Starship On-orbit refueling - Options and Discussion

[DanClemmensen]

To dock the nose of a Depot to the tail of another ship, the Depot nose could use a 6DOF soft-dock ring similar to the IDSS soft-dock but 9 meters in diameter.

I can't see this working. Starship's nose is about 13-14m long, say 12m to reach the opposite ring hardware. Assuming the soft-dock interstage-ring hybrid is around 2m, that means you need six 10m long extendable struts. Plus a receiving ring around the engine skirt of the other ship with all the clamping hardware. Plus you need entirely new plumbing on both ships. Plus you need to add the extendable docking ring to the nose of the depot after launch, which means you need another pre-installed flush-mounted docking system before it has this docking ring (a docking ring to let you dock a docking ring.)

Depot would not be designed for EDL. It would have a nose that looks like the top of a Booster, covered by a fairly blunt disposable fairing. the struts in the ring truss would all be extensible, and the extended length will be shorter than the probes we see in the renders of the probe-and-socket tandem scheme. You only need six struts to achieve 6DOF, but they can use additional struts to achieve more strength and stability as needed.

Ship does not need an additional receiving ring. Ship already has a hard docking ring that mates it to the top of the Booster for launch.

As I said, I think the actual fluid transfer would use an arm that the Depot extends after it achieves the hard nose-to-tail dock, conceptually similar to the "claw" that connects a Dragon capsule to its trunk, but a whole lot more complicated.  The complexity and mass of this are are the biggest problems with this entire hare-brained scheme. This arm mates to the Ship's existing QD plate and couples to it using the same mating interface as the SQD on on the tower.

[Paul451]

[...]

You still haven't explained what you are gaining (or saving) from this thing. What is the advantage over SpaceX's model?

As I said, SpaceX is going to develop the least-parts version until they hit an unforeseen show-stopper. The nature of that show-stopper will decide which alternative models would then be considered. What possible failure modes does theirs have that yours doesn't?

[DanClemmensen]

[...]

You still haven't explained what you are gaining (or saving) from this thing. What is the advantage over SpaceX's model?

As I said, SpaceX is going to develop the least-parts version until they hit an unforeseen show-stopper. The nature of that show-stopper will decide which alternative models would then be considered. What possible failure modes does theirs have that yours doesn't?

Potential gains:
    nose-to-tail may simplify the actual propellant transfer
    allows Depot-to-Depot transfer
    zero modifications to Ship
    use of a Stewart platform for 6DOF operation and docking is fully characterized and demonstrated
    Stewart platform kinematics is a lot simpler than four independent probes
    The Ship ring is already reinforced. No need for extra structural reinforcement.
    Depot can do propellant settling acceleration for the combined stack with no contribution from Ship.
    Depot can remain attached and provide full boost.
    Depots can be daisy-chained.
Potential drawback:
   The arm is probably more complex than the equivalent on probe-and-socket design, but we don't know yet.

Please note: I suspect probe-and-socket is the the actual current plan of record, but we have not seen quite enough hardware to be sure.

[rsdavis9]

Potential gains:
    nose-to-tail may simplify the actual propellant transfer

So the current plumbing and qd plate position currently works fine in 1g fill and drain.

[DanClemmensen]

Potential gains:
    nose-to-tail may simplify the actual propellant transfer

So the current plumbing and qd plate position currently works fine in 1g fill and drain.

Both the quad probe-and-socket concept and the crazy nose-to-tail concept appear to use the existing QD plate on the Ship. I think both concepts assume that the existing QD plate and plumbing will work under 1g axial acceleration (like on the pad) and still work under reduced acceleration. We don't know how low that acceleration can go.

[OTV Booster]

Potential gains:
    nose-to-tail may simplify the actual propellant transfer

So the current plumbing and qd plate position currently works fine in 1g fill and drain.

Both the quad probe-and-drogue concept and the crazy nose-to-tail concept appear to use the existing QD plate on the Ship. I think both concepts assume that the existing QD plate and plumbing will work under 1g axial acceleration (like on the pad) and still work under reduced acceleration. We don't know how low that acceleration can go.

To quote Churchill, "I'm warming to this idea."

Pros:
- The propellant load stays axial avoiding off axis ullage thrust and system torques.
- when loading the depot the fluid flow is always from the bottom of the tanker tanks and flow to the depot is always down hill.

Cons:
- transfer to the target is always up hill. Pumps on the depot could mitigate this.

