On-Orbit Refueling Primer

I’ve talked a lot about the concept of on-orbit propellant transfer, and how important it is to reducing the costs of interplanetary and cislunar transportation. However, I realize that there aren’t a lot of easy to find resources discussing the problem, and many readers may not know all of what’s involved in the process. While I’m not an expert in the area by any stretch of the imagination, I think I know enough about the problem and the various ideas suggested that I should be able to give a brief introduction for those unfamiliar with the topic.

Rendezvous and Docking
The first step in on-orbit propellant transfer is actually getting the tanker to the vehicle it is refueling (or the orbital propellant depot). I won’t spend too much time on this, because although this is a definitely complicated task, it is one that has been discussed thoroughly elsewhere. The basic idea of orbital rendezvous if for the vehicle being launched to try and match positions and velocity vectors with the object it is wants to dock with.

There has been a lot said about the difficulty of orbital rendezvous, but one needs to keep in perspective the fact that the US has been doing this for decades now. NASA did have some problems with their DART spacecraft, but that stems more from the fact that they were trying to do a fully autonomous rendezvous, as opposed to a rendezvous that is either performed by a pilot on board, or remotely by telepresence. For on-orbit refueling in cislunar space, there really isn’t a need to do the rendezvous autonomously, because communication lag is so short that even if you don’t have a pilot on-board, telepresence is adequate.

A quick distinction also ought to be made between the two standard ways of mating two vehicles in orbit,docking and berthing. Docking is basically flying the two vehicles together using a heavy mechanism that latches the two together after they’ve contacted. Berthing uses some sort of robotic arm to connect the two vehicles together in a much gentler manner, requiring a much lighter connection system. The Russians have prefered docking systems in the past, while the US has been more a fan of berthing systems. A orbital propellant depot might very well have a robotic arm to allow for berthing of tankers and visiting vehicles, while a refueling system that connects directly to the vehicle might just use a docking system. It might also be possible to have a small robotic arm or set of small arms on the tanker if that turns out to be a useful idea.

Once the two vehicles have been mated, some method for connecting the propellant tank on the tanker to the receiving vehicle must be implemented. A manned tanker might have manually attached plumbing umbilicals, while an unmanned tanker might have automatically connecting ones. Depending on how good the alignment is during docking/berthing, this could be a relatively complex or relatively simple. The number of connections that need to be made depends on how many fluids are carried on one tanker. A tanker could carry just one propellant, or it could carry several liquids and gasses. The umbilicals may need to be able to transfer gas from the ullage of the tank being filled back to the tank being emptied. Lastly, umbilicals will also likely need to provide at least some data to the tanker if possible to let it know when to start and stop.

Propellant Management
Probably the most difficult part of on-orbit propellant transfer is the propellant management. One of the difficulties in designing propellant tanks that function in zero-G is controlling where the liquid is within the tank. The reasons why propellant location within the tank is important are:

  • Many rocket engines can be damaged if their feedlines are sucking gas in, particularly if the engine is turbopump fed.
  • If the vent line isn’t uncovered, you could end up venting liquids as well as gasses–this is wasteful, and could be dangerous depending on the liquid.
  • If the propellant is floating around unconstrained, the vehicle’s CG can move substantially, particularly for high mass ratio vehicles like lunar transfer tugs. Sloshing propellants can make it very tough to dock with, among other problems.

With these reasons in mind, you generally want to keep the liquid near the outlet (usually at the engine end of the tank), and the gas near the vent port, especially during docking and propellant transfer, and immediately prior to engine firing.

This is particularly important for propellant transfer. As you transfer propellant from one tank to the other, it is a lot easier if you can also either vent the excess gas from the one tank, or suck it off and use it to pressurize the transfer tank. This requires making sure that the outlet of the transfer tank is always covered with liquid (so you aren’t just passing gas between the tanks–that would be impolite to say the least), and that the vent port on the receiving tank is always uncovered.

For storable (ie room temperature or non-cryogenic) propellants, there are some rather easy ways of dealing with this problem. The best being using a flexible diaphragm. The diaphragm is basically a flexible sheet of plastic or elastomer that separates the gas from the liquid. If the diaphragm is impermeable, you can always assure that the gas and liquid are where they’re supposed to be. This also gives you a lot more flexibility with how you pressurize the tank.

