Starship On-orbit refueling - Options and Discussion

[Greg Hullender]

If the gendering issue does not get fixed, then tankers can never be target fuelers.  They can be rovers between two target fuelers, but they can't slurp methalox from an accumulator, rove up to HEEO or LO, and fuel a target.¹

Well, not quite. Since there will be two depots, make one male and one female. Make your tankers an even mix of male and female. Make your mission ship female, and fuel it from the male depot, once it's full. Use a male tanker to fuel from the female depot at the same time. Launch them together into HEEO, transfer from the tanker to the mission ship.

Again, that assumes it's trivial to put male vs. female QD hardware on any given Starship and that it's no problem for a launch tower to have two different genders of GSE. If you think those are unreasonable assumptions, I'd like to understand why. I'm not totally convinced that gender changes can't be done with a fairly simple adaptor that converts a female QD port to a male one, at which point it's really simple.

If you want something that can BLT back and forth to NRHO without EDL, then (I claim) you need an orbital drydock. If you want to send something disposable to NRHO, then, sure; use a modified depot as a mission ship.

Hey! Maybe combine the orbital drydock with a multi-depot layout and bill it as Elon's Orbital Garage! "Best prices for Meth and Lox above the Karmen Line!"

[butterwaffle]

Elon's Orbital Garage! "Best prices for Meth and Lox above the Karmen Line!"

Next stop for methalox in 1.3 light minutes!

[TheRadicalModerate]

If the gendering issue does not get fixed, then tankers can never be target fuelers.  They can be rovers between two target fuelers, but they can't slurp methalox from an accumulator, rove up to HEEO or LO, and fuel a target.¹

Well, not quite. Since there will be two depots, make one male and one female. Make your tankers an even mix of male and female. Make your mission ship female, and fuel it from the male depot, once it's full. Use a male tanker to fuel from the female depot at the same time. Launch them together into HEEO, transfer from the tanker to the mission ship.

So now you have a male depot for targets that need VLEO refueling, and a female depot for filling rover tankers?  What happens when you have a mission that needs an integral number of lifters, but one of the lifter's prop needs to be split between the VLEO depot and the rover?

This is insanity.

If it really, really turns out that androgyny is impossible, then the much, much simpler plan is to make the depot the rover for HEEO, and to use a tanker as the rover between VLEO and LO depots. 

You're manufacturing an engine life problem that probably doesn't exist.  And if it does exist, it won't for very long.  They're engineering Raptors for hundreds of restarts and hours of burn life--otherwise, boosters aren't going to work.

Again, that assumes it's trivial to put male vs. female QD hardware on any given Starship and that it's no problem for a launch tower to have two different genders of GSE.

Well, I've never heard of a problem with cryogenic plumbing leaking after replacing its fittings, so everything should be great.

[Greg Hullender]

So now you have a male depot for targets that need VLEO refueling, and a female depot for filling rover tankers?  What happens when you have a mission that needs an integral number of lifters, but one of the lifter's prop needs to be split between the VLEO depot and the rover?

You fly one extra tanker. Not the end of the world--especially if the prop stays around for a while. (Less good if it tends to boil off between missions.)

Well, I've never heard of a problem with cryogenic plumbing leaking after replacing its fittings, so everything should be great.

It's not like you to make bad-faith comments. You know very well I'm not proposing making changes on the pad.

Anyway, I don't think we can resolve this argument without actual data. We'll just have to watch SpaceX and see what they do. Until then, we'll just have to agree to disagree. On the bright side, it's clear that there are at least three different ways to do it (each with its own costs). That's a lot better than problems that appear to have no solution at all!

[OTV Booster]

I think you want to put a stake in the ground, insist that depots never boost into HEEO (or fire their engines at all after reaching final orbit), and bend everything else to make that constraint work.

How eager you are to transform optimization goals into straight-jackets.

Absolutely! It's one of the best ways to explore a space where there are a lot of variables. Fix one or two and see what happens to the rest.

