If you catch the audio on this from the nasa space flight forum (I think?), it was pretty funny. They were speculating if it was a lox purge (vs a methane purge). They assumed lox, since if it were methane, the flare would..... BOOOM.
A bit later they run the tape back frame by frame and determined the fire definitely didn't come from the flare stack. (Though in realtime it happens so fast they weren't sure.)
That video was almost certainly taken with a telephoto lens a long distance away, and telephoto lenses tend to compress distance. The flare could’ve been a long way away from the rocket, and a telephoto lens would make them look much much closer.
It was not intentional. There must be some mechanism to disconnect the external propellant lines before lift-off and they obviously disconnected unintentionally after the static fire. The best guess based on the way the flame spread was that both lines disconnected and there was a mixture of methane and oxygen just waiting for the right fuel-oxygen ratio before it was ignited on any number of post-burn hot spots. (I haven't seen any new info recently so this explanation may have been already superseded by more recent analysis.)
From just the video it didn't look like a very fast burn. I wouldn't be surprised if it was just methane, or at least a non-stoichiometric mixture. (Could also just be poorly mixed).
Scott Manley posted on Twitter [0] that it looks like it was a proper detonation (supersonic flame front, etc.) based on some math given the known size of the test rig and rocket.
Disappointing to be sure, but it's why they already have SN5 (nearly?) complete and SN6 is well on its way.
I like how Elon Musk contrasted past and currently planned human space activity as 'Apollo' and the potential future as 'D-Day'.
The goal is to cheaply and quickly mass manufacture interplanetary space craft. I think that imploding/exploding early serial numbers are an expected early part of that path.
Not disappointing - the more things that go wrong during the extensive series of tests developed and cataloged over the past several decades, the more we can correct and cement into permanence.
I'm looking forward to SN747 (the number corresponds with one of Boeing's standardized atmospheric flight models) sometime in 2024. I'm probably not thinking big enough though.
Musk was pretty clear in an interview that he does want things to blow up. He wants the testing to push and exceed the limits so those limits become well defined. It sounds quite reasonable.
> Musk was pretty clear in an interview that he does want things to blow up.
I get that fanboys are excited and they see the Musk world through rose-tainted glasses.
But SpaceX was planning on flying this vehicle. And it exploded.
How does anyone anywhere believe in that? Obviously this outcome was not planned. If anyone wants to fly a vehicle to test other stuff, they don't want it to blow up prior to it.
I'll reiterate a point I made deeper down: obviously the failure wasn't planned or intended. But the overall process of being failure tolerant and having a high rate of failures, was.
You can spend your resources on doing careful modelling and over-engineering and conservative testing to always stay within limits, or you can go fast and loose and discover those limits the kinetic explodey way. SpaceX chose the latter strategy.
Also, not a fanboy. On a personal level the guy's a prick. But the strategy that SpaceX is employing is sound and tested. It's the same R&D strategy that was employed by Skunkworks, which is arguably the most successful aerospace R&D story in the entire history of our species.
> You can spend your resources on doing careful modelling and over-engineering and conservative testing to always stay within limits, or you can go fast and loose and discover those limits the kinetic explodey way. SpaceX chose the latter strategy.
Again, this baseless assertion is just plain wrong on many levels, and flies in the face of basic engineering practices.
At best, you're confusing a consciencious choice of allocating piles of resources to avoid bottlenecks and time constraints, such as losing a prototype which is a project death sentence to your general aerospace research project, with plain old incompetence.
Engineering 101 teaches that when in doubt err on the safe side. Your assertion is the exact opposite because... Because what? What do you believe is the trade-off of wasting time and resources trying to fly intentionally half-baked designs that define the critical path of the project?
Please spend a moment thinking about what you've said to try to see how absurd it is to waste time and resources for nothing at all.
> Engineering 101 teaches that when in doubt err on the safe side.
Nonsense. You can't discover limits if you always stay within them. Commercial engineering, sure. But R&D is by definition the process of exploring an unknown parameter search space. And considering that spaceX have achieved what STS has failed to do with 1/10th of the STS budget, I'd say they're not doing terribly badly with their chosen R&D strategy. You seem to have a deeply ingrained misconception about what R&D actually entails and how it's different from both commercial engineering and designed experiments.
