Rapid Orbital Propellant Transfer

guest blogger john hare

The ability to refuel in space is one of the critical technologies for opening up the high frontier. Without the ability to refuel in microgravity, destinations will be restricted to either low DeltaV missions or planets with in situ refueling  capabilities. Any high DeltaV missions will have to be extremely valuable with massive launchers for tiny robotic explorers. These points have been  made so many times by knowledgeable people that they should be beyond dispute. Unfortunately that is not the case. People are still designing programs and missions based on the assumption that propellant transfer in orbit is just too hard and dangerous to be in the critical path.

While I haven’t studied this problem to the extent Jon has, his posts got me started thinking about the problem. I think most of the approaches use just too much finesse. This seems to be caused by the attitude that mass in orbit is so expensive that minimum mass solutions must be sought even at the expense of time. Gravity gradient settling with a small tether, slow spin settling, magnetic attraction settling, and some type of tank vortex generator.  Sometimes a sledgehammer is the answer. Put some equipment in orbit that yanks liquid, gas and all out of the delivery tank, replace enough helium to keep it from collapsing on reentry, and send it back for another load.

A bad case delivery scenario (not worst case, it can always get worse) would be an SSTO tanker has to rendezvous, deliver propellant, and reenter in one orbit. It carries the propellant delivered in the same tank as used to supply the vehicle main engines so that you have 10 tons of LOX drifting in random blobs around a 100 cubic meter tank mixed with helium and the GOX that vaporized on the way up. It takes 180 degrees of longitude to launch and rendezvous, and needs 60 more degrees to reenter, leaving one third of an orbit to empty the tank. We have 30 minutes to empty the tank of useful propellant and excess helium, and replace enough helium to prevent tank collapse of the delivery vehicle during reentry.

At 1,800 seconds to harvest 10,000 kg of LOX, a 6 kg per second (~90 gallons per minute) transfer pump would have margin to spare in a well defined pumping situation. Or the receiving tank could be at a lower pressure for a vacuum transfer with no pump at all. If settling of whatever flavor could be guaranteed, then that is all that is needed and there wouldn’t even be a question of orbital refueling feasibility. 

This concept assumes that settling cannot be guaranteed in the allowable time frame, and that the delivery vehicle does nothing to cooperate. The cooperation assumed not offered  include GOX vapor pressurizaton to eliminate helium. Dedicated delivery tanks so that only a relatively small tank would need to be emptied. Bladders to separate liquid from gas or sliding cylinder bulkheads. Tanks shaped to facilitate vortex generation for automatic separation during transfer. And so on through dozens of things that could be done to make the orbital storage facilities’ life easier.

If a very large pipe is used off center in the delivery vehicle, rapid emptying of the tank could create a vortex that would mostly yank the denser liquid first. This requires having equipment that will rapidly separate gas from liquid and do something with the gas so that it doesn’t interfere with the rest of the process. If two thirds of the gas must be removed in the process of removing the LOX, it has to be separated into oxygen and helium and sent to the proper tank without   interfering with the main liquid transfer. After the liquid is out of the delivery tank, the remaining gas is helium and the GOX that boiled out during the rapid pressure drop of initial liquid removal.

There is a lot of gas to be removed during this process. During the initial LOX transfer phase, two thirds of the helium and GOX might be removed. By having a compressor remove the gasses from the vortex separator in the receiving vehicle, a liquid transfer pump could move the LOX to the holding tank without problems. The compressor sends the GOX and helium mixture through a heat exchanger to cool the gas well below the boiling point of LOX. A second vortex separator after the heat exchanger sends the LOX to the orbital tank and helium to another compressor that forces it into the orbital helium reservoir. After the initial bulk transfer, the compressor keeps sucking the delivery tank to nearly vacuum and sending the gas through the heat exchanger and secondary separator until pressure is so low that there is no useful quantity of GOX remaining. Then the delivery tank gets some helium back for reentry.  

A 1,000 gallon per minute pump could empty the liquid in two and a half minutes with a clean intake. Assume 5 minutes and it should work. The compressor (or vacuum depending on which side of the machine you stand) needs to suck enough gas through the separator out of the delivery tank fast enough to  to keep some vortex action working. I arbitrarily chose 30 seconds as the amount of time the compressor should empty the delivery tank to keep a strong enough swirl going to get the liquid out early. It works out to a 7 inch diameter compressor inlet in the initial vortex separation chamber. During the 5 minutes of initial transfer, the compressor sucks in a volume of gas equivalent of 10 times that of the delivery tank. Most of that will be GOX that boiled out during the pressure drop. If the right equipment is employed, there is a very small penalty in liquefying the GOX and sending it to the station tank. The pressure will be fairly low by this time, so there will need to be more compressor stages online to compress the oxygen to a temperature that allows it to condense in the heat exchanger.

