Starship On-orbit refueling - Options and Discussion

[Twark_Main]

That's what I did, I took the equilibrium and "goal seeked" it.  You can make a table with the two going in opposite directions.  You can write a Python script for that.  modern LLMs can write a python script (but I always verify the result).

Since you're making the table anyway, I just figured you'd use your favorite spreadsheet's Goal Seek feature.

I just make an error cell that calculates the difference between the radiant and conductive flux, then Goal Seek the surface temperature to make the error cell zero.

I had already assumed the heat shield faced earth shine - using the boiled methalox for that kind of positioning.

Do we have an order-of-magnitude of how much propellant this would require?

Note you just point the nose at the sun ALL THE TIME and I'm pretty sure that's the optimal orientation, and then you rotate around the axis to keep the heat shield pointed at the earth.  which is at its worst when 90 degrees from the sun (at the terminator).  When you are on the opposite side of the sun the nose is facing the earth, when you are on the side of the sun the tail is facing the earth.  You angle it just enough to hide the bare steel section, no more.  That's why I went with an average of 200W/m^2.

Yes, this matches my assumptions about the attitude control strategy.

As far as TUFI, I used the Shuttle tiles, which was 0.05 W/m-K at room temperature - is TUFI really that much worse?  Maybe you used the high temperature thermal conductivity?

Source is the NASA page I linked.

[InterestedEngineer]

I had already assumed the heat shield faced earth shine - using the boiled methalox for that kind of positioning.

Do we have an order-of-magnitude of how much propellant this would require?

well the boil rate is about 0.2kg/sec, at an Isp of 70 that's a thrust of 137N.

That should be plenty.  I would assume most of it will be wasted on offsetting outgassing.

[TheRadicalModerate]

Assume we face the sun, so the sunshine never hits the tanks, so we only have to deal with earthshine.

Note that this requires some amount of attitude control to oppose tidal forces.  It's small but probably not trivial.

Well, the boiled methalox has to go *somewhere*, attitude control seems like a good place for it.

A better place for it to go is into a cryocooler.

[OTV Booster]

Assume we face the sun, so the sunshine never hits the tanks, so we only have to deal with earthshine.

Note that this requires some amount of attitude control to oppose tidal forces.  It's small but probably not trivial.

Now the tank area facing the earth is about 270m² facing 100W of earthshine.  assuming none of this is radiated back out (it will be), that's a rate of heat absorption of 27kW.

Just taking the Google AI answer at face value, IR re-radiation from Earth is 150-350W/m², average 230W/m².  Average Earth albedo is about 0.3,¹ which would translate to ~400W/m².  So your average luminance from Earth is 630W/m².

OTOH, stefan-boltzmann for the entire starship at the boiling point of methane (130K) is -10kW/m².

The tiles themselves will have a heat gradient on them and stay at 170K while conducting little heat to the inside, so half the ship is really at -20kW/m².   So that's 34kW/m² net just to have the entire starship at boiling point of methane (an equilibrium).

I'm on shaky ground here, but I think you're doing this wrong.  Yes, the skin itself will be at 90K-130K, but you can't really calculate the emittance without knowing the thermal resistance of the insulation (SOFI, MLI, who knows?), which should give the gradient, and eventually the outer surface temperature.  The net result should give a better result than what you're estimating, i.e., the outside of the insulation should be hotter than you estimated.

________
¹I'm unclear if that number is for the whole Earth, or just the illuminated side.  If the former, then my back-of-napkin should be about right.  If the latter, divide by roughly 2.

For solar panel calcs total insolation at earth distance is 1.3kW in orbit and 1.0 at the surface. This may include near IR - or maybe not.


Reflected light would be for half the orbit while reemitted IR would be for the entire orbit. The 230W/m^2 average sounds like a useful BoE place holder for the entire orbit but cloud cover, specular reflections and high vs low beta orbits would add so much variability that every situation would be a situation.

[OTV Booster]

Assume we face the sun, so the sunshine never hits the tanks, so we only have to deal with earthshine.

Note that this requires some amount of attitude control to oppose tidal forces.  It's small but probably not trivial.

Well, the boiled methalox has to go *somewhere*, attitude control seems like a good place for it.

A better place for it to go is into a cryocooler.

