Starship On-orbit refueling - Options and Discussion

[steveleach]

Why not have the transfer 'legs' and 'socket' crossed. What I mean by that is one leg and one socket on each side? It looks like the legs extend out so they must be usually stowed inside during assent. By having them crossed then any type of Starship could dock with any other type.

If all the complexity is kept on the Tanker side it can be serviced on the ground when a problem occurs.

I wouldn't be surprised if the depots also return to Earth after they are depleted.

If the plan is to position the depots where they are needed for each mission (rather than plan for the missions to travel via wherever the depots are), then you'd need to reposition them between missions, and it could be easier to simply return them to Earth before launching them into the next location.

Depot should not be designed for EDL. EDL capability takes a lot of mass: TPS, control surfaces, header tanks, increased mass to support the catch, and increased mass to support the actual re-entry stresses. But Depot needs to be as light as possible to support efficient propellant transport.

Moving a near-empty Depot to a new orbit is cheaper than EDL and relaunch.

Are you just assuming it is cheaper, or has someone run the numbers?

[StraumliBlight]

Indeed yes. I thought I recalled an actual analysis somewhere earlier in this 167-page thread, but maybe it was in the "Ship boosters to KSC" thread.  It was associated with the concept of just putting a disposable nosecone on a Booster and sending it to KSC.

Eager Space covered it, though the numbers may have changed with Starship V3/V4.

[DanClemmensen]

Indeed yes. I thought I recalled an actual analysis somewhere earlier in this 167-page thread, but maybe it was in the "Ship boosters to KSC" thread.  It was associated with the concept of just putting a disposable nosecone on a Booster and sending it to KSC.

Eager Space covered it, though the numbers may have changed with Starship V3/V4.

The video is interesting and maybe worth watching. It uses only very basic and oversimplified equations that I suspect will cause an actual rocket scientist to cringe. It concludes that Starship could easily make the dogleg hop if boosted by SH. It also concludes that SH can make the dogleg hop. The conclusions are based solely on energy computations.

It does not address re-entry heating of superheavy at all.

[OTV Booster]

Why not have the transfer 'legs' and 'socket' crossed. What I mean by that is one leg and one socket on each side? It looks like the legs extend out so they must be usually stowed inside during assent. By having them crossed then any type of Starship could dock with any other type.

If all the complexity is kept on the Tanker side it can be serviced on the ground when a problem occurs.

I wouldn't be surprised if the depots also return to Earth after they are depleted.

If the plan is to position the depots where they are needed for each mission (rather than plan for the missions to travel via wherever the depots are), then you'd need to reposition them between missions, and it could be easier to simply return them to Earth before launching them into the next location.

Depot should not be designed for EDL. EDL capability takes a lot of mass: TPS, control surfaces, header tanks, increased mass to support the catch, and increased mass to support the actual re-entry stresses. But Depot needs to be as light as possible to support efficient propellant transport.

Moving a near-empty Depot to a new orbit is cheaper than EDL and relaunch.

The depot would be even lighter with fewer engines. What might be the minimum engine count to make it from stage separation to LEO? Hmmm. That calls for some arm waving on final mass. That never stopped us before.


Three Vac and one SL?  One of each? Three Vac with no TVC, no SL and differential thrust for control?


None of this would happen until late in the game and some options would call for serious hot end mods.

[OTV Booster]

Why not have the transfer 'legs' and 'socket' crossed. What I mean by that is one leg and one socket on each side? It looks like the legs extend out so they must be usually stowed inside during assent. By having them crossed then any type of Starship could dock with any other type.

If all the complexity is kept on the Tanker side it can be serviced on the ground when a problem occurs.

I wouldn't be surprised if the depots also return to Earth after they are depleted.

If the plan is to position the depots where they are needed for each mission (rather than plan for the missions to travel via wherever the depots are), then you'd need to reposition them between missions, and it could be easier to simply return them to Earth before launching them into the next location.