If the assumption is made that no LOX is pulled through the pipe and all the material is sucked through the compressor, then this size would pull about 7 pounds of air per second at STP (standard temperature and pressure). Since the temperature is much much lower than STP, and the initial pressure double or more, The compressor would initially be capable of removing ~70 pounds of GOX per second. The problem with just planning to take it all that way is that some of the LOX will freeze into possibly inaccessible  solid chunks during the pressure drop. There may not be time to thaw it and recover the material before the tanker leaves.

As always, I make a lot of assumptions that would need investigation. My first cut estimate is that the extraction equipment and heat exchanger would mass on the order of a ton. Better case scenarios could cut this to a fraction. Cooperative tankers, longer available time on orbit, aerospace class minimum weight engineering, and such. If a ton of equipment could do the job with low development risk, it would seem a cheap way to enhance capabilities. The more elegant ways would  be much better if they can be brought to use economically and reliably.

Testing this might be cheaper and easier than some of the better methods. As fast as the reaction takes place, testing could be done on parabolic flights. Each phase could be tested in turn during one roller coaster series. Since the early applications would probably be smaller tanker vehicles, full scale extraction equipment could do a serious shakedown before being sent to orbit. Some fairly basic simulation would come before that to make sure the physics is valid.

If this means for emptying the delivery tank are valid, then an FBC method could be an inflatable tank in orbit of large relative size. If a 1,000 cubic meter inflatable tank were in vacuum, it could suck the delivery tank nearly dry in a minute or so without the complex methods mentioned here. Then the delivery tank with no liquid and 10% of the mixed gas remaining could disconnect and reenter after a fractional orbit. The inflatable could be then emptied into permanent tanks at leisure.

The following two tabs change content below.


I do construction for a living and aerospace as an occasional hobby. I am an inventor and a bit of an entrepreneur. I've been self employed since the 1980s and working in concrete since the 1970s. When I grow up, I want to work with rockets and spacecraft. I did a stupid rocket trick a few decades back and decided not to try another hot fire without adult supervision. Haven't located much of that as we are all big kids when working with our passions.

Latest posts by johnhare (see all)

This entry was posted in Uncategorized. Bookmark the permalink.

12 Responses to Rapid Orbital Propellant Transfer

  1. James says:


    One of the things that made computers much easier to design and build is microprocessors with standardized interfaces.

    Why not standardize an interface to a collection of tank sizes and design refuelable vehicles around those? Refueling involves disconnecting the empty tank and reconnecting the full one. Much easier than pumping liquids around zero G.


  2. James,
    I’m of the opinion that pumping propellants in space will be cheaper than making swapable tanks. Spacecraft tanks typically for integral parts of their structures, and making them swapable would likely be both expensive, complicated, and very mass inefficient. There are ways of pumping propellants without having to deal with the zero-G problems per se, and I’m of the opinion that those will be what wins out in the end.


  3. John,
    I have to admit that this sounds a lot more complicated than many of the approaches I’ve been looking at. Though I’m not 100% sure they could work in such a short (one-orbit) delivery timeframe, this seems a little overly complicated. Just my opinion of course.


  4. john hare says:


    The single orbit and the rest was about as bad a scenario as I could realistically believe as a requirement. As far as I can figure, the difficulty of orbital propellant transfer is a problem that ain’t. So this one is pretty much my answer to the people saying that it is just too difficult a problem. Even if all the rational transfer methods were to fail, this brute force solution could be a back up.

    This method could be checked out on the cheap by retrieving liquids from a tank in a paint shaker at maximum rpm. Then in a parabolic 30 seconds. Then you can go to the program and mission designers that say microgravity propellant transfer is not feasible and have ammunition to get the mindset changed. After the mindset accepts that the fueling station is technically feasible, funding for alternative methods would be easier to shake loose.

    The methods that you and other people are checking out are certainly a far better way to go. I just can’t believe that orbital propellant transfer is a real problem. I think the real problem here was in James’ last sentence, ” Much easier than pumping liquids around zero G.”. This is a sincere belief that we both disagree with, by a rational space buff. I think the method I described would be a brute force method of working on that belief.

    Given a reasonable time frame and architecture for extracting materials, I certainly agree that simpler and more efficient methods can be used. But if they don’t, we have alternatives

  5. john hare says:


    Interchangeable tanks is frequently argued and seems logical on the surface. It may even be the way things are done by some organizations. The problem is that it is based almost solely on the premise that pumping liquids in microgravity is very very difficult, if not impossible to do on useful scale. Very similar to the argument that VTVL was not feasible before DCX.