That's a shoo in for the depot, but for a tanker or an end user ship?


For an end user ship, if not needed elsewhere, leave the depot attached for cryo services. This could be some combination of circulating propellant for cooling and providing shade until its time for the end user to light off and go.


During a depot filling campaign the most reasonable goal for the tanker is to git 'er in and git 'er transferred as fast as possible. My takeaway from the settling discussion of the last week or two is the tanker approach has to be slow and gentle. Twisting about to minimize thermal input seems to be at variance with this.


Sounds like a new restriction on the tanker approach. Ideally boiloff would be exactly equal to that needed via hot & cold gas thrusters for settling and approach - with no wild gyrations. I seriously doubt the gods will smile on this.


So, a new dimension in the trade space is the time for preliminary tanker settling and approach vs acceptable boiloff specifically during this maneuver.


Related would be tank pressure, specifically in the tankers. Has Elon shared any numbers from later tank tests to destruction? More mass towards higher tank pressure would give a higher boiloff temp but enough to be worthwhile? The hassle of special handling in the sausage factory has to be part of the equation.

[Twark_Main]

I had already assumed the heat shield faced earth shine - using the boiled methalox for that kind of positioning.

Do we have an order-of-magnitude of how much propellant this would require?

well the boil rate is about 0.2kg/sec, at an Isp of 70 that's a thrust of 137N.

That should be plenty.  I would assume most of it will be wasted on offsetting outgassing.

Ideally you want it to be a similar order-of-magnitude, so you're not wasting anything on either side.

That's 1080 kg per orbit, or 17.3 tons per day. That number is fine for up-and-down tankers, a drag for short-term depots (5-10%), and a showstopper for long-term depots to "flatten the curve" of Mars launch windows.

As for needing to offset outgassing, don't they use a non-propulsive vent? Or are you talking about volatile material sticking to the outside of the ship?

[TheRadicalModerate]

A better place for it to go is into a cryocooler.

That's a shoo in for the depot, but for a tanker or an end user ship?

I thought we were talking about long-term storage/accumulation for the depot.  The tanker will do whatever a vanilla Starship does.

The problem with cold gas is how much impulse you have before you have to wait for more stuff to boil off.  There's also a limit on autogenous pressurant for combusting liquid thrusters, but it's much higher.

For an end user ship, if not needed elsewhere, leave the depot attached for cryo services. This could be some combination of circulating propellant for cooling and providing shade until its time for the end user to light off and go.

This might be useful as a nanny strategy for HLS Ships, which need to loiter in lunar orbit (or VLEO, if we're in the D2+HLS transit situation) long after they've boiled dry, but then the depot is providing all the attitude control.

I don't think this happens any time soon.

During a depot filling campaign the most reasonable goal for the tanker is to git 'er in and git 'er transferred as fast as possible. My takeaway from the settling discussion of the last week or two is the tanker approach has to be slow and gentle. Twisting about to minimize thermal input seems to be at variance with this.

It certainly needs to be no faster than current prox ops are.  I don't think slosh is a major problem during the prox ops phase.  But even if docking is 10x as slow as, say, ISS, that's a matter of an extra few tens of minutes, not days.

Related would be tank pressure, specifically in the tankers. Has Elon shared any numbers from later tank tests to destruction? More mass towards higher tank pressure would give a higher boiloff temp but enough to be worthwhile? The hassle of special handling in the sausage factory has to be part of the equation.

Combusting gas makes all of this moot.

[TheRadicalModerate]

For solar panel calcs total insolation at earth distance is 1.3kW in orbit and 1.0 at the surface. This may include near IR - or maybe not.

Reflected light would be for half the orbit while reemitted IR would be for the entire orbit. The 230W/m^2 average sounds like a useful BoE place holder for the entire orbit but cloud cover, specular reflections and high vs low beta orbits would add so much variability that every situation would be a situation.

I think you're mixing apples and oranges here.  There are two separate issues:

1) How much irradiance (flux) is the Starship receiving, in W/m²?  Three contributors to this:
a) Solar flux (1362W/m² average).  Received when not in eclipse.
b) The outgoing longwave radiation (OLR) emitted by the Earth. Received always in LEO.  This is the ~230W/m².  Received all the time in LEO.
c) Albedo.  Mostly shortwave, and measured as a coefficient = incident/reflected radiation = ~0.3.  My question is whether that's an average across all phases of Earth illumination, or whether it's normalized to the full Earth.