Pumps can make for a fast transfer. This is semi irrelevant for a ship waiting for crew to show up but there are use cases where transfer with crew makes sense if proven safe.

Slow transfer driven by g load and pressure differential, tanker to depot, could be anything from irrelevant to a major PITA depending on how fast tankers can show up.

Depot shows every promise of being a heavy and uniquely complex variant. Should this be avoided, embraced or ignored? IDK.

[Twark_Main]

Check my logic:

The "optimal" pumping rate is achieved when the tanker boil-off from solar heating is balanced by the frictional heating from the pipe and pump losses.

If you pump any slower, you have more heating due to the Sun hitting the (presumably unchilled) tanker. If you pump any faster, the friction in the pipe will cause more heating. You want to hit the minimum of those two curves.

Based on this, knowing the pipe sizes and lengths and bends, we should be able to calculate the total pipe loss and therefore the "optimum" pumping rate.

Thoughts?

Naturally, when doing the above math, the total pressure drop (plumbing + pump) will be twice the pressure drop in the plumbing. This is because (as we all know) current in a circuit is maximized when the load resistance matches the battery internal resistance.   

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

[mikelepage]

Using the Super Heavy as the fuel depot is really an underexplored idea. 

Using the mass and ISP numbers from Wikipedia (who knows if they're any good) the Super Heavy is on the cusp of SSTO if it were flown without a starship/payload.

It's certainly not optimized for the job, but if they could get it up there it's a water tower version of a starting platform.

Interesting idea. TWR of SH as SSTO would be too high for launch anyway, so the mods you would make to SH to optimise this function would be similar to what you would do for the pusher-depot.  I’d be curious how many sea-level engines you could lose, to make SSTO with no payload easier. Maybe even include one to three Vacuum raptors?

    Depots can be daisy-chained.

This is the really interesting part IMO. Three stage TJI from LEO with two full superheavy depots thrusting behind a starship. Enough for a substantial Europa lander? 

EDIT: On second thought, the engine replumbing would be serious. I guess the only “minimal” change would be to keep the outer 20 engines and replace the inner 13 with 3 (gimballed) raptor vacs for a liftoff TWR of ~ 1.2. And that would exclude daisy chaining.

[InterestedEngineer]

Using the Super Heavy as the fuel depot is really an underexplored idea. 

Using the mass and ISP numbers from Wikipedia (who knows if they're any good) the Super Heavy is on the cusp of SSTO if it were flown without a starship/payload.

It's certainly not optimized for the job, but if they could get it up there it's a water tower version of a starting platform.

Interesting idea. TWR of SH as SSTO would be too high for launch anyway, so the mods you would make to SH to optimise this function would be similar to what you would do for the pusher-depot.  I’d be curious how many sea-level engines you could lose, to make SSTO with no payload easier. Maybe even include one to three Vacuum raptors?

    Depots can be daisy-chained.

This is the really interesting part IMO. Three stage TJI from LEO with two full superheavy depots thrusting behind a starship. Enough for a substantial Europa lander? 

EDIT: On second thought, the engine replumbing would be serious. I guess the only “minimal” change would be to keep the outer 20 engines and replace the inner 13 with 3 (gimballed) raptor vacs for a liftoff TWR of ~ 1.2. And that would exclude daisy chaining.

Y'all keep missing that a fuel depot in LEO needs an eartshine shield or it'll just boil everything off.  Lot of math in this long thread says this is so.

So a booster isn't going work - you'd have to give it a way to deploy a shield.

A water tower is a water tower - just make starship so many rings longer and have its cargo bay spit out and deploy a sun/earth shine shield and solar panels (also needed, batteries aren't going to last weeks)

custom starships are going to be a thing.  Might as well get going on it.

[mikelepage]

Using the Super Heavy as the fuel depot is really an underexplored idea. 

<snip> It's certainly not optimized for the job, but if they could get it up there it's a water tower version of a starting platform.

Interesting idea. TWR of SH as SSTO would be too high for launch anyway, so the mods you would make to SH to optimise this function would be similar to what you would do for the pusher-depot.

EDIT: I guess the only “minimal” change would be to keep the outer 20 engines and replace the inner 13 with 3 (gimballed) raptor vacs for a liftoff TWR of ~ 1.2. And that would exclude daisy chaining.

Y'all keep missing that a fuel depot in LEO needs an eartshine shield or it'll just boil everything off.  Lot of math in this long thread says this is so.

