Random Thoughts/Orbital Access Methodologies VII: Air-Launched Glideforward TSTO with Exo-atmospheric Suborbital Refueling

Ok, I’ve been toying with another orbital access methodology, but I wasn’t sure whether to file it under Random Thoughts (which tend to be my more half-baked, far-out ideas) or with the rest of the Orbital Access Methodologies series (which I’ve tried to keep a lot more professional/high-brow).  This idea is actually an offshoot of two ideas I’ve posted about previously (air-launched glideforward TSTO and Fleet Launched Orbital Craft), along with the Boom Rendezvous idea I just wrote about.

Carrier Craft By-the-Slice
Anyhow, the key thing that led to this concept was the realization that for air-launched vehicles, you really want to be able to buy the carrier aircraft “by-the-slice”, instead of having to own it outright.  Ie, you want to be able to call up Virgin Galactic and say “what’s your schedule for this week…ok, can I buy the noon-5pm slice on your WK2 next Thursday?  Usual ammenities.  We’ll meet you on the flight line at 12 o’clock sharp.  Thanks!”

Especially for early-generation airlaunched RLVs, not having to pay the full burdened cost of an aircraft in addition to the rocket stages is really important.   It’s even more important to avoid having to pay the development cost of a custom carrier craft.   If you only have to rent the carrier craft when you need it, you are much more able to handle ups and downs in demand, which will likely be fairly volatile for the first few years.  It also means that you can use the carrier plane during development without having to have that big of a capital investment early on.   IIRC, this was one of the keys to making Zero-G’s parabolic services possible–the same aircraft can be converted to cargo-hauling overnight.  I don’t know if they still use this capability, but not having to pay for the full aircraft just out of space revenues makes it easier to charge an attractive price.

The problem is that there aren’t any airlaunch carriers that you can “buy by the slice” that are big enough for an manned orbital RLV.  That’s where exo-atmospheric refueling could come in.

Exo-Atmospheric Suborbital Refueling
Many years ago, people in the military realized that being able to refuel aircraft in-flight could greatly expand their operational capabilities.  Of course, at the time that the idea had first been raised, mid-air refueling sounded about as crazy as exo-atmospheric refueling sounds today.  The first mid-air refueling involved linking two biplanes, and having a guy walk from one aircraft to the other carrying a 5lb gas can!  Of course, over time, much safer procedures and techniques were invented, and nowadays its easy to forget how insane this idea must have looked before it had been tried.

That said, while it might be possible in some cases to eliminate the need for mid-air refueling by using a bigger vehicle with larger propellant tanks, there are some missions that would be flat-out impossible without the capability.  The typical response I get when I suggest something crazy is, “why don’t you just build a bigger rocket”.  While there are some cases where you’d be better off just “building a bigger rocket”, there are some times where additional operational complexity more than pay for themselves.  And I think the added complexity in this case is worth it if it buys you the ability to use an off-the-shelf carrier plane, bought by-the-slice, and keeping your rocket stage sizes small, and the rocket engine sizes in the low 10s of klb range, while still being able to feasibly deliver people to orbit.

So, what do I mean by exoatmospheric suborbital refueling?

Basically, it means that at some time after the vehicle leaves the main sensible atmosphere, it hooks up with another vehicle on the same trajectory, propellant is transferred from one vehicle to the other, and then the first vehicle continues on to orbit (while the mostly-empty tanker vehicle reenters and lands). [Note: Gary Hudson reminded me that Mitchell Burnside-Clapp of Pioneer Rocketplane investigated just such an enhancement to their system, during their work on the RASCAL project.]