2) Emittance from Starship.  That's strictly a function of temperature at the exterior of the Starship, which is a function of the gradient, which is a function of tank temperature and the thermal conductivity (or resistance, take your pick).

Heat absorbed by the Starship is (irradiance - emittance)*A.

[InterestedEngineer]

For solar panel calcs total insolation at earth distance is 1.3kW in orbit and 1.0 at the surface. This may include near IR - or maybe not.

Reflected light would be for half the orbit while reemitted IR would be for the entire orbit. The 230W/m^2 average sounds like a useful BoE place holder for the entire orbit but cloud cover, specular reflections and high vs low beta orbits would add so much variability that every situation would be a situation.

I think you're mixing apples and oranges here.  There are two separate issues:

1) How much irradiance (flux) is the Starship receiving, in W/m²?  Three contributors to this:
a) Solar flux (1362W/m² average).  Received when not in eclipse.
b) The outgoing longwave radiation (OLR) emitted by the Earth. Received always in LEO.  This is the ~230W/m².  Received all the time in LEO.
c) Albedo.  Mostly shortwave, and measured as a coefficient = incident/reflected radiation = ~0.3.  My question is whether that's an average across all phases of Earth illumination, or whether it's normalized to the full Earth.

2) Emittance from Starship.  That's strictly a function of temperature at the exterior of the Starship, which is a function of the gradient, which is a function of tank temperature and the thermal conductivity (or resistance, take your pick).

Heat absorbed by the Starship is (irradiance - emittance)*A.

There's also the idea that the tank is what matters.  The cross section of steel between the nose cone and the tanks is very small and 304L stainless is a poor heat conductor as well, so that's why you can round off the solar flux by pointing the nose at the sun.

[TheRadicalModerate]

The new HLS info published by SpaceX and the updated "moon" section contain the following two attached pictures.

Two important things to note:

1) The depot has the sockets and the tanker has the struts, which is kinda surprising to me.  I can't figure out why this would be useful.  Maybe the sockets weigh more?

2) The tanker and the depot are offset from one another, so the tanker will be farther forward than the depot when they're docked.  I assume that implies that there really are two QDs on the depot, one for launch and one, farther forward, for refueling.

Note that there's no sign of a QD in the artwork.

[TheRadicalModerate]

There's also the idea that the tank is what matters.  The cross section of steel between the nose cone and the tanks is very small and 304L stainless is a poor heat conductor as well, so that's why you can round off the solar flux by pointing the nose at the sun.

Yes, I agree that solar irradiance should be modest, which is why I focussed in the earlier post on just OLR and albedo.  OLR I'm pretty confident I have right.  It's albedo where I don't know if the average should be 0.3 or 0.15.

[Twark_Main]

The new HLS info published by SpaceX and the updated "moon" section contain the following two attached pictures.

Two important things to note:

1) The depot has the sockets and the tanker has the struts, which is kinda surprising to me.  I can't figure out why this would be useful.  Maybe the sockets weigh more?

When this came up in renderings the first time, my favorite hypothesis was that putting the active (and probably yes heavier) hardware on the tanker side is mostly about risk mitigation and maximizing the pace of hardware inspection and iteration.

Refilling is still early days. If a problem crops up, they can print a new part in days. If the hardware is damaged they can just fix it and re-fly it, and they don't lose the entire depot worth of propellant (bringing their refilling campaign back to square one).

I expect once the hardware is mature (in the usual three iterations), they'll swap the hardware over for a nice increase in tanker performance. They might even do that upgrade in V2.

2) The tanker and the depot are offset from one another, so the tanker will be farther forward than the depot when they're docked.  I assume that implies that there really are two QDs on the depot, one for launch and one, farther forward, for refueling.

I think it's just a trick of the camera angle. If you trace out the ellipse the separation plane on both vehicles, it looks like they're just next to each-other. Just a couple grain silos on flat ground. 

Note that there's no sign of a QD in the artwork.

Refilling through the HSR struts and docking struts confirmed. 

[Nevyn72]

Note that there's no sign of a QD in the artwork.

Refilling through the HSR struts and docking struts confirmed. 

