Today (2020-01-19) SpaceX had a successful flight abort test. As far as I can understand, at a specific altitude, Crew Dragon fired its engines and escaped from the rocket.

This test looks like a very safe abort test, Dragon commanded Falcon to shutdown, and only then fired the Super Draco engines; if the booster engines were still firing, would it collide with Dragon? (i.e. the acceleration of the booster without dragon would be more than the acceleration of the Dragon by its Super Draco engines?)

My question is why did this test not use actual failures to trigger that abort? Like the booster stage explodes, and it is detected by Crew Dragon and triggers the escape system. That will be able to validate actual failure detection too.

  • 3
    $\begingroup$ @user3528438, As far I understand there no other test, the next one will be real flight. Also as far i can see booster shutdown just before super Draco fired. $\endgroup$ Commented Jan 19, 2020 at 22:10
  • 10
    $\begingroup$ There is an acknowledged point to testing where the testing becomes financially or physically unfeasible - once you start going down the path of "what if..." there are thousands of scenarios which would require their own individual tests. What if the booster exploded? What if the booster suddenly veered off course by a significant amount, causing lateral acceleration? What if something in the trunk exploded? What if both craft suddenly lost all electrical power? Etc etc etc. You have to accept that sometimes ideal testing is the only testing you are realistically going to get. $\endgroup$
    – Moo
    Commented Jan 19, 2020 at 23:12
  • 9
    $\begingroup$ @uhoh possibly because a non-immediate-RUD power loss situation might be thought to be the most common scenario that the crew need to escape from? Often the RUD comes later (see the Soyuz inflight abort - issue, followed by escape, followed by issue escalating into RUD). This would cover anything from a pump failing, engine gimbaling failing, unplanned engine shutdown, inability to stage etc etc etc. $\endgroup$
    – Moo
    Commented Jan 20, 2020 at 0:20
  • 13
    $\begingroup$ An engine failure often cascades up to the structure of the rocket. But it can take time to do so. If you watch a lot of rocket failure videos, there is often a few seconds between malfunction and boom. The capsule can't outrun a shock wave, but it can get out while the cascade is happening. This test simulates very common scenarios and some in which they may escape the body exploding. Also the rocket was traveling at more than Mac 2, hardly an easy task at that speed. $\endgroup$ Commented Jan 20, 2020 at 1:16
  • 6
    $\begingroup$ @aml: The hosts in the webcast mentioned that Dragon initiates the whole sequence, including commanding F9 to shut down. It was a bit ambiguous whether Dragon would then go on detecting the loss of thrust and initiate escape or whether that's just the next step that happens after booster shutdown. $\endgroup$
    – Joey
    Commented Jan 20, 2020 at 6:12

4 Answers 4


It was a real failure (albeit triggered externally rather than accidentally), just not the only failure that can happen. and it is the worst case of a series of the most likely failure scenarios: multiple engine failure.

If you want to test every conceivable way a rocket can fail, you're looking at thousands if not hundreds of thousands of possible failure scenarios and will never make it into orbit, so they picked one that they and the customer could agree on was typical.

  • 5
    $\begingroup$ Thank you for your answer, Yes it is done during Max Q, But calm and quit all engine shutdown is the worst-case scenario? It ensures very easy separation(Almost like 2nd stage separation). But it is understandable if idea was to test a typical sensio instead of the most difficult one the escape system is designed for. $\endgroup$ Commented Jan 20, 2020 at 10:08
  • 11
    $\begingroup$ @JithinJose sudden shutdown of all thrust is pretty much worst case. And the flight conditions were far from perfect, with winds close to the limits for the Falcon 9. Pretty much the only worse thing would be a sudden thrust surge after triggering the escape sequence, causing a sudden rapid accelleration driving the booster directly into the path of the escaping capsule. And the thrust of the escape system is designed to guard against such a scenario, computer simulations no doubt were done to verify that. $\endgroup$
    – jwenting
    Commented Jan 20, 2020 at 11:54
  • 1
    $\begingroup$ Blue Origin managed to test it's Max Q abort with the booster engine firing... youtube.com/watch?v=_zWkvm7HpH8 $\endgroup$
    – user20636
    Commented Jan 20, 2020 at 13:22
  • 8
    $\begingroup$ Most people might think of rocket explosions as worst case. That might be true, but they almost always happen within seconds of takeoff while the rocket is still traveling relatively slowly. If you want to test your escape system at Max Q, you pick the worst case scenario likely to happen at Max Q. $\endgroup$
    – Turksarama
    Commented Jan 20, 2020 at 22:36
  • 3
    $\begingroup$ I think what they're asking is: why not just slam the self destruct button in the middle of the flight and see if the capsule can/can't survive the rocket exploding while still attached to it (since you can't anticipate a spontaneous detonation by definition), and make away from the fireball without having been broken up by the force, and without the inside temperature soaring to cooking levels? $\endgroup$ Commented Jan 21, 2020 at 6:24

You seem to think they were testing in ideal conditions. That's as far from truth as you can get. The abort happened at the moment in flight with worst aerodynamical conditions (called maxQ), when booster flies still low enough in atmosphere for significant drag to be present, yet fast enough already. If Dragon can escape at this moment, it can escape at any other moment in flight.

I also don't think that Dragon commanded Falcon to shutdown. My understanding is that engine shutdown command was sent remotely from control center. Dragon only then recognized abnormal conditions and decided to abort on its own - that was important aspect to the test.