The configuration that I’ve used in my analysis is a pair of TSTO vehicles launched off of a WK2.  The first stage in both stacks is identical, and are roughly 25klb wet, 5klb dry.  The orbiter and tanker stages are about 10klb wet each, with the orbiter upper stage launching a little more than half full, and the tanker stage having smaller tanks fully-loaded on takeoff.   The tanker and orbital stage would be built to more aggressive mass fraction targets than the first stages. [Note: you can find a copy of the spreadsheet I used, in case it's helpful, here --Jon]

The operational procedure would be something like this:

  1. Both WK2′s head uprange to the appropriate drop zone, spaced as close as is reasonably possible–maybe 1km apart?
  2. The two TSTO stacks drop and light at the same time.
  3. During the first stage burn, the two stages slowly close the relative distance between each other, so that at staging they’re maybe 50-100m apart.
  4. As soon as the dynamic pressure is low enough (possibly even before staging), booms are extended between the two orbital stages.
  5. The first stages separate from the upper stages, and glide forward from the staging point to the launch field for landing and reuse.
  6. The upper stage and tanker stage would start their engines, while finalizing the boom connection.  During this phase, the upper stage would be following its own trajectory, and the tanker stage trying to match.
  7. The booms have built in propellant transfer hoses, and a quick disconnect possibly like the one ULA proposed for cryo prop transfer based on their slip-joint duct design.  The QD would engage, and as soon as it is sufficiently engaged, propellant would start flowing between the vehicles, either pump-fed or using differential pressure.
  8. At some point the upper stage gets tanked all the way up.  If there’s still propellant to be transferred, the stages may stick together for a short period of time, with the transfer pump operating throttled back in a way so that the upper stage is not using any of its own propellant.
  9. Before the vehicles separate, the QD is disengaged, and the booms reeled back out a bit.
  10. Vehicles separate, upper stage continues on to orbit.  Lower stage is still within a glide-based RTLS maneuver of the initial starting site, if the propellant transfer operation can be kept quick enough.

The nice thing about such a setup is that if you do things right, most worst-case failures result in an aborted mission, not a loss of vehicle.  If one of the TSTO pairs doesn’t ignite when air-dropped, you abort (with the upper stage from that TSTO combo having enough propellant to make it home, and you only have to figure out what to do about the first stage).  If you can’t mate up in time, you abort.  If the QD doesn’t work, you abort.  If you can’t keep the vehicles together exoatmospherically, at worst the boom/hose fails, and you use hydraulic fuses to keep that from becoming a loss of vehicle event.  Now, there are many more things that can cause an abort in this scenario, but many of them are things that should get more reliable with practice.  The nicest thing is that many of them can be practiced with first-gen suborbital RLVs without even requiring an air-launch.

Performance and other Observations
Here are some observations from my super low-fi simulations:

  1. Performance-wise, this system behaves like a quasi-3STO, giving you a bit of a benefit over a pure TSTO.  For instance, in order to put the same sized upper stage into the same velocity and altitude as the upper stage has at the point the tanker leaves it, you would need something like an 82klb stack instead of two 35klb stacks.
  2. The quicker you can get propellant flowing, the better.  The longer you have to throttle back to stretch out the refueling phase, the further you have to carry the tanker stage, which impacts performance.  Ideally you’d like to have propellant transfer done within 90-120s of when the first stages separate.
  3. You probably want to throttle-back during prop transfer, this makes it so you’re blowing through less propellant during this phase when you still have a lot of dry mass you’re accelerating.  This would come at the cost of either more lofting earlier on, or more gravity losses.
  4. Obviously more exotic propellants for the upper stages tend to provide better payloads, but the quasi-3STO benefits decrease.
  5. You want as much of the fueling subsystem mass on the tanker side as possible, since it’s the part that doesn’t get hauled all the way to orbit.
  6. You have to be able to pump enough propellant through the transfer hose over the course of the transfer to overwhelm the amount of propellant being used by the main propulsion system.   You are pumping this against the tank backpressure.  Both requirements suggest you probably want an actual pump instead of trying to use pressure feed.

At the end of the day, this would give you an orbital capable system, using the existing WK2 that had all stages fully reusable, and could carry at least 2-3 people.

