Wednesday, June 15, 2011

Aerial Propellant Transfer.

There are I guess four basic types of schemes for getting into orbit (putting aside the nuclear option which for political reasons and genuine safety reasons isn't worth wasting time discussing), there's the disintegrating totem poles we've always used (eg. Falcon, Shuttle), there's the really big capital investments schemes with banks of lasers, or banks of microwaves emitters to get the effective Isp of some tiny launch vehicle up enough to get it to orbit, thers's space elevators, and then there's some sort of fully reusable self-contained launch system (eg. Skylon).

If you want to see airline type operations into orbit forget Falcon, expendable or nominally re-usable - more like refurbishable - launch systems will never get the cost down low enough, they're like throwing 747's away after a single use.

Maybe one day we'll see the laser launcher or a space elevator, but if we do it'll only be after there's already lots of stuff going up via other means, people need to be realistic that big mind numbingly huge capital investments aren't going to get the ball rolling.

So we're left with the Single Stage To Orbit (SSTO) and Two Stage to orbit (TSTO)fully reusable options that have been proposed again and again over the last 40+ years.

Why hasn't one of the many variations on these methods ever been seen through?

SSTO has the problem with mass ratios, even with the most energy dense rocket propellant in use, 90% of your launch mass has to be propellant and the other 10% is structure, leaving ~0% as payload. You can look to improving that mass ratio with air breathing engines, and if jet engines could be built with the same thrust to weight ratios as rocket engines have, SSTO would be a done deal, but the reality is that the best jet engines only have T/W ratios of about 13:1 (verses about 50:1 for rocket engines), and you end up taking all the mass of those engines to orbit, and one tonne more of engine means one tonne less of payload.

TSTO has the problem of shear size, building a winged orbiter to carry 20 tonnes to orbit would in itself be a manageable financial expense, but to build the ~800 tonne winged booster for a twenty tonne payload isn't quite so manageable. When they were looking for a reusable system to replace expendable launch vehicles in the 70's the designs were often of two stage fully reusable vehicles, and Congress always said "no - too expensive".

If you've seen many of my comments on various blogs recently you probably know that I've become a bit obsessed with Aerial Propellant Transfer for spaceplanes. The reason is that I see it as a route to achieve today what could not be done 40 years ago. APT means you can dispense with that 20 or 30 billion dollar specialized booster and replace it with a $350 million dollar (plus modification) commercial aircraft.

Probably one of the first reactions to the suggestion of a "booster" that only goes to Mach 0.85 and 12,000 meters is that it's not enough to make a difference. But it actually counts for more than commonsense would suggest; when the orbiter begins its climb to orbit its engines are working with far less atmospheric back pressure and so are more efficient, the orbiter can take off from the ground with only enough propellant on board to reach the tanker, and so requires lighter landing gear and possibly smaller wings, and the atmospheric drag from 12,000 meters is far less than that from the ground half the atmosphere is below you.

That start from altitude is worth about 1km/s, it reduces the propellant mass ratio from about 90% to about 85%, and if 11% of your light-off mass is orbiter your payload goes from less than 0% to 4% of the orbiters total mass, even better, aircraft landing gear typically makes up about 2% of lift-off mass, if you're refueling in the air the weight of the landing gear can be halved - so you've increased payload by another 1% of the orbiters total mass to 5%.

If want to get a 20 tonne payload to orbit, how big does the orbiter need to be? 20 tonnes X 100/5 = 400 tonnes, of which 85%, or 340 tonnes is propellant and 40 tonnes orbiter.

So we need a tanker that can transfer up to 340 tonnes (minus any excess that the orbiter has carried up) to the orbiter at an altitude of 12,000 meters.

That's a lot of propellant, it's more than can be carried by a 747-8, and even the A380 freighter only has a take-off weight of 590 tonnes total, of which 250 tonnes is plane, which would leave only 340 tonnes for both rocket propellant and the A380's own fuel.

I've tried sounding aircraft engineers out on this, the A380 as a tanker has far more space in the fuselage than would be required, my thoughts are that perhaps a fuselage based on that of an A340 - some models of which are actually longer than the A380 - could be attached to the flight surfaces of the A380, that would shave about 80 tonnes off the total mass, but even doing this would still require that some of the rocket propellant (LOX) be carried in the tanker aircrafts modified inboard wing tanks as the structural payload (payload that can be carried in the fuselage, even with the lighter fuselage, would still only be 230 tonnes.

So while heavy lift to orbit using APT might be possible now, the aircraft to do it didn't exist just a few years ago, and even the A380 would need big modifications for a heavy lift orbiter.

