Starship On-orbit refueling - Options and Discussion

[Greg Hullender]

What this means is that any change in the sign of acceleration in any of the 3 translation or 3 roll axes will be followed some 10s of seconds later by a hard-to-predict series of wobbles or bumps in the orientation and acceleration of the starship - as tons of fuel splashes down onto the other side of the tank or turbulently changes its flow direction. Needless to say, those uncontrolled wobbles and bumps are hugely dangerous for prox-ops.

.
.
.

The nose-to-tail geometry, both pointing in the same direction, appears to allow a safe prox-ops approach.  The trailing starship would approach from below, rising up as it translates along the orbital path into position, while pitching its nose towards the tail of the leading ship (continuously decelerating in pitch rate until grapples activate).  I don't see a safe prox-ops for the other geometries.  Do you?

Since the accelerations in question will be on the order of 20 micro-g, this may not be a bad as you think. Check this document for "Low Acceleration Settling."

[TheRadicalModerate]

If there are structural members within, say, a meter of the hoses, would that work?  I don't know.

On the other hand, if you have one structural member latched onto the QD and one latched several meters away, I'm pretty confident the pair would bear the loads pretty easily.

The rare (or maybe not so rare) "Argument from I Didn't Feel Like Doing The Math."

Challenged accepted--sorta.  See below.

NASA seems to have no problem with structural connections close to fluid connections (see: all their docking standards), so why should TheRadicalModerate?

It's not a question of structural connections close to fluid connections.  It's a question of whether you can build a structural connection that couples reliably and can still support the worst-case torsion of the system.

You want to put the connection waaaaaaaay at the back of the vehicle, far from the center of gravity, and you want a soft-dock instead of a grapple/berthing.  (You can't do a grapple without an arm, and that really will mess up the aerodynamics, even if your androgynous QD plate doesn't.)  That subjects the vehicle to two separate torsional risks:

1) An RCS error (stuck thruster, software problem, etc.).  This is dependent on the RCS thruster, for which we don't know the thrust.  But to have decent control authority over a 270t vehicle that's 50m long, it's fair to say that the thrust isn't going to be as small as on most other vehicles.

2) Much more serious IMO, you have a big soft-capture problem, because all of the translation velocity errors are going to generate big torques.

Let's continue with your Orbiter comparison, because we know that the Orbiter docked away from the center of mass:

Orbiter Length:  37m
Docking position (from nose): 11.6m (derived from measuring pixels)
Nominal Center of Mass: 67% of length = 24.8m (I got this from here)
Dry Mass: 78t
Max Payload for ISS docking:  Couldn't find this, but I'd be surprised if it was >8t.

So, assuming that the up-payload was placed on the CoM, that would give you a moment of inertia of (86t)*(24.8-11.6)² = 1.5E7kg-m².

(Note that I'm assuming that the ISS has infinite mass for this little exercise.  I'll also be assuming that the depot Starship has the same infinite mass.  So the Orbiter and the tanker Starship are the sole active components in the system.  Obviously not true, but close enough for government work.)

We can't really compute torques without knowing a whole bunch about stiffness of the vehicles and the torsional springs in the docking system, but we can compute an angular momentum in Orbiter-normalized units, where ⍵_orb = (24.8-11.6) * (maxTranslate).  This would be the rotational speed after contact if the Orbiter and ISS were completely rigid.

Let's assume that maxTranslate is the same for the Orbiter and a tanker, so if we know the CoM of the tanker and the docking point, we can get an apples-to-apples comparison.

I have a model that shows that a Starship tanker with 150t of prop and 120t dry mass has a CoM that's about 30m from the nose.  If we assume that the QD is 49m from the nose, then I_tanker = (270t)*(49-30)² = 9.7E7kg-m², and ⍵_tanker is 1.4x higher than the Orbiter.

All told, angular momentum for the Starship tanker is 9.1x larger than that on the Orbiter, given similar translational velocity limits.

