SpaceX recently test fired vacuum optimized raptor engines of starship. Doesn't vacuum optimized engines disintegrate when operated in atmosphere? If yes, what additional modifications are made to the engines to test them in sea level?

  • 2
    $\begingroup$ Where did you get that idea? $\endgroup$
    – GdD
    Oct 22, 2021 at 8:38
  • $\begingroup$ I read that somewhere. $\endgroup$
    – Ashvin
    Oct 22, 2021 at 8:45
  • 1
    $\begingroup$ @Ashvin You might have read it on this site in the answer to a question you asked linast January, Can the space shuttle use OMS engines during landing? $\endgroup$ Oct 22, 2021 at 10:15
  • 4
    $\begingroup$ @GdD There are many questions and answers on this site about grossly overexpanded nozzles. In an overexpanded nozzle the exhaust pressure is less than ambient pressure (atmospheric pressure at the rocket's altitude). In a grossly overexpanded nozzle, the exhaust pressure is so much less than ambient that the exhaust flow separates from the nozzle before exiting the nozzle. Engines have blown up because of this. $\endgroup$ Oct 22, 2021 at 12:35
  • $\begingroup$ @DavidHammen "flow separation will occur, which can potentially damage or destroy the engine" and "vacuum optimized engines disintegrate when operated in atmosphere" are rather different statements. $\endgroup$ Oct 22, 2021 at 12:49

4 Answers 4


Flow separation can occur in a rocket nozzle that is overexpanded.
This can cause quite severe turbulence and thus buffeting of the rocket nozzle.
The SSME used a special rocket nozzle shape to partially compensate for this.

Apparently, the Raptor just bulls its way past the problem by virtue of very high chamber pressure (Meaning the nozzle is not so very overexpanded):

screenshot of Twitter of Elon Musk regarding Raptor vac testing at sealevel

  • 8
    $\begingroup$ The problem isn't with an overexpanded nozzle per se. It's with a grossly overexpanded nozzle. Almost all engines are overexpanded at launch. In many launches one can see the first stage exhaust suddenly bloom when the rocket reaches a high enough altitude. That's the altitude at which the exhaust transitions from overexpanded to underexpanded due to the reduced pressure at that altitude. $\endgroup$ Oct 22, 2021 at 10:29
  • $\begingroup$ Hm... Elon Musk generally knows his stuff, even when it sounds inconsistent, but... how does this explain anything? Of course higher chamber pressure means you're going to have more expansey nozzles. It means even the sea-level nozzles will have more expansion than those of other booster engines. But surely it also means that an efficient vacuum nozzle is going to have even more expansion, at which point flow separation should again be just as much of an issue? $\endgroup$ Oct 23, 2021 at 0:23
  • 1
    $\begingroup$ @leftaroundabout take two engines with the same expansion ratio and different chamber pressures. The one with the higher chamber pressure will have the higher nozzle exit pressure. It would appear that the vacuum raptor isn't that overexpanded at all - presumably making it so would make the nozzle so vast it would either be uneconomic or wouldn't fit in the ship. $\endgroup$ Oct 23, 2021 at 1:10
  • 1
    $\begingroup$ I wonder whether the "higher chamber pressure" was a special condition for a sea-level test. It makes sense to test beyond the normal operating envelope to verify margins, and that would also help mitigate the trouble with over-expansion. $\endgroup$
    – John Doty
    Oct 24, 2021 at 19:50
  • 1
    $\begingroup$ @JohnDoty partially plugging the nozzle would do more to invalidate the engine test than sawing off the whole nozzle, that's like testing a racing car's stability under speed by having the car tow a train, so it doesn't go as fast. $\endgroup$ Oct 25, 2021 at 0:41

There are several test stands with altitude capability of up to 100 K ft (30.5 km or 8 Torr) for engine firings using the steam ejector system and up to 250 K ft (76 km) non-firing capability with vacuum pumps. 8 Torr is 1 kPa or about one hundredth of the sea level air pressure.

The NASA White Sands Test Facility Propulsion Test Stands and the Glenn Research Center, Plum Brook Station.

At the White Sands Test Facility there are 6 altitude test stands and 3 ambient pressure units.

See https://ntrs.nasa.gov/api/citations/20180005322/downloads/20180005322.pdf and https://www.nasa.gov/sites/default/files/files/WSTFTestStands.pdf

There is also an ESA testfacility in UK: https://www.esa.int/ESA_Multimedia/Images/2021/06/Maintaining_vacuum

  • 1
    $\begingroup$ another non-SI unit I didn't know about. 1/760 standard atmosphere, because why not $\endgroup$
    – njzk2
    Oct 23, 2021 at 22:10
  • $\begingroup$ @njzk2 Torr or mm Hg is not a SI-unit but it is still used for blood pressure. Torr was used for vacuum technology during many decades, even some centuries. So if your blood pressure was ever measured you missed the opportunity to learn about Torr. $\endgroup$
    – Uwe
    Oct 25, 2021 at 20:39

The Arnold Air Force (AEDC) base at Tullahoma Tn has major altitude rocket firing test cells. Witnessed Apollo Service Module firing tests there in the mid 60s. That engine was around 25k lbf. It had an expansion ratio of 62.5. I believe the capabilities of that facility have expanded greatly to larger engines since then.

Seem to recall an altitude firing cell being constructed at NASA Stennis. Do not know it's current status.

Engines designed for use at altitude must be test fired at simulated altitude or the nozzle will collapse inward during a sea level firing for a flight weight nozzle.


The main (only!) difference between them is the length of the nozzle (the vacuum one is a bit longer).

Because of flow separation, as pcman correctly pointed out, some engines would see buffeting and might be damaged/destroyed by running at sea level pressure.

There are several ways to work around this:

  1. Test in vaacum. This is done for the RL 10 engine
  2. Tweak the nozzle to work in both condition. This was done for the space shuttle SSME which had to work in both conditions.
  3. For the raptor, it seems that an increased chamber pressure did the trick

A nice article that goes deeper. With videos of flow separation: Ars technica

  • 8
    $\begingroup$ False. Flow separation in an over-expanded rocked nozzle is a significant added stress to the engine , and can indeed cause sufficient buffeting to destroy the nozzle (and thus the engine) $\endgroup$ Oct 22, 2021 at 9:14
  • 2
    $\begingroup$ @user44085 Engines designed to operate in vacuum typically have a very long nozzle. The very long nozzle allows the gas to expand more. Using an engine designed for vacuum operations at sea level can result in the engine tearing itself apart, or even worse, such as the bell collapsing in itself. This answer is incorrect. $\endgroup$ Oct 22, 2021 at 10:20
  • 1
    $\begingroup$ yes @DavidHammen, but with a "shorter" nozzle which wont over expand the exhaust, that isn't the case. by not having the nozzle extension, over expansion isn't an issue when testing a vaccuum engine at sea level. $\endgroup$
    – user44085
    Oct 22, 2021 at 10:26
  • 3
    $\begingroup$ @DavidHammen - here is SpaceX testing a vacuum merlin at sea level without nozzle extension: youtube.com/watch?v=wkdReoxGHG8 $\endgroup$
    – user44085
    Oct 22, 2021 at 10:31
  • 2
    $\begingroup$ @BrendanLuke15 I'm confused. That the nozzle extension has no plumbing was my point. If they only tested the engine head they did not test the engine. With no plumbing there are potential serious heating issues (as in the nozzle extension can melt heating issues) that remain untested. $\endgroup$ Oct 22, 2021 at 12:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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