Depot should not be designed for EDL. EDL capability takes a lot of mass: TPS, control surfaces, header tanks, increased mass to support the catch, and increased mass to support the actual re-entry stresses. But Depot needs to be as light as possible to support efficient propellant transport.

Moving a near-empty Depot to a new orbit is cheaper than EDL and relaunch.

Are you just assuming it is cheaper, or has someone run the numbers?

If the loss of capacity due to having EDL capability is included in the cost calculation it makes a big difference. All the SpaceX renders point to no EDL.

Add in all the special kit expected for a full blown depot and it promises to have the highest dry mass and cost in the fleet, except maybe the human rated ones. No EDL means not returning it but replacing it.

Opinion: Early builds will be EDL capable tankers adding on depot like characteristics until at some point the EDL gear will be stripped and it really is a depot. In the SX tradition, early depots will be only crude approximations of what they need be and will be splashed after one campaign.

In the short term, there won't be enough operational use to get deep into the refinement process or justify reuse. A thin maybe for a one ship 2026 mars shot, a 2028 mars campaign with an unknown number of ships, and up to three lunar landings.

Do all of that and the design might be mature enough to make orbital storage and repositioning worthwhile. In between tanker campaigns launch facilities will be in low gear and the marginal cost of a tanker flight or two for repositioning a depot would be small change.

Alternatively, four or more depots in different planes opens up the options for TLI through the month if lunar traffic becomes high enough to justify it.

Another one for GSO or GSI makes sense too.

If/when Musk starts sending increasingly large fleets to Mars every 26 months ISTM (without hard numbers) that learning how to build a depot robust enough to last years to form an infrastructure backbone will be more cost effective than building a new depot fleet every 26 months.

[rsdavis9]

Moving a near-empty Depot to a new orbit is cheaper than EDL and relaunch.

And that is the question right there. I think that edl and relaunch is what spacex is shooting for to be cheap and easy and frequent. Whats the delta-v required for a 25deg orbit change versus relaunch. And remember the fuel to the orbit change is 50 times(100t/5000t) more expensive in energy because somebody had to get it to orbit.

[DanClemmensen]

Moving a near-empty Depot to a new orbit is cheaper than EDL and relaunch.

And that is the question right there. I think that edl and relaunch is what spacex is shooting for to be cheap and easy and frequent. Whats the delta-v required for a 25deg orbit change versus relaunch. And remember the fuel to the orbit change is 50 times(100t/5000t) more expensive in energy because somebody had to get it to orbit.

Basically, It will take one launch and landing (of the Depot) to change orbits via a launch and re-landing. It will take about one launch and landing (of a Tanker) to make the same change using a plane change. But Tanker is optimized for this and Depot is not.

Clearly, Actual tradeoffs will depend on the the two planes and on lots of other considerations.

[Kenm]

For example, the legs might be able to extend and compress over a length of (say) 2 meters, with ball joints with a thirty-degree range of motion, and with variable force in tension and compression  up to (say) one tonne.

OK, maybe it's not as bad as I thought, but it's still pretty serious.
Momentum (a vector) is mass times velocity: p = mv.
Consider the momentum of the lighter ship relative to the heavier ship. If the lighter ship is 400 tonne and the closing velocity is 1 cm/s, then the momentum is 4 tonne*m/s. To stop it using one leg it needs to push back at 400 tonne for one second or 4 tonne for 100 seconds, etc. At 1 tonne for 400 seconds, it will compress by 2 meters.

Of course, we have four legs, but we are correcting in six dimensions, so some may be in tension, which throws more compression at the other legs. Compression will also be variable and change gradually, so use one tonne as the average. We are also still sloshing and perhaps getting assistance from the thrusters on both ships, but I suspect we are basically OK here. After the ships are brought to rest relative to each other, the legs can slowly retract to compete the hard docking.