I gather you dislike partial derivatives too? :-)

Again, I don't see why it's that hard for a depot to just have two different QDs...

It doesn't solve the problem.  If you want tankers to be able to connect directly to target Starships, then every tanker--which is the thing you most want to mass-reduce--needs the double QDs.  The plumbing is non-trivial.

No, you just need to make some number of male tankers and some number of female tankers. As far as I can tell, the difference amounts to swapping out a plate on the outside of the vehicle. (This may be my fatal misassumption though.) :-)

I agree with Twark in this case.  Depots are the cheapest components in the system. If there are restart or burn time limits, they're probably worth finding early.

Cheaper than a single metal plate? And the best limit is no limit. Never restarting those engines eliminates a whole class of potential problems--and losing a full depot is very expensive indeed. In fact, having two different ways to drain a full depot is a plus.

If it's going to be done, it's better to do it before the beginning of the refueling orbital test campaign, rather than the end.  Accumulating data on QD reliability is going to be a major requirement for making NASA comfy--and for silencing the naysayers.  You want as big a stable sample size as possible.

Again, the best change is no change. Use the same connectors you have now. Make the depot hermaphroditic, make the tankers mixed sex. I really do think that solves everything we've been talking about--and it's much simpler in operation. The question is how much does it cost?

A QD plate has plumbing, valving and latching associated. The idea of a variant with two QD plates works if the mass is low enough. Which variants gets them depends on the mission architecture.


IMO, subject to change, is over and under would add unwanted complexities. Which leaves ventral/dorsal. Which means ships without heat shields. Known ships are LSS and depot. A speculative super light weight, non EDL transfer tanker would work too.

[Greg Hullender]

A QD plate has plumbing, valving and latching associated. The idea of a variant with two QD plates works if the mass is low enough. Which variants gets them depends on the mission architecture.

IMO, subject to change, is over and under would add unwanted complexities. Which leaves ventral/dorsal. Which means ships without heat shields. Known ships are LSS and depot. A speculative super light weight, non EDL transfer tanker would work too.

I just spent a while studying Hungry Hungry Hopper: Starship's Propellant Distribution System, January 3, 2024, by Jax on Ringwatchers. Those (beautiful!) renders make it look as though one could add an additional QD to a Starship with very little additional plumbing, particularly since there will be no header tanks.

Can you think of any reason a male QD would cause problems? E.g. could it be more likely to leak? Has anyone looked closely at what happens when they remove the fuel from a vehicle today? Is it a fairly clean process, or does it leak a lot?

The only other thought I had was that it might have air-friction problems during launch, but I'd think there would be a lot of ways around that problem.

[TheRadicalModerate]

You fly one extra tanker. Not the end of the world--especially if the prop stays around for a while. (Less good if it tends to boil off between missions.)

You can fly one extra tanker and use the depot as an HEEO rover. 

FYI, for a v2 LSS, the difference between a v2 depot that does propulsive return to VLEO and a v2 tanker that does an EDL return is about 120t of prop.  Many assumptions on dry mass, Isp, FPR, and boiloff involved in that number.  A v3 LSS with a v3 depot doesn't need to refuel in HEEO at all.

Well, I've never heard of a problem with cryogenic plumbing leaking after replacing its fittings, so everything should be great.

It's not like you to make bad-faith comments. You know very well I'm not proposing making changes on the pad.

Well, I'll admit it was snarky, but it wasn't intended in bad faith.  If you change the plumbing on half the Starships but not the other half, then you have to change the GSE half of the time.  That requires changing the plumbing on the pad.

Update: This isn't much of a problem if you always have half the pads, each with exactly the same operational tempo, gendered one way or the other.  But that's a pretty big constraint.

I just spent a while studying Hungry Hungry Hopper: Starship's Propellant Distribution System, January 3, 2024, by Jax on Ringwatchers. Those (beautiful!) renders make it look as though one could add an additional QD to a Starship with very little additional plumbing, particularly since there will be no header tanks.