Erring on the safe side is what they're doing with their manned missions. Cargo missions are medium risk and occasionally do dirt-cheap launches to deliberately try risky scenarios. R&D ops is deliberately high risk and low process overhead. They rather someone just try something and see what happens instead of spending 6 months writing an experiment plan and getting it validated etc. as you would see in the pharma industry for example. I'm not sure why you're struggling to comprehend this fairly simple and fairly standard strategy.
In R&D things blow all the time.
You just usually don't see it. In labs when they are looking for new drug or vaccine - thousands of variants fail. And it's ok. When you are developing anything new - you do thousands of try and fail. IF you don't - then you will even ever invent anything new. The best you can do is a little bit upgraded iteration.
And here a few fails and it is disappointing?
Nope. Its the only way to discover something new.
There are world's of difference between a prototype accidentally blowing up because a design fault was unnoticed up to that moment, and this absurd spin on this accident about how these accidents are sought after and desirable. They aren't. They are always problematic and a setback, and in some projects even project-killers.
So, enough with the bullshit about how this accident is good news. It isn't. Even if the project can easily recover from this setback, it's still a setback.
And most fresh engineering graduates know nothing about real engineering.
You don't intentionally half bake the design, but you do intentionally try to exceed whatever your maximum design specification is.
The wing flex test, for example. 150% of the expected maximum flex the structure will experience during real operation.[1]
Even with software, you can't stop at "well I think this is the maximum load this service will handle in production." You need to know what will happen if that expectation is exceeded. Maybe nothing, or maybe your initial estimate was too high to begin with.
> And most fresh engineering graduates know nothing about real engineering.
And that's ok, that's why they are supervised by experienced engineers, who in turn in relevant projects with some complexity answer to senior engineers.
But even then screwups might happen, and sometimes it's ok. Nevertheless, unlike the spin that is being given to this disaster, that doesn't mean accidents are desirable or sought after.
Particularly when the accident consisted of exploding the prototype that was already scheduled for flight tests.
> Engineering 101 teaches that when in doubt err on the safe side.
This sounds like lecturing birds on how to fly (Nassim Taleb quote below.) They'll possibly fly people to the space station today. Maybe they have engineering 101 down? If they're breaking guidelines in a 101 level course, then maybe they're just making amendments to the rules. Maybe the textbook writers need to make a revision.
The greatest problem in knowledge is the “lecturing birds how to fly” effect. -Nassim Taleb
You care too much about the vehicle. What they are testing isn't the vehicle, they are testing the production processes.
I'm confident the vehicle design will work once they get a handle on the production. They got really good simulation tools for the vehicle flight part of the problem.
> You care too much about the vehicle. What they are testing isn't the vehicle, they are testing the production processes.
No, they clearly are not.
They are designing a vehicle. One of the design requirements might be a better production process, but as it is very obvious to everyone their goal is not to produce a vehicle that blows up unexpectedly.
And the vehicle blew up unexpectedly, when it was scheduled for a flight test.
The goal of SpaceX is not to implement the broken window fallacy.
He does want things to blow up, but he vary rarely wants this thing to blow up.
If things don't blow up, they're spending too much time being careful, since the cost of being reckless at this stage of development is less than the cost of things blowing up on a regular basis.
If this thing blows up, well now that was just unfortunate. That was a not-entirely-free test article or lawn ornament and now the test failed so they can't be sure that it blowing up didn't hide another issue 10 seconds down the line. Obviously it would have been better if the person on the ground just remembered to check the whatchyamaculit before the test and prevented it from blowing up (which doesn't change the fact that it's not worth it to pay people to check every whatchyamaculit at this stage of development).
That's the point - they're discovering the limits.
The other way to do this would be to over engineer, have more conservative testing, etc. But Musk has been clear that they've explicitly chosen a strategy of fast iteration, and breaking things instead. And that's one of the primary things that gives them an edge over NASA.
You don't need to be an aerospace engineer to be aware of the spaceX R&D strategy which has been spelled out pretty clearly by Musk and others in various interviews.
I'm not saying the failure was intended, I'm saying the reason there are so many failures because they opted to go hard and fast, while government funded shops like NASA can't afford to have such a high failure rate because of the optics and politics. This modality to R&D isn't at all unique to aerospace and is more of a business/management issue so I'm not sure why you think an engineer would have anything more meaningful to say.
As Kelly Johnson of skunkworks said, you can't have innovation unless you are prepared to see failures.
