Starship On-orbit refueling - Options and Discussion

[Twark_Main]

It's not uncommon to model a "highlight" feature in higher fidelity for the scene it's being highlighted in, then use a lower-fidelity model for other shots. But the fact that they did model a more detailed version is probably not an accident.

This design makes a lot of sense to me. The LOX and CH4 ports are maximally far apart, minimizing the chances of detonation in case of a leak.

The "standpipes" bring ullage GOX down from the top of the oxygen tank, and bring liquid fuel up from the bottom of the methane tank. So we're actually looking at multiple pipes (3-4), even though they may not be modeled as such.

The docking procedure is nice and symmetrical, without any large torque or slosh concerns. The docking hardware is located maximally in the lee of the superheated reentry plasma, minimizing wear and tear. The docking hardpoints are located far apart to efficiently transfer torque, and they're both located at a dome-wall interface which is already stronger to begin with. Very nice.

The only thing I'm not sure about is, why run the standpipes on the outside? It seems like they would be more protected inside. It's the same propellant inside the pipe as in the tank, so it's not about separating fuel and oxidizer. So why? 


Edit: Ok, I'm an idiot. All the pipes I mentioned are located inside the tanks.

The standpipes are actually the contingency transfer pipes. If one of the two cryo fluid + ullage gas connectors fails, they can simply purge the connector to vacuum, throw a couple valves in each ship, and use the other connector to transfer the stranded propellant.    It takes over twice as long, but this way the entire tanker mission won't be a waste. Genius!

This explains why they're on the outside, and why each appears to be one continuous pipe directly connecting the fuel and oxidizer ports, and why there are two (liquid and gas).

By doing this they can minimize mission risk during the development of their cryogenic propellant transfer technology.

[LMT]

EVLEO on-orbit ISRU manufacture of ASCENT propellant (formerly AF-M315E) for tugs, payloads, etc. could be accomplished within a closed electrochemical nitrogen cycle.  Notably, ASCENT vacuum-manufacture is easier in space than on the ground.

Production of ammonia and nitrate is only part of the greater challenge, but it's remarkable that high efficiency is reported in recent experiments, e.g., nearly 100% efficiency in electrochemical ammonia synthesis. 

Q:  What is the most efficient electrochemical nitrate synthesis method?

Each constituent of an ASCENT tank load -- pressurant and ionic liquid -- can be sourced in this nitrogen cycle, with 96% of the mass potentially accessible in the thermosphere.

Some fraction of the remaining feedstock (H2) is potentially sourced from docking Starships and other ships.  E.g., H2 can be reduced out of excess, waste Starship gCH4 pressurant, prior to undocking and EDL.

N2 is an abundant feedstock molecule, so in this application only two reactions would be needed from the electrochemical nitrogen cycle: 

1.  nitrogen reduction reaction (N2RR) for ammonia, and

2.  nitrogen oxidation reaction (N2OR) for nitrate.

Refs.

Du, H.L., Chatti, M., Hodgetts, R.Y., Cherepanov, P.V., Nguyen, C.K., Matuszek, K., MacFarlane, D.R. and Simonov, A.N., 2022. Electroreduction of nitrogen with almost 100% current-to-ammonia efficiency. Nature, 609(7928), pp.722-727.

McLean, C.H., Viesca, S.G., Deininger, W.D., Unruh, B.W., Spores, R.A., Frate, D.T., Yim, J.T., Johnson, W.L., Aggarwal, P.K. and Reed, B.D., 2014, March. Green propellant infusion mission program overview and status. In 2014 IEEE Aerospace Conference (pp. 1-20). IEEE.

Yang, X., Mukherjee, S., O'Carroll, T., Hou, Y., Singh, M.R., Gauthier, J.A. and Wu, G., 2023. Achievements, Challenges, and Perspectives on Nitrogen Electrochemistry for Carbon‐Neutral Energy Technologies. Angewandte Chemie, 135(10), p.e202215938.
 

[InterestedEngineer]

Did anyone consider surface tension methods?

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

[Greg Hullender]

Did anyone consider surface tension methods?

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

I think a weakness of the prior discussion is that everyone pretty much assumed that the general problem of extracting cryogens from a tank in zero or micro-gravity was already solved, so there hasn't been much said about it. (Or else I somehow missed it.)

Using search terms from the Wikipedia article, I found a more recent conference article, "A Detailed Historical Review of Propellant Management Devices for Low Gravity Propellant Acquisition," which is very informative. It suggests that capillary systems are only necessary if the ullage burn is under 1E-4 g. (Of course, you might need it to feed the ullage engine just to get the burn started, I suppose.) Otherwise, "a hole in the bottom of the tank" could be all you'd need.