How to Develop Exo-Atmospheric Refueling
There are three major challenge areas for fielding this technology:

  1. The formation flying GN&C development
  2. The rapid boom rendezvous system
  3. The actual propellant transfer interface

The interesting thing is that the second two systems can at least be tested out in the near-term without even reaching space.  For instance, using two VTVL hovering vehicles (say Xombie and Xoie, or Xoie and SuperMod), you could fly the two together, demonstrate the boom connection and even swap some propellants operationally.   These just require a relatively precise hovering mission, the two vehicles don’t even necessarily need to be actively cooperating.  The next level would involve figuring out how to fly the vehicles in close formation at high speeds and altitudes.  Finally once you had those, you’d reintroduce the exoatmospheric boom rendezvous step.  And if it really looks like it can hook up fast enough, you can demo a little fluid transfer up near apogee, where the air density is low/practically-nonexistent, and the velocities are low.  Only once you’ve developed and demonstrated that, do you try and build the full-up system.

The nice thing is that the incremental cost of flight tests for reusable suborbital vehicles should be really cheap compared to building orbital stages and TPS and other things, so once you have those systems available, it only makes sense to try it (especially if you can find someone crazy enough to pay you to try).

Other Benefits
There are a few other benefits to trying something crazy like this:

  • Once you’ve demonstrated exoatmospheric fuel and LOX transfer, people won’t be able to honestly question if orbital propellant transfer is feasible.  This is the acid test.
  • The rapid rendezvous techniques end up being very similar to what is needed for a suborbital RLV + MXER tether concept, so by developing this, you open the door for the latter.
  • If demand picks up enough eventually to allow you to do a bigger carrier aircraft and bigger system so you don’t need exoatmospheric refueling, having the technique in-hand allows you to launch even bigger payloads in a pinch without having to develop a bigger system.

Parting Thoughts
Complexity should generally be avoided, but sometimes complexity can make things easier, not harder.  A comparable ground-launched TSTO system would likely weigh upwards of 300-500klb wet, and would likely cost more to develop.  It would be a lot simpler, and a lot operationally easier, but its not clear that it would actually be more cost effective or profitable.  Mid-air refueling is still mostly used by the military, but it’s an indispensable part of military operations today, even though it adds complexity.  I think the case of exoatmospheric suborbital refueling will likewise be one of those crazy things that we wonder how we ever lived without.

This entry was posted in Launch Vehicles, Orbital Access Methodologies, Technology. Bookmark the permalink.

34 Responses to Random Thoughts/Orbital Access Methodologies VII: Air-Launched Glideforward TSTO with Exo-atmospheric Suborbital Refueling

  1. Pete says:

    I do like all the technologies this might demonstrate, but I wonder if this approach is otherwise entirely necessary – or even necessary for demonstrating these technologies.

    You suggest 2-3 people, presumably one WK2 might launch a one person say Siamese TSTO launch vehicle. Would this not be a higher flight rate, easier to coordinate, and lower cost system? Such a Siamese system could also presumably similarly and perhaps more easily demonstrate exo-atmospheric refueling if desired.

    I suspect more than 90% of payload to ISS (including construction) is not people. Even assuming an extensive tourist market, I am not sure this non people launching rate would get below 90%. Point being, I am not sure sizing the payload to carry a few people is necessary. Considering the hoop jumping required, carrying people may not even be desirable for the first generation.

    If a larger payload is on occasion needed, how about a standard type ELV TSTO launch vehicle where the first stage is reusable and the rocket engines and avionics (modular) from the second stage can be returned in the smaller air launched reusable launch vehicle using the otherwise underutilized down mass capability?

  2. Pete,
    I’m not positive, but I don’t think the scaling works out like that. I don’t think you can deliver a single person to orbit on a vehicle launched off of a WK2. As it is, 2-3 people is probably optimistic for this system. VG is looking at barely being able to do 200kg with a fully expendable system off of WK2.


  3. More to the point, a fully reusable orbital vehicle of *any* positive payload capacity is barely feasible off of a WK2.


  4. Robotbeat says:

    Sounds easier if you just launch 2 upper stages seperately on suborbital trajectories (each from their own WK2 and first stages), they meet up and dock (exatmospherically), and then one of the two stages acts as a 2nd stage and the other as the third stage.

    BTW, I like the idea of using suborbital vehicles to test things like orbital prop transfer, etc. It will have the side-benefit of making you have to approach and dock within only a couple minutes, versus the eternity it takes for the ISS (perhaps for good reasons, but necessity and a rapid build/test cycle are the mother and father of invention). However, propellant transfer from Progress spacecraft happens regularly on the ISS for fuel for reboosting, so I don’t see why the naysayers of the technique have a leg to stand on even right now, except perhaps for cryogenic propellants.

