An Insane, But Interesting Idea: Fleet Launched Orbital Craft

Ok, before I go into this latest blog post, I want to put a disclaimer up front:

  • This idea is crazy.
  • I’m not posting this because I think it’s the greatest idea since sliced bread.
  • I don’t think that this the one and true way to get to space.
  • Don’t try this at home.
  • I don’t necessarily thing this is better than .
  • Oh, and did I mention I thought this idea was a bit crazy?

But even if it’s crazy, it is interesting. And it was developed by a guy with the same last name as me, Allen Goff of Novatia Labs (Sacramento, CA), so while it’s a bit crazy, it’s definitely clever.

The idea is what Allen called Fleet Launched Orbital Craft, or which another author calls “Separated Ascent Stage Launch Vehicles” (I’ll call it by Allen’s term “FLOC” from here on out). The idea is somewhat related to bimese and trimese launch vehicles. In a bimese launch vehicle, you have a TSTO vehicle where both stages are identical to each other. Both stages ignite for a vertical takeoff, and the ideas typically use propellant crossfeed to guarantee that one stage is still full when the other one’s propellants are almost depleted. At that point the empty stage separates and returns home while the other one proceeds on to orbit. Trimese approaches are similar, but use three identical stages. The theory being that by designing just one stage and using it several times, you can get a lower development cost, even if both stages are somewhat suboptimal.

FLOC is just the logical extension (or maybe reductio ad absurdum) of the Bimese/Trimese concept. You could technically call it an n-mese concept, where instead of only 2 or 3 similar stages you instead have “n” similar stages. John Carmack’s looking at one version of the n-mese concept (like OTRAG did in the past), but this one is different from John’s modular concept. The big difference between FLOC and a more traditional n-mese approach like John’s is that for FLOC, not all of the stages are attached when the vehicles are on the ground. An illustration from Chris Taylor’s AIAA article on the economics of the approach (AIAA 2006-4783) might clarify things a bit:

Basically, you have 2^n stages at takeoff, each of them paired together into a bimese configuration. They all takeoff together, all from right near each other (ie they probably all launch within 1-2 seconds of each other, and within 1-2km of each other). They fly as close together as is physically safe. They use propellant crossfeed to guarantee that one stage on each of the bimese pairs is still full when the other one runs dry. When those stages run dry, each bimese pair stages. The “empty” stages all return to the launch site (possibly using an airbreathing engine once back down to subsonic speed to cruise back). The full stages then perform a…wait for it…exoatmospheric rendezvous with each other, mechanically hook-up so they can operate as a new bimese pair, reestablish propellant crossfeed, and then continue on their way. You then lather, rinse, and repeat until your final stage ends up in orbit.

I mean, what could possibly go wrong?

Seriously though, as Chris points out in his AIAA paper (linked above), while the idea is totally crazy, it does have some interesting ramifications. Chris points out that using a fleet of 8-32 launch vehicles, you can place extremely large payloads into orbit using stages that have a propellant fraction similar to a 747 without having to use cryogenic propellants. Also, you can “tune” your payload to orbit (or your suborbital performance) by adding more or less bimese pairs. Using a 747 sized launch vehicle, they were predicting up to 200 metric tonnes (!!) of payload to orbit in a single launch campaign using 32 vehicles. That’s about twice as much as a Saturn V, with a stage design that’s so low performance it’s more like a commercial airplane than a rocketship. As I said previously, smaller payloads could be done as well using smaller numbers of stages.

How hard can doing a dozen exoatmospheric rendezvous be? I mean, you have about a 2 minute window for each rendezvous operation. That’s plenty of time….

Needless to say, there would need to be a lot of operations development and practice before such a system could become practical. The rendezvous happen outside of the atmosphere, which definitely helps a lot. And the launch vehicles can all launch from the same area at the same time, which also helps a lot. It’s a lot easier to do a rendezvous when you’re already flying along an almost identical trajectory at the same time. You’d probably need some sort of automated hold-down mechanism for the bimese pairs, or some way of starting up all the rockets in idle and automatically checking to make sure things are working on each vehicle before you commit to launch. You’d probably also want to do things like building in more thrust than the vehicles actually need (so that underperformance on one engine, or a premature engine shut down doesn’t risk the mission). Having a redundant bimese pair or two (depending on the number of stages you’re launching already) could also help. Another point that should be made is that in the 2 minute rendezvous window, you only have to get the vehicle pairs mated back together–you can actually hook up the propellant crossfeed after the engines have lit, so long as it happens before you burn about say 1/3 of the propellant of the combined vehicles. This allows you to take the process in two steps instead of having to do both all at once.