That article suggests that, of the different surface-tension methods, only the "Screen Channel Liquid Acquisition Devices" aka "Gallery Arms" have the ability to manage the large propellant flows that a depot would require. (p. 15 "C. Advantages and Disadvantages.") The challenge would apparently be to make one that was robust enough for repeated use.

[Greg Hullender]

I think you could get useful amounts of artificial gravity if you simply tethered two depots together with cables no more than 500m long. While ullage burns are probably fine, this does suggest an alternative way to do it.

Suppose you have two depots of equal mass connected by a cable of length 2L such that the cable is always on a line through the center of the Earth. Assume the midpoint of the cable is in a circular orbit at a height, h, of 100 km, so the ensemble has a period of about 85 minutes and the angular velocity, ω, will be about 0.0012 radians/sec.

At distance L from the center of mass, the centrifugal acceleration will be ω²(R+h+L), where R is the radius of the Earth. The acceleration due to gravity will be -μ/(R+h+L)². Using the fact that L is much smaller than R+h, the acceleration at the two depots is very close to 3ω²L. If you wanted 100 micro-gravities, L of 222 would suffice, for a total cable length of 444m. For a milli-g, you'd need a 4.4 km cable, which is still not unreasonable. (Someone should check my math though.)

The rotation should be slow enough that a vehicle could dock with either one without any need to stop the rotation.

Obviously there are stability issues, but if you make one depot heavier than the other, it will tend to stay down. Or just make a counterweight that's heavier than a full depot plus Starship. Maybe that's still too much work just to avoid ullage burns, but it doesn't seem to add all that much complexity. Or so it seems to me.

[Brigantine]

I think you could get useful amounts of artificial gravity if you simply tethered two depots together with cables [...]

1) Depending how you do the cables, you can control the rotation of 1 depot around its CoM independently of its motion about the cable - at least for short periods << 2π/ω, at much lower cost than the docking tanker/ship constantly thrusting to 'hover' and dock. So your ω can be perhaps 6x faster (depending on how long you need to rendezvous and dock)
[EDIT: you lose the favourable tidal forces, but get much better ω². Solving for 10mms⁻² = ω²L - (tidal force with cable horizontal)... For 15 min rotation, ω = π/450. Tidal force is μ/(R+h)² * L/(R+h) i.e. gravity vector misaligned by L/(R+h) radians. 2L ~= 425m.]

2) I feel there should be a much easier solution, just involving the 2 rendezvousing vehicles spinning, and attention paid to where the joint centre of mass is (as it migrates) relative to places in each tank you can pump from. (Or open a valve from, while you pump gas into the tank)
[EDIT: Tankers/depots would need an extra sump/outlet in a certain place in the forward dome and near the top of the aft dome. For 10mms⁻², ω is on the order of 0.03 rad/s (3 min rotations), relevant distances from CoM in the range ~10 - 40m. Note that only rotation about the axis of highest moment of inertia is stable, so cartwheeling over each other]

The surface tension methods are growing on me, but if they need a bit of acceleration to squeeze the last couple of tonnes out of a tanker... well those last tonnes won't be worth it via linear acceleration if the depot is near full. There will be waste.
[EDIT: Job done, the tanker needs those last few tonnes of fuel for EDL anyway.]

1st priority - just get something that works so HLS can progress - linear acceleration
2nd priority - make a simple, reliable system that isn't too wasteful when the depot is near full - surface tension methods
3rd priority - get every last drop out of a tanker as efficiently as possible, tolerating complexity - rotation methods

[edzieba]

For spin or gravity-gradient settling, what you gain in not using propellant for settling thrust you may well lose in power needed for pumping. A microgree settling thrust and side-by-side transfer means a very small pressure differential - achievable by ullage venting - is sufficient for total propellant transfer from one tank to another. At the ends of a >200m tether, you now have a non-negligible pump head to deal with, along with an orientation change. There is also the propellant needed to set up for the transfer: for spin-settling, that's spinup and spindown thrust. For gravity-gradient, you still are effectively performing orbit raising/lowering burns to move the vehicles to their respective offset altitudes, though for most transfers you could probably omit return burns (though the next tanker would still be burning incrementally for initial insertion longer as the depot orbit raises).

[rsdavis9]

I think you could get useful amounts of artificial gravity if you simply tethered two depots together with cables [...]