  5. A_M_Swallow says:

    Try using a jumbo jet as the carrier craft. You will need to add wheels onto your launch vehicle to turn it into a towed glider. You may be able to release the wheels (and wings) at carrier separation.

    The carrier craft can be used as a cargo aircraft between missions.

  6. Pete says:

    “More to the point, a fully reusable orbital vehicle of *any* positive payload capacity is barely feasible off of a WK2. ”

    I would think the same to then also be true of the dual WK2 approach, especially with the added refueling booms and associated systems. I can not see anything here that drastically does not scale down to half size?

    17 ton to 50,000ft, or there abouts for the WK2 with 200kg to LEO suggested. The Falcon 1e is ~35 ton and capable of ~1000kg to LEO, inferring perhaps ~750kg if scaled down and air launched. Reusability will seriously reduce payload – for both single and dual systems.

  7. Itokawa says:

    A rocket engine has a very powerful turbopump with which to pump liquid from propellant tanks into the engines. It still takes several minutes for a rocket engine to drain a tank. Do you think you can find a pumping system that will be able to pump liquid from the tanker to the orbiter in a sufficiently small amount of time, smaller than that of a normal rocket engine?

    One thing that will likely be necessary is a very large transfer pipe. But you’ll probably need a hefty turbopump to achieve the transfer quick enough. Big pipes, extra turbopumps == weight. Do you concur?

  8. Anom says:


    You could attempt a single-stage-to-orbit (SSTO) demonstration of your XA-1.0E vehicle that weighs ~ 30,000 lbs (fueled) and ~ 1,600 lbs empty, and that uses a cluster of your new ~ 3,000-lb thrust engines.

    An organization like DARPA/AFRL/MDA could do a 3-year program with you at Masten, where you build and test 4 of these SSTO vehicles in 4 flight tests where you would probably lose each SSTO vehicle within each flight test (i.e. no serious effort for payload fraction or reusability, but a serious effort to demonstarte orbital velocities in a single stage). This new SSTO program would probably cost ~ $10 Million over 4 flight tests versus the ~ $200 Million flight test program for programs like the AFRL/DARPA X-51A or AFRL/DARPA FALCON hypersonic and orbital research demonstrator programs. These programs will typically test and destroy 2 to 4 vehicles each to accomplish their goals, and I would think that Masten could rapidly build, test, and destroy 2 to 4 SSTO vehicles over the next 3 years for ~ $10 Million.

    This “crazy idea” for a SSTO program to reach orbit is probably a lot more closely related to what Masten could accomplish along its current technical path, and it is probably a lot more interesting for a lot of potential customers than the logistical nightmare of trying to refuel sub-orbital rocket stages to reach orbit.

  9. “More to the point, a fully reusable orbital vehicle of *any* positive payload capacity is barely feasible off of a WK2. ”

    Not according to Burt Rutan:

  10. Anom says:


    I also heard Burt Rutan say the same thing at the same time in 2004 on an American television show called “60 Minutes”. Burt Rutan was clearly not refering to a fully reusable orbital vehicle.

    On that show, Burt Rutan showed his interviewer, Ed Bradley, an orbital design which was clearly 2 or 3 stages to orbit. This orbital vehicle had a long expendable rocket stage connected to SpaceShipOne, and it was dropped from a carrier aircraft larger than WK1.

    You can probably find a YouTube video clip of that Burt Rutan interview from 2004, and you can see the picture of that orbital vehicle for yourself. It is definitely using at least one expendable rocket stage connected to a winged orbital vehicle, and it looked somewhat similiar to the “X-37 like” picture of their WK2 orbital launcher that also has wings, and that was displayed on your Hyperbola blog. This design looks a lot like a Boeing picture within the magazine Aerospace America that shows an X-37 connected to an expendable booster and launched to orbit from the top of a 747 aircraft.

  11. Jonathan Goff says:

    A typical rocket pump is pumping it’s volume against a very high backpressure (trying to force the propellant through cooling channels and injectors into a high-pressure chamber). If you’re only fighting say the 20-50psi pressure in the propellant tanks, the same pump power could deliver several times the flow rate.