I think a bit of a mix-up with units has caused a overestimate of the forces needed.
A 400 tonne force acting on a 400 tonne object will cause a one g acceleration which will change the velocity by 9.8 m/s or 980 cm/s if applied for 1 sec.
For the 400 tonne object moving at 1 cm/s it's momentum is 4 tonne*m/sec or 4000 kg m/s as you correctly pointed out. To bring it to a stop in one second a force of 4000 nt or about the weight of  a 400kg mass on earth is needed. Note that in this case the ship will only move 0.5 cm while being brought to a stop. If we use the full 2 meter stroke slowing down in 400 sec then only a 10 nt force (about 1 kg weight) is needed. So it looks like lighter legs can be used but they will still need to be stiff enough to damp out any swaying between the ships and the sloshing in the tanks.

[Twark_Main]

For example, the legs might be able to extend and compress over a length of (say) 2 meters, with ball joints with a thirty-degree range of motion, and with variable force in tension and compression  up to (say) one tonne.

OK, maybe it's not as bad as I thought, but it's still pretty serious.
Momentum (a vector) is mass times velocity: p = mv.
Consider the momentum of the lighter ship relative to the heavier ship. If the lighter ship is 400 tonne and the closing velocity is 1 cm/s, then the momentum is 4 tonne*m/s. To stop it using one leg it needs to push back at 400 tonne for one second or 4 tonne for 100 seconds, etc. At 1 tonne for 400 seconds, it will compress by 2 meters.

Of course, we have four legs, but we are correcting in six dimensions, so some may be in tension, which throws more compression at the other legs. Compression will also be variable and change gradually, so use one tonne as the average. We are also still sloshing and perhaps getting assistance from the thrusters on both ships, but I suspect we are basically OK here. After the ships are brought to rest relative to each other, the legs can slowly retract to compete the hard docking.

I think a bit of a mix-up with units has caused a overestimate of the forces needed.
A 400 tonne force acting on a 400 tonne object will cause a one g acceleration which will change the velocity by 9.8 m/s or 980 cm/s if applied for 1 sec.
For the 400 tonne object moving at 1 cm/s it's momentum is 4 tonne*m/sec or 4000 kg m/s as you correctly pointed out. To bring it to a stop in one second a force of 4000 nt or about the weight of  a 400kg mass on earth is needed. Note that in this case the ship will only move 0.5 cm while being brought to a stop. If we use the full 2 meter stroke slowing down in 400 sec then only a 10 nt force (about 1 kg weight) is needed. So it looks like lighter legs can be used but they will still need to be stiff enough to damp out any swaying between the ships and the sloshing in the tanks.

At first I wondered where the KenM joke was, but then I saw "nt" instead of "N" used as the abbreviation for newtons... 

(Yes "newtons" in lower case is correct; upper case would refer to the person and only the abbreviation "N" is capitalized, same goes for watts, joules, hertz, etc)

[Overtone]

Moving a near-empty Depot to a new orbit is cheaper than EDL and relaunch.

And that is the question right there. I think that edl and relaunch is what spacex is shooting for to be cheap and easy and frequent. Whats the delta-v required for a 25deg orbit change versus relaunch. And remember the fuel to the orbit change is 50 times(100t/5000t) more expensive in energy because somebody had to get it to orbit.

Basically, It will take one launch and landing (of the Depot) to change orbits via a launch and re-landing. It will take about one launch and landing (of a Tanker) to make the same change using a plane change. But Tanker is optimized for this and Depot is not.

Clearly, Actual tradeoffs will depend on the the two planes and on lots of other considerations.

One of those other considerations, in my view, is that if EDL is not required, you could put half a meter of spray on foam insulation all around the outside of the depot. We know from shuttle missions that SOFI survives launch, but it would never survive reentry. Insulation would keep boil off low, which has significant knock-on benefits in all aspects of mission design.

[TheRadicalModerate]

One of those other considerations, in my view, is that if EDL is not required, you could put half a meter of spray on foam insulation all around the outside of the depot. We know from shuttle missions that SOFI survives launch, but it would never survive reentry. Insulation would keep boil off low, which has significant knock-on benefits in all aspects of mission design.

