Sorry for the possibly dumb question, but I don't know much about rocket technology.

As I gather, the main problem in the recently failed Elon Musk's Starship launch was that the Starship "capsule" couldn't separate from the booster and that forced the control center to activate the self-destruct sequence.

Now, what perplexes me is that I would assume (in my technical ignorance of the subject) that a basic function such as firing the pyrotechnics to separate two stages should be one of the most reliable part of the design, especially at this stage of Starship development and after decades of space missions in which multistage/multi-part space vehicles were employed.

In other words, with so many things that could go wrong, is it really so likely that the failure was that of separating two stages? When I heard that, my engineering gut feeling was "it smells of sloppy design or sloppy implementation". I would have assumed that something bad could indeed happen, but just immediately after the separation, not that the separation wasn't going to happen.

Is it really that the reason for the mission failure, or was a journalistic simplification? Did the Starship team disclosed the actual causes of the failure (to their current knowledge)? Did I miss something? Is my "gut feeling" completely wrong due to my ignorance in the field and "separation failure" is indeed a reasonably likely failure in a new rocket design?

Moreover, can independent experts say something on the actual reliability and sensibleness of the design from this failure and the data Musk's company has released until now about the project?


Some of the comments seemed to imply that my assumption of the failure being caused by a failed separation was not warranted. However I have heard and read Italian news reports that mentioned that as apparently one of the causes.

Here is an excerpt from "Il Sole 24ore", maybe the most authoritative economics newspaper of Italy (emphasis mine):

Nel primo test in volo, la nave Starship della SpaceX non si è separata dal lanciatore e non tutti i motori del razzo Super Heavy si sono si sono accesi correttamente, ma secondo l’azienda di Elon Musk il test è stato comunque un successo. «Il successo deriva da ciò che apprendiamo e il test di oggi - scrive la SpaceX in un tweet - ci aiuterà a migliorare l’affidabilità di Starship». «Congratulazioni alla squadra di SpaceX per l’eccitante lancio di prova di Starship! Abbiamo imparato molto per il prossimo lancio di prova, che avverrà tra qualche mese», è stato il commento su Twitter di Elon Musk, amministratore delegato di SpaceX, dopo il lancio della Starship.

La Starship è stata fatta esplodere in volo per motivi di sicurezza, perché la mancata separazione dal primo stadio del lanciatore la rendeva altamente instabile. Subito dopo il lancio, la navetta ha cominciato a ruotare in modo disordinato e i tecnici della SpaceX hanno deciso di distruggere la navetta perché il suo rientro a Terra sarebbe avvenuto in modo incontrollato.

This is my translation:

In its first flight test, the SpaceX ship Starship didn't separate from the launcher and not all engines of the SuperHeavy rocket ignited correctly, but according to Elon Musk's company [...]

Starship has been made explode during flight for safety reasons, because the failed separation from the first stage of the launcher made it highly unstable. Right after the launch, the ship has begun spinning in a chaotic way and SpaceX technicians decided to destroy the ship because its reenter on Earth would otherwise have happened in an uncontrolled manner.

That's why I also mentioned "journalistic simplification" as a possible source of my doubts.

  • 19
    $\begingroup$ I'm not sure where I should begin: 1) the investigation has just started, nobody, including SpaceX, knows what happened. 2) SpaceX uses autonomous flight termination systems, no-one in the control center was "forced" to "activate" it – the rocket does it on its own. 3) SpaceX never uses pyrotechnics, since they are not reusable. 4) the separation system is of a completely novel design that has never been attempted before. 5) Why would you think stage separation was the problem? I mean, 6 engines failed, the hydraulic power units exploded, without the hydraulic power, the vehicle spun out … $\endgroup$ Commented Apr 22, 2023 at 12:13
  • 5
    $\begingroup$ … control, isn't any one of those things much more likely to be a problem than a failed stage separation that, as far as anyone can tell, never actually happened? $\endgroup$ Commented Apr 22, 2023 at 12:14
  • 6
    $\begingroup$ my engineering gut feeling was "it smells of sloppy design or sloppy implementation". If this was SpaceX's first ever rocket launch, then you'd be absolutely right. But... SpaceX has successfully launched rockets 230 TIMES. Thus, definitely not sloppy design or sloppy implementation. $\endgroup$
    – RonJohn
    Commented Apr 22, 2023 at 19:37
  • 1
    $\begingroup$ @ErinAnne a vote to close and prevent further answers to "Is the failure... a "dumb failure" and does it tell something about the project reliability?" will likely reduce the visibility of Slarty's +31 answer and the others by re-directing readers to the proposed duplicate. Do you think the answer there answers this question better than all these answers combined? (cf. Have there been analyses to see if views of well-received answers are reduced by closure as duplicate which try to control for other factors?) $\endgroup$
    – uhoh
    Commented Apr 30, 2023 at 9:23
  • 1
    $\begingroup$ In this particular case I don't think that closing improves things, so voting to leave open. $\endgroup$
    – uhoh
    Commented Apr 30, 2023 at 9:23