The problem is that it’s hard to find diaphragms that are both compatible cryogenic propellants (particularly LOX), while still being sufficiently flexible at those temperatures to avoid cracking and eventually leaking. While the flourocarbon that XCOR is using as the matrix for their LOX compatible composites might just do the trick, there’s no way of knowing how many cycles it will last for. Not to mention that cryogenic liquids tend to have boiloff issues as heat enters the propellant tank from the outside environment. This can quickly create gas bubbles on the liquid side that now need to be dealt with.

There are fortunately several avenues that could be explored for solving this problem, depending on what sort of propellant transfer scheme is used. Since this is just an introduction, I’ll just list a few toss a few out (some conventional, some rather wacky) without elaborating too much for now:

  • Gravity Gradients or Tethers–When a vehicle is said to be in a specific orbit, it is actually the CG of the vehicle that is in that orbit. Any portion of the vehicle closer to the gravity well is actually going slower than the orbital velocity of a chunk of matter at it’s precise distance from the center of the planet, while any portion of the vehicle further out than the CG is actually going a little bit faster. This acts like a very slight outward acceleration on anything past the CG, and a very slight downward acceleration on anything planetside of the CG. In small vehicles, these forces are almost negligible. Just enough that some ultraprecise microgravity experiments can get thrown off, but not enough to do too much for short-term propellant settling. These settling forces need to be strong enough to offset any unsettling forces caused by things like equipment vibrations. Over the space of days or weeks, even a constant 1 millionth of a G (in abscence of any other disturbing forces) is sufficient to settle propellant tanks, but in the presence of disturbing forces, and with the time constraints inherent in economically viable propellant transfer, the accelerations need to be bigger.

    For very long propellant tanks, large propellant depots oriented with their axis pointing down toward the center of the earth, or vehicles hooked to a sufficietly long tether, the accelerations can be sufficient to settle the propellants in a reasonable amount of time. This requires at least a dozen or more meters long between the two ends of the vehicle, or between the two ends of the station (or the end of the tether and the opposite end of the vehicle). For permanent propellant depots, it is even better if the tether is of the electrodynamic sort, being used for reboost. The constant, but small, acceleration caused by an electrodynamic tether is more than sufficien to insure adequate propellant settling for quick propellant transfers.

    A quick google search on tethers and propellant settling can get you more details.

  • Tank Pistons–If the propellant tanks are cylindrical, lightweight pistons can be used instead of a diaphragm. Propellant tanks on space vehicles tend to be very light compared to the liquid they hold. It isn’t unheard of for LOX tanks to be only about 1% of the mass of the LOX they hold for low pressure tanks. Adding an extra tank segment long enough to put a tank piston/float that won’t cock will only add another 20-30% to the tank mass, plus the mass of the piston. This might make the tank 2-3% of the propellant mass instead of just 1%. Liquid Hydrogen is low enough density, and requires enough insulation that tanks for it tend to be a lot heavier per lb of propellant, and getting good seals that work at LH2 pressures is more challenging, but this still may be a valid solution to the problem.
  • Surface Tension Screens–These are the preferred method on most satellites, but usually end up being a lot more complicated than they sound. Basically, screens, baffles, and other structures are placed througout the tank so that the propellants will stick to the screens. Unfortunately, these can get heavy fast, tend to result in large amounts of unused propellants stuck in the tank at the end of firing, and can cause boiling issues with low temperature cryogens like LH2. But they are still an option
  • Fans–one could use a fan with a magnetic coil outside the tank (think flowmeter in reverse) to create sufficient force to send the liquid to one end of the tank. Besides the problem of having moving parts, this might also result in lots of gas entrainment in the liquid if the velocity is too high. It might also augment heat transfer between the liquid and the gas (which may or may not be bad). This would probably need to be used in conjunction with a surface tension screen, but could be used to reduce the complexity of that system.
  • Metallic diaphragms–there have been ultra thin-walled metallic diaphragms that could work down to cryogenic temperatures. However, due to wrinkling and other issues, these tended to be single-use items. With modern materials, though, this may now be feasible.
  • Spin Gravity–if you are docking the tanker directly to a vehicle instead of depot, it may be possible to spin the two end over end, using the vehicles’ RCS systems, producing some centrifugal forces. Only a very slow rotation is needed, probably on the order of .001-.01G might be sufficient, which implies very low RPMs.
  • Propulsive Venting–if one of the propellants is LH2, the boiloff from that can possibly be vented through a low-pressure cold-gas thruster to produce just enough thrust to settle propellants a bit. If you’re going to toss it anyway….
  • Ion drives–ion drives have such low T/W that you could possibly fire an ion drive to settle tanks, and so long as you transfer the propellant fast enough, the actual orbital velocity change should be relatively minute. Not to mention that you don’t waste as much propellant. Could cause some interesting issues with ion jet impingement….