More importantly, I’d also like to see the rapid rendezvous and mating demonstrated a lot of times on a subscale basis before trying this for real. You could demonstrate at least some of the basics using cold-gas thrusters on armed robots on something like a Zero-G flight. That would at least allow you to get some of the basics of grappling and mating down in an environment where you could easily get 50-100+ attempts for only a few tens of thousands of dollars. The next step would be demonstrating with two subscale suborbital vehicles that you could consistently do the full rendezvous and mating operation exoatmospherically. With suborbital vehicles, you could start out with a more relaxed rendezvous window of say 5 minutes, and work your way down as you get the bugs worked out.

The good news is that a single one of these stages (or a single bimese pair) should have enough performance to perform a suborbital flight with a decent sized payload. You can then slowly work your way up from there. With a 4 stage configuration, you should be able to at least get to orbit with some payload (something light, probably in the 1-2 tonnes range) once you’ve demonstrated and debugged doing a single exoatmospheric rendezvous mission. After that, it’s mostly operations from there, working your way up to the point where you have enough reliability to reliably pull off larger missions.

Anyhow, for more details, read the paper. I just thought that while this idea was crazy, it was a very fun and interesting form of crazy, and does actually get you thinking.

Anyway, enjoy!

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.
Jonathan Goff

About Jonathan Goff

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 Launch Vehicles, Technology. Bookmark the permalink.

28 Responses to An Insane, But Interesting Idea: Fleet Launched Orbital Craft

  1. Brian Dunbar says:

    I had to pun ..

    Sorry ’bout that. Very interesting idea.

  2. Mike Puckett says:

    Certainly the most ‘Out of the Box’ concept I have heard in a while.

  3. Gary C Hudson says:

    As part of the Phase 1 RASCAL concept studies this was considered by the Pioneer Rocketplane / HMX team in 2003 or 2003, can’t recall exactly when. But we considered only n=2, both vehicles being derivatives of the Rocketplane concept of that time. I believe there were about ten RLV concepts evaluated and this one stands out in my mind as the least feasible, but Mitchell Clapp might recall better, as he was running that part of the study. In other words, nothing new under the sun…

  4. Iain McClatchie says:

    Why is there a need for rendezvous?

    Instead, start with 2^n stages all stuck together, propellant crossfeeds in place and operational. Burn until half the fleet is out of gas and the other half is full, and then drop half the fleet.

    Then you reroute the propellant, since half the stages just sucked someone else dry, and need to change propellant supply to a different vehicle.

    It seems to me the bimese seperation, then subsequent rendezvous is just an elaborate way of handling the plumbing. I see obvious benefits and very little loss from just adding the plumbing in the first place.

    You end up with a design where, starting with 2^n vehicles, the guy left in orbit starts with n crossfeed inputs. One might argue that so many crossfeeds would increase mass. But since just only one crossfeed runs at a time, you can have most of the plumbing on other vehicles.

    In fact, I’m pretty sure all you need on each vehicle is two inputs and one output, such that the output is fed from either internal tanks or one of the inputs. I’m gonna guess that that amount of plumbing weighs less than something that can establish a connection while in flight.

    You didn’t mention, are the vehicles doing their rendezvous while under full ascent thrust?

  5. roger jacobs says:

    Wont work?
    If you want to launch a large payload of 200MT with 32 bimese pairs, you have to split you payload in 32 parts!
    Probably makes sense for bulk payloads like fuel.

  6. redneck says:

    >>>>>>Wont work?
    If you want to launch a large payload of 200MT with 32 bimese pairs, you have to split you payload in 32 parts!
    Probably makes sense for bulk payloads like fuel.<<<<<<<
    May I suggest that you read the paper, or at least the post before being critical about a problem that does not exist.

    I found it interesting that this paper seemed to be more honest than many proposals I have read. The main savings was more in the relaxed demands in construction than any increased performance. One thing that struck me as a construction worker is that large numbers of identical launch pads would be cheaper and easier than individual pads of massive proportions.

    This paper is worth reading for the other material it goes over before getting to the main idea.

  7. Anonymous says:

    An interesting project would be to try this thing on model scale, like with HPR hybrids. However, guidance, IMUs etc would have to be very well done and even then in atmosphere rendezvous is potentially hard enough as to be a showstopper.
    That would be a nice followup to the CPSS Starboosters project

  8. Rand Simberg says:

    I think you’re misusing the word “rendezvous” here and making it sound harder than it is (not to imply that it’s easy). As you note yourself, they’re really flying in formation, so no rendezvous is required (that word generally means two vehicles)(or persons) meeting coming from different destinations). It’s mostly prox ops and docking. The difference between this and in-flight refueling is that it’s necessary to not only make a fluid connection (and a pressurized one) but also to hard-mount structurally. It is something that definitely should be demonstrated and practiced a lot before attempting actual flight operations with it.