Can you think of any reason a male QD would cause problems? E.g. could it be more likely to leak? Has anyone looked closely at what happens when they remove the fuel from a vehicle today? Is it a fairly clean process, or does it leak a lot?

The current QD is in the attic, below the LOX dome, and the fill/drain plumbing pierces the LOX tank dome to get the fill/drain lines into the LOX and LCH4 sumps.  If you were to add a second connector, it needs to be on the mid-line, or the docking gets a lot more complex.  So you probably have to pierce the LOX tank wall itself with all lines, which gets a little weird.  Cryofluids and electrical and gas lines aren't a great combination.

I can't think of a reason why a male Ship-side QD would be more likely to leak.  I was objecting to the need to change the GSE out.

Note that an androgynous QD is likely reflected about the midline, so you might have something like LOX (male), LCH4 (female), GOX (male), GCH4 (female), etc., in a vertical column on the left side, and then (LOX) (female), LCH4 (male), GOX (female), GCH4 (male), etc., in a vertical column on the right side.  Note that you need to do the same thing with other gases, your power/data lines and (most importantly) your grappling/tensioning posts and sockets.

In effect, this is all of the plumbing for two connectors.  They're just smooshed together in a somewhat more compact form, and they remain in the attic, on the midline.  The question is whether the resulting plumber's nightmare in the attic itself is manageable.  Lots of joined piping.  And of course getting a QD with double the number of connectors still needs to be reliable, which is likely the biggest problem.

Edited for some semblance of clarity.

[OTV Booster]

A QD plate has plumbing, valving and latching associated. The idea of a variant with two QD plates works if the mass is low enough. Which variants gets them depends on the mission architecture.

IMO, subject to change, is over and under would add unwanted complexities. Which leaves ventral/dorsal. Which means ships without heat shields. Known ships are LSS and depot. A speculative super light weight, non EDL transfer tanker would work too.

I just spent a while studying Hungry Hungry Hopper: Starship's Propellant Distribution System, January 3, 2024, by Jax on Ringwatchers. Those (beautiful!) renders make it look as though one could add an additional QD to a Starship with very little additional plumbing, particularly since there will be no header tanks.

Can you think of any reason a male QD would cause problems? E.g. could it be more likely to leak? Has anyone looked closely at what happens when they remove the fuel from a vehicle today? Is it a fairly clean process, or does it leak a lot?

The only other thought I had was that it might have air-friction problems during launch, but I'd think there would be a lot of ways around that problem.

You lost me on "no header tanks."  Have I missed something?

[TheRadicalModerate]

You lost me on "no header tanks."  Have I missed something?

There's a certain amount of header downcomer jiggery-pokery that needs to be integrated with the fill/drain system.  An extra QD wouldn't help.

However, if you're going to make some tankers hermaphroditic, those do need header tanks.

[Greg Hullender]

A v3 LSS with a v3 depot doesn't need to refuel in HEEO at all.

Yes, that simplifies things a lot, doesn't it? Or it just lets you deliver more payload. Also a good thing.

If you change the plumbing on half the Starships but not the other half, then you have to change the GSE half of the time.  That requires changing the plumbing on the pad.

Or just have two hoses for each pad.

The current QD is in the attic, below the LOX dome, and the fill/drain plumbing pierces the LOX tank to get into the LOX and LCH4 sumps.  If you were to add a second connector, it needs to be on the mid-line, or the docking gets a lot more complex.  So you probably have to pierce the LOX tank itself, which gets a little weird.  Cryofluids and electrical and gas lines aren't a great combination.