Another point I want to emphasis is that using even the A380 as a booster for air launch wouldn't put anything like a 20 tonne payload into orbit, even with sticking on an A340 fuselage to reduce weight, the maximum weight of the fueled orbiter would be 230 tonnes and the increased drag and high center of gravity would bring in more complications, maybe 9 tonnes of payload could be carried to orbit.

ATP has an advantage over a system which uses a direct ascent from a runway, there are far more launch windows because the ascent from 12,000 meters can take place anywhere within perhaps a thousand kilometers of the runway.

So how much would each launch cost?
The cost of flying a jumbo on a long haul flight is around $500,000, a third of which is fuel, that's also about the cost of a USAF KC135 flight to refuel other aircraft a thousand km from base. So excluding the cost of the propellant transfered, that's probably a reasonable figure for the tanker.

The cost of the propellant LH2 and LOX will be around $250,000

The cost of the orbiter excluding propellant will be servicing and payback on capital invested, with the servicing aerospace engineers tell me that with the experience gained from the SSME's it's possible to build H2/O2 that can be turned around in with almost no servicing between flights if that applies to all of the orbiters systems, servicing of $450,000/flight seems reasonable (that's about 20 times the servicing cost/tonne of airliners).

With the payback on capital invested, the development of aerospace vehicles that don't require radically new technology is around a hundred million dollars a ton dry mass, the numbers quoted for Skylon are much higher than this, but everything about Skylon is new technology, from the engines, to the active thermal control system, to the truss frame construction. The ATP would weigh about 45 tonnes dry, so I'm putting development cost at $4.5 billion, if 10 are initially built and per unit manufacturing cost is $100 million that's $550 million each. If they're good for one thousand flights that's $550,000/flight.

Adding those all together works out at $1.5 million per flight plus the operators profit. Which comes to $75,000/tonne or $75/kg payload.

That seems ridiculously cheap compared to the SpaceX Falcon, but then they're still throwing hardware away with every flight.

It also seems ridiculously low compared to what REL are talking about for Skylon, by then Skylon is really pushing the envelope with new technology, and it's the last 10% that takes 90% of the effort, an ATP using the Skylon structural system would in theory weigh half as much dry and carry twice the payload.


  1. Hello Andrew,

    How much time is allowable for fuel transfer? 360 m3 per hour is 1800 usgpm, or an 8 inch pipe. And i'm certain you'd need to go much faster and that the fuel is less dense than water...
    Could you fly up a tank (rather like the shuttle one) and have the orbiter pick it up?


    Michel Lamontagne

  2. Could you fly up a tank (rather like the shuttle one) and have the orbiter pick it up?

    I'm after a system that's 100% reusable and doesn't require reassembly after each flight. Do you see a way to achieve those objectives in what you have in mind?

    The fuel transfer is something that has several different ways in which it could be done, they might all be possible and I'm not sure which would be best.

    My favorite is for the tanker and spaceplane to rendezvous at low altitude, for the initial connection to be a towline with both LOX and LH2 feed pipes attached, it would only be after the tow line was connected and under tension that the fuel and oxidizer connections would swivel down, one after the other to make firm seals with the spaceplanes plumbing. most of the load would be on the tankers engines with the less efficient rocket engines only making whatever contribution was necessary to get them both to separation altitude.

    At the other extreme there's the option of no tow load going on and only the LOX being transfered across, with all the LH2, both to rendezvous and on to orbit being carried by the spaceplane from the ground.
    In the oiginal Black Horse study it was proposed that it be done this way to keep things simple and because, as the oxidiser made up the vast majority of the propellant, most of the advantage of APT could be gained without bothering with the fuel being transfered.

    I think there was a bit of a phobia about transferring both fuel and oxidizer that I don't think is warranted, as, as long as connections are made and separated in series (one after the other) LOX and LH2 will disperse quickly in the air flow with no possibility of them having any chance to go bang, after all, air forces already do fuel transfers in an O2 rich atmosphere.

    A towline has already been demonstrated as possible by Kelly Space and Technology, who did a tow from the ground of an idling F106 by a C141.

  3. As I said it's a complicated issue, if a tow is to be used the location of the anchor points will need to be near the center of mass of both aircraft, because the spaceplane gets very heavy relative to the tanker when nearly full, this introduces more complications with the spaceplane because that's where the doors of the payload bay will need to be, and with the towing aircraft the anchor point will need to be under the aircraft well forward of the tail.

  4. Hello Andrew,

    I scaned wikipedia a bit and looking at the shuttle external tank, I realised it's a kind of aerial refueling system itself. There are 2 17 inch pipes between the tank and the orbiter, with the oxygen pipe being quite long. So I was completely wrong when I worried about fuel transfer times. Since I have found that most of the blogs out there annoy me with people stating opinions as facts, I was kind of mortified of having done it myself.
    So I made a drawing of an horizontal take off SSTO beeing fueled by an airbus. It's on my site. Hope you enjoy. Please comment if you feel anything is off, such as the orbiter camber, model, wings, type of boom, or wathever!