I never got an answer from you (other than something cryptic and non-responsive) about whether your "tilt the QD plate" scheme tilted it in the x-y plane or x-z plane.  Either way, though, a docking miss is going to result in a collision:  tilted in the x-y plane, you're going to have elonerons colliding, and in the x-z plane, the fuselages will collide.

The more I think about it, I'm pretty sure that any kind of traditional soft-dock is fraught with peril.  I think they'll use grappling--kinda like berthing but with just enough stability to connect the plumbing.  Given that you can't put active grapples on the Starship during launch or landing, they'll have to be deployed.  And if you're going to do that, deploying the QD-to-QD adapter seems like the least difficult thing to do.

[Overtone]

What this means is that any change in the sign of acceleration in any of the 3 translation or 3 roll axes will be followed some 10s of seconds later by a hard-to-predict series of wobbles or bumps in the orientation and acceleration of the starship - as tons of fuel splashes down onto the other side of the tank or turbulently changes its flow direction. Needless to say, those uncontrolled wobbles and bumps are hugely dangerous for prox-ops.

.
.
.

The nose-to-tail geometry, both pointing in the same direction, appears to allow a safe prox-ops approach.  The trailing starship would approach from below, rising up as it translates along the orbital path into position, while pitching its nose towards the tail of the leading ship (continuously decelerating in pitch rate until grapples activate).  I don't see a safe prox-ops for the other geometries.  Do you?

Since the accelerations in question will be on the order of 20 micro-g, this may not be a bad as you think. Check this document for "Low Acceleration Settling."

Thanks for the very interesting reference!   However, I don't think it directly addresses my concern.  Those low accelerations are for settling propellant.   If starships relied on those low accelerations during docking ops, wouldn't the time from <at relative rest at standoff distance> to <grappled and ready to connect to the QD> be excessively long?

Do you happen to have a reference that shows typical accelerations used during precise navigation in the last 100 meters of current docking operations?

[edzieba]

What this means is that any change in the sign of acceleration in any of the 3 translation or 3 roll axes will be followed some 10s of seconds later by a hard-to-predict series of wobbles or bumps in the orientation and acceleration of the starship - as tons of fuel splashes down onto the other side of the tank or turbulently changes its flow direction. Needless to say, those uncontrolled wobbles and bumps are hugely dangerous for prox-ops.

.
.
.

The nose-to-tail geometry, both pointing in the same direction, appears to allow a safe prox-ops approach.  The trailing starship would approach from below, rising up as it translates along the orbital path into position, while pitching its nose towards the tail of the leading ship (continuously decelerating in pitch rate until grapples activate).  I don't see a safe prox-ops for the other geometries.  Do you?

Since the accelerations in question will be on the order of 20 micro-g, this may not be a bad as you think. Check this document for "Low Acceleration Settling."

Thanks for the very interesting reference!   However, I don't think it directly addresses my concern.  Those low accelerations are for settling propellant.   If starships relied on those low accelerations during docking ops, wouldn't the time from <at relative rest at standoff distance> to <grappled and ready to connect to the QD> be excessively long?

Do you happen to have a reference that shows typical accelerations used during precise navigation in the last 100 meters of current docking operations?

Closing from 100 metres with an acceleration cap of 20 microgee = ~1400s for a brachistochrone approach: accelerate halfway to about 0.15m/s, then decelerate for the other half to a stop. Or less than 25 minutes.

[Greg Hullender]

Closing from 100 metres with an acceleration cap of 20 microgee = ~1400s for a brachistochrone approach: accelerate halfway to about 0.15m/s, then decelerate for the other half to a stop. Or less than 25 minutes.

I get the same answer. (I don't know about you, but I always like it when someone double-checks my math--at least if they get the same answer.) :-)

Fuel use seems to be really small too. If I'm doing this right, (and assuming equal specific impulse), the fuel used to accelerate for one second at one full g should last almost 14 hours at 20 µg.

This must be why they're not really worried about settling burns wasting a whole lot of fuel. (Lots of people have said this, but I never ran the numbers before.)