SOFI is designed to make boiloff not-terrible for the amount of time needed to fill the tanks on the pad, launch them, and maybe store prop remnants for the remainder of the stage life, which is no more than 5 hours, and may be as little as 45 minutes.¹  That's at least 3 orders of magnitude too short for depot use.

I got kinda half-convinced by somebody somewhere that it would be possible to attach an aeroshell backed by MLI to the outside of the tanks.  In Starship's case, it might even be possible to build tiles like this, and use the existing clips to do the attachment.  (You'd need to add clips to the dorsal side as well.)

Since only cylindrical parts of the Ship need insulation, it might be possible to build larger panels, which could still be snapped on pretty easily.  You need some kind of aeroshell collar around the front of the whole thing to keep flow separation from tearing the whole thing to bits, but that's easy--especially since there's a cylindrical barrel forward of the tanks.

_____________
¹What does Centaur use for insulation?  Is it SOFI, or MLI?  Both the Centaur III for the Atlas V 4xx series and all the Vulcan Centaur V's have hydrolox tankage that acts as fairing.  Surely that must be insulated somehow.

[Robotbeat]

You just need a sunshade, maybe an earth shade, to keep methalox boiloff low.

Heck, a flat plate shielded from direct sunshine, painted the Earthside white with 95% reflectivity and the space-side at 0 albedo, the equilibrium temperature is around 115-120K at 500km or so.

A cylinder is more complicated, but a deployable shade really helps.

[OTV Booster]

Closeups of the upper strut sockets. First the right, then the left. Both pics sharpened a bit.

There is what looks like reinforcement on the right at about its 6:30 and none on the left... sorta. There is a corresponding dark something at its 5:30 but it looks different. Maybe the lighting? Maybe in the middle of a repair to replace a piece that didn't work out?

The sockets are in the midst of internal stringers which answers to reinforcement and inside we can see the skin and stringer welds. There appears to be no socket structure below the skin.

At the bottom of each socket is a vertical slot crossed by a thin horizontal rod. One image hints at a thicker structure in the middle of the rod. It appears to be a latch point but I'd rate that somewhere between possible and probable. If this is a latch my seat of the pants judgment is it's not strong enough for vigorous strut control input.

Another tentative thought is that all high power strut manipulations and fluid transfers might be through the lower struts. More likely, these may be first iteration place holders for a fit check on a busy airframe and for judging aero issues.


Edit to add: when the airstream hits them just right I bet they whistle.

[SpaceLizard]

When the airstream hits them just right I bet they whistle.

A whistling rocket?...that could be very interesting during lift-offs(if you can hear it over the engines) and the banking maneuver during landings.

[InterestedEngineer]

When the airstream hits them just right I bet they whistle.

A whistling rocket?...that could be very interesting during lift-offs(if you can hear it over the engines) and the banking maneuver during landings.

[mikelepage]

I've been wondering if anyone else thought this when they saw the slide below... that "starship 3" on the right (stretched, 9 engines) is actually the tanker variant, and every other variant (that does EDL) will look more like the "starship 2" form factor. Or at least, that there will be that kind of size differential between the tanker and everything else. Reason being to reduce the number of tanker flights for each HLS (or Mars-bound) flight to the numbers (4-8) that Elon has said.

If you have 250 ton of prop per tanker to 1500 tons of prop per vehicle, that's 6-7 tanker flights per mission. Which, given the trade between efforts to reduce boil-off prior to TLI, and just getting the prop up there faster, you could see 6-7 flights in under two weeks (comparable to F9 today). Higher numbers of tanker flights seem to me to be less of a technical challenge than a logistical one.

EDIT: I realise prop is dense, and wouldn't actually need a stretched stage. But as with all of SpaceX presentations, the mass numbers are more important than the graphics.

[Twark_Main]

Moving a near-empty Depot to a new orbit is cheaper than EDL and relaunch.

And that is the question right there. I think that edl and relaunch is what spacex is shooting for to be cheap and easy and frequent. Whats the delta-v required for a 25deg orbit change versus relaunch. And remember the fuel to the orbit change is 50 times(100t/5000t) more expensive in energy because somebody had to get it to orbit.