The whole fundamental concept of innovation is that there is no plan. If there's a clear path to get to where you want to go, then you don't need to innovate. Innovation is the act of exploring the unknown regions of the problem search space. No, there's no guarantee they picked an optimal strategy. If such a guarantee were possible, they wouldn't need to be trying and failing in the first place.
What there is clear evidence of, is you generally reach your goal faster if you're more risk tolerant. And that's very well understood by now. Various r&d folk from NASA have also said that they wish they could go harder and faster. But because they're micromanaged by congress, every failure is very expensive politically.
The stated attitude of a company as public as SpaceX is carefully tuned and measured. The actual attitude internally is likely to be substantially different.
> The stated attitude of a company as public as SpaceX is carefully tuned and measured. The actual attitude internally is likely to be substantially different.
Its still a Musk corp, and as I was told at Tesla 'Elon gets, what Elon wants.' Its really odd philosphy but not entirely surprising, because I was going for Operational Support roles, not Engineering where its understood he has full reign, and it was felt in those department's leads/directors.
As for SN4, well, what's the saying: Progress is messy. Aerospace has lots of failures. Onto SN5!
Please don't let these rather aggressive comments from Ruminator and others stop you commenting.
You put forward perfectly reasonable observations and it is frustrating when someone tries to silence others using the appalling Credentials Fallacy.
It is perfectly logical to say at the macro level our "take risks, move fast" strategy will produce more failures, whilst at the micro level being very disappointed at each failure.
Now, if this was a manned mission with life at stake I would expect the risk approach to be modulated accordingly. But even so, astronauts are not civilian passages and even they knowingly embrace flying at high risk. It would be interesting to know how (and if) SpaceX has approached derisking manned flight. Because the PR from blowing up humans is not good whether you're NASA or a private company.
2. look for lapses in construction technique and QA
In car racing, you'll never learn to be a fast driver unless you can drive right at the limit, and you'll never know where the limit is unless you exceed it now and then.
Destructive tests are planned accordingly, and the outcome is known beforehand. The questions to be answered are a) which failure mode, b) quantify limits of a failure mode.
This was not a destructive test. Their plan was to fly this same vehicle next. Now they can't because it blew up.
This wasn't a test to destruction. It was a static fire test. While I'm sure there's good data to leverage, I highly doubt there were any details in the test plan for characterizing catastrophic failure.
Yeah, but if you dont want it to explode and it does - it's not a good thing. Destructive testing is something different, like the mission abort test they did a while ago.
> Not disappointing - the more things that go wrong during the extensive series of tests developed and cataloged over the past several decades, the more we can correct and cement into permanence.
To some extend, but just as with software, I’m super happy if my program just has zero bugs after all the testing is done.
This has never happened, but it’s nonetheless something to strive for.
Would it have really killed their schedule so much to push this until after tomorrow's crewed flight? They're obviously unrelated, but it sucks to have this is the headlines the day before they launch astronauts.
This was the fifth static fire of the same rocket and frankly it wasn't seen as high risk at the time. If it was, I am sure they would have postponed it, as they did for the planned short-hop.
Of course "high-risk" is a relative term. They had a 1.5 mile exclusion zone for the test and the handful of village residents a few miles away were warned to leave their houses to prevent injuries from broken glass, so there was obviously some risk.
This was really a just a potential PR problem for click-baity new organizations and casual observers not aware of all the different projects SpaceX is working on. No one from NASA going to say maybe we should scrub Saturday's launch because they know the two not related in any way.
Also, there are many Elon haters are out there. I had a few try to blame Wednesday's scrub on Musk or try to imply that it was a failure. (Of course the fan boys can be just as annoying.)
The astronauts are both veterans and one would assume they understand that this was not only a prototype ship but a prototype engine. Falcon has 50+ successful flights over four years since its most recent failure.
I get cynical about people sometimes, but I have to believe the average person can understand the difference between a bleeding edge test and the launch of a proven system. They don't stop driving when someone dies trying to break the land speed record. I don't see a PR problem.
I thought the same thing at first, but I doubt very much this will be reported very widely. Despite it being huge in our circles, most people don’t follow this stuff very closely. Unless both SpaceX and NASA are pushing out stories at max volume with included press kits, anything they do will be basically niche at best.
Recall on the FH test flight how the loss of the center core was not announced immediately, and as a result of SpaceX delaying the announcement of the loss by an hour or three, very nearly 100% of the reporting on the topic was about their successful launch and landing of the two (heh) boosters, with no mention whatsoever of the fact that the launch also contained a failure.
