This engineering question got me wondering. If cross feed technology were sufficiently developed, and stresses could be managed (these are big if's, and there are more I'm sure), how would these various configurations roughly compare?

Are there numbers out there, or could someone ballpark them, or tell me a simple way to ballpark the missing values?

The first two are copied from Wikipedia's Falcon Heavy. They are tons to LEO. They also agree with the numbers in this Reddit which is probably not independent, but that table also gives the third number as well.

F9 FT:      22.8
FH:         54.4
FH w/cf:    64.5 
FQH:         ?
FQH w/cf:    ?

In my very unofficial acronyms, FQH = 'Falcon Quad Heavy' - four F9-like boosters, and w/cf = 'with cross feed'.

Cross feed for FQH might have boosters 1 and 3 feeding into 2 and 4, and dropping off first, while boosters 2 and 4 are also simultaneously feeding the core. It's a mess of engineering - I'm not proposing it as a real solution - it's more of a best possible (or best imaginable) upper limit.

edit: I'm starting to rethink this question; The potential benefit of the "FQH" might be better realized by optimizing throttling and even staggering engine starts. In other words, maybe this is more difficult to even "ballpark" than I first realized.

  • 1
    $\begingroup$ Looks at this link: m.reddit.com/r/spacex/comments/4ibtcb/… . It is not a reliable source, maybe could be as a start to find some values. The payload value is only for FH w/cf (60 t). I am curious to know what would be the payload for FQH and FQH w/cf and also a possible launch cost for these versions. $\endgroup$
    – Mark777
    Commented Jul 24, 2016 at 5:58
  • $\begingroup$ @MarkBoghdani looks promising, although I get dizzy looking at 120+ numbers. Is the answer to my question there? If so, can you post it? Thanks! $\endgroup$
    – uhoh
    Commented Jul 24, 2016 at 6:01
  • $\begingroup$ i would really like but i am not sure if it is.The person that has made that table with thouse values,says that has made calculations for those and has being supposing for some numbers(to match right with the payload and delta v of FH without cross-feed)to get to these results.Even him is not sure if the number that he has found and supposed are correct.Knowing the exact numbers(for example for the fuel,stages weight,average Isp)probably could be found the payload not only for FH w/cf,but even a hypothetical payload near the real payload values that would have FQH and FQH w/cf. $\endgroup$
    – Mark777
    Commented Jul 24, 2016 at 6:58
  • 1
    $\begingroup$ @MarkBoghdani OK so the two numbers in my question agree with that spreadsheet, so I suppose I could put the 64.5 in as my third number. I'll go ahead and do that now - thanks! $\endgroup$
    – uhoh
    Commented Jul 24, 2016 at 7:30
  • $\begingroup$ At the same link of reddit that i put in my 1st comment look again and you will find a discusion in the last comments with the author of that table.He gives FQH with no optimization a payload by 70 t, FQH where center core is turned off or different throtling 85 t, and FQH w/cf 93 t.He explains how he get this numbers, method that he uses.Of course he says that for more accurate numbers launch simmulations should be done including even atmospheric drag,gravity loses or Isp at different altitudes.I can't say how accurate are these numbers but could help you to have an idea what payloads will be $\endgroup$
    – Mark777
    Commented Aug 3, 2016 at 10:28

1 Answer 1


Do you know KSP? It simulates unrealistic perfect crossfeed capability which allows for Asparagus staging - a very effective technique in the simulation (look at some videos, I remember one showing a rocked with stage separation ocurring each 3 seconds or such all the way to orbit).

The problem with it is that fueling of 1 F9 booster takes roughly 30 minutes and using that fuel through engines only a bit more than 150s (based on MECO times of GTO launches where the remaining fuel was minimal). For the crossfeed to work optimaly in quad or more side boosters configuration, you need to "refuel" the "inner" boosters with a flow roughly 4 times bigger than during fueling (for quad, n-fold more for each additional "layer" of boosters). You soon find out that the pipes get quite wide and the pumps very heavy.

But to the question - I don't have exact numbers, but my understanding is that without crossfeeed you are looking at a problem with quickly diminishing returns. You either have too much thrust and lose to drag (or lose the vehicle), or you throttle down (or turn off) the "inner" engines and lose to gravity:

  • if you just lower the thrust of all or some engines, you will get almost drained fuel tanks after separation.
  • if you disable some engines and start them later to preserve fuel in "inner" boosters, you turned the thing into normal 3-(or-more)-stages rocket, just with small stages on the side instead of bigger one at the bottom.

With super-crossfeed you can model the entire situation a bit simpler:

  • FH w/cf turns to a fully fuelled F9 after side booster separation but with all the height and velocity already imparted by the boosters. You are effectively air launching F9 (but with the advantage of not needing to change velocity from horizontal to vertical). To get the numbers you need to compute the delta-v of two cores of propelant with thrust of 3 cores (but it will have to throttle down for Max-Q etc.) and payload of full F9. You can then add the delta-v of "normal" F9 to get the ability of entire stack.
  • FQH w/cf - analogically, you are air-launching FH, you just now have fuel of 2 boosters to feed 5 sets of engines (= 45 Merlins) so the staging will happen sooner.

My "launch-math" is not good enough to provide any reasonable estimates for actual numbers unfortunatelly. I just know from my KSP tests that the difference between no crossfeed and full crossfeed is really substantial - where without crossfeed the rocket did not even get out off the atmosphere, the crossfeed variant was able to get straight up to escape velocity.

  • $\begingroup$ Wow, thanks for taking the time to think all of that through! OK I'll give this some more thought, will probably take a few days now that the week has started. $\endgroup$
    – uhoh
    Commented Jul 17, 2016 at 19:23
  • $\begingroup$ OK this will be a long-term project, one of many so I'll close this one out. Thanks! $\endgroup$
    – uhoh
    Commented Jul 19, 2016 at 11:37
  • 3
    $\begingroup$ That KSP rocket can be seen here: youtube.com/watch?v=F92l2s_bO-k&t=584s It starts out with 108 stages. $\endgroup$
    – Hobbes
    Commented Nov 20, 2016 at 18:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.