There’s probably more ways to skin the cat, but this article is just supposed to be an intro, and it’s already too long as it is.


The following two tabs change content below.
Jonathan Goff

Jonathan Goff

President/CEO at Altius Space Machines
Jonathan Goff is a space technologist, inventor, and serial space entrepreneur who created the Selenian Boondocks blog. Jon was a co-founder of Masten Space Systems, and is the founder and CEO of Altius Space Machines, a space robotics startup in Broomfield, CO. His family includes his wife, Tiffany, and five boys: Jarom (deceased), Jonathan, James, Peter, and Andrew. Jon has a BS in Manufacturing Engineering (1999) and an MS in Mechanical Engineering (2007) from Brigham Young University, and served an LDS proselytizing mission in Olongapo, Philippines from 2000-2002.
This entry was posted in Uncategorized. Bookmark the permalink.

21 Responses to On-Orbit Refueling Primer

  1. Ed says:

    It is probably safe to assume that an orbital fuel depot would be a rather large structure, say on the order of the size of a space shuttle external tank. For a long structure like that, it would most likely end up priented with its long axis pointing to the earth due to tidal forces, and hence with the maximum cross-sectional area normal to its orbital speed. There isn’t a lot of atmosphere up in any likely orbit, but with that orientation any drag will be at a maximum, tending to lower the orbit over time. For that reason, I like the electrodynamic tether idea, to use electricity produced by received sunlight to maintain the orbit by pushing against Earth’s magnetic field.

  2. Anonymous says:

    The storage tank could be kept at very low temps for the propellant being transfered.
    Then a small pump/compressor could
    transfer the liquids or gasses to the storage tank that is cold enough to keep them liquid. In the
    final stages of transfer, the device would be a pure compressor
    operating at well under 1 psi to get the last ounces of propellant into the storage tank.

    Given that this is not transfer during acceleration, a fairly small
    pump/compressor could spend a day or more scavaging every last drop from a simple tank.


  3. Anonymous says:

    On second thought, I would not
    want to attempt a two phase pump
    /compressor. To hold the delivery
    tanks to the simplist possible
    configuration, two phase extraction with the complexity in the recieving storage tank seems desirable.

    Suggest a heat exchanger in the pipeline to chill the incoming two phase material to all liquid, followed by a small pump to force the liquid into the storage tanks.

    Only when the pressure drops to an unreasonable level would a compressor be desirable to force the remaining gas into the heat exchanger. A compressor to raise pressure in a chilled gas from 0.2 psi to 5 psi need not be that heavy.


  4. Jon Goff says:

    Yeah, LEO propellant depots really want to use some sort of electrodynamic tether to offset drag losses (and to settle propellants). Propellant depots in deep space (like L-1 lagrange points), or in orbit around planets without magnetospheres will need different means for settling the propellants. For depots in orbit around large moons or planets, just a normal, non-electrodynamic tether should be sufficient to hold orientation the way that is desired. For deep space systems, an ion drive might be preferable for propellant settling.

  5. Jon Goff says:

    Using some sort of slower pump (ie slower than the main engine pumps) to transfer propellant makes perfect sense. After all, so long as the transfer takes less time than the time between tanker deliveries, it should be fine. Taking 1 hr instead of 1 minute to transfer the same amount of propellant allows you to use a much smaller, lighter, simpler pump. Possibly electrically driven using solar power or something.

    But I’d definitely try to avoid two-phase flow in the pump. Adding a cryocooler/heat exchanger upstream of the pump might work, but propellant settling would likely be better.

  6. Randy Campbell says:

    I haven’t finished the primer yet, but I noted that there was another couple of ‘fuel-transfer’ ideas on the Dunn Engineering site:

    The self Pressurized rockets article.
    (Interesting to note this seems to be a version of the VAPAK system used by T/Space)

    And the Space Transfer of Propellent .pdf:

    And space storage of LH2/LOX:

    As for ED tethers and such, the depot ‘might’ be a good use for the rotating tethers, or gravity stabilized ‘hypersonic’ tethers which would allow a bit more payload with the same launchers.


  7. Iain McClatchie says:


    great post. I don’t mean to harp on this point, but could you suggest with a little detail a mission for which propellant transfer would have an obvious benefit over whole-tank transfer?

    Your previous justification was flexibility… making it possible for more than one launcher to put up fuel for a mission. Do you see *any* short-term need for this flexibility?

