sideways, without ever dropping the acceleration.  And at that point, start moving fuel until done.

So the entire operation from before docking to after separation, is under this slight baseline acceleration (and the docking accelerations are lower)

Otherwise the effects of 100 ton propellant blobs hitting the sidewalls at random delays, off center, will be insane.

[InterestedEngineer]

I have been following this thread for a while, but have not seen this refueling method discussed. I apologize if I missed the discussion.

Gee, it's only 131 pages over the past five years--read faster! :-)

Seriously, the things that soured me on all rotational approaches in the foreseeable future were:
a) You only need maybe 50 micro-g of acceleration to settle the tanks, so even a very long ullage burn (e.g. hours) won't waste very much fuel.

b) Cryogenic plumbing is the very devil to make work, so anything that needs <i>new</i> plumbing is probably off the table for a long time. Any practical system needs to work with Starships more-or-less as they are today--no new pipes and/or outlets. You <i>definitely</i> want to avoid anything that would add extra mass to each and every tanker Starship--adding mass just to the depot is probably okay.

c) No one has even tried to make a rotating system work in space, much less one that would potentially involve huge changes in moment of inertia from docking, fuel transfer, and sloshing. You could imagine it taking many years to get the kinks worked out.

The ullage-burn approach is so easy and so cheap it's just impossible to beat. Or so it seems, anyway. We'll know more when they start actually trying to do it.

This, and as its just over pressure in tanker pushing liquid gas so gas voids is not an issue, just pump gas back to tanker.
Most of ullage is probably boil off anyway and you can simply burn some methane and oxygen gas for more trust once you nailed it. Not needed for the prototype version however, none has done this before.
Its like its 1916 and you are building something called an aircraft carrier. It will not be an very good one but it was unlikely to get into carrier vs carriers fights.

I'm with you on all this, and I'll reemphasize the topic of docking.

We're not used to docking spacecraft where over half the mass is loose in the tanks.  That's not slosh, that's a couple of independent tsunamis in there.

The only predictable way to do this is get the ships on your fractional-g recipe ahead of time, never going into drift.  Align courses, and dock sideways, without ever dropping the acceleration.  And at that point, start moving fuel until done.

So the entire operation from before docking to after separation, is under this slight baseline acceleration (and the docking accelerations are lower)

Otherwise the effects of 100 ton propellant blobs hitting the sidewalls at random delays, off center, will be insane.

What if they use the slosh to move the propellant?

[Twark_Main]

Ive got a rich suite of background processes working on orbital refueling and two concerns have popped out of the que. The current discussion on propellant temperature needs is one. QD sealing is another.


Cryogenic seals are a tough problem. Adding in multiple sealing cycles can only make it tougher. Our discussion has focused on using either the current QD layout or something similar. This puts the oxidizer and fuel seals in close proximity. If both start leaking liquid, local conditions determine the possibility of vaporization. If vapor is present only the lack of an ignition source prevents ignition.


I suspect the current state of material science is inadequate for perfecting the seals. I will gladly defer to wiser minds on this.


AIUI, the flow of the propellants themselves will build up triboelectric potential differences. This isn't insurmountable but it has to be dealt with.


Up until now I've been a strong advocate of working with current QD layout but I'm having second thoughts. Might it be best to keep the LOX connection where it is and having a second QD plate up by the methane tank? This would keep leakage separated and make an ignition source a non problem. It's a PITA but so is unintended consequences. Could this arrangement be worked to ease some of the temperature concerns?


Or am I just pissing in a tea pot and creating a tempest? (Did I get that right?  )

Falcon 9 put the fuel and oxidizer QDs 180 degrees apart for IMO good reason.

I proposed a similar scheme a while back while trying to kremlinologize this SpaceX rendering. TL;DR the external pipe could be used as a "shunt" to send fuel/oxidizer flow between ships using the other connector in case of seal failure.     Of course you'd be purging a good long time to hard vacuum in between.

[Twark_Main]

That is, it looked as though an HLS fully refueled in LEO, sent to Gateway and then down to the moon would be unable to return to Gateway even with zero payload.