So a booster isn't going work - you'd have to give it a way to deploy a shield.

A water tower is a water tower - just make starship so many rings longer and have its cargo bay spit out and deploy a sun/earth shine shield and solar panels (also needed, batteries aren't going to last weeks)

custom starships are going to be a thing.  Might as well get going on it.

My hunch is that we’re going to see various functions deployed from chines on the outside of starship - especially essentials like roll-out solar arrays/radiators etc - reserve the payload bay for customer payloads.

So I think this could work equally well for a superheavy converted to depot-pusher - deploy your earth-shine shield from a chine that runs the length of SH.

[Twark_Main]

Y'all keep missing that a fuel depot in LEO needs an eartshine shield or it'll just boil everything off.  Lot of math in this long thread says this is so.

So a booster isn't going work - you'd have to give it a way to deploy a shield.

As mentioned in this same long thread, just use MLI with a beta cloth aerothermal cover a la the Shuttle blankets.

No fancy deployment mechanism needed.

[wes_wilson]

Using the Super Heavy as the fuel depot is really an underexplored idea. 

Using the mass and ISP numbers from Wikipedia (who knows if they're any good) the Super Heavy is on the cusp of SSTO if it were flown without a starship/payload.

It's certainly not optimized for the job, but if they could get it up there it's a water tower version of a starting platform.

Interesting idea. TWR of SH as SSTO would be too high for launch anyway, so the mods you would make to SH to optimise this function would be similar to what you would do for the pusher-depot.  I’d be curious how many sea-level engines you could lose, to make SSTO with no payload easier. Maybe even include one to three Vacuum raptors?

    Depots can be daisy-chained.

This is the really interesting part IMO. Three stage TJI from LEO with two full superheavy depots thrusting behind a starship. Enough for a substantial Europa lander? 

EDIT: On second thought, the engine replumbing would be serious. I guess the only “minimal” change would be to keep the outer 20 engines and replace the inner 13 with 3 (gimballed) raptor vacs for a liftoff TWR of ~ 1.2. And that would exclude daisy chaining.

Y'all keep missing that a fuel depot in LEO needs an eartshine shield or it'll just boil everything off.  Lot of math in this long thread says this is so.

So a booster isn't going work - you'd have to give it a way to deploy a shield.

A water tower is a water tower - just make starship so many rings longer and have its cargo bay spit out and deploy a sun/earth shine shield and solar panels (also needed, batteries aren't going to last weeks)

custom starships are going to be a thing.  Might as well get going on it.

A filled booster holds roughly 2.2 starships worth of fuel, a filled starship holds exactly 1 starship worth of fuel.  Boiloff matters on starship because the depot and the recipient have the same capacity. With a filled super heavy in leo, how long would it take to boil off 2.2 starships of fuel to 1 starships worth of fuel remaining?

A shine shield is necessary for a starship depot, it's likely a "someday" optimization for a superheavy depot.  Mass buys margin with superheavy as the depot.

I agree there will be starship variants.  However, I'm not clear that starship variants are any more/less desirable than booster variants.  They have two working vehicles forming the core of their architecture, it's reasonable they could have variants of each.

[wes_wilson]

Interesting idea. TWR of SH as SSTO would be too high for launch anyway, so the mods you would make to SH to optimise this function would be similar to what you would do for the pusher-depot.  I’d be curious how many sea-level engines you could lose, to make SSTO with no payload easier. Maybe even include one to three Vacuum raptors?

EDIT: On second thought, the engine replumbing would be serious. I guess the only “minimal” change would be to keep the outer 20 engines and replace the inner 13 with 3 (gimballed) raptor vacs for a liftoff TWR of ~ 1.2. And that would exclude daisy chaining.

Ignore daisy chaining for a moment, another "minimal" change to address thrust to weight would be to add rings and increase the weight/capacity. 

[Paul451]

This arm mates to the Ship's existing QD plate and couples to it using the same mating interface as the SQD on on the tower.

I had assumed you intended the refuelling connections to be within the nose-tail docking area, not this.

Given this, how does the QD arm get from the nose area (where it presumably has to be tucked away during launch for aerodynamics) to outside the docking rings? I can think of convoluted mechanisms to do so, but it would be incredibly complex and fragile. Is there a gap in the docking ring for it to pivot outside? Or does it "step" over the docking ring hardware, then reconnect to plumbing on the outside (kind of like the Canadarm inch-worming around ISS)? Or is it flexible enough to wrap around (under & outside) the docking ring during launch, and unwrap and rotate back vertically after the nose is jettisoned?