    Think of it this way. Suppose NASA decides to launch a deep-ocean submarine to explore Europa. They want to deliver 100,000 kg to Europa, so they’re going to need a lot of propellant in LEO. I have no idea what the numbers are, lets assume they need 400,000 kg. Consider the two options:

    1) engineer a propellant transfer system, as you sketch in your post. Multiple companies bid on launching propellant, maybe two actually do it.

    2) Assume a cluster of tanks, let companies bid on launching the whole cluster, each company picks a different sized tank to make up the cluster. Now just one company wins the bid.

    Even though I’m not in the biz, I’m going to claim that (2) has less design risk and engineering cost. You get some competition either way. (2) has the downside of locking your mission in to the launch provider, but you were already locked in to the launch provider for the submarine, so is this an added problem?

    This propellant transfer thing smells like a solution in search of a problem.

  8. Jon Goff says:

    I don’t mean to harp on this point, but could you suggest with a little detail a mission for which propellant transfer would have an obvious benefit over whole-tank transfer?

    If cryogenic propellant transfer proves to be reasonably feasible, I honestly think it would have the leg up in almost all situations. With full tank transfer, you have to design custom tanks for every single thing you want to do, you end up with lots of extra plumbing, you end up with lots of extra umbilicals and connections, you end up with more other subsystems required. And since these tanks would likely need to be custom designed for each application, you no longer really have any standardization.

    Once you have a good standard interface for propellant transfer (which interface can be completely open source), any transfer vehicle and any tanker can work together. You don’t need to redesign stuff. If you want to add two or three ports for very big systems that want to refuel faster…go right ahead. Basically, you don’t end up having to do anywhere near as much design work, you have less things that can go wrong, and you have more flexibility.

    I may be biased, but I really don’t see the advantage of tank transfer. There are people who disagree, and the reality is that there almost never is a perfect single way to do things, but at least for the metrics that are important to me, propellant transfer seems be substantially better.

  9. Anonymous says:

    One of the advantages to propellant transfer is the ability to retrieve any excess from any delivery vehicle. A dedicated tanker could easily be one of the most mass efficient vehicles possible with no cargo bays. The propellant remaining would be the idea cargo for companies attempting to prove SSTO.


  10. Paul Dietz says:

    Here are a couple of other propellant settling ideas.

    Dielectric: the liquid propellant will have a dielectric constant greater than that of vacuum (or its vapor), so it will be attacted to regions with electric fields. Put a couple of nested electrodes where you want to propellant to go and agitate the liquid some to make it end up there. This will also drive away bubbles from the liquid.

    Similarly, liquid propellants will be either paramagnetic (LOX), and therefore be attracted to regions of strong magnetic fields, or diamagnetic (most everything else, particularly hydrocarbons with aromatics), and therefore be repelled from regions of strong magnetic fields. Diamagnetic repulsion has been used on Earth to levitate water against 1 gravity; the force is proportional to B * grad B. If the propellants are cryogenic, superconducting coils could be used to produce the magnetic fields.

  11. Kelly Starks says:

    As far as propellant settling ideas, why not just have a paddle wheel like stirrer inside circular or cylindrical tanks? Rotating inside the tank, it would act like a centrifuge. Place vents by (or in) the axis, drains on the outside. Use in pairs to cancel torque. Possibly form to double as anti slosh baffles for launch (in a tanker).

  12. murphydyne says:

    Ambivalent wondered “could you suggest with a little detail a mission for which propellant transfer would have an obvious benefit over whole-tank transfer?”

    I’ve wondered about that myself. When I was researching the amount of dead mass in GEO (over 600 mt back in 2000, near as I could reckon) one thing that did surprise me was the large number of large Russian kick stages still floating around in GEO. Part of the EML-1 business plan is dropping into GEO for engineering services as well as picking up dead mass for forensic analysis. These Russian kick-stages used silane (a potential Lunar fuel), and irrespective of what propellant they carried they had propellant tanks. Could these be tested and reused as part of a bolt-on apparatus for a Trans-LEO CEV vehicle (or space-constructed probe, or whatever)?

    But I’m not a big fan of frequent boltings/unboltings on our hardware, and would honestly rather have built in tanks, but bolt-on tanks are nevertheless a compelling concept. All the same, those bolt on tanks need to be filled when needed, and that’s why you have a large central depot. It is possible to have more than one space mission going on at one time. So you ship up the water from Earth in bulk, crack it when and as much as you need, and save the rest for the next fill-up. Introducing the launch of lots of small, bolt-on tanks is inefficient because you’re always shipping up the mass of the tankage as well as the propellant from the Earth’s gravity well.