I thought the consensus was that such a mission plan would require fully fueling in HEEO, not LEO.

I thought that was just with the V1 Starship. I still have your "spreadsheet of doom," by the way; I just don't have the numbers to put into it.

With apologies for the doom!  Truly an accursed Excel horror beyond our comprehension, if ever there was one...   

[Twark_Main]

I have been following this thread for a while, but have not seen this refueling method discussed. I apologize if I missed the discussion.

Depots and propellant supply ships would both have internal ribcage-like manifold in each tank. Outbound ships would not need the manifold.

Ships dock back-to-back.  Flywheel on both ships, spin axis parallel to the long axis of each ship, spin up, imparting a very slow spin to the joined ships, settling propellant along the belly of each ship.

Increase supply-side tank pressure above that of other vessel, and open valves. Slow rotation keeps pressure head of the fluids low, so tank pressure should dominate. Propellant enters manifold  on the supply vessel from ventral side, and leave from the dorsal side on the receiving ship.

Center of mass, rotational axis, and moment of inertia will all shift during transfer, but perhaps flywheels can compensate.

After transfer, flywheels cancel out spin of the vessels, and vessels unlock.

It may be possible to so something similar using only thrusters instead of flywheels.

Interesting concept, the problem is that any axis not aligned with the maximum moment of inertia is unstable. The two ships will start tumbling until they are rotating in the only stable configuration which should be around an axis perpendicular to the symmetry plane going through both ships long axis (and that plane will want to be parallell to the orbital plane).

That configuration might work with your suggestion instead as most of the propellant would pool on the dorsal top and bottom corners of their respective tanks.

I was wondering about the rotational stability, but the tennis racket theorem suggests that rotation is stable about the principal axis associated with the minimum moment of inertia. Of course, that result assumes fixed moments of inertia, which wouldnot apply given the dynamic nature of the mass distribution in the system.

Indeed, it does specifically not hold for any object that is able to dissipate some rotational energy, the classic example of which is tanks with mixed fluids in them. This experiment can be done at home (perhaps outside) with a half full water bottle.

Previous discussion, right down to the digression about intermediate axis stability and damping. 

https://forum.nasaspaceflight.com/index.php?topic=50157.msg2514406#msg2514406

 (hint hint, use "Print" to search the whole thread on a single page) 

[OTV Booster]

I have been following this thread for a while, but have not seen this refueling method discussed. I apologize if I missed the discussion.

Gee, it's only 131 pages over the past five years--read faster! :-)

Seriously, the things that soured me on all rotational approaches in the foreseeable future were:
a) You only need maybe 50 micro-g of acceleration to settle the tanks, so even a very long ullage burn (e.g. hours) won't waste very much fuel.

b) Cryogenic plumbing is the very devil to make work, so anything that needs <i>new</i> plumbing is probably off the table for a long time. Any practical system needs to work with Starships more-or-less as they are today--no new pipes and/or outlets. You <i>definitely</i> want to avoid anything that would add extra mass to each and every tanker Starship--adding mass just to the depot is probably okay.

c) No one has even tried to make a rotating system work in space, much less one that would potentially involve huge changes in moment of inertia from docking, fuel transfer, and sloshing. You could imagine it taking many years to get the kinks worked out.

The ullage-burn approach is so easy and so cheap it's just impossible to beat. Or so it seems, anyway. We'll know more when they start actually trying to do it.

This, and as its just over pressure in tanker pushing liquid gas so gas voids is not an issue, just pump gas back to tanker.
Most of ullage is probably boil off anyway and you can simply burn some methane and oxygen gas for more trust once you nailed it. Not needed for the prototype version however, none has done this before.
Its like its 1916 and you are building something called an aircraft carrier. It will not be an very good one but it was unlikely to get into carrier vs carriers fights.

I'm with you on all this, and I'll reemphasize the topic of docking.

We're not used to docking spacecraft where over half the mass is loose in the tanks.  That's not slosh, that's a couple of independent tsunamis in there.

The only predictable way to do this is get the ships on your fractional-g recipe ahead of time, never going into drift.  Align courses, and dock sideways, without ever dropping the acceleration.  And at that point, start moving fuel until done.