All the rest of our argument seems to just come down to your personal incredulity vs mine. You think the four probe'n'drogue system is self-evidently heavy and complex and risky, while I think a 9m docking ring is self-evidently much, much worse.

For eg:

    use of a Stewart platform for 6DOF operation and docking is fully characterized and demonstrated
    Stewart platform kinematics is a lot simpler than four independent probes

By the standards that suggest a 9m docking ring is "characterised", probe'n'drogue is completely proven in a variety of forms and scales. From early capsule docking, to aerial refuelling, to PDGFs and robot arms on ISS, to virtually every time anyone's ever plugged a thing with a nozzle into a thing with a hole.

Four 1DOF telescoping arms and ball-locking catches. The arms telescope pull back linearly to push the flat plate of the protruding box against the QD plate of the other ship. If they aren't able to achieve the control fidelity to put the balls in the slots, then IMO they aren't going to be able to line the docking petals with the receiving ring. Which was my point: Nose-tail docking doesn't solve the things that would cause probe'n'drogue to fail. (That's why I mentioned the length of the nose section, I was assuming you were, essentially, using the entire nose as a giant probe, since that's the only way a docking ring requires less control authority than probe'n'drogue. There's a reason why probe'n'drogue docking came first in the space program.)

    zero modifications to Ship
    The Ship ring is already reinforced. No need for extra structural reinforcement.

Passive-side docking rings are more than just a ring. They are complex (and heavy) systems in their own right. OTOH, the method of launch integration to SH is (AIUI) gravitationally, not secured. Unsuitable for in-orbit docking. There must be modifications to the engine bay ring; at the very least, petal alignment guides and locking catches. I doubt you are saving anything compared to four comparatively small cones.

No need for extra structural reinforcement.

Look at the locations of the four arms/cones. Two are at the top of the upper tank, where it mates with the nose section and underneath the Pez hatch, an area which is already reinforced. The other two are just below the lower tank, where the thrust structure is, again, already reinforced. I doubt that's a coincidence. Hence I doubt there's any particular additional reinforcement.

Think about how and where the flaps are attached, and the loads on them.

    Depot can remain attached and provide full boost.
    Depots can be daisy-chained.

These are the only advantages I can see. And it's apparently not enough for SpaceX to be considering it ahead of the simpler version.

Please note: I suspect probe-and-socket is the the actual current plan of record, but we have not seen quite enough hardware to be sure.

We've seen no hardware except that which matches the probe'n'drogue images. If they were building a 9m wide active docking ring, even an early simplified mock up, it would have attracted immediate attention from the 24/7 ring-watchers.

[Paul451]

Weaponized Incompetence

Oh good. Social media has taught you a new way to belittle people. Just what we needed.

[OTV Booster]

Check my logic:

The "optimal" pumping rate is achieved when the tanker boil-off from solar heating is balanced by the frictional heating from the pipe and pump losses.

If you pump any slower, you have more heating due to the Sun hitting the (presumably unchilled) tanker. If you pump any faster, the friction in the pipe will cause more heating. You want to hit the minimum of those two curves.

Based on this, knowing the pipe sizes and lengths and bends, we should be able to calculate the total pipe loss and therefore the "optimum" pumping rate.

Thoughts?

Makes sense if optimizing for transfer losses. If that is the thing to optimize for is another question.

[OTV Booster]

Using the Super Heavy as the fuel depot is really an underexplored idea. 

Using the mass and ISP numbers from Wikipedia (who knows if they're any good) the Super Heavy is on the cusp of SSTO if it were flown without a starship/payload.

It's certainly not optimized for the job, but if they could get it up there it's a water tower version of a starting platform.

Interesting idea. TWR of SH as SSTO would be too high for launch anyway, so the mods you would make to SH to optimise this function would be similar to what you would do for the pusher-depot.  I’d be curious how many sea-level engines you could lose, to make SSTO with no payload easier. Maybe even include one to three Vacuum raptors?

    Depots can be daisy-chained.

This is the really interesting part IMO. Three stage TJI from LEO with two full superheavy depots thrusting behind a starship. Enough for a substantial Europa lander? 

EDIT: On second thought, the engine replumbing would be serious. I guess the only “minimal” change would be to keep the outer 20 engines and replace the inner 13 with 3 (gimballed) raptor vacs for a liftoff TWR of ~ 1.2. And that would exclude daisy chaining.