    Military aircraft use drop tanks on a lot of aircraft. They also use in flight refueling. In flight refueling gives global range while the drop tanks give continental ranges at best. It is also cheaper to refuel from a flying tanker than to expend the drop tanks. When drop tanks are expended, new ones must be built, attached, and checked out. Even though the built part may be years old by the time it is bolted to a plane, it is still a major purchase item that must be transported and stored in a proper manner. Occasionally drop tanks will still exhibit some problems that must be dealt with. The internal fuel capacity of the aircraft is lighter and more reliable than drop tanks. The tanker aircraft can refuel any aircraft fitted with the right refueling gear from light foreign strike planes to USAF B52 heavy bombers. This would not be possible if the fuel supply planes had to carry an assortment of full tanks to fit different classes of aircraft. Handing off full fuel tanks in flight would be a daunting task to do with any regularity.

    The separate tank on spacecraft launchers has a more potent argument in the economy of launching dedicated tankers. A tanker that had to carry a specified standard propellant tank must also carry shrouds, interconnect structure, and payload monitoring system. A tanker launch vehicle that pumped the cargo can carry the delivered propellant in the same tank as its’ own propulsion requirements. The dedicated tanker could be simpler with no external structure, shrouds, and payload control systems. Simpler is cheaper. Without the weight of the extra gear, delivered payload is higher at $?k per pound. Turnaround is faster without the extra gear to attach and check out.

    With propellant transfer, it is possible to buy from any launch vehicle and sell propellant to customer whether their architecture matches yours or not as long as the pipes connect.

    This is a long and continuing argument for another time I suppose.

  6. John Carmack says:

    I am somewhat inclined towards going farther than even swappable tanks and delivering complete propulsion modules to be combined together. Assume a fully loaded space operating propulsion module has a mass ratio similar to a cheap upper stage, at least six or so. Compared to just pulling liquid out of a tank, this means you are probably carrying an extra 10% in inert hardware for the engine and various other bits, and you get that much less mass transferred with each launch. That amount is negligible.

    In the case of a one-shot mission being assembled, you probably even save money on hardware, as many small modules generally cost less than one big vehicle that needs to be fueled up with a transfer system. If you are looking at something like a lunar cycler that would get refueled many times, there would be a cost if you dumped the modules after each use, but it is conceivable that they could be reentered and refurbished. However, it probably doesn’t make much difference — I remain confident that modules can be fabricated for way under $100 / pound of contained propellant, so it would take a very good RLV indeed to make that the dominant cost of delivering impulse to orbit.

    I also lean towards having an astronaut at the assembly point to just use a wrench to connect them together, rather than any automated process.

  7. Rand Simberg says:

    I don’t understand what would drive a single-orbit scenario. Once in orbit, why would there be such a rush to come back?

    By the way, you are aware, of course, that this is only possible in an equatorial orbit?

  8. Jonathan Goff Jonathan Goff says:

    John C,
    Thanks for the comment!

    Full propulsion modules would be easier than swapping tanks, but I still think there are several benefits of going with a propellant transfer (pumped or blowdown) concept instead.

    1. Without being able to top-up on orbit, you really can’t use cryogenic propellants for most missions–that includes LOX. While LOX and methane and propane are all technically “space storable”, it still actually takes a lot of work in practice to get a low-cost passive thermal system that won’t have boiloff issues. Which means that if you’re doing more than one module, you’re probably going to have unacceptable boiloff losses during assembly. Sure you can go with pure storable propellants like say peroxide kerosene, but I think that the added complexity of fueling is greatly exaggerated compared to the cost of not being able to use LOX.

    2. There are some applications like space proxops tugs that really want to be able to tank back up. Making generic propulsion modules with all the features a prox ops tug would need to operate compared to just a transfer stage is going to drive up the costs, and require more standard module sizes.

    3. For low cost propellant delivery to orbit, you’re probably going to be talking about light payload RLVs for the foreseeable future. Which means that for sizeable missions, you might have to cluster tons of these modules together, and have them work correctly together…

    There are applications where prefueled propulsion module assembly probably does make more sense than propellant transfer, but I think that actual propellant transfer has enough advantages for enough markets that it’s just worth it. Plus, I think people continue to overrate the difficulty of this problem. Boeing/MDA/Ball demonstrated non-cryo transfer with Orbital Express. Dallas Bienhoff was telling me that the fluid transfer system they had was rated for cryo temperatures. That still leaves the propellant settling for cryo transfer, but there are several proven, soon to be proven, or under research options, none of which are particularly onerous. I actually had one I was working on with Frank Zegler that we almost got an SBIR for last year that would be particularly elegant. But Dave told me to hold off on reproposing it until next year since it’s pretty far off of our critical path right now.