Sorry. I don't see why it needs to do anything different from what the current one does--just on the opposite side of the depot. Can you elaborate? Remember that the depot has no header tanks (so that plumbing is gone), no flaps, and no heat shield. In SpaceX's drawings, it looks like a long, thin, featureless bullet.
[/quote]

Note that an androgynous QD is likely reflected about the midline, so you might have something like LOX (male), LCH4 (female), GOX (male), GCH4 (female), etc., in a vertical column on the left side, and then (LOX) (female), LCH4 (male), GOX (female), GCH4 (male), etc., in a vertical column on the right side.  Note that you need to do the same thing with other gases, your power/data lines and (most importantly) your grappling/tensioning posts and sockets.

In effect, this is all of the plumbing for two connectors.  They're just smooshed together in a somewhat more compact form, and they remain in the attic, on the midline.  The question is whether the resulting plumber's nightmare in the attic itself is manageable.  Lots of joined piping.  And of course getting a QD with double the number of connectors still needs to be reliable, which is likely the biggest problem.

Yes, the androgenous connector essentially doubles everything. Instead of a single LOX port, every ship has to have a male one and a female one, mirrored across the midline of the QD. I always ruled this out on the grounds that it adds a whole lot of extra plumbing to each and every vehicle. Better to just have everything be single gender except for depots, which are hermaphroditic, not androgenous. (I'm willing to consider the GSE to be a "depot on the ground," so it's hermaphroditic too.) :-)

So I think that gives us three solutions: roaming, androgenous, hermaphroditic, bisexual, and roaming hermaphroditic.

The roaming solution is nice because nothing changes at all. Only depots are male, so if a mission needs a refill in HEEO (or above), a depot flies out with it, fills it, and then flies back. Trouble is, with a depot, there probably is no "back." So you either expend it, figure out how to gently aerobrake it back, or else go with one the other three solutions. Also, it can't ever be serviced, so if there is between-flight maintenance required, you just have to expend it.

The androgenous solution is nice because everything plugs into everything else, but it adds a per-ship cost in terms of mass and complexity.

The hermaphroditic solution is nice because it only complicates the design of depots (and GSE), but it does mean at least a few tankers need to be male.

The bisexual solution is nice because it doesn't require any design changes at all, other than letting vehicles be male. But the gender ratio needs to be 50-50, and if even launch pads are gendered, it's pretty restrictive.

The roaming hermaphroditic solution is like the roaming depot, except instead roaming depots, it uses a handful of hermaphroditic tankers. Simplifies life for everyone who doesn't need a second refueling. When V3 makes it possible to avoid this step, you can just quit using that kind of tanker.

Did that about cover it?  (And, uh, is there any hope of better nomenclature?) ;-)

[Greg Hullender]

You lost me on "no header tanks."  Have I missed something?

I thought they were only needed for EDL. Depots don't EDL, of course. Or do they serve another purpose I'm unaware of?

[OTV Booster]

You lost me on "no header tanks."  Have I missed something?

I thought they were only needed for EDL. Depots don't EDL, of course. Or do they serve another purpose I'm unaware of?

Gotcha. Somehow i got a general SS drift, not depot specific.

[OTV Booster]

Ive got a dumb question that looks off target but does tie in to refueling.


How hard is it to rig a vacuum Raptor to gimbal? A depot, especially a depot awaiting a "depot kit" will be light. Once on orbit it will only need power to boost its decaying VLEO orbit, or in some conops, boost to HEEO or NRHO.


Could a depot make orbit with no SL raps and a single centrally mounted vacuum engine? It need not have a full 15deg swing but it will need a skirt extension or an interstage.


Would the simplified plumbing and lower weight justify the engineering effort?

[Greg Hullender]

Ive got a dumb question that looks off target but does tie in to refueling.


How hard is it to rig a vacuum Raptor to gimbal? A depot, especially a depot awaiting a "depot kit" will be light. Once on orbit it will only need power to boost its decaying VLEO orbit, or in some conops, boost to HEEO or NRHO.


Could a depot make orbit with no SL raps and a single centrally mounted vacuum engine? It need not have a full 15deg swing but it will need a skirt extension or an interstage.


Would the simplified plumbing and lower weight justify the engineering effort?