Y'all keep missing that a fuel depot in LEO needs an eartshine shield or it'll just boil everything off.  Lot of math in this long thread says this is so.

So a booster isn't going work - you'd have to give it a way to deploy a shield.

A water tower is a water tower - just make starship so many rings longer and have its cargo bay spit out and deploy a sun/earth shine shield and solar panels (also needed, batteries aren't going to last weeks)

custom starships are going to be a thing.  Might as well get going on it.

Shades, radiators and PV can be mounted behind fairings along its length. This would ease deployment but still leaves dV and other practicality questions.


Hmm. While idly noodling a couple SRBs (no, I don't think this is a good idea) the unbidden image of an SS booster with a couple of F9 boosters strapped on settled in behind my fevered brow. Totally impractical but interesting.


Fighting gravity loss favors high thrust over high ISP during initial boost. OTOH, how to attach this and what mods would the launch table/tower need?


Naw, even with well characterized hardware like the F9 it'd be a diversion from solutions that promise faster results.


Still... 

[meekGee]

...

When you've eliminated all rational arguments, what remains, however irrational, must be transcendental. Or in Europe.

[catdlr]

author=Twark_Main link=topic=50157.msg2766178#msg2766178 date=1772727165
author=meekGee link=topic=50157.msg2766164#msg2766164 date=1772722982
author=Paul451 link=topic=50157.msg2765852#msg2765852 date=1772644528




Moderation,

OK, now let's get back to the topic, which is being derailed once again with heartless comments. Please focus on what is being said, not disparage people who say it. (Chris B).

Tony

[OTV Booster]

Weaponized Incompetence

Oh good. Social media has taught you a new way to belittle people. Just what we needed.

You're welcome. Now you can play the victim (your favorite  ) and avoid the rational arguments you took extra time to delete while quoting me. 

You can just say you don't have a counter-argument. That's perfectly fine! Everything's made up and the points don't matter...

Everything's okay as long as you avoid Paul-Doesn't-Like-Your-Idea Engineering LLC. Personally I have to try extra hard to problem-solve the other side, otherwise how do I know I'm not fooling myself? It's too easy to get the answer you want by unconscious sandbagging.

Check my logic:

The "optimal" pumping rate is achieved when the tanker boil-off from solar heating is balanced by the frictional heating from the pipe and pump losses.

If you pump any slower, you have more heating due to the Sun hitting the (presumably unchilled) tanker. If you pump any faster, the friction in the pipe will cause more heating. You want to hit the minimum of those two curves.

Based on this, knowing the pipe sizes and lengths and bends, we should be able to calculate the total pipe loss and therefore the "optimum" pumping rate.

Thoughts?

Makes sense if optimizing for transfer losses. If that is the thing to optimize for is another question.

Indeed. The hard part is always transforming the real-world problem into math, not turning the mathematical crank. Figuring out the correct optimization parameter is extra hard.

This does seem to be the correct "zeroth approximation" parameter to optimize for. It's naturally limited in time (because solar heating is relatively powerful), and it should also minimize the transfer hardware mass. So any second-order effects that pull in those directions shouldn't drag it too far down the curve.


To be clear I'm proposing this as a method to reverse-engineer the (back-of-the-envelope) transfer time by looking at the sizing of the plumbing SpaceX built. This isn't necessarily the optimal forward-engineering tactic that you'd use to size the plumbing, since AIUI it's under-constrained for that purpose.

The numbers would be interesting if only to suggest the utility of this optimization. There are a wide range of other as yet undefined impacts that might favor fast or slow transfer.


Assume a Mars bound ship needing four tanker loads. One option is to launch one tanker every 24+/- hours. This would allow but not necessarily demand a slow transfer. Another option is to launch two each from BC and the Cape and have them queue up awaiting their turn. This would almost but not quite demand a fast transfer.


If the depot has great thermal control and the tankers don't, fast transfer with either option looks good. At the other extreme, if neither has good thermal control slow may be the only savings to be found. Then there's everything in between.


And there's other issues. High vs low beta and pad availability being two variables that come to mind.


As you say, building the model to apply the math to is the hard part.


A note on the plumbing. With absolutely no proof, I think it was designed to service a stacked ship at one g. We (and probably SX too) don't know with absolute certainty that this plumbing will be used for transfer. I'm sure it'll be tested first but well, this is SpaceX we're talking about.