1) Depending how you do the cables, you can control the rotation of 1 depot around its CoM independently of its motion about the cable - at least for short periods << 2π/ω, at much lower cost than the docking tanker/ship constantly thrusting to 'hover' and dock. So your ω can be perhaps 6x faster (depending on how long you need to rendezvous and dock)
[EDIT: you lose the favourable tidal forces, but get much better ω². Solving for 10mms⁻² = ω²L - (tidal force with cable horizontal)... For 15 min rotation, ω = π/450. Tidal force is μ/(R+h)² * L/(R+h) i.e. gravity vector misaligned by L/(R+h) radians. 2L ~= 425m.]

2) I feel there should be a much easier solution, just involving the 2 rendezvousing vehicles spinning, and attention paid to where the joint centre of mass is (as it migrates) relative to places in each tank you can pump from. (Or open a valve from, while you pump gas into the tank)
[EDIT: Tankers/depots would need an extra sump/outlet in a certain place in the forward dome and near the top of the aft dome. For 10mms⁻², ω is on the order of 0.03 rad/s (3 min rotations), relevant distances from CoM in the range ~10 - 40m. Note that only rotation about the axis of highest moment of inertia is stable, so cartwheeling over each other]

The surface tension methods are growing on me, but if they need a bit of acceleration to squeeze the last couple of tonnes out of a tanker... well those last tonnes won't be worth it via linear acceleration if the depot is near full. There will be waste.
[EDIT: Job done, the tanker needs those last few tonnes of fuel for EDL anyway.]

1st priority - just get something that works so HLS can progress - linear acceleration
2nd priority - make a simple, reliable system that isn't too wasteful when the depot is near full - surface tension methods
3rd priority - get every last drop out of a tanker as efficiently as possible, tolerating complexity - rotation methods

Since you only need microgees I like a hub that starships would connect to nose first. Nice and stable and nose first keeps acceleration same as on ground. Hub could support 6 starships all nose in easily. Hosing could be flexible from hub to QD plate. Small flyers with thrust or maybe long RMS (remote manipulator system) like system.

[Brigantine]

For spin or gravity-gradient settling, what you gain in not using propellant for settling thrust you may well lose in power needed for pumping.

True, but also electric power is free

At the ends of a >200m tether, you now have a non-negligible pump head to deal with

Assuming the other end of the tether is simply a spare depot acting as a counterweight, then depending on docking geometry relative to the tether, you could be 'pumping' downhill

There is also the propellant needed to set up for the transfer: for spin-settling, that's spinup and spindown thrust. For gravity-gradient, you still are effectively performing orbit raising/lowering burns

1) For the gravity-gradient scenario being discussed, the only setup is the docking maneuver with one depot in an already set up system, i.e. a small amount of 'hover' thrust on the final approach. (though docking to the top one does mean going to a higher energy orbit)

2) For spin (or pumping along a tether), true there is such a cost. The premise is that propellant transfer is a slow process and set up acceleration << total acceleration i.e. the propellant transfer time t >> 2/ω. Is this a valid assumption?

the next tanker would still be burning incrementally for initial insertion longer as the depot orbit raises

That... is actually very interesting. I had assumed linear acceleration was wasted ∆v, but perhaps not. If propellant transfer is always done prograde at perigee, then that's the direction that fuel was eventually going to go anyway on basically any mission that needs refueling. Some mission profiles actually require a top-up in a very high orbit. So you have depots going through cycles of
- start in a low orbit
- mainly getting filled up and the orbit drifting to a mid-height apogee
- mainly getting depleted and the orbit drifting to a high apogee
- one final top-up for a particularly high ∆v mission, leaving the depot with just enough fuel for
- empty depot orbit lowering, potentially employing aerobraking?
- start the cycle again

There is still the problem of ullage thrust having a horrible ISP (can the depot have an extra tiny methane engine? with electric turbos?), and that the extra orbital energy might not be in the exact direction the missions need. But it still reduces linear thrust settling from a total waste to only a partial waste.

[edzieba]

There is still the problem of ullage thrust having a horrible ISP

From the NASA NEA mission proposal (https://ntrs.nasa.gov/citations/20230003852), the hot-gas thruster ISP is 295s. Not the greatest ISP, but not terrible either.

[Barley]

Note that only rotation about the axis of highest moment of inertia is stable, so cartwheeling over each other]

Don't get carried away by the intermediate axis theorem.  You want an engineering definition of stability not a pure math one.  I.e. with time constants and the amount of effort to stabilize "unstable" rotations.  Then compare that the effort for the "stable" alternative.

On the one hand avoiding an instability is good, on the other hand there could be many other factors that are more important than avoiding a weak instability.