An absolutely masterful PR decision, and one I learned a lot from, but also evidence of how fickle, inaccurate/incomplete, and flighty/shallow media attention is. For the vast majority of people and media outlets: nobody really pays attention to rockets in general, especially while the nation is burning/dying/being impoverished. You have to be insanely loud to penetrate that fog.
When you put it that way, that masterful PR decision stands in stark contrast to this one. Any outlet that was planning to cover the launch on Wednesday and would never have covered SN4, but now plans to cover the Demo2 launch again today might be more likely to cover the SN4 failure.
This is the 4th Starship that has imploded or exploded. They only lost 3 Falcon 1's before their first successful flight, and Elon has publicly stated that a 4th failure would have bankrupted the company.
Those three Falcon flights were attempts to reach orbit with rockets that weren't prototypes so the failures were more significant, and they probably blew up a bunch of Falcon prototypes behind closed doors prior to the launch attempts, but it's interesting how three failures on Falcon was a big deal but four failures on Starship isn't. SpaceX is certainly in a different and much less precarious position now than it was in 2006-2008.
They lost many more than 3 Falcon 1 tanks during development. They only lost 3 Falcon 1 launches to orbit. There have been 0 starship launches to orbit - they haven't even built a first stage prototype yet.
The difference between some test parts on a stick vs. a full rocket is so vast that I don't think you can make any coherent/interesting comparison of the numbers.
It's like comparing 4 millimeters to 3 inches. It doesn't matter at all that 4 > 3.
It's interesting that a company with a revenue stream can tolerate more failure of declared prototypes more than one without can tolerate operational (if early stage) deployments?
I mean yeah, it's a viable business now. In 2008 they had no successful launches, no revenue stream, and had almost exhausted their seed funds. That's pretty normal.
Boeing and Lockheed Martin got a billion dollars (together, so half a billion each) from the Air Force to develop Delta IV and Atlas V.
Then ULA got monopolist prices ($400 million for a Delta IV Heavy launch!) for years and a sweet billion dollars per year for maintaining the capability even if no launches happen that year.
Falcon 1 was developed using private money.
Falcon 9 and Dragon 1 were developed using 396 million of government money from NASA COTS program, and about a billion dollars of private money (for upgrades and reusability).
Falcon Heavy was developed using half a billion of private money.
That’s how it goes with building something properly large, isn’t it? The speed range on first gear is more narrow than second, and third, and so forth.
I’m just glad SpaceX has the cash flow and team to be able to pursue something as ambitious as BFR. Humans are going to need many thousands of rockets like that one if we are to survive.
> Humans are going to need many thousands of rockets like that one if we are to survive.
There is no chance in the next hundred years (I would bet more like 200-500) of a self-sufficient colony anywhere away from the earth. If humanity can't survive without rockets, it can't survive at all.
Not to mention that it is basically impossible to imagine that any technology that could allow a self-sufficient colony on another planet couldn't much more easily allow humans to live on earth after any catastrophe you could imagine short of cosmic ray events. It's much, much harder to live on Mars than on a post-apocalyptic earth, regardless of which apocalypse you care to choose (nuclear war, nuclear meltdowns in all nuclear power plants, catastrophic global warming, super volcanoes, meteor impact the size of the Cambrian event, you name it).
Sure, in the enormously long term there may be a need to leave the earth before the sun ages too much, or to try to have a chance against planetary collisions or gamma ray bursts.
I notice all the broken bits of rocket are in square panels.
That indicates the welds failed rather than the steel sheets in the explosion.
If you want to design something to be as strong as possible for the weight, you make the welds as strong as the sheets. In most cases, that means making the weld lines thicker than the rest, since the weld is typically formed in less ideal conditions so weaker.
There are tricks to doing that like zig-zag weld lines, overlapped sheets at the weld, etc.
Surprising spaceX hasn't done this, considering strength to weight is critical in rocketry.
Two possibilities: the first is that for prototypes (even flight capable prototypes) they're optimizing for cost/iteration time more than performance. Given their apparent plan (which is just to build lots of them sand not worry too much if they blow up), going with simpler methods seems reasonable.
The other possibility is that they straight up are willing to eat the weight penalty, Starship is being put together with different constraints/assumptions than most other systems - like it's literally a stainless steel reusable rocket. Maybe their gaining something from that tradeoff that they prefer more.