  12. Jonathan Goff says:

    I don’t think Masten could come anywhere near an SSTO RLV with our current technology. Xoie was definitely good enough for the job, but getting to those kinds of mass ratios is really non-trivial. We won’t be doing orbital RLVs of any flavor for some time yet.


  13. Jonathan Goff says:

    Your link says nothing about being able to do an orbital RLV off of a WK2 carrier platform. If you relax that constraint, and go with a WK3, then of course you could make it work without doing anything fancy (I said as much). My point was that if you were trying to get to orbit from WK2, with a fully reusable system, that some sort of trickiness is required.


  14. A_M_Swallow says:

    A 747 can cruise at 35,000 feet at a speed of 0.845 Mach = 560 mph (901 km/h) whilst carrying 248,300 lbs (112,630 kg). More if not flying 4,445 nautical miles.

  15. Paul Breed says:

    VT hook them together before they take off……
    This then looks a lot like a reusable otrag….

  16. Pete says:

    Jon, I have never found any rough performance numbers on XCOR, Masten or Armadillo rocket engines. T/W? ISP? How might these numbers extrapolate to an orbital vehicle?

    Presumably this information is not yet public… :-)

  17. Anom says:


    Expendable SSTO and not an RLV.

    This should massively reduce your materials requirement. You can have it de-orbit to destruction over the ocean.

    If your vehicle empty weight could reach ~ 5% of its fueled weight then you would have a chance with the engine technology available to you.

    This would be a demonstrator and a confidence builder, and not an operational vehicle. This would give a confidence level of the costs to maybe build an orbital RLV as a follow-on.

  18. john hare says:

    I wonder if you could do an early demonstration with just one Xoie class VTVL. Boost it to ~mach 2 at 30 km and release two separate micro upper stages. They would have about two minutes to demonstrate formation and docking capabilities even with no real accelerating propulsion system. The advantage would be less risk for your primary revenue vehicle, and the ability to develop the capability in house with one vehicle.

    The capability would also allow you to use a single bus revenue upper stage later on with variable altitude and payload capabilities provided by free flying tankers.

    Another possibility is launching with a seriously underfueled crew upper stage that would then have the T/W to be an escape system if required by circumstances in the lower stage.

  19. Doug Jones says:

    I sorta agree with RobotBeat- exoatmospheric docking turns into a bimese twin stage. If the gimbal range is good, both sets of engines can stay at full throttle even as the tanker craft gets light, the mated pair just goes to a rather high AOA, which doesn’t matter at low Q.

    Achieving a fast safe docking argues for some arms on the tanker to grab the orbiter as both accelerate.

    Actually, this whole idea sounds like it’s single stage to NASA funding :)

  20. kert says:

    That reminds me of the Single Stage to Tether concept, where a subrobital would be helped to orbit by a rotating orbital tether.
    see spacetethers.com

  21. Pete says:

    “I’m not positive, but I don’t think the scaling works out like that. I don’t think you can deliver a single person to orbit on a vehicle launched off of a WK2.”

    Then the first and foremost design challenge is to “make” the scaling work out like that. The alternative is billion dollar development programs that forget where they are before they even get one prototype built. Is not fifty years of such boondoggles enough?

    There is an awful lot to be learned, many prototypes are yet required. Assuming ~$300 million overall funding maybe $10 million dollar prototypes that take six months are possible without people losing their way. Simply throwing more money at this problem has specifically not solved it countless times in the past.

    $3 million prototypes would seem to be a lot safer scale, especially to start with (plan on vehicle growth). How large a reusable rocket vehicle might you prototype for $3 million? At $10,000 a kilogram of dry mass I get maybe 50-100kg of payload (very low dry mass fractions and reentry systems will be required). Maybe 5 ton air launched GLOW – well within WK2′s capability.

    Five years before you have to start paying the investment back (definitely no more than ten). Ten prototypes in five years means one on average every six months, and the last few prototypes will probably have to enter service. Skimping on prototype number would probably not be a good idea.