I agree they could have booster engines on during abort, but there we meet what others said in comments, there are infinite number of failure scenarios, and they can't test for all of them.

  • 7
    $\begingroup$ Dragon did command shutdown. It was clarified in the press conference. $\endgroup$
    – Tim
    Commented Jan 20, 2020 at 9:47
  • 2
    $\begingroup$ @Tim: There's two different engine shutdowns, one was part of the test and one is part of an abort sequence. Dragon will command shutdown as the first step in an abort. This specific abort was triggered by shutting down the engines when the vehicle reached a certain speed/altitude, so they were already shut down when the abort sequence started. It's possible they used the Dragon to send the commands to do so. $\endgroup$ Commented Jan 20, 2020 at 13:30
  • $\begingroup$ @ChristopherJamesHuff Interesting, So F9 will do a normal shutdown, Which is sensed by Dragon, then it commands F9 to shutdown again(i assume maybe for safe separation) then Super Draco fires. Possible to share any reference document/video? $\endgroup$ Commented Jan 20, 2020 at 14:41
  • $\begingroup$ @ChristopherJamesHuff I’m unsure that’s true - the draco thrusters we’re already firing before F9 flameout. Is it just that it takes a few seconds between shutdown command and flameout? $\endgroup$
    – Tim
    Commented Jan 20, 2020 at 15:26
  • 4
    $\begingroup$ @Tim: The SuperDraco thrusters can certainly respond faster than the Merlins (being designed as an abort mechanism), and the abort sequence likely started when the thrust dropped below some threshold (likely dependent on how far along in flight it was) or started a rapid enough decline to indicate an unrecoverable problem, not when it reached zero. $\endgroup$ Commented Jan 21, 2020 at 0:30

In the post-launch media event with Jim Bridenstine and Elon Musk, a similar question was asked and Elon said that the capsule would be able to "fly through the fireball" and explained how that even in a critical failure, the Falcon 9 doesn't really explode, rather it causes a huge fireball (no big pressure wave). Additionally, it was mentioned that the capsule under super-draco thrust is capable of reaching 6 gees of acceleration which is more than enough to get the capsule away from still-firing first stage engines.


I think you are working from a faulty assumption here.

...if the booster engines were still firing, would it collide with Dragon?

If the booster engines are still firing, there's no reason to trigger the escape -- other than severe control malfunction (a la Ariane 5), which wasn't what was tested here. Any kind of failure of the booster that is not a control malfunction, right up to complete conflagration (e.g. due to ruptured tanks), will result in a loss of thrust.

My question is why did this test not use actual failures to trigger that abort? Like the booster stage explodes, and it is detected by Crew Dragon and triggers the escape system.

The (realistic) failure mode here was "unexpected loss of thrust". The other failure mode would have been "catastrophic control malfunction". They picked one.

  • 2
    $\begingroup$ Assume there was an unbalanced thrust or engines that went into some gimble lock. Like Ariane 5, I know a similar thing happened for Soyuz and GSLV. Most of them did not seem to result in controlled balances loss of thrust, either it will lead to an explosion or went to high AoA. Maybe I am wrong but I did not seem rocket failures, that shut down all it's booster engines smoothly at the same time. Please share with me any such info if available., I will stand corrected. $\endgroup$ Commented Jan 21, 2020 at 12:09
  • 1
    $\begingroup$ @JithinJose: Unbalanced thrust equals reduced thrust. Unbalanced thrust and gimbal lock equal high AoA equal severe control malfunction. Explosion of booster equals sudden loss of thrust. Sudden loss of thrust is one, perfectly valid, failure mode. Catastrophic control malfunction is the other. They picked one. When all engines are at full thrust and on course, you don't trigger the escape. If the booster starts going into high AoA, you trigger the escape and the booster doesn't follow the capsule's path. Either way, you don't escape a 100% thrust booster on its path. $\endgroup$
    – DevSolar
    Commented Jan 21, 2020 at 12:15
  • 1
    $\begingroup$ I disagree on never triggering the escape when the engines are burning properly. Think back to the Challenger disaster. The way SpaceX instruments everything I suspect they would have detected the burn-through, if a camera could eyeball it I can see ground control realizing the danger in time and telling them to get out of there. And I thought the escape motors were powerful enough to run away from the booster when it was still burning. $\endgroup$ Commented Jan 21, 2020 at 16:17
  • 1
    $\begingroup$ @LorenPechtel: As I said, excess thrust to pull away from a functional booster can be tested statically. I disagree on the Challenger scenario: You don't hit the big red button the second some sensor goes bonkers, and you certainly don't auto-escape on it. From the drop-off of external tank pressure (the first indication that something was really wrong) to disintegration of the shuttle was 7 seconds. Anyway... if the Challenger had an escape system, it would still have been sufficient to pull away at the sudden throttle loss of T+73... $\endgroup$
    – DevSolar
    Commented Jan 21, 2020 at 16:33
  • 1
    $\begingroup$ @LorenPechtel: You mean the "seeing" that no one at mission control so much as commented on, and which required the review of the footage from various angles post-fact to figure out? Again, you don't hit that abort button on a whim, trowing away the mission. They survived the disintegration of the craft, as would be expected. Escape motors firing at T+73 would have been plenty sufficient. $\endgroup$
    – DevSolar
    Commented Jan 21, 2020 at 19:10

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.