However in this render there's an obvious raised box like structure below the 'launch QD' on the tanker...

[InterestedEngineer]

Is this the first time there's been an official render of a depot?

I seem to recall Elon saying there was no need for a depot.  But here it is, in all its glorious solar paint.

[Roy_H]

Related would be tank pressure, specifically in the tankers. Has Elon shared any numbers from later tank tests to destruction? More mass towards higher tank pressure would give a higher boiloff temp but enough to be worthwhile? The hassle of special handling in the sausage factory has to be part of the equation.

If the methane tank is 1 bar, and the O2 tank is 6 bar, then both are liquid in the 95°K to 100°K range. The tanks are designed to operate at 6 bar as they are pressurized to feed the turbo pumps.

I think it is a given that Depots and HLS will have Multi Layer Insulation, and active cooling with solar panels and radiators. The MLI will be wrapped around the hull and reflect heat from both Earth and the Sun. The goal is zero boil-off.

I believe the Depot and Starship Tanker should have pumps to pressurize the gas in the providing tanks and draw excess gas from the receiving tanks. This only needs to be about 1/2 bar differential. This way fuel transfer can be done with no loss. A Howden Roots Tri-RAM Model 409 would be a good choice. It's rather heavy at 152kg and that does not include the electric motor to drive it. A customized version could be built with lighter weight materials and save about 40%.  https://www.pdblowers.com/product/roots-tri-ram-model-409/?srsltid=AfmBOopLFBhBcK3PKg3Gjc1eRBZbF3fLMquL_FQVAbf8B3N0U2NNNak0

This is a positive displacement pump and can run in either direction. It does not matter if there are globules of liquid. Also, the vanes do not have to have seals that wear on them, small clearance as some blow back can be tolerated because only low pressure required. This avoids issues like possible cavitation if trying to pump the liquid directly. The liquid transfer at the bottom should be as simple as possible, just tubes and ball valves. I have attached a simple diagram showing the plumbing and the pumps as circles. A very low level thrust is required to keep the fuels settled on the bottom.

[TheRadicalModerate]

Is this the first time there's been an official render of a depot?

Nope, but it's the first time we've see a render with the struts and sockets in a stowed postioned, with an view straight at them.

I think it's just a trick of the camera angle. If you trace out the ellipse the separation plane on both vehicles, it looks like they're just next to each-other. Just a couple grain silos on flat ground.

I'm talking about the renders of the three different stacked variants.  There is no ellipse; the separation plane is a line.  The render viewpoint is normal to the vehicle's x-axis.  The boosters are all the same height, and if you pixel-count, the tops of the struts (which are in their stowed position on the tanker) are a lot lower down than the sockets on the depot.

[StraumliBlight]

The Depot Starship has 4 extra panels that are not visible on the HLS Starship, are these fold out radiators?

[TheRadicalModerate]

The Depot Starship has 4 extra panels that are not visible on the HLS Starship, are these fold out radiators?

Well, now you've dug just enough of the rabbit hole for me to dive into it:

We need to talk about which pictures are of what sides of which variants.  In the picture you've attached, we see the ventral (-z) side of the HLS.  How do we know this?  Because we don't see any docking probes, which we know are on the dorsal side of a vanilla Starship¹ or tanker, because the ventral side is covered with TPS.

On the other hand, we really have no clue which side of the depot we're looking at in your picture, other than we can assume that it has the sockets on it.  We've been assuming that's the dorsal side, but then we get this little tidbit from the Wall of Text, which has a lot to parse:

Starship V3 vehicles come equipped with docking ports and can be configured to act as tanker vehicles with the addition of docking probes. Starship also has a connection point where propellants are loaded onto the vehicle in preparation for launch that has been updated to enable on-orbit propellant transfer.

So many observations from two measly sentences:

1) First, note we have new nomenclature:
a) "Probes", FKA struts or docking thingies. 
b) We have "docking ports", which are ambiguous; I'm assuming that they're FKA "socket thingies", since a true IDSS-like docking port would be singular, not plural, and there's no evidence that either vanilla Starships or tankers have those, since they'd be blocked by the headers.
c) Then we have the "updated connection point", which is almost certainly the QD, modified for both vanillas/tankers and the depot.