    Michel Lamontagne

  5. Hi Michel.

    obviously a 17 inch pipe will give you 4 times the flow of the 8 inch pipe you started with, I expect that the lower viscosity of LOX and lower viscosity and density of LH2 would allow the propellants to flow fasted than would water. I don't think you need to feel mortified.

    Thanks for the picture, the orbiter as I imagine it would be similar in layout to the shuttle, though more streamlined as it would be about 55 meters long with a 6m diameter fuselage, about the same size as an A300, delta wings like the shuttle.

    One thing that did occur to me though was that it might be possible to have the payload bay doors open as "gull wings" hinged off a ~two meter wide dorsal spine running over the top of the payload bay, the spine would be the anchor point for the tow. The actual bay would only be 4.5 meters deep and have a flat floor nearly 6 meters wide and up to 12 meters long at about the level of the top of the wing (wing bracing running across the fuselage under the payload bay floor connecting the two wings, a stronger arrangement that structurally separate wings) I think it should be possible to have "wet wings" with LOX in them with the LH2 stored in the fuselage in tanks ahead and behind the payload bay (as done in Skylon).

    But heck, if it's ever built, who knows what the engineers would come up with?

    Best wishes,


  6. Hello Andrew,

    So a 10t payload to orbit spacecraft, weighting 20 tons dry and requiring 170 tons of fuel, for a total of 200 tons? 55m long.

    I made a sketch, that you can find on my website. Seems about right?

    I wonder: if the payload was brought down to 5 tons, couldn’t we have 100 T orbiter, pre-fuelled, sitting on top of the A380, rather like Enterprise on the Boeing 747?

    I guess you would then need some kind of orbital assembly facility, but at 300$ per kg, it could make sense to build one…, ground handling would require a large crane as well.

    Can the orbiter engines be started in flight? I vaguely remember that being one of the problems, back when they designed the shuttle.


    Michel Lamontagne

  7. Hi Michel,
    I'm pretty sure the SSME's igniter is a once only device. If you were going for a smaller air launch orbiter you'd maybe use a late version of the J2, the J2 was used on the 2nd and 3rd stages of the Saturn V, they were obviously started in flight and were optimized for use at altutude, I don't see why that wouldn't work, Boeing have worked on something similar, and they were looking into putting a SSME in the tail of the 747, to get that extra altitude, my thoughts are though that the more complexities you bring in are going to push up price more than payload.
    Yep, what you've drawn there is pretty close to what I imagined it would look like, the possibility LOX stored in the wings still interests me, big delta wings have a lot of volume.

  8. "I'm pretty sure the SSME's igniter is a once only device." Nope, looks like the entire igniter unit is re-usable, the older expendable rocket engines used to use what are basically solid rocket motors to start the liquid fueled engines, the SSME's use 6 spark igniters, 2 in each of the 2 preburners and 2 in the main combustion chamber.

  9. Hello Andrew,

    I've added a calclation for the launcher to my web site, with a propane powered version as an alternative, to try to avoid that huge hydrogen tank. The propane ISP is quite low though. Probably not practical. Sigh.
    I've also updated the moon rotovator spreadsheet, and added choices to the material selection. It turn out that Xylon can be used fo the cable, nanotubes are not absolutely required. The center of gravity of the cable beeing halfway down the cable, the induced sttrain is lower than I thought. However, then motor HP goes up a bit.


    Michel Lamontagne

  10. FYI:
    The SSME "igniter" IS reusable, the only problem is it's part of the launch pad and NOT on the Orbiter :)

    IE: No "restart" for the SSME. They had the same issue when the idea of an "air-start" for the SSME was suggested. The SSME igniters are the things producing those huge shower of sparks that are thrown at the engines so mounting "one" on a booster was a no-go.


  11. Hi Randy, good of you to visit, but I think you probably should read this thread:

  12. Hi Michel, thanks for that. We've been thrashing this about at NASA spaceflight, (which is why I've been a bit quiet here):

  13. Hello Andrew,

    Interesting discussion over at the forum, but it's quieter here, and I can't bring any expertise to the fueling boom issue!

    Why do the payloads have to be so heavy? The design of the orbiter system seems a lot simpler if the payload is smaller. A 2 to 5 ton payload for example. I understand that this would then require orbital assembly, But at a few hunded dollars per kg instead of thousands and tens of thousands it should be possible to develop some kind of orbital assembly system using the difference in cost. It would give them something to do at the ISS! (although remote controlled assembly is perhaps simpler) How small do you think the minimum building blocks could be for a communication satellite assembled in orbit?