For example: the Apollo barbeque roll, around the long axis with a 20-minute period, was "unstable".  This did not appear to cause any angst.  It probably did cause a few reaction control system operations.

[Brigantine]

Note that only rotation about the axis of highest moment of inertia is stable, so cartwheeling over each other]

[...] For example: the Apollo barbeque roll, around the long axis with a 20-minute period, was "unstable".  This did not appear to cause any angst.  It probably did cause a few reaction control system operations.

That's a good point. Starship tankers will be less rigid body motion and more fluid dynamics, so that complicates things. Also if one ship has a CoM more forward than the other (e.g. because the tanker has header tanks) then I worry the other axes will be very unstable... But I grant you that this only means "I don't know how to analyze it" as opposed to "it's impossible". Maybe there's a way to e.g. deliberately transfer fuel and LOX at different rates to manage ballast and somehow make it all work within acceptable RCS use.

Is there a verdict on whether 1 milli g is enough? Would it be sensitive to vibrations?

Maybe the best result is a combination of subtle gravity and very localized surface tension features. surface tension has a short distance and so good gradient around the outlet to control bubbles, and gravity just makes sure it has enough liquid in the vicinity to work with.

[edzieba]

Note that only rotation about the axis of highest moment of inertia is stable

It is only the intermediate axis (if present, which is not always the case for some axisymmetric objects) that is unstable, the other two axes are stable. Spin-stabilised satellite stacks (particularly long narrow ones with a heavy SRM at the bottom, with the CoM moving forwards as the motor burns, and perturbations from the SRM attempting to rotate the stack being the reason for the spin in the first place) work, after all.

[Greg Hullender]

For spin or gravity-gradient settling, what you gain in not using propellant for settling thrust you may well lose in power needed for pumping. A microgree settling thrust and side-by-side transfer means a very small pressure differential - achievable by ullage venting - is sufficient for total propellant transfer from one tank to another. At the ends of a >200m tether, you now have a non-negligible pump head to deal with, along with an orientation change. There is also the propellant needed to set up for the transfer: for spin-settling, that's spinup and spindown thrust. For gravity-gradient, you still are effectively performing orbit raising/lowering burns to move the vehicles to their respective offset altitudes, though for most transfers you could probably omit return burns (though the next tanker would still be burning incrementally for initial insertion longer as the depot orbit raises).

I guess I wasn't very clear in my original description. What I'm discussing might better be called "tidal settling," since the idea is to use a gentle settling force--just like with an ullage burn--but without actually venting any propellant. There would never be a need to spin down, since it's not spinning fast enough to matter. And there's definitely no pumping prop along the tether! Where did you get that idea?

To simplify the idea as much as possible, imagine a large counterweight made of lead with a cable that connects it to the nose of a depot. As long as the counterweight is heavier than the depot, tidal forces will tend to keep it down (towards the Earth) and keep the depot up. Because the system orbits the Earth, it's technically spinning, but only once per orbit. This has none of the instability problems that make rotational settling impractical.

Tidal forces generate a very small amount of gravity in the depot--the same as an ullage burn, if you get the cable length right. A Starship would dock with the depot (not the counter weight) side-by-side and transfer propellant exactly the same way it would during an ullage burn. No extra plumbing.

Obviously there are details to work out. (E.g. unless the counterweight is really huge, there's going to be a significant change in the center of mass whenever a vehicle docks/departs, which will likely cause oscillations. It's unclear whether those can be tolerated or not.) The biggest drawback that I can see is that the cost of lifting the counterweight is probably prohibitive. Although replacing it with a space station might be fun to discuss--probably in a different thread.

Anyway, does that make the idea any clearer?

[Brigantine]

It is only the intermediate axis [...] that is unstable, the other two axes are stable. Spin-stabilised satellite stacks (particularly long narrow ones with a heavy SRM at the bottom [...] work, after all.

Maybe this is only a semantic quibble, but:
the long axis is stable in a rigid body because it's the highest energy state (given fixed angular momentum). In a system with damping, once you have minor peturbations (or dynamic imbalance?) energy can be lost, allowing angular momentum to bleed into other axes. (though this takes a lot of time, compared to the intermediate axis case).

On the topic of damping, liquid is very damp 

[Paul451]

[tidally stabilised tether]
As long as the counterweight is heavier than the depot, tidal forces will tend to keep it down (towards the Earth) and keep the depot up.

I don't think that's necessary. It will work on any combination of masses, as long as it's vertical to Earth.

Having a heavier counterweight will increase the g-load felt by the depot, reducing the required tether length beyond the mid-point (which reduces the risk of MMOD breakage), but doesn't itself add (let alone cause) tidal stability.