Basically, It will take one launch and landing (of the Depot) to change orbits via a launch and re-landing. It will take about one launch and landing (of a Tanker) to make the same change using a plane change. But Tanker is optimized for this and Depot is not.

Clearly, Actual tradeoffs will depend on the the two planes and on lots of other considerations.

One of those other considerations, in my view, is that if EDL is not required, you could put half a meter of spray on foam insulation all around the outside of the depot. We know from shuttle missions that SOFI survives launch, but it would never survive reentry. Insulation would keep boil off low, which has significant knock-on benefits in all aspects of mission design.

In this application, SOFI will benefit greatly from simply covering it with solar white paint (similar to STS-1 and STS-2). For example AZ-93 absorbs only 15% of sunlight, which compares favorably to uncoated SOFI which absorbs 49% of sunlight. So to the first approximation that should be good for a >3x improvement in boil-off.

I wonder which manufacturing process is cheaper: SOFI + paint, or quilting MLI together with (ala Shuttle) an outer fiberglass aerocover fabric and gluing it on with RTV.

Curiously MLI underperforms at low temperatures, so if you want to try a hybrid approach then it's best to put the MLI over the SOFI. SOFI covers the low temperature part while MLI handles the majority of solar radiation.

Z-93C55 used on early Dragon https://ntrs.nasa.gov/citations/20130014266

AZ-93 specs https://www.aztechnology.com/product/1/az-93

SOFI Coating Study https://ntrs.nasa.gov/citations/20190032050

Shuttle blanket https://airandspace.si.edu/collection-objects/thermal-blanket-shuttle/nasm_A20100123000

Cryogenic Insulation for Mars Surface (mentions SOFI-MLI hybrids) https://ntrs.nasa.gov/citations/20180006152/

[Twark_Main]

You just need a sunshade, maybe an earth shade, to keep methalox boiloff low.

Heck, a flat plate shielded from direct sunshine, painted the Earthside white with 95% reflectivity and the space-side at 0 albedo, the equilibrium temperature is around 115-120K at 500km or so.

A cylinder is more complicated, but a deployable shade really helps.

"Sunshade," eh? Well I guess I never thought about it before, but technically paint does cast a shadow on the surface immediately below it... 

A deployable system is doable (certainly for SpaceX), but R&D tends to be slow. I expect they prefer a static system.

"The best part is no part. The best process is no process."

Deployment can't fail if there's no deployment.

[Vettedrmr]

"Sunshade," eh? Well I guess I never thought about it before, but technically paint does cast a shadow on the surface immediately below it... 

I know you're kidding, but just for the sake of completeness, the paint conducts heat because it's in physical contact with the insulation, while a sunshade doesn't.

[jarmumd]

Closeups of the upper strut sockets. First the right, then the left. Both pics sharpened a bit.

There is what looks like reinforcement on the right at about its 6:30 and none on the left... sorta. There is a corresponding dark something at its 5:30 but it looks different. Maybe the lighting? Maybe in the middle of a repair to replace a piece that didn't work out?

The sockets are in the midst of internal stringers which answers to reinforcement and inside we can see the skin and stringer welds. There appears to be no socket structure below the skin.

At the bottom of each socket is a vertical slot crossed by a thin horizontal rod. One image hints at a thicker structure in the middle of the rod. It appears to be a latch point but I'd rate that somewhere between possible and probable. If this is a latch my seat of the pants judgment is it's not strong enough for vigorous strut control input.

Another tentative thought is that all high power strut manipulations and fluid transfers might be through the lower struts. More likely, these may be first iteration place holders for a fit check on a busy airframe and for judging aero issues.


Edit to add: when the airstream hits them just right I bet they whistle.

It really looks to me like this is the support structure for the refueling system.  I would expect cones or probe arms mechanisms to fit inside these.  Supported by the bolt circle on the outer perimeter.