6 Answers 6


It is far too early to say for sure exactly what happened during the Starship launch. And, to my knowledge, SpaceX have not provided any explanation so far. That said it is highly likely that SpaceX will have sufficient telemetry and other data to figure out exactly what went on fairly soon.

So I don't know, but here is my guess at what happened: The rocket spent far more time on the launch pad between ignition and lift off than would be expected - perhaps 8 seconds. It is unclear why this happened, but I imagine that the original plan had not intended 33 Raptor engines to be throttled up to 90% thrust for 8 seconds directed at the concrete base before lift off (maybe 2-3 seconds) and the concrete simply could not take the shock. The result was a storm of concrete sent in all directions as can be seen here:


The Raptor engines were a lot closer to the concrete than the car in the above video. And with so much concrete flying around it is likely that some was basted into the base of Superheavy where all manner of damage could have been caused to pipe work, cabling and the engine bells.

In my opinion this was probably the ultimate cause of what happened next. Fortunately the rocket did eventually get off the pad but as soon as the SpaceX feed started showing the raptor status there were already 3 engines out. Over the next minute or more various other engines cut out leaving the rocket under powered.

A lot of bright flashes and orange tinted plumes can be seen which suggests that other bad stuff was going on during the ascent (as might be expected if the base had been sandblasted with concrete).

It is unclear exactly why Starship didn't separate from Superheavy but by that point there were plenty of suspects. The hydraulics may have been damaged preventing the clamps from opening, the Superheavy was still providing thrust so may also have prevented separation. The vehicle was much lower and slower than planned and Starship might not have been able to reach orbit from that position so the software may not have activated the clamps or it might simply have been more desirable to have one giant rogue rocket spiraling out of control than two.

Some points to note the "capsule" known as Starship is actually a 50m tall second stage rocket with a large payload bay.

No pyrotechnics are used to separate Superheavy from Starship. I believe the clamps are all hydraulic.

"can independent experts say something on the actual reliability and sensibleness of the design from this failure" they surely can and will with great fanfare. Whether they are right or not is another matter (myself included).

The flight was terminated by an AFTS (Autonomous Flight Termination System) so no technicians or flight safety officers were involved during the flight itself. The system aboard detected that the vehicle had strayed too far from its correct flight path and so triggered the explosives to destroy it.

"Sorry for the possibly dumb question, but I don't know much about rocket technology." no problem. There are no dumb questions, only dumb answers and everyone is still learning.

  • 2
    $\begingroup$ Thanks for the short and spot-on failure analysis! BTW, this demonstrates that a little bit of research could have been made those crappy new articles worth reading without taking up much column space. This clears up many of my doubts and misconceptions about the mission. Even if your analysis ended-up being proven partially wrong, it all makes sense. So thanks again. $\endgroup$ Commented Apr 22, 2023 at 19:51
  • 8
    $\begingroup$ thx np: latest tweet from Elon Musk today: "Still early in analysis, but the force of the engines when they throttled up may have shattered the concrete, rather than simply eroding it. The engines were only at half thrust for the static fire test." $\endgroup$
    – Slarty
    Commented Apr 22, 2023 at 22:04
  • 3
    $\begingroup$ @nasch If you fire all the engines at full power then the rocket will launch. It's sort of a fundamental part of the design. $\endgroup$ Commented Apr 24, 2023 at 15:30
  • 3
    $\begingroup$ @wizzwizz4 My--probably incomplete--understanding is that they're trying to build with materials that are simple so that they will be able to build the same thing on Mars with limited facilities. I think it's simple management over-optimism/wishful thinking/toxic positivity/head-in-sand nonsense... $\endgroup$ Commented Apr 24, 2023 at 18:30
  • 4
    $\begingroup$ Per Musk, they expected more severe erosion like they saw in the half-thrust test. They thought it'd be good enough for one launch, then they'd proceed with the planned upgrades. Good for one launch != good for a full-throttle static fire + one launch. And it was probably for the best that they didn't do such a static fire...the pad would not be improved by having a leaking, burning booster sitting on top of it. $\endgroup$ Commented Apr 24, 2023 at 19:29

The quoted article from the Italian source looks like there a combination of poor word choice, limited technical background and possibly even a bit of truth. First problem is that at time of writing it is less than 48 hours since the launch attempt so almost certainly not even SpaceX really knows everything that happened.

