DM-2's first launch attempt was scrubbed at about T−17 minutes due to bad weather at the instantaneous launch window, even though the weather was predicted to clear up as early as T+10. It was explained on the live stream that even though the trajectory can be adjusted to make up for a few minutes, Falcon 9's launch window is nevertheless a single second because its cryogenic fuel cannot be held once fully loaded.

My wife asked why, when it was realised that a delay by a mere ten minutes would have allowed the launch to proceed, the fueling wasn't slowed down in order to be completed once the weather would have cleared.

Does the technology simply not allow for this, have such procedures simply not been planned, was the necessary delay beyond what could be corrected for later, or is there an entirely different reason this wasn't done?

  • 4
    $\begingroup$ Launch window does not depend on fueling, it depends on the orbital position of the ISS. $\endgroup$
    – Uwe
    Commented May 31, 2020 at 18:24
  • 2
    $\begingroup$ @Adám: 10 minutes is not "a few minutes". The window due to orbital mechanics is very short, only 5 minutes long for the Shuttle (with manual checklists carefully arranged so humans could actually hit that window reliably), and the highly-automated launch of the Falcon 9 targets a single point within it because there's no reason to complicate things beyond that. $\endgroup$ Commented May 31, 2020 at 19:22
  • 2
    $\begingroup$ This is an interesting question but it contains the premise that "Falcon 9's launch window is nevertheless a single second because its cryogenic fuel cannot be held once fully (loaded)." I don't think the 1 second launch window is such because "cryogenic fuel cannot be held", See for example answers to Why would sub-cooled LOX tanks need to “topped-off” until the last minute or so? As launch vehicles, F9's do not require instantaneous launch windows. However it might have been chosen in this case for specific reasons. $\endgroup$
    – uhoh
    Commented May 31, 2020 at 20:41
  • 6
    $\begingroup$ @Uwe, they also mentioned on the livestream that Falcon 9 has sufficient extra engine performance to handle a delay of up to 5 minutes either way, and that the instantaneous launch window is due entirely to the use of densified propellant. $\endgroup$
    – Mark
    Commented May 31, 2020 at 21:11
  • 1
    $\begingroup$ @Mark citation needed; can you mention a specific video link and timestamp? Thanks! $\endgroup$
    – uhoh
    Commented Jun 1, 2020 at 0:34

3 Answers 3


For a flight to the ISS, the Falcon 9 should launch into the orbital plane of the ISS. To change the plane later requires extra fuel. There are some reserves for safety but not for launch delays.

The first stage should be reused and land on the barge with the remaining fuel. If a plane change maneuver is done, the first stage would leave the optimal trajectory for a soft landing on the barge. A launch delay could cause the loss of the first stage by missing the landing barge.


SpaceX uses a different methodology for fueling, colloquially called "load and go". The idea is you load supercooled fuel (for maximum density) onto the rocket 35 minutes before launch. It's a methodology that has its risks

SpaceX uses load-and-go for its satellite and cargo Dragon missions currently, starting the fueling process just 35 minutes before liftoff. The company has adopted that approach because it uses "supercooled" propellants that are denser, improving the vehicle's performance.

That approach, though, attracted scrutiny after the September 2016 explosion of a Falcon 9 on the pad at Cape Canaveral during preparations for a static-fire test prior to the planned launch of the Amos-6 spacecraft. That accident, which destroyed the launch vehicle and satellite, was blamed on the failure of a composite overwrapped pressure vessel in an upper stage propellant tank.

So, why can't they wait? It's because the fuel gets too warm

That's because liquid oxygen is pumped into the Falcon 9 at a very low temperature: 340 degrees Fahrenheit below zero. That keeps it liquid and densifies the fuel, a type of kerosene called RP-1, which allows SpaceX to cram more of it into the rocket and squeeze more performance out if the machine.

Once inside the rocket, however, the fuel begins to warm up, expand, and boil off. That fuel loss starts the launch clock ticking.