    So what technologies and development tricks (like sub orbital stepping stones), are necessary to develop orbital launch vehicles at these individual prototype cost/time scales? How much might the prototype cost/time, and hence overall development cost/time be reduced?

  22. nick says:

    Intriguing idea. Here’s a much simpler version:

    (1) launch two rockets from two White Knight 2s, as before, but the first is a one-stage rocket carrying payload, shroud, and optional orbital transfer stages, and the second is a two-stage rocket;
    (2) jettison the consumed first stage from the first rocket, leaving the payload/shroud/etc.;
    (3) jettison the consumed first stage from the second rocket;
    (4) dock the second stage of the second rocket with the payload/shroud.

    No worry about fuel lines, pumps, pumping rates, etc.

  23. Bob Steinke says:

    When people say “why don’t you just build a bigger rocket” remind them of the Panama canal and supertankers. No matter how big you build it someone will come along the next day and ask for a little bigger.

  24. john hare says:

    A weird thought on suborbital refueling. Aries with no upper stage carries a honking big propellant tank. After SRB termination, an Atlas heavy refuels for a massive payload boost. Keeps everybody employed too.

  25. Doug Jones says:

    A lot of these ideas- Rick’s “jack up the payload and insert a two stage rcoket” in particular- remind me of going skydiving without a parachute, you just grab one from your buddy after you go out the door. Yeah, you can make it work, but the failure modes are ugly…

  26. john hare says:

    You don’t need a parachute Doug, just learn to fly, the rest of your life to learn.

  27. Tim says:

    Hi John, Itokawa’s comment and your response got me wondering. Could you use the transfer pump to temporarily increase the tank pressure, and pass the pressure rise downstream to give an increased chamber pressure, and using the resultant increased performance to both reduce propellant consumption during refueling and increase the refueling window? You’d be trading a low pressure rise high flow pump for a high pressure rise low flow pump, and I’m not sure that makes sense. I suspect there is a flaw in my logic here, but I’m not sure where it is.

  28. Tim,
    Higher pressures are mostly useful at low altitudes (where they allow you to takeoff with a bigger nozzle expansion ratio). At higher altitudes like you would get with an air launched stack, it’s only a second order benefit. You get more thrust out of the engine, but the Isp gain is pretty small. That said, you’d want to run a sim to actually quantify the benefit, to guide you on whether or not it’s worth the hassle.


  29. Martijn Meijering says:

    According to this link http://rt.com/Russia_Now/Russiapedia/Those_Russians/tkonstantin-tsiolkovsky.html Tsiolkovsky himself considered suborbital propellant transfer for launch applications! He apparently thought it was necessary to reach orbit. The article doesn’t give a time frame so it’s not clear if he beat Baron von Pirquet to the idea of propellant transfer.

  30. Jeff Wright says:

    So an HLLV that allows for propellant transfer safely on the ground Rand hates, but this Rube Goldberg stuff he likes.


    Dock three times, touch your nose, clap three times, do the Macarena…and get an astronaut killed. The OCD way to space–make it as complicated and as many steps you can just to keep from admitting that it is best to–I don’t know–MAKE BIGGER ROCKETS

  31. Jeff,
    I hadn’t heard that Rand liked this particular idea. I would’ve thought that he also would think this suborbital refueling concept was more than a bit Rube Goldberg (I came up with it and I sure think it’s Rube Goldberg). Did Rand actually suggest or endorse this particular approach somewhere? I would definitely think some of the earlier Orbital Access Methodology approaches would be a much more realistic fit for a depot-architecture. But really you don’t even need that for depots to make sense. They already make sense even with normal LVs like Atlas and Falcon.


  32. Roy Paul says:

    Here’s a variation on your original idea that might work. Use three TSTO vehicles rather than two and join them together at launch rather than join together in the atmosphere. The two outer TSTOs would pump fuel into the center one all the way up to say, 20 miles, then drop away. Then center one, fully fueled, would then continue on to orbit. This version has a name: Falcon Heavy.

  33. Jonathan Goff says:


    Thank you for pointing that out! I would’ve never thought of the idea of crossfed strap-on boosters without your insightful help!


Leave a Reply

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

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>