I suspect it would be easier to build a depot with no vacuum engines. Sea-level engines are probably enough to lift an empty depot to orbit. If the idea is that you'll never fire them again (using some other thrusters for station keeping and ullage), that'd probably be acceptable.

Is it worth it, though?

[TheRadicalModerate]

Or just have two hoses for each pad.

The hose isn't the problem.  You either replace the plate on the QD arm, or you need two complete arms in almost the same place.  How are you going to engineer that?  Why do you want to engineer that?

Sorry. I don't see why it needs to do anything different from what the current one does--just on the opposite side of the depot. Can you elaborate? Remember that the depot has no header tanks (so that plumbing is gone), no flaps, and no heat shield. In SpaceX's drawings, it looks like a long, thin, featureless bullet.

Nope, sorry, you're trying to assume the fleet with the split gendering as a predicate, and I don't accept the predicate.  The manufacturing, operational, and logistical costs of splitting the fleet in two are huge.  They're much larger than taking the hit for an extra 120t of prop to use the depot as the rover--especially when there are two different solutions that eventually mitigate the problem:

1) An androgynous QD, so tankers can be rovers.
2) A depot that can aerobrake.

I sorta had this up-thread, but let's look at three different cases, which are all for v2 hardware, all with a 200 x 26,600km HEEO, which is about the lowest subsync orbit I've seen.  (Any lower and you're deep into VAB #2.  You can save prop by going for the VAB 1/2 gap at 13,000km, but I'm pretty sure that comes close to maximizing the time you spend in the most intense parts of VAB #1.)

Depot goes to HEEO with full propulsive return:  1640t of prop
Tanker goes to HEEO, with EDL return: 1520t
Depot goes to HEEO, aerobrakes and circularizes back to VLEO: 1470t

Dry masses, prop capacities for tankers/depots, Isps, etc. are all open for debate, but these are apples-to-apples numbers.  The target is a v2 LSS for Option A.

Yes, the androgynous connector essentially doubles everything. Instead of a single LOX port, every ship has to have a male one and a female one, mirrored across the midline of the QD. I always ruled this out on the grounds that it adds a whole lot of extra plumbing to each and every vehicle.

It's more complex plumbing, but there's not a lot of extra mass.  Every pair of lines would merge together almost immediately.  The plate is obviously bigger, with more connectors, which may be a problem.  But I'd be surprised if you're adding more than 100kg.

The roaming [depot] solution is nice because nothing changes at all. Only depots are male, so if a mission needs a refill in HEEO (or above), a depot flies out with it, fills it, and then flies back. Trouble is, with a depot, there probably is no "back."

Why?  It propulsively went to HEEO when it was heavy, and now it propulsively returns to VLEO when it's light.  See the numbers above.  Your concern about the Raptors is a non-problem, or at worst an extremely minor problem.  (Note that a single RSL will work fine for moving a depot, so you need at least two permanent failures before you have to think about deorbiting the depot.)

With the exception of occasional one-off wild and crazy interplanetary missions, HEEO is as far as the VLEO depot needs to rove (and it doesn't even need to go there if it's v3 hardware).  Mars missions stage from VLEO, and an increasing number of lunar missions will refuel with a depot permanently stationed in LO, where a tanker can act as a rover between two depots.

The androgynous solution is nice because everything plugs into everything else, but it adds a per-ship cost in terms of mass and complexity.

The hermaphroditic solution is nice because it only complicates the design of depots (and GSE), but it does mean at least a few tankers need to be male.

The bisexual solution is nice because it doesn't require any design changes at all, other than letting vehicles be male. But the gender ratio needs to be 50-50, and if even launch pads are gendered, it's pretty restrictive.

No, a hermaphroditic depot solution doesn't complicate GSE, because you always use the female QD to fill/drain on the ground.  But the problem with it is that it implies the "bisexual" solution, both for rovers (tankers) and targets (LSS, etc.).  (And yeah, the nomenclature is getting pretty creepy.)