"That changes how much performance you get carrying into orbit, and we don't want to cut into those margins," Insprucker said.

For a satellite, you can just pump the warm(er) fuel out and add new chilled fuel. But the ISS is a moving target

But [detanking fuel] is not an option for Demo-2, since the ISS flies over Earth's surface in a winding path, and at great speed.

"In the case of the International Space Station, an hour and a half from now, it's nowhere where we need to be to get to orbit," Insprucker said.

  • $\begingroup$ This still doesn't answer why the fuel couldn't be pumped in slightly slower once a delay or scrub is inevitable. Indeed, with such a fast fueling, they should be able to delay beginning the fueling when a delay or scrub is inevitable. $\endgroup$
    – Adám
    Commented Jun 1, 2020 at 13:36
  • 2
    $\begingroup$ @Adám Actually it does. Even if you slow the pumping in, the fuel you've put in the tanks will still be getting warm. Falcon 9 is built around the fuel being in a very specific temperature range $\endgroup$
    – Machavity
    Commented Jun 1, 2020 at 13:38
  • $\begingroup$ True, but there must be some margin, especially since payload to LEO is 22,800 kg but the full Crew Dragon is only 15,525 kg. (Yes, the inclination was very different that the nominal, but this Crew Dragon also didn't appeared very full.) $\endgroup$
    – Adám
    Commented Jun 1, 2020 at 13:46
  • $\begingroup$ Yes, there is a margin, and they want to keep as much of that margin as possible for manned flights. John Insprucker made it clear it was about keeping the performance margins: "...that changes how much performance you get carrying into orbit. And we don't want to cut into those margins." (space.com/spacex-demo-2-astronaut-launch-delay-explained.html) $\endgroup$ Commented Jun 1, 2020 at 14:15
  • $\begingroup$ To be a bit more precisely, it is not the fuel RP-1 which boils off. It is the cryogenic oxidizer LOX that boils off. The cooled fuel RP-1 will get warmer and expand its volume, the tanks may overflow. $\endgroup$
    – Uwe
    Commented Jun 1, 2020 at 14:59

A non-instantaneous launch window means the vehicle has to be reprogrammable with regard to its trajectory. Making the vehicle reprogrammable in this regard is extremely expensive and is thus something that SpaceX has eschewed, at least for the most part.

  • $\begingroup$ A rough order of magnitude guess: I estimate that eliminating the required reprogrammability needed to make launches non-instantaneous saves SpaceX millions of dollars per flight in software development costs alone. $\endgroup$ Commented Jun 1, 2020 at 4:58
  • $\begingroup$ Flight software is amazingly expensive. When one adds all of the hours spent by people who directly created the flight software for the Shuttle or the ISS, the hours spent by people creating test software for that flight software, and the hours spent by people who can't program their way out of a paper bag but can come up with ingenious ways to make flight software behave badly. The Shuttle and the ISS flight software development processes proceeded at a rate less than one line of code per person per week. SpaceX does better than that, maybe an order of magnitude better than that. $\endgroup$ Commented Jun 1, 2020 at 5:06
  • $\begingroup$ But an order of magnitude improvement still means one line of code per person per day. Human-rated flight software is ridiculously expensive.The cheapest approach is not to do it. Instantaneous launch windows represent one way to not do it. $\endgroup$ Commented Jun 1, 2020 at 5:08
  • 2
    $\begingroup$ -1 for now. Are you sure that SpaceX has not had wider-than-instantaneous launch windows, and that Falcon 9's are not "reprogrammable" because it would be "extremely expensive"? I find both of these unsupported items difficult to believe. I'm having trouble parsing "has eschewed, at least for the most part." Last dozen? 90% of all of them? $\endgroup$
    – uhoh
    Commented Jun 1, 2020 at 6:22
  • 1
    $\begingroup$ Software for real-time trajectory adjustment already exists, and is even human-rated: it was used on the Space Shuttle. $\endgroup$
    – Mark
    Commented Jun 1, 2020 at 20:26

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.