2) It's still unclear whether the depot has a dedicated male QD for prop transfer, or whether the entire QD system has been made androgynous.  But from what we can tell from the depot picture, it's a eunuch.

3) The blurb says that "Starship V3 vehicles come equipped with docking ports," which implies that all the sockets have to be on the dorsal side, or they'd interfere with the TPS.  That's consistent with the picture of the lander picture, where they're not shown.  It's not inconsistent with the depot picture, as long as we assume we're looking at its dorsal side (and the QDs have been elided from the artwork).

4) But that makes the tanker picture inconsistent, because it's definitely the dorsal side (no TPS visible), but it only has probes.  Two possibilities:

4a) The probes replace the "ports" (sockets).  But I'm almost certain that the probes are mounted lower than the corresponding spots for the probes.  Maybe the blurb is incomplete, and the ports can be removed, with different mounting points for the probes.

4b) The artwork set is simply inconsistent.

5) Another reason to think the artwork is inconsistent or incomplete:  There's no evidence of any kind of QD on the side of the depot shown, and that must be the refueling side, irrespective of whether it's dorsal or ventral, because of the sockets/ports.  I guess this might support Twark's theory that the probes/sockets are the refueling pathways, but that's inconsistent with the statement that the QDs are being modified for refueling purposes.

My bet is that they're simply not ready to reveal the QD arrangement on the depot, so they omitted it/them from the artwork.

6) Bottom line:  I can't make the blurb and the pictures be completely self-consistent.  Am I missing something?


As for the question of the additional fold-outs:  If we're looking at the dorsal side of the depot and the ventral side of the HLS (which is the most likely explanation), then you don't really know what the dorsal side of the HLS looks like.  Maybe those fold-outs are present.

Note also that, in your picture, everything on the depot is retracted:  there are no visible solar or radiator arrays.  One would think it'd be weird to have to retract the arrays every time a tanker or HLS came for refueling, since that's likely to be failure-prone system, and a failure could take out the entire depot, along with gobs of previously-accumulated prop.

Again, I think it's likely that the depot picture is intentionally incomplete.

This brings us back to the infamous David Willis artwork:  If it's correct, the solar arrays form the basis of a half-assed sun shield, which could be made more fully-assed if the arrays included sun shades that folded out of the array to plug the gaps between them.  Or, maybe the extra doors are for sunshades to plug the gaps.

Still many questions.

_________
¹Note that there's no such thing as a vanilla Starship yet.  I've always assumed that the "vanilla" flavor would be the version that could deploy arbitrary payloads into space, and that would indeed need to be refuelable, based on the orbit requirements.  But we've only seen the Pez, which never needs to be refueled.  However, it's quite possible that the Pez will be the basis for the ship-to-ship transfer tests.

[Roy_H]

I had already assumed the heat shield faced earth shine - using the boiled methalox for that kind of positioning.  Note you just point the nose at the sun ALL THE TIME and I'm pretty sure that's the optimal orientation, and then you rotate around the axis to keep the heat shield pointed at the earth.  which is at its worst when 90 degrees from the sun (at the terminator).  When you are on the opposite side of the sun the nose is facing the earth, when you are on the side of the sun the tail is facing the earth.  You angle it just enough to hide the bare steel section, no more.  That's why I went with an average of 200W/m^2.

This is not as easy as you think. Either a massive gyro system (mass penalty), or near constant small thrusters (fuel cost). It would be interesting to do the actual calculations, but shouldn't be assumed to be cheap.

[Twark_Main]

I had already assumed the heat shield faced earth shine - using the boiled methalox for that kind of positioning.  Note you just point the nose at the sun ALL THE TIME and I'm pretty sure that's the optimal orientation, and then you rotate around the axis to keep the heat shield pointed at the earth.  which is at its worst when 90 degrees from the sun (at the terminator).  When you are on the opposite side of the sun the nose is facing the earth, when you are on the side of the sun the tail is facing the earth.  You angle it just enough to hide the bare steel section, no more.  That's why I went with an average of 200W/m^2.

This is not as easy as you think. Either a massive gyro system (mass penalty), or near constant small thrusters (fuel cost). It would be interesting to do the actual calculations, but shouldn't be assumed to be cheap.

Gyro is 1 metric ton. Since solar heating would >double the flux, this pays for itself in 1 orbit.