At this time probably the most authoritative source is the livestream. There it is pretty clear that a lot of debris was flying during engine start, and by the time the engine status display appears several engines are inoperative. Take off is slow and during ascent more engines fail and the engines have visibly different plumes, suggesting problems with the remaining engines (possible burning out running 'engine rich').

Assuming the Methane and Oxygen indicators in the stream are correct the LOX is running out much faster than Methane, possibly indicating leaks or other issues that would complicate the planned separation.

While engines are still running it starts to tumble, and then several explosions, at least one claimed to be range safety system by commentator. Camera cut to the separation mechanism during this and it appears to remain static through to loss of signal, suggesting that any failure was to activate the system, not system partially operating.

Various you tubers have released videos such as this one by Scott Manley showing the shear amount of damage done to the pad during the launch, and also doing some math to compare observed vs expected performance that suggest the flight was in trouble throughout ascent.

So it is possible there was a failure of Starship to separate, but if so it happened on a rocket already well off nominal and would not be expected to be confirmed one way or another for a couple of months until official reports get written. It certainly would be premature at this point (48 hours after launch) to be making amateur judgments that the design of the separation system being dumb.

The quoted article seems to have missed a number of serious issues apparent just watching the live stream even before the explosions happened, suggesting caution with anything else included in the pieace. It does not appear to mention the official SpaceX position on the test that anything that cleared the tower could be counted as a success on this first test.

In terms of backseat engineering there is probably more to be said about the launch pad/stage 0 design.

  • 11
    $\begingroup$ A quick look at the timeline shows that stage sep happens after BECO and a quick look at pretty much any of the dozens of live streams shows BECO never happened, so I am honestly flabbergasted how one could even think that there was a problem with stage sep if it clearly never even got to that step. I looked up the newspaper the OP cited and it is a financial newspaper. While I have no reason to doubt that the people working there are trustworthy experts in their field, there is also no reason to assume that they know anything about spaceflight. $\endgroup$ Commented Apr 22, 2023 at 14:31
  • 3
    $\begingroup$ Probably best to get your info from a technical source rather than than one that specialises in economics. I recommend nasaspaceflight.com $\endgroup$
    – Slarty
    Commented Apr 22, 2023 at 14:56
  • 2
    $\begingroup$ @JörgWMittag it’s my understanding that the sequence of events is supposed to be (1) initiate flip, (2) BECO, (3) stage sep, (4) booster and Starship each ignite engines and go their separate ways. Either the flip we saw on stream was uncontrolled, or there was an issue in steps 2-3 that I mentioned before. And I’d be comfortable calling a problem happening in the sequence of events involved in separating the stages as “a problem with stage separation”. Technically it didn’t undergo the previous step, but there’s supposed to be literally 3 seconds between BECO and stage separation. $\endgroup$ Commented Apr 22, 2023 at 17:52
  • 3
    $\begingroup$ @LorenzoDonatisupportUkraine Unfortunately scientific and technical reporting by main stream media (yes, I'm talking to you, Spiegel online) is not to be trusted in Germany either. $\endgroup$ Commented Apr 24, 2023 at 9:06
  • 4
    $\begingroup$ @JörgWMittag Missing/failing BECO is surely the reason here: Starship can not separate in a controlled manner as long as the booster engines are running.-- My guess is that they were kept running as long as there was fuel and hope to re-stabilize the vessel. Why it went out of control is anybody's guess; but asymmetrically missing engines and unknown (to us) damage to others and the gimbling mechanism and control is a safe guess. $\endgroup$ Commented Apr 24, 2023 at 9:13

Now, what perplexes me is that I would assume (in my technical ignorance of the subject) that a basic function such as firing the pyrotechnics to separate two stages should be one of the most reliable part of the design, especially at this stage of Starship development and after decades of space missions in which multistage/multi-part space vehicles were employed.

Most of the other answers are focused on the technicals, but let's step back for a second and consider the bigger picture here. This is SpaceX. They are not afraid to fail in trying new things. They literally made a video about it

We often joke about it, but the truth is rockets are hard things to get right. NASA blew up a ton of early rockets in the 50s and 60s before getting things right enough that they could put people on top of them. And they got so successful at doing things, they forgot that things can still go very, very wrong. SpaceX also is far from alone here

An uncrewed test flight of a Boeing spacecraft designed to carry NASA astronauts may have narrowly avoided catastrophic failure in December. A software error that could have resulted in loss of the spacecraft was discovered and fixed while the capsule, known as Starliner, was in orbit, and not long before it returned to Earth.