    Of course, this is assuming that we want to build an architecture where we’re not throwing everything away (a la the current NASA Lunar return architecture).

    Ambivalent also took a position of “This propellant transfer thing smells like a solution in search of a problem.”

    Oh, there’s a good reason to learn this stuff. If we’re going to be space-faring and not just space-visiting, then we need to learn as much as we can about all kinds of activities in space, like how to refuel our vehicles and other assets. Such as when we bring our Sun and Star Watchers back from their Libration stations for servicing, upgrade & refueling a la Hubble. Such as when a CEV is making a service run down to GEO. Such as when a Lunar bus is preparing to drop to the surface. The list can go on and on.

    Question for Jon: From what I understand the wicking action works in space. Can that serve in some way to at least provide an intial pool for firing up the engines (which solves the rest of the problem)?

  13. Jon Goff says:

    From what I understand the wicking action works in space. Can that serve in some way to at least provide an intial pool for firing up the engines (which solves the rest of the problem)?

    Yeah, that’s more or less similar to the surface tension screens idea. I’m not sure how well that will work with all propellants, and it’s probably too dense for much more than the initial engine light, but it’s not a crazy idea.

  14. Jon Goff says:

    Thanks for the interesting ideas. I never really thought about taking advantage of the magnetic properties like that.

    Paddle wheels are an interesting idea, but full tank wheels start getting big, heavy, high maintenance, etc. Might still be doable, but seems a little overkill to me.

  15. Kelly Starks says:

    Well you need slosh baffles etc anyway, and it seemed a lot lighter and simpler then spining the tanks out on cables and such.

    You might not need paddles extending across the full tank if you keep them turning or largely filled. If the paddelle just came in a 3rd to quarter in from the outter edge of the tank, it would catch and spin a floating glob in the middle of the tank. Surface tension would pull the ininer volume if you didn’t speedthe paddles up to fast.

    Or if you routinely keep it rotating, it would keep any mass in the tank spun outwards.

    Just a thought.

  16. Paul Dietz says:

    NASA has done some work on explointing the paramagnetism of LOX for pumping; google has some pages on this. The magnitude of the magnetic susceptibility of LOX is quite a bit higher (factor of 100?) than that of diamagnetic liquids, so it would probably be the place to start. Dewar himself, IIRC, was startled to observe that LOX would leap out of a container onto the poles of a strong magnet, where it would cling as it boiled away.

  17. Kelly Starks says:

    > The magnitude of the magnetic susceptibility of LOX
    > is quite a bit higher (factor of 100?) than that of
    > diamagnetic liquids, so it would probably be the place
    > to start. Dewar himself, IIRC, was startled to observe
    > that LOX would leap out of a container onto the
    > poles of a strong magnet, where it would cling as it boiled away.

    Interesting. That would suggest a few magnetic coils around or in a LOx tank would relyably throw it onto a intake – or even be usable for a magnetic pump. Given kerosine (high likely primary fuel) can just be handeled in a bladder, that would handel zero g refueling easily.

    How much heating does the LOx get from inductance?

  18. Kelly Starks says:

    just a nit, but since I have a blogger account why do I also have to enter in the word verification also?

  19. Jon Goff says:

    NASA has done some work on explointing the paramagnetism of LOX for pumping; google has some pages on this.

    Quite interesting stuff! While I’d still prefer just a normal electric pump (or a free-piston pump maybe) for pumping the stuff, using magnets to settle the LOX sounds like a really, really good idea. For LOX/Kero designs, the kerosene can be held in a bladder, and the LOX constrained by the magnetic field.

    Combine that with diamagnetic repulsion for the other propellant, and you may have a winning solution. The real key will just be having a reasonable magnetic coil, and making sure that coil doesn’t muck with any of your other sensitive equipment. Of course, having a coil like that around might also be nice for other reasons…..

  20. Jon Goff says:

    just a nit, but since I have a blogger account why do I also have to enter in the word verification also?

    Not sure. As it is, even *I* have to do the word verification to post comments on my own blog…..maybe that’s something we ought to have Blogger look into–allowing those with blogger accounts to bypass word verification requirements (just have a word verification used in account creation instead)…..

  21. Paul Dietz says:

    How much heating does the LOx get from inductance?

    LOX is an insulator, so there should be no inductive heating.

    There will be heating from the irreversible process of the LOX falling ‘downhill’ into the potential created by the magnetic field, but this should be very small, as the potential well needn’t be very deep.

Leave a Reply

Your email address will not be published. Required fields are marked *