[TheRadicalModerate]

Closing from 100 metres with an acceleration cap of 20 microgee = ~1400s for a brachistochrone approach: accelerate halfway to about 0.15m/s, then decelerate for the other half to a stop. Or less than 25 minutes.

I'd expect to spend a lot more propellant on attitude control during prox ops than on actually closing and braking.

Overtone's note about propellant slosh isn't as serious as it would be at higher acceleration, but it still puts limits on your attitude and translation errors, especially just before docking/grappling.  I wonder if a cryogenic propellant management device could tolerate a large enough acceleration to be useful?

[Greg Hullender]

I had a thought about making an adaptor work without having to fly it down from the cargo bay. Suppose it just lay on its side right next to the QD, attached to the body of the rocket by a hinge. I visualize it lying along the body of the rocket from the QD forward.

Once in orbit, it would rotate out 90 degrees, plugging itself into the QD in the process. The adaptor could then be as long as necessary--or even be a robot arm--with a suitable connector at the end for the other Starship.

This presupposes a few things that I'm not really sure of. First, is it possible to design something like that that rotates into place and yet still makes a tight-enough seal? (And doesn't interfere with the GSE that fuels it on the ground.) Second, is this something small enough to be covered with a "shroud" or something like that and still keep the rocket aerodynamic enough?

It still has to be powered and controlled somehow (and, presumably, the pumps that control the actual refueling operation would go inside the adaptor somehow), but it at least seems reasonable to me. (Apologies if someone has already proposed this.)

[OTV Booster]

Closing from 100 metres with an acceleration cap of 20 microgee = ~1400s for a brachistochrone approach: accelerate halfway to about 0.15m/s, then decelerate for the other half to a stop. Or less than 25 minutes.

I'd expect to spend a lot more propellant on attitude control during prox ops than on actually closing and braking.

Overtone's note about propellant slosh isn't as serious as it would be at higher acceleration, but it still puts limits on your attitude and translation errors, especially just before docking/grappling.  I wonder if a cryogenic propellant management device could tolerate a large enough acceleration to be useful?

During a low acceleration maneuver surface tension would probably be the dominating force influencing propellant moving as a large glob vs splashing.


If it's moving in a glob might real time modeling predict the resulting thrusts needed to keep it as a single body and still achieve the objective of the maneuver? Am I overthinking this?

[Paul451]

I had a thought about making an adaptor work without having to fly it down from the cargo bay. Suppose it just lay on its side right next to the QD, attached to the body of the rocket by a hinge. I visualize it lying along the body of the rocket from the QD forward.

You could certainly do this with a simple adaptor plate. (Or just have a second (male) QD plate on the depot.) But my understanding was that the "adaptor" was proposed to carry all the depot-specific hardware, including the cryocooler/docking-systems/etc, allowing any tanker to serve as the depot/accumulator. Something that size, probably can't be kept in in the engine-skirt, but would have to be carried into orbit in the nose. (This is the case, whether the depot hardware is carried as a free-flyer/crawler, or built in permanently.)

[Greg Hullender]

I had a thought about making an adaptor work without having to fly it down from the cargo bay. Suppose it just lay on its side right next to the QD, attached to the body of the rocket by a hinge. I visualize it lying along the body of the rocket from the QD forward.

You could certainly do this with a simple adaptor plate. (Or just have a second (male) QD plate on the depot.) But my understanding was that the "adaptor" was proposed to carry all the depot-specific hardware, including the cryocooler/docking-systems/etc, allowing any tanker to serve as the depot/accumulator. Something that size, probably can't be kept in in the engine-skirt, but would have to be carried into orbit in the nose. (This is the case, whether the depot hardware is carried as a free-flyer/crawler, or built in permanently.)

I was thinking of the adaptor handling the docking system, but not the cryocooling. I could see the cryocooler sitting at the top of the Starship, assuming it can cool the tanks from there--via conduction, maybe.

[TheRadicalModerate]

I had a thought about making an adaptor work without having to fly it down from the cargo bay. Suppose it just lay on its side right next to the QD, attached to the body of the rocket by a hinge. I visualize it lying along the body of the rocket from the QD forward.