The additional software problem, first publicly reported Thursday during a meeting of NASA’s Aerospace Safety Advisory Panel, is the second major flaw known to have marred Starliner’s trip in December, the first orbital flight of the spacecraft. During its trip to orbit, the spacecraft set its clock to the wrong time, causing it to deplete its propellant. A planned docking at the space station was called off, and the mission was cut short, to two days instead of eight.

This is to say nothing about all the other non-NASA rocket programs that have failed...

Risk aversion

To some extent, NASA is a lot more risk averse than a private company would be. That risk aversion carries its own costs. The Space Launch System went up successfully on its first try... but it took more than a decade and a lot of budget to get there

A product not of NASA’s leadership but of congressional legislation, the SLS will be the space agency’s first human-rated, heavy-lift launch vehicle since the Saturn V. At its first launch, it will be the most powerful rocket in the world. However, SLS development has been undermined by serious cost overruns and schedule delays. Originally claimed to cost \$6 billion and launch in 2016, the SLS will instead launch in 2022 after expenditures of more than \$23 billion. The SLS is neither reusable nor cheap, costing between \$876 million and $2 billion per launch, depending on how one accounts for its related overhead costs.

The SLS is, basically, a "zero failure" system. It has to be, with $2B per launch and human spaceflight beyond orbit on the line. You could make sure you triple-check everything over and over, but the catch there is you make it a lot more expensive to do business.

TL;DR - What does this say about reliability?

At some point, the only way to see if these things actually work is to fire them up and watch what happens. There is no teacher like failure. Yes, rockets explode (thanks Woody), but as Falcon 9 has proved, they are necessary on the road to success and reliability.

  • 2
    $\begingroup$ Thanks for this fundamental aspect of the issue: risk aversion. I have red several technical articles on space exploration, but their source were almost always either of NASA or ESA derivation (they were either taken from NASA or ESA or they were written by knowledgeable people who analyzed NASA or ESA documents). So I didn't actually realize that, being SpaceX a private enterprise, this project is led with very different management criteria. $\endgroup$ Commented Apr 25, 2023 at 9:34

" ... That a basic function such as firing the pyrotechnics to separate two stages should be one of the most reliable part of the design ..."

SpaceX do not like to use pyrotechnics if it can be avoided, they prefer reusable systems that can be tested before they are flown and refurbished after landing.

To separate from the Super Heavy rocket, Starship does not use an active mechanism like other rockets. It simply unscrews itself (like twisting off a bottle cap) to open the connection clamps and then waits for centrifugal force to force separation. That seems to be the mechanism that failed during this flight test.


This is not a full anwser, just to clarify a misunderstanding in OPs question. Otherwise I agree with the preliminary assessment that debris kicked up during launch was the root cause of the failure.

  • $\begingroup$ Another example of this design approach are the Falcon-9 landing legs, which are deployed using pressurized helium instead of e.g. frangible nuts. $\endgroup$ Commented Apr 24, 2023 at 11:57

A detail in the lifestream attracted my attention. There was a little graphics, that showed the inclination angle of the rocket. At arround T + 2:30 the graphical rocket suddenly turned around almost 180 degree. After that, the rocket started to turn around.

It is just an assumption, but if the information for the graphical rocket came from the inclination sensor of the real ship, it could mean that the sensor delivered wrong data. In such a case the regulation systems would try to correct the "wrong" inclination, which would lead to turning rocket and loss of control in the end.

  • $\begingroup$ Your answer could be improved with additional supporting information. Please edit to add further details, such as citations or documentation, so that others can confirm that your answer is correct. You can find more information on how to write good answers in the help center. $\endgroup$
    – Community Bot
    Commented Apr 25, 2023 at 15:14

The challenges are comparable to the ones of the old Soviet Moon rocket, N1. N1 was also very large, powerful (more than Saturn V), had a big number of engines in the first stage and was developed under time and budget pressure.

Interestingly, Starship started where N1 ended. Last, the most successful flight of N1 lasted till the time of first stage to separate that did not happen, the rocket failing at this point instead. The first flight of Starship/SuperHeavy was very comparably successful.

Hence Starship is at least obviously more successful than N1. I think the basic competence of the N1 team was never questioned.


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