We can agree to disagree on this one, but that just sounds like a terrible burden to carry forward.

[OTV Booster]

Ive got a dumb question that looks off target but does tie in to refueling.


How hard is it to rig a vacuum Raptor to gimbal? A depot, especially a depot awaiting a "depot kit" will be light. Once on orbit it will only need power to boost its decaying VLEO orbit, or in some conops, boost to HEEO or NRHO.


Could a depot make orbit with no SL raps and a single centrally mounted vacuum engine? It need not have a full 15deg swing but it will need a skirt extension or an interstage.


Would the simplified plumbing and lower weight justify the engineering effort?

I suspect it would be easier to build a depot with no vacuum engines. Sea-level engines are probably enough to lift an empty depot to orbit. If the idea is that you'll never fire them again (using some other thrusters for station keeping and ullage), that'd probably be acceptable.

Is it worth it, though?

Yeah, is it worth it. The idea came from noodling optimizations. The reason for noodling a vac engine is the better ISP. A SL would be a lot easier. And only one to keep mass down if one could get a bare bones depot to orbit.

The idea of a depot never leaving its station isn't settled wisdom. There are reasonable arguments for one somewhere around the moon or HEEO. Best to keep options open while architecture is in flux.

I've a suspicion that with large enough filters the engines can last quite a while. In general, IC and turbine engines' most immediate maintenance consists of changing out filters, replacing heat damaged fluids and decoking where necessary. The little coking raptor faces can be handled procedurally and it doesn't have hydraulics or an oil sump. Propellant contamination is where big filters are needed. A lot is construction crud in the tanks and this should drop way down as star factory procedures tighten up. The combustion ice? Please tell me this is a temporary kludge and will disappear with V3.

Another optimization idea. As long it's aligned with the ship(s), settling thrust doesn't really care where the vector is pointing. Settle prograde when too low and antigrade when too high. It may not be much but like the say, a meter/second saved is a meter/second earned.


Edit: cleanup

[OTV Booster]

Or just have two hoses for each pad.

The hose isn't the problem.  You either replace the plate on the QD arm, or you need two complete arms in almost the same place.  How are you going to engineer that?  Why do you want to engineer that?

Sorry. I don't see why it needs to do anything different from what the current one does--just on the opposite side of the depot. Can you elaborate? Remember that the depot has no header tanks (so that plumbing is gone), no flaps, and no heat shield. In SpaceX's drawings, it looks like a long, thin, featureless bullet.

Nope, sorry, you're trying to assume the fleet with the split gendering as a predicate, and I don't accept the predicate.  The manufacturing, operational, and logistical costs of splitting the fleet in two are huge.  They're much larger than taking the hit for an extra 120t of prop to use the depot as the rover--especially when there are two different solutions that eventually mitigate the problem:

1) An androgynous QD, so tankers can be rovers.
2) A depot that can aerobrake.

I sorta had this up-thread, but let's look at three different cases, which are all for v2 hardware, all with a 200 x 26,600km HEEO, which is about the lowest subsync orbit I've seen.  (Any lower and you're deep into VAB #2.  You can save prop by going for the VAB 1/2 gap at 13,000km, but I'm pretty sure that comes close to maximizing the time you spend in the most intense parts of VAB #1.)

Depot goes to HEEO with full propulsive return:  1640t of prop
Tanker goes to HEEO, with EDL return: 1520t
Depot goes to HEEO, aerobrakes and circularizes back to VLEO: 1470t

Dry masses, prop capacities for tankers/depots, Isps, etc. are all open for debate, but these are apples-to-apples numbers.  The target is a v2 LSS for Option A.

Yes, the androgynous connector essentially doubles everything. Instead of a single LOX port, every ship has to have a male one and a female one, mirrored across the midline of the QD. I always ruled this out on the grounds that it adds a whole lot of extra plumbing to each and every vehicle.