You could certainly do this with a simple adaptor plate. (Or just have a second (male) QD plate on the depot.) But my understanding was that the "adaptor" was proposed to carry all the depot-specific hardware, including the cryocooler/docking-systems/etc, allowing any tanker to serve as the depot/accumulator. Something that size, probably can't be kept in in the engine-skirt, but would have to be carried into orbit in the nose. (This is the case, whether the depot hardware is carried as a free-flyer/crawler, or built in permanently.)

I was thinking of the adaptor handling the docking system, but not the cryocooling. I could see the cryocooler sitting at the top of the Starship, assuming it can cool the tanks from there--via conduction, maybe.

That would require new plumbing. If you take boil off from the QD repress lines, you can reliquify and feed the liquid back through the fill/drain lines.

If you’re going to add plumbing, the nose-to-tail configuration might make more sense. It’s a lot easier to deploy.

[john smith 19]

I was less than clear. Floppy stuff=PV, radiators and MLI. Deploying this stuff is always a point of concern. Folding it back up for later reuse would be... Well, interesting.

Absolutely.

Hence my belief they would be the key issues for a Mars SS flight, and my belief they'd have to take a "minimum viable vehicle" IE no radiators, no PV, if they wanted to even try for a launch this year.

The "Space mechanisms" conference proceedings make very interesting reading for people who like seeing (to coin a phrase) "tricky mechanical gadgets." One of my favourites was the system that deployed the "aero-disk" on the front of the Trident missile to lower its drag. An amazing mechno-explosive subsystem. Stunning. 

One theme that repeatedly comes up in these volumes are the issues around  friction.
 Incompatible materials, faulty lubrication etc stopping something deploying (or less frequently stowing). Thermal cycling also seems to have caused an inordinate amount of trouble over the years. Parts distorted so they can't  move. Parts (specifically long booms for example) unevenly heated so they bend, and the mass at the end starts to cause the boom to rotate (in a very peculiar motion) going into and out of Earth eclipse.

While I expect there are several options that could be made to work eventually keeping it as simple as possible sounds like a good idea. Also keeping things fairly close together seems like a good way to keep things stiff, and hence less prone to distortion.

[LMT]

"NASA’s Initial Artemis Human Landing System"

w/depot.  h/t Chris Bergin.

[Hog]

h/t= hat tip = tip of the hat

[Greg Hullender]

"NASA’s Initial Artemis Human Landing System"

w/depot.  h/t Chris Bergin.

Well, that was educational! Still lots of issues to talk about, but that seems to remove a lot of degrees of freedom.

So if I read this correctly, the plan is for a depot with no active cryocooler--just "exterior optical properties" and lots of spare capacity to allow for tolerance of some amount of boiloff. That simplifies things a lot.

Presumably something like Solar White would suffice for the HLS, since it only needs to "loiter" in the NRHO.

And, as others have said, the tanker is just a regular starship with larger tanks and little or no cargo space.

Instead of (or maybe in addition to) a QD plate, I'd expect the depo to have something like the existing GSE that's used to fill/drain the starships on the launchpad. Exactly how they'd squeeze that in there is a good question, but it seems to make the most sense, given everything else.

At least, that's my read on it.

[spacenut]

I still think the tail to tail docking for fuel transfer was the most optimum safety and good solid docking of all connections. 

[Stan-1967]

Closing from 100 metres with an acceleration cap of 20 microgee = ~1400s for a brachistochrone approach: accelerate halfway to about 0.15m/s, then decelerate for the other half to a stop. Or less than 25 minutes.

I'd expect to spend a lot more propellant on attitude control during prox ops than on actually closing and braking.

Overtone's note about propellant slosh isn't as serious as it would be at higher acceleration, but it still puts limits on your attitude and translation errors, especially just before docking/grappling.  I wonder if a cryogenic propellant management device could tolerate a large enough acceleration to be useful?

I appreciate all the thought going into this discussion on docking & prop slosh. 