So the entire operation from before docking to after separation, is under this slight baseline acceleration (and the docking accelerations are lower)

Otherwise the effects of 100 ton propellant blobs hitting the sidewalls at random delays, off center, will be insane.

What if they use the slosh to move the propellant?

How would that work?

[StevenOBrien]

Interesting interview posted yesterday with someone from HLS, saying that SpaceX is planning to begin the ship-to-ship prop transfer campaign by March 2025 and have it completed by Summer 2025:

[Brigantine]

begin the ship-to-ship prop transfer campaign by March 2025

Does "begin the campaign" mean flights, or just starting construction of ships with docking hardware?

I don't think it can mean flights, unless that hardware is already around somewhere waiting to go on S34, at least the passive side.

[Clarification: I'm trying to work out what part of the design/build/assemble process we should be looking at right now, in order to read the tea leaves about which On-orbit refueling options SpaceX is actually going with. Capabilities of the docking hardware put limitations on what you can do to achieve prop transfer, and then there's the QD issue. There might already be parts lurking in Starfactory, maybe even on the S34 aft section and 2nd mid-LOx section]

[InterestedEngineer]

begin the ship-to-ship prop transfer campaign by March 2025

Does "begin the campaign" mean flights, or just starting construction of ships with docking hardware?

I don't think it can mean flights, unless that hardware is already around somewhere waiting to go on S34, at least the passive side.

These are program managers. It probably is the starting mark for "moar meetings"

[KilroySmith]

I'm not sure they could construct two ships and finish a flight campaign by, say, August, if they started construction in March.   
In a best-case scenario, I see three flights:
1. Launch the receiver into orbit
2. Launch a tanker into orbit, dock, transfer propellant, deorbit
  a. Note that completing HLS doesn't REQUIRE catching OR re-using tankers, if SpaceX is willing to build enough of them
3. Launch a second tanker, dock, transfer propellant
  a. Validates prop transfer to a partially full receiver, where the dynamics may be different from a mostly empty

Add in a flight or two for contingencies, and at a launch rate of perhaps 1/month, this suggests a June/July completion if first flight is in March.

[Greg Hullender]

I worry that the microgravity estimates for settling propellant derived from restarting Centaur engines won't apply to refueling. This is because, in the case of a restart, nothing else is stirring the propellant, and the ullage burn only had to get enough propellant to the bottom to make it safe to start the engines. Once the engines started, very robust acceleration quickly settled the remainder, without any worry about sloshing.

But when you're pumping propellant from one tank to another, you're constantly stirring the tanks. I think it's likely that there's a tradeoff between speed of propellant transfer vs. required microgravity. On the other hand, the prop transfer can probably tolerate a lot more gas in the mix than an engine restart can. And the right sort of propellant-management device at the bottom of the tank might make things easier (i.e. it may not matter how much slosh you have above the PMD, just as long as the PMD stays wet.) Or can you get pumps that handle a mix of liquid and gas and just pump mostly-liquid in the bottom and suck mostly-gas out the top until you estimate you've got enough propellant in the target tank?

Has anyone tried to enumerate the whole trade space here?   

[dondar]

begin the ship-to-ship prop transfer campaign by March 2025

Does "begin the campaign" mean flights, or just starting construction of ships with docking hardware?

I don't think it can mean flights, unless that hardware is already around somewhere waiting to go on S34, at least the passive side.

I think you in the wrong subforum.
Ship to Ship Propellant transfer demonstration campaign will start in march and SpaceX intends to complete by the end of summer. They intend to fly every two week (*unclear, from every tower or cumulative from Boca).

My take that this "2025" will start not in next January because they need to complete this round (going orbital and catching the starship and it will take at least two flights (next will be still suborbital). Mentioned March indicates most probably the start of the launch carousel.

[Twark_Main]

begin the ship-to-ship prop transfer campaign by March 2025

Does "begin the campaign" mean flights, or just starting construction of ships with docking hardware?

I don't think it can mean flights, unless that hardware is already around somewhere waiting to go on S34, at least the passive side.