Can you provide references supporting thicker welds as well as reinforcing patterns? I have done introductory welding and am genuinely curious of the ramifications of both.
Wouldn't strength only need to get to a certain "good enough" point rather than "as strong as possible?" Maybe the shell is already at that point? And after that point, maybe all there is left to concern yourself with is how it behaves in an explosion. As with any thing designed to explode (grenade, bomb, etc.) Maybe they want for the thing to come apart in a certain way rather than holding together?
My first guess at an explanation, assuming the first poster is correct about the panels and welds, would be that they haven't narrowed down their specifications of the panels and welds yet, not that they have purposely left the welds as the weakest portion.
Another guess would be that they are using identical panels in many locations and have thinner welds in some of them. But that is sort of the same thing as above, because they'd likely make different panels as production volume increased.
Could be, considering we're talking about Elon Musk. If you know the thing is going to get caught on video exploding, then might as well make a good show of it.
> would be that they haven't narrowed down their specifications of the panels and welds yet
Sure, it's a prototype. Maybe they'll revisit welds on the next one. Or maybe they'll figure out a way to make explosions look even more cool.
I know these kinds of explosions aren't disasters per se, but at what point in development do these kind of destructive results become very unwanted? Or was that classic Calvin and Hobbes comic about bridge testing less of a joke than I had though?
When you start making the same mistakes over and over. As long as they learn new things with each iteration it doesn't matter if they blow up 5 or 50 ships. The actual material cost of these prototypes is probably in the low 6 figures. The biggest cost is the labor, but by iterating rapidly they are learning new manufacturing techniques, so if they weren't blowing up these prototypes they would be stacking them in a field or tearing them down anyways, since the value is in the process, not the final product, at this point.
Also, there are only an estimated 300 employees building Starships (although there are probably more working on engineering), so on-site labor costs are probably about a million a week. Since they can pop out a new iteration every few weeks they cost is probably only about a few million per ship. That's a lot of money in some industries, but it's a rounding error in rocketry. They could build dozens or even hundreds of prototypes for less than the cost of a conventional disposable rocket, like SLS.
The idea that the cheapest way to build a rocket is to blow up dozens of prototypes is a bit counterintuitive, but I believe it's the future. Organizations that move more slowly to avoid failures will fall further and further behind, since the tortoise only wins if the hare falls asleep, which SpaceX shows no signs of doing.
Civil engineers are able to use mathematical models to develop bridges because we've been building bridges for thousands of years, and they can build in huge safety factors. Fully reusable rockets have never been built before and can't afford the luxury of large safety factors due to the Rocket Equation, so testing to failure is the fastest way to learn, and when almost all of the cost is labor, fastest = cheapest.
> The latest from the President and COO of SpaceX is that there are 900 employees working at Boca Chica alone.
Yeah, they expanded fast at Boca. Gwynne Shotwell doesn't get the attention she deserves unless you're in the know, but she really is a total badass COO. I really wanted to see her at the Cape for this momentous occasion but things are Hawthorne are he priority.
I still remember that shot from her and the Team at HQ going crazy after the first stage recovery, I threw my hands in the air like that watching the Live stream for ORBCOMM 2 and just kept yelling 'they did it!'
All links to Techcrunch go to
https://guce.advertising.com/collectIdentifiers?sessionId=1_... uuid> and Blockada [1] blocks them on my Android devices. I stopped visiting Techcrunch since they started to do that. I can always google the same news on a different site.
Yeah given the delayed and soon upcoming actual human scheduled flight, this would have been maybe nicer to put off a couple days... just optics but still...
"It's actually good optics that it exploded, because it shows they aren't changing their schedules for the sake of optics. Political pressure to make the schedule fit the public's feelings was a major contributor to the Challenger disaster. So, having a tank blow up the day before a manned launch shows that SpaceX has overcome the challenges of the past."
(See? That spin makes the turbopump look like a stator. If I have a point it's that talking about optics is pointless, because you can make the optics anything you want by casting the right light.)
Funny article, it stressed at least 4 times this is different from the normal flights ones they do regularly, otherwise covidiots/flat-earthears/conspiracists will fill forums with their cries. And I'm not sure they wont anyway.
The timing was very perfect.
EDIT: https://twitter.com/NASASpaceflight/status/12664420878489600... has the audio with it.