    Anyway, thanks again for the comment! Good luck in Las Cruces!

    ~Jon Goff

  9. Eric Collins says:

    I was actually thinking about the tank settling problem a few days ago. I figure that given a properly designed tank, one fairly easily keep the liquid fuel at the tank outlet, thus avoiding gas in the feed lines. The simplest of these is a tapered cylinder, with the large end towards the aft, and a single baffle inside the tank. With this arrangement, a very modest roll rate would settle the fuel to the outside and bottom of the tank where it can then be tapped from an appropriately offset valve.

    Just in reading this post, however, I had another thought. Suppose you arrive in orbit with whatever fuel you’ve left over in your RLV’s tanks. In the payload bay, you have a collapsed fuel bladder (think Bigelow module, but with the minimal structural and thermal characteristics to get the job done). Since the bladder is empty at first, you don’t need to worry about it being stiff from the cryo-fuel while you are inflating it. Transfer the remaining fluid into the bladder, and once you’ve attained a nearby compatible orbit with the fuel depot, simply detach the bladder and let it float near the depot until it can be fetched by a prox-ops tug. Thus you have a drop tank without the aerodynamic and structural penalties associated with a more rigid arrangement. Also, the transfer interface on the bladder can be a standardized quick-release/hookup variety that Jon has advocated elsewhere.

    I’m also tempted to run the numbers on John’s rapid de-tanking scenario to see what is the likelihood that the LOX would freeze up due to the rapid pressure drop and subsequent fuel vaporization. At first I didn’t think it would be much of a problem, but then I checked and saw that the heat of vaporization is ten times that of the heat of fusion for oxygen (i.e. for every mole of oxygen evaporated, enough heat is removed to freeze ten moles or to drop a 127 moles of LOX by 1K). I also noticed that the freezing and boiling points are fairly close together (50.5K and 90.33K respectively). This boiling point is probably for 1 atm. of pressure, so as the pressure drops, so too would the boiling point. So, it looks as if freezing could indeed pose a problem if one were to rapidly depressurize the tank.

    Disclaimer – All of the above numbers are based on a quick Google search and very simple back of the envelope calculations. I apologize in advance for any egregious errors I may have committed due to inaccurate data, improper assumptions, or botched calculations. That said, I look forward to running these numbers in greater detail later, if I can find the time.

  10. john hare says:

    John C,

    That 10-15% does matter some though and does cramp your style some for using widely varied vehicles for widely varied purposes. For your company, modules make excellent sense, others may want to use other vehicle types than you are set up to deliver. An astronaut on site would be good.

    Using thousands of modules once does tend to put you back to the ELV problem that each unit may have infant mortality problems. When you get to distant destinations, the ability to use ISRU will imply the ability to refuel. Some of those vehicles will eventually want to be traveling to various destinations that don’t necessarily have Earth as a stopping point to refuel.

    Good luck in Las Cruces. Don’t put up with any of Murphys’ crap.

  11. john hare says:


    I was trying to contrive an unreasonably bad requirement that could be overcome. A rational propellant transfer scheme with plenty of time wouldn’t need the brute force redneck method here. The point is that if a ton of equipment in orbit could make it unnecessary to build gonzo launchers, it is insane to build gonzo launchers. What fraction of zee Griffinshaft budget would it take to fully develop a propellant transfer system or John Carmacks propulsion module system?

    With air launch and cross range reentry, one pass could be done from modest latitudes above the equator. Twelve hours would be an easier requirement that should be able to get the launcher home for supper. With twelve hours, I just don’t see why it would be much of a challenge to develop the transfer system.

  12. john hare says:


    The fuel left over in the RLV tanks is the initial delivery system in my mind. Build them oversize just enough that full the GLOW is the same as having a full payload. Any emergencies or performance shortfalls have the payload propellant to draw from if necessary. Sell whatever propellant you can rendezvous with. After shaking down the vehicle, you have a real good idea of the quirks and capabilities for payload matching.

    After entering revenue service, the RLV can use propellant in the tanks as both margin and automatically adjustable secondary payload. Any minor efficiency problems cost, and improvements boost the bottom line.

    The detachable bladder sounds like something to look into. The delivered propellant could be available to the RLV if needed until the last possible second. Settling could be by the RLV OMS at a fractional gee while the bladder is filling with liquid. It wouldn’t need a large special cowling or other heavy connections.

Leave a Reply

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