These are program managers. It probably is the starting mark for "moar meetings"

You're confusing SpaceX for Boeing. 

"Begin the campaign" means flights. There will no doubt be some delays, but they mean flights.

[InterestedEngineer]

begin the ship-to-ship prop transfer campaign by March 2025

Does "begin the campaign" mean flights, or just starting construction of ships with docking hardware?

I don't think it can mean flights, unless that hardware is already around somewhere waiting to go on S34, at least the passive side.

These are program managers. It probably is the starting mark for "moar meetings"

You're confusing SpaceX for Boeing. 

"Begin the campaign" means flights. There will no doubt be some delays, but they mean flights.

It was a NASA project manager speaking.  They'll have lots of meetings.  SpaceX probably has someone whose job it is to make the PMs happy, but has no authority inside SpaceX.

At least, that's how I'd run it.

[Twark_Main]

begin the ship-to-ship prop transfer campaign by March 2025

Does "begin the campaign" mean flights, or just starting construction of ships with docking hardware?

I don't think it can mean flights, unless that hardware is already around somewhere waiting to go on S34, at least the passive side.

These are program managers. It probably is the starting mark for "moar meetings"

You're confusing SpaceX for Boeing. 

"Begin the campaign" means flights. There will no doubt be some delays, but they mean flights.

It was a NASA project manager speaking.  They'll have lots of meetings.  SpaceX probably has someone whose job it is to make the PMs happy, but has no authority inside SpaceX.

At least, that's how I'd run it.

I was basing it on the negotiated NASA milestones between the two companies. Boeing's milestones were mostly meetings, while SpaceX's milestones were mostly actual work getting done.

[OTV Booster]

Ive been noodling all the refueling schemes and one point keeps popping up. All depots have a lot of engine dry mass they don't need. It's based on StarShip and it launches like StarShip. Starship has 6 or nine engines so the depot has 6/9 engines.


The impact is that any maneuvering BLEO consumes more propellant than it needs to. That's ok for LEO and early depots that will probably be one use builds. We're a long way from the massive on orbit wrenching needed to remove engines for BLEO missions. What to do?


This will be heresy, but what about an only sorta, kinda SS based solution? An 8m dia. OTV with one gimbaled RVac (and/or electric thrust) launched by SS. Tankage launched separately. A stainless build like SS to keep development and build costs low. Maybe one central tank for CH4 and 2 side tanks for LOX. Some thermal problems go away with no downcomer.


A big question is, how hard would it be to dependably do the one time plumbing connections for long term reuse without a human hand? Another question is ROI. It's a lot of trouble for an occasional trip to HEO but it's a different story for NRHO or LLO. It might cast some of the lunar refueling strategies in a new light.


Another place it might show advantage is early mars missions before ISRU is dependable. They'd still need to land propellant in a traditional tanker for returning ship(s) but only enough to make orbit for the big gulp.

[JIS]

Ive been noodling all the refueling schemes and one point keeps popping up. All depots have a lot of engine dry mass they don't need. It's based on StarShip and it launches like StarShip. Starship has 6 or nine engines so the depot has 6/9 engines.

Propellant depot should really be closer to HLS rather than tanker SS.
It means heat insulation of tanks and three gimbaled vacuum raptors.
No reentry and recovery systems or sea level engines are required.
Of course the HLS "nose cone" will be unique.

Edit: Although on second thought more engines are benefitial to minimize gravity losses and lower ISP os sl Raptors is not such a big deal for LEO.

[Twark_Main]

In no particular order

Atlas demonstrated disconnecting engines in-flight.

Why should we expect connecting engines will be easier than disconnecting them?

Why separate engines and tankage vs using on-orbit refilling?

[OTV Booster]

In no particular order

Atlas demonstrated disconnecting engines in-flight.

Why should we expect connecting engines will be easier than disconnecting them?

Why separate engines and tankage vs using on-orbit refilling?

I guess I wasn't clear. The assemblage of OTV and tanks would be the BEO depot rather than using an SS variation. It saves dry mass for missions to NRHO, LLO, mars and anywhere else a depot is needed other than LEO.