  9. gravityloss says:

    I love it. 🙂
    The first burn has to have enough delta vee to get out of the atmosphere, otherwise this will be deadly. But I’ll read the paper.

  10. Anonymous says:

    I think one of the most interesting elements of the paper comes at the end:

    the concept might still be useful as a way to
    reduce the cost of future interplanetary (or even interstellar) missions where high delta-V’s are also needed but the
    time and atmosphere problems would be absent.

    Put four identical vehicles in orbit, two provide the delta-v to get the other two to Mars to land. Once mars exploration is done, both take off, with one providing the launch, and the second providing delta-v back to earth.

    Obviously, this is overly simplistic of the difficulties (landing on Mars full of fuel, launch from Mars without a launch pad, etc.) but it is intriguing.

  11. ザイツェヴ says:

    If Iain’s line of reasoning to be followed through, we end with OTRAG.

  12. Anonymous says:

    Forgive me if this is a silly question but, would it be possible avoid direct docking by having one of the ships in a pair “tow” the other using some kind of tether? Maybe the grappling mechanisms considered for MXER tethers could be used…

  13. Jon Goff says:

    Yeah, I took out the mention of Clapp’s proposal during my final edit before posting. This isn’t a new idea, just one that most people (myself) hadn’t heard about before last week.


  14. Jon Goff says:

    The problem is that a 32-stage cluster like that is first off going to be a drag nightmare, second off is going to expose different stages to completely different loads and thermal environments, third off it makes aborts and crew escape more tricky, fourth it probably requires a lot more structure, fifth, it makes TPS for the stages a lot more complicated as the stages now have to be structurally linked in more than one direction, etc.

    Basically it’s switching one complexity for a different one.


  15. Jon Goff says:

    Good catch on the wording re: rendezvous. I was having a hard time figuring out a better term.

    As for mechanical contact being necessary, one possible alternative version of this is to just transfer propellants instead of docking and mating the two vehicles. But I think you get more out of making the physical connection.

    But yeah, lots and lots of demonstration and practice.


  16. Rand Simberg says:

    As for mechanical contact being necessary, one possible alternative version of this is to just transfer propellants instead of docking and mating the two vehicles.

    Now you have to continuously match velocities at the same time as you’re depleting propellants from the tank of the “tanker” vehicle.

    As you said, what could go wrong? 😉

  17. Jon Goff says:

    I meant conducting prox-ops, transferring all the fuel from one vehicle to the other, and then separating before relighting the engines on the stage continuing onward. The devil’s still in the details though. For that to work you have to make sure that the exact right amount of propellant is left in each stage at the time of the separation/refueling dance.


  18. Iain McClatchie says:


    The principal benefit of this idea over the usual idea of serial staging is that you get to use the upper stage engines at takeoff. That’s a win because if you added those same engines to the lower stages you would increase the mass of the vehicle.

    By how much would vehicle mass increase? Assume we drop half the vehicles at each staging event, and assume T/W for LOX/kero engines is around 100. Assume also (this is a bit lame) that T/W for a fuelled stage unit is 1.0, so that the engine is nomally 1% of the fuelled mass of the FLOC unit.

    Now build another rocket, which uses serial staging, where the first stage is a cluster of 8 units, second is 4, third is 2, and final stage is a single unit. If each of those units has twice the engines of the FLOC unit, the rocket is 101% of the mass of the FLOC but has slightly more thrust. If we delete one engine from each stage (we wouldn’t, but it makes the comparison easier), the rocket varies in weight penalty versus FLOC and has the same thrust:

    engines engines start
    stage firing carried mass
    1 8 15 100.88%
    2 4 7 100.75%
    3 2 3 100.50%
    4 1 1 100%

    The FLOC will have some weight penalty for plumbing, which will negate some of the above penalty. This caps the benefit of the FLOC concept, and I think the benefit is small.

    I understand that you also don’t place hope in the FLOC concept, Jon, but you left out this limit and I think it’s essential to understanding why the idea is not so wonderful.

    There is another problem too: A FLOC’s engine has to work from the ground, versus a serially staged vehicle which uses low-back-pressure-optimized engines.

  19. gravity loss says:

    I still haven’t read the paper, but I think it must end up with the problem that the one stage having the payload must be able to fly out of the atmosphere before the first “dance move” and that sets a minimum isp / propellant fraction / size for it. And it probably isn’t very small.

  20. Carl says:

    Iain’s point about not using engines optimized for exoatmospheric work is a big one I think. Some technology improvements with altitude-compensating engines can probably be done; but I don’t put much hope in it.