Would it make sense to dock using primarily V-Bar approaches between the tanker & depot? If you align the longitudinal axis of the ships with the velocity vector, you would be settling the prop  towards the aft end of each respective vehicle during your approach & hard dock.  You would get some minimal sloshing from counter thrusting in the radial vector.  It seems like prop baffles already in the bottom of the LOX/CH4 tanks should dampen the transient forces when hard dock occurs.

The last 10 meters or so may be better done on an R-Bar, as I think the slosh from the lighter vessel that undergoes a greater differential acceleration at the instant of hard dock would see the prop translating such that the tank walls would absorb the forces in compression or tension along the outer radius of the tank walls, vs. having prop slosh against the domes?

[Greg Hullender]

I was thinking of the adaptor handling the docking system, but not the cryocooling. I could see the cryocooler sitting at the top of the Starship, assuming it can cool the tanks from there--via conduction, maybe.

That would require new plumbing. If you take boil off from the QD repress lines, you can reliquify and feed the liquid back through the fill/drain lines.

Well, that's what you do if you're trying to be perfect, but I was thinking of simply cooling the stainless steel top of the upper tank as a way of sucking out some of the excess heat. Without adding any plumbing below that point.

Looking at a cut-away view of Starship, I see the methane tank is above the oxygen tank, which messes this idea up, but imagine that, in the depot, the oxygen tank were at the top, not the bottom. (Is there a reason the methane tank is on top?) Then you could cool the top of the LOX tank to just above the freezing temperature of oxygen. There would be a temperature gradient, with the tank getting warmer the further down you went, but steel conducts heat pretty well, so there might not be that much difference from the top to the bottom. Even if it didn't work perfectly, that'd be okay if the goal were merely to reduce boiloff--not eliminate it entirely.

You wouldn't want the methane tank to freeze, of course, but there's already insulation separating them, and adding heat (at the very bottom, of course) is a much easier problem. Also, you could do an ullage burn from time to time, which would pull the colder tank contents to the warmer bottoms and lift the warmer contents (and all the gas) to the cold upper surfaces.

No extra plumbing required. However, I suspect swapping the oxygen and methane tanks would count as a major change. Pity. Otherwise, this would offer a way to add cryocooling with (what seems to me to be) a simple incremental change. 

[DanClemmensen]

I was thinking of the adaptor handling the docking system, but not the cryocooling. I could see the cryocooler sitting at the top of the Starship, assuming it can cool the tanks from there--via conduction, maybe.

The cryocooler is cooling the cryogenic liquid inside a large steel tank. Cooling of the tank walls will be by convection whether you want it to or not.
(Edited to fix the quoting.)

[OTV Booster]

I was less than clear. Floppy stuff=PV, radiators and MLI. Deploying this stuff is always a point of concern. Folding it back up for later reuse would be... Well, interesting.

Absolutely.

Hence my belief they would be the key issues for a Mars SS flight, and my belief they'd have to take a "minimum viable vehicle" IE no radiators, no PV, if they wanted to even try for a launch this year.

The "Space mechanisms" conference proceedings make very interesting reading for people who like seeing (to coin a phrase) "tricky mechanical gadgets." One of my favourites was the system that deployed the "aero-disk" on the front of the Trident missile to lower its drag. An amazing mechno-explosive subsystem. Stunning. 

One theme that repeatedly comes up in these volumes are the issues around  friction.
 Incompatible materials, faulty lubrication etc stopping something deploying (or less frequently stowing). Thermal cycling also seems to have caused an inordinate amount of trouble over the years. Parts distorted so they can't  move. Parts (specifically long booms for example) unevenly heated so they bend, and the mass at the end starts to cause the boom to rotate (in a very peculiar motion) going into and out of Earth eclipse.

While I expect there are several options that could be made to work eventually keeping it as simple as possible sounds like a good idea. Also keeping things fairly close together seems like a good way to keep things stiff, and hence less prone to distortion.

A tangent, not to be taken too far. Most folding structures appear to be light, high precision, close tolerance designs. Would there be any gains in a heavier, low tolerance and sloppy structure? It's not like an extra 20 kilos is going to hurt much.