    From the spreadsheets, I find that the real killer obstacle for this type of launch system is development cost. If there is no volume, the costs per kg become as inflated as those of the shuttle. To me, the obvious "volume of traffic" application is solar satellites. I've seen recently in the news that a new pipeline from Canada to the US for petroleum from the tar sands will require 7 billion dollars in financing. A smaller launch system might bring the total cost closer to the one large corporations are willing to pay. Specially if the initial solar satellite is not too big itself. Is this the general consensus?


    Michel Lamontagne

  14. Yeah, I'm coming around to the idea that getting something flying that's fully-reusable, even if it only carries a few people or less than a tonne of cargo, might be the way to go to at least get the ball rolling, if that worked out it would be far easier to get finance for something bigger.

    I think a system with a larger P/L (at least 15 tonnes) would be needed fairly soon though to get per flight overheads down to make space based power affordable.

  15. On lower payloads-to-get-the-ball-rolling;

    A big factor here is the lower development and production costs of a smaller spaceplane help close the financials a bit. Jim over at DOES have a point that it won't be anywhere near where the "market" is for communications satellites as they are now, however you have to keep in mind that's the only CURRENT "market" most people can see. Small-Sats, Micro-Sats, and Nano-Sats are not considered a "market" right now because they pretty much rely exclusivly on hitching rides on "normal" launches. There is no real market for putting people into orbit or payloads of smaller than 10,000lbs because there is no where for people to go and no "useful" satellites of that size. But that COULD change if regular access was available.

    And I think that MIGHT actually be major "selling" point for a smaller vehicle: Fast turn-around and more regular operations.

    If you have a vehicle that can "only" put 1000lbs into a 100nm orbit, but can be turned fast enough to do so every 3 or 4 days you have almost 10,000lbs in LEO at the end of a month. And that is 'just' a single vehicle!

    Now this is NOT something you can walk into an investment service and "sell" unfortunatly since they would simply call an "expert" on the phone who would tell them that this won't serve the communications satellite market and then repeat the rest of my list above and then tell them it's not "viable" and hang-up. (Though he probably charged them a couple hundred dollars for that "assesment" at least SOMEONE got paid :o)

    But I have a 'hunch' that there is a possiblity here for a "business" case if the particulars can be worked out enough to shop around FOR a market.


  16. "But I have a 'hunch' that there is a possiblity here for a "business" case if the particulars can be worked out enough to shop around FOR a market."

    This is one area where you and I are on different wavelengths, I look at hundreds of people booking space flights with Virgin Galactic and see a demonstrated demand, if not quite a demonstrated supply, whereas you seem to look at almost zero space tourism now and don't see a certain market.

    Of course, the capacity of carrying ~7 people to orbit is hardly a novel idea, what with Dragon, CST-100, and Dream Chaser all vying to serve that market.

    I've been thinking about what past or present aircraft most resemble, in terms of wing area and shape, and the aircraft overall shape including fuselage, a ~7 person APT spaceplane, just to try to better visualize it, and think the B58 Hustler and the Mirage IV are closest.

    And wouldn't it be cool to have a full fighter canopy, rather than just the traditional passenger cabin layout :)

  17. Hi,

    I want to congratulate and thank you for publishing this article. For some odd reason we have been squandering billions of dollars for over 30 years because we can't seem to apply our physics education and mathematical literacy to the problem of the mechanics and economics of spaceflight.

    You have correctly identified the most cost effective and fastest scheme for achieving low-cost, routine access to space. One only has to run the numbers to see it and I have no idea why people are spending so much time and effort on obvious dead-ends.

    Anyway, I'm involved in a project you can read about here:
    I am writing Reference Designs for a project for a consortium of the petroleum industry by which modified Boeing 747-8s will refuel orbiters in a daisy chain manner. They would refuel three orbiters at about 60,000 feet and mach 0.95. After that the orbiters would climb to 145,000 feet and 4000 mph on rocket power (methane and Hydrogen Peroxide). Then, the two orbiters would refuel the third which would continue on to orbit. The other planes would return to Earth and land.

    We chose the 747-8 because it carries more weight - or can be made to carry more weight - than any other plane in existence ... yep, more than the 380 or C-5 Galaxy. We can load over a million pounds of oxidizer to an orbiter in this manner and put 50,000 lb. cargo per flight into orbit.

    There is nothing magic about this at all and it is indeed very routine. The only caveat in that is that the aerial refueling will involve some pretty advanced development work. But in the scheme of spaceflight boondoggle things its rather pedestrian.

    It is refreshing to read this rare gem you've written and you and I think a lot alike. This is a great blog and I'll be following it closely.

    - kk