    When talking about proximity-operations testing, what occurred to me as a way to experiment with this quickly and cheaply, would be to simplfy it to 2D by having all the robots driving up the inside of a freely-rotating drum. (like a very wide hamster wheel). That would at least let you do a first-order approximation of your necessary routines.

    — Carl.

  21. redneck says:

    For cheap 3D testing maybe drop pairs of dense rc vehicles from an airplane. They manuever with rcs thrusters and try to link before reaching terminal velocity. Depending on the plane and the unit masses, you should be able to get a few dozen tests per flight.

    Under those controled conditions, it should be possible to dock in 10 seconds or less. A couple thousand feet of altitude should do it. Chute recovery of the models.

  22. Anonymous says:

    I like redneck (aka John Hare)’s idea- do drop tests from a tower, with cushions below, reasonably tough test vehicles can take the impact. One time when doing some two-way relative work from a Cessna we got such a good exit that I pinned the other guy about 2 seconds out the door- man did his eyes go big & round…

    So it’s just subterminal automatic RW with cold gas thrusters. Piece of cake 🙂

    Recharge the air bottles & batteries, hoist ’em up, and do it again fifteen minutes later.

    Doug Jones, wistful former skydiver

  23. Sam says:

    Doug: Or set up a vertical wind tunnel and have them repeatedly mate and unmate and transfer fluids back and forth.

    That sounds like a good honeymoon activity for people, not just a way to test machine coupling.

    One can probably get some of the efficiency back for the upper atmosphere ISP by having half the fleet optimized to be going onward.

    I think that there is some built in redundancy here for all but the payload module; about half of them will need male and female components and of those, they are pretty interchangeable so you only net one of the wet lock directions to take. Or one can have the plumbing be an interstage to save some weight.

    I think that if you miss the wet lock but have a dry lock, one can just burn all the fuel in one of the pair. That is, instead of propellant transfer, some of the matchups can just be momentum transfer which should provide a way to get much of the use of the propellant if not optimal.

  24. Sam says:

    With a hydrogen or natural gas system could one just transfer the LOX?

  25. redneck says:

    This one is just too much fun to leave alone. At work today I thought of another possibility for early use.

    Start with Dave Salts’ 4 ton payload vehicles that have to be refueled leo before delivering a geo bird. After Dave has proved that he can quickly dock and move fuel in orbit, Just start moving the refueling process back toward the release from the lower stages. After a dozen or so launches, the mating could be early enough to seriously add to the fuel remaining in orbit. Two ships delivering a satelite and 6 tons of fuel instead of a satelite and 4 tons of fuel.

    It also allows the tanker stage to be completely drained with no requirement for unusable reserve. He could eventually deliver 10 ton satelites instead of 4 ton ones.

    This could provide early use with minor changes in the upper stages. If this adds something useful, then I dub it Partnered Upper Stages. Probably just use the three letter acronym.

  26. Anonymous says:

    I find this concept fascinating.

    Running the numbers with likely propellant combinations gives very promising results.

    I suppose a baseline non-cryogenic design would use 98% peroxide and 1,2-butadiene as propellants. The chamber would run at about 120bar and the ground nozzle would only expand it about 16:1, for an expected sea level ISP around 275-285. During the first exo-atmospheric rendezvous, which would occur about 80km directly above the launch pads, you would also deploy a nozzle extender, increasing the expansion ratio to about 100:1, giving you an expected vacuum ISP of 330-340. I foresee difficulties trying to loft a payload bigger than about half the dry mass of the plane module, but that should be enough to make the concept profitable.

    It does seem that the rendezvous operation itself would have to be exceptionally low probability of failure, but I see no reason to expect that the required reliability would be impossible.

    Also, the question of where the upper high-velocity sub-orbital stages would be able to land would become a significant issue. I don’t believe that they could all fly directly back to the launch site, unless they were refueled in mid-air.

    What I like about it most is that it scales directly from sub-orbital test flights to orbital missions without changing the design at all.

  27. Axel says:

    After seeing Pixel hover I wondered: why not have many of them, loosely connected, hovering side by side, doing fuel crossfeed. Similar idea, no rendevous required, but very good formation flying capabilities needed. You can find some details on

    In the end I’ve been told fuel crossfeed is a lot more dangerous than a layman like me would expect. Crazy idea, but it was fun.

  28. Pingback: Selenian Boondocks » Blog Archive » Random Thoughts/Orbital Access Methodologies VII: Air-Launched Glideforward TSTO with Exo-atmospheric Suborbital Refueling

Leave a Reply

Your email address will not be published.