It's more complex plumbing, but there's not a lot of extra mass.  Every pair of lines would merge together almost immediately.  The plate is obviously bigger, with more connectors, which may be a problem.  But I'd be surprised if you're adding more than 100kg.

The roaming [depot] solution is nice because nothing changes at all. Only depots are male, so if a mission needs a refill in HEEO (or above), a depot flies out with it, fills it, and then flies back. Trouble is, with a depot, there probably is no "back."

Why?  It propulsively went to HEEO when it was heavy, and now it propulsively returns to VLEO when it's light.  See the numbers above.  Your concern about the Raptors is a non-problem, or at worst an extremely minor problem.  (Note that a single RSL will work fine for moving a depot, so you need at least two permanent failures before you have to think about deorbiting the depot.)

With the exception of occasional one-off wild and crazy interplanetary missions, HEEO is as far as the VLEO depot needs to rove (and it doesn't even need to go there if it's v3 hardware).  Mars missions stage from VLEO, and an increasing number of lunar missions will refuel with a depot permanently stationed in LO, where a tanker can act as a rover between two depots.

The androgynous solution is nice because everything plugs into everything else, but it adds a per-ship cost in terms of mass and complexity.

The hermaphroditic solution is nice because it only complicates the design of depots (and GSE), but it does mean at least a few tankers need to be male.

The bisexual solution is nice because it doesn't require any design changes at all, other than letting vehicles be male. But the gender ratio needs to be 50-50, and if even launch pads are gendered, it's pretty restrictive.

No, a hermaphroditic depot solution doesn't complicate GSE, because you always use the female QD to fill/drain on the ground.  But the problem with it is that it implies the "bisexual" solution, both for rovers (tankers) and targets (LSS, etc.).  (And yeah, the nomenclature is getting pretty creepy.)

We can agree to disagree on this one, but that just sounds like a terrible burden to carry forward.

AAARRRrrrggghhh! This makes me crazy. RadMod, please, what is VAB (surely not Vehicle Assembly Building)?


I have no memory of seeing VAB in this discussion and can not divine its meaning from the context. Acronyms are intended to streamline communication. If the cost of efficiency is incomprehensibility, it's a net loss.


Please, please, take pity on us mere mortals.

[TheRadicalModerate]

AAARRRrrrggghhh! This makes me crazy. RadMod, please, what is VAB (surely not Vehicle Assembly Building)?

I have no memory of seeing VAB in this discussion and can not divine its meaning from the context. Acronyms are intended to streamline communication. If the cost of efficiency is incomprehensibility, it's a net loss.

Please, please, take pity on us mere mortals.

Sorry.  Van Allen Belt, in this context.

[TheRadicalModerate]

Yeah, is it worth it. The idea came from noodling optimizations. The reason for noodling a vac engine is the better ISP.

From TRM's Cryptic Prop Mass Estimator for Lunar Missions, using the same baseline v2 LSS as above for the target, and a propulsive return to VLEO for a depot with:

Isp=365s (approx average for 3 RSL v2's and 3 RVac v2's): 1643t of prop to VLEO.
Isp=370s (v2 RVac): 1626t
Isp=353s (v2 RSL in vacuum): 1692t

So it's not nothing either way, but it's not a huge amount.  All told, you'd get more bang for your buck engineering a depot that could aerobrake (see values up-thread).

A SL would be a lot easier. And only one to keep mass down if one could get a bare bones depot to orbit.

I doubt that a single RSL would have the T/W to get even a minimally fueled depot to VLEO.

As for a gimbaled RVac, it's a non-starter, not because the gimbaling is hard (although it might be) but because you can't put an RVac in the center of the thrust puck:  it's too tall.  It'll hang out below the skirt.  I guess you could build a taller hot-stage ring for the booster, but...

The idea of a depot never leaving its station isn't settled wisdom. There are reasonable arguments for one somewhere around the moon or HEEO. Best to keep options open while architecture is in flux.