[Barley]

Note that only rotation about the axis of highest moment of inertia is stable

It is only the intermediate axis (if present, which is not always the case for some axisymmetric objects) that is unstable, the other two axes are stable. Spin-stabilised satellite stacks (particularly long narrow ones with a heavy SRM at the bottom, with the CoM moving forwards as the motor burns, and perturbations from the SRM attempting to rotate the stack being the reason for the spin in the first place) work, after all.

The first corollary to the Intermediate axis theorem is that the low moment axis is also unstable on a different time scale, because it is an energy maximum.  As Brigatine suggests this is particularly relevant to a floppy, fluid filled tanker.  Even the high moment axis is suspect if fluid is not rotating as a rigid body. 

IIRC you can also derive most of this from the topology of the phase space.  But the topology does not have a measure of distance or time, so cannot tell you how fast things happen.  Since you really care how long the states are stable, and/or how much active control is needed to stabilize them, you should do the CFD or run experiments rather than relying on abstract theorems that do not address these important questions.

[mikelepage]

I'm just trying to remember, where did this thread get to with the concept of using an accumulation tanker as a booster stage for TLI burns, in scenarios where HEO refilling is currently proposed? I vaguely recall something like this was discussed previously, but I can't find it.

The idea was the prime starship would have been refilled by a (first) accumulation tanker in LEO. Then there would be a second, modified, accumulation tanker, with a removable nose cone, so it can dock to the prime starship in the same position as the superheavy booster.  The TLI burn is two stage and the accumulation booster stage retains enough prop to burn back into LEO for reuse.

[eriblo]

I'm just trying to remember, where did this thread get to with the concept of using an accumulation tanker as a booster stage for TLI burns, in scenarios where HEO refilling is currently proposed? I vaguely recall something like this was discussed previously, but I can't find it.

The idea was the prime starship would have been refilled by a (first) accumulation tanker in LEO. Then there would be a second, modified, accumulation tanker, with a removable nose cone, so it can dock to the prime starship in the same position as the superheavy booster.  The TLI burn is two stage and the accumulation booster stage retains enough prop to burn back into LEO for reuse.

Why would you do serial staging with a custom stage instead of parallel staging with a standard tanker?

To answer your question it was discussed on the first few pages of this thread, for example:

In my opinion the second stage of development of the concept should be a permanently orbiting 'tank' which can be repositioned in orbit as required.
The prime candidate for this would be a modified SH which would be comparatively easily developed given it's an iteration of an existing design.

I envisige you would replace the SH thrust structure with a SS thrust structure. The SH (tank) could even supply fuel to the SH (launch) to help with the increased takeoff weight. You would then be using the proven SS attachment system for In-Orbit fuel transfers.

This is similar to an idea that Twark_Main and I discussed (thread is here).  I'd pretty much assumed that you just stretched the SS with a few extra tank rings and cannibalized some of the payload bay space, but you could probably start with a set of SH tanks as well.

The most demanding mission is getting a heavy cargo + crew to the lunar surface and back without lunar surface refueling.  You have to avoid refueling in HEEO to keep the crew out of the Van Allen Belts during refueling, and you have to carry all your prop with you for the return.  The "pusher/tanker" described in the link gets launched almost empty, uses regular tankers (whatever those are--likely just no-payload SSes) to acquire however much prop is needed, then docks nose-to-tail with the payload SS.  It transfers some of the prop to the payload SS, but then just acts as a first stage for most of the TLI delta-v.  Once in TLI, it either does a direct or free return to LEO, either aerobraking or propulsively inserting into whatever orbit the next mission needs.

The big deal here is risk reduction for crewed missions.  You only need one rendezvous/docking operation in the relatively low-radiation environment of LEO, and one jettison of the pusher/tanker.  Pusher/tanker isn't quite as prop-efficient as a series of refuelings in LEO and one refueling in HEEO, but prop efficiency probably isn't the most important criterion for a crewed mission.

[Greg Hullender]

I'm just trying to remember, where did this thread get to with the concept of using an accumulation tanker as a booster stage for TLI burns, in scenarios where HEO refilling is currently proposed? I vaguely recall something like this was discussed previously, but I can't find it.

I think we abandoned terminology like "accumulation tanker" when SpaceX released a document to the effect that there would just be "tankers" and "depots" to refuel the HLS. I thought we dropped discussion of HEO refueling when someone pointed out that SpaceX could just make the depot a little bigger and thus avoid problems with the Van Allen Belts. But I might have missed something.