I don't even think it's a good idea for a v2 depot servicing a v2 LSS.  However, a decent-sized nit to pick.  When we talk about the "location" of a depot, we're usually talking about energy, not its actual orbit.  It turns out that, for a VLEO circular orbit or some kind of lunar orbit, "energy" and "orbit" are semi-decent synonyms.¹

For an HEEO, they're not synonyms.  HEEOs, because they're eccentric, need to have RAANs and arguments of perigee that point in the right place at TLI.  (That in turn affects the inclination, but that's workable for a fairly small delta-v penalty.)  The best way to rejigger those parameters between missions is to bring the depot down to VLEO and start over.

There's another reason to return depots to VLEO each mission:  It's much, much, much cheaper to send lift tankers to VLEO to refill the depot than it is to send them to HEEO.  Spending the delta-v to boost the depot and recover it will be repaid many times over when filling a depot with 10-15 lift tankers' worth of prop.

I've a suspicion that with large enough filters the engines can last quite a while. In general, IC and turbine engines' most immediate maintenance consists of changing out filters, replacing heat damaged fluids and decoking where necessary. The little coking raptor faces can be handled procedurally and it doesn't have hydraulics or an oil sump. Propellant contamination is where big filters are needed. A lot is construction crud in the tanks and this should drop way down as star factory procedures tighten up. The combustion ice? Please tell me this is a temporary kludge and will disappear with V3.

The other reason to keep a full complement of RSLs on a depot is that it gives you all kinds of contingency procedures as engines begin to fail.  You can do VLEO-HEEO and HEEO-VLEO maneuvers to your heart's content with a single RSL, and you have three of them.  As long as the anomalous shutdown procedures for the engines are good enough to preclude big explosions, that means you don't even have to think about deorbiting a depot until it's already had two RSL failures.

And depots are cheap.  If they have even a 10-mission lifetime, that's going to be about $1.5M/mission in costs.  Optimizing around engine life is silly.

BTW:  I agree that this water ice thing had better be a passing phase.  I'd guess they'll be putting the right heat exchangers on the v3 to pull pre-combustion CH4 and O2 out of the system.

________
¹For NRHO this is true, because you can get to any lunar inclination from NRHO, and any Earth orbit inclination as well, due to weird stuff you can do with the flyby.  Things are a bit dicier for a frozen LLO (I think there's one near a polar inclination), but still manageable.  If you put a depot in a prograde circular orbit (PCO) or an elliptical frozen orbit (ELFO), things are dicier still.  I think you're OK with a distant retrograde orbit (DRO), but that's even higher energy than NRHO.

Note that arbitrary LLOs are not accessible for all missions. 

[Greg Hullender]

The idea of a depot never leaving its station isn't settled wisdom. There are reasonable arguments for one somewhere around the moon or HEEO. Best to keep options open while architecture is in flux.

I know this, but I'm trying to see what happens if you make that a requirement. I'm fully prepared to learn it's not a good idea, but I want to explore it far enough to understand why that might be.

As TheRadicalModerate points out, you can't really have a depot stationed in HEEO (they can only visit), but you could certainly have one in NRHO, for example. That is, I'm okay with the depot doing whatever it has to do before it gets to its designated station. I just want them to sit tight after that.

As long as the anomalous shutdown procedures for the engines are good enough to preclude big explosions, that means you don't even have to think about deorbiting a depot until it's already had two RSL failures.

Yes, if you're sure an engine failure isn't going to blow up the depot you just spent ten launches to fill up. it may take quite a bit of experience with reuse before anyone is confident of that though.
 

And depots are cheap.  If they have even a 10-mission lifetime, that's going to be about $1.5M/mission in costs. 

How do you figure this? Estimates I find just searching the web suggest it would cost ~$30 million to build one plus (if it's full) the cost of all the launches to fill it up, say, another $100 million. So if you lose a full one, you're out $130 million plus whatever the cost of the delay getting another one up and fueled. That's assuming it doesn't require lots of custom stuff like extensive solar panels for active cooling.