We get it. You hate Musk. Get over yourselves.
The Mk1 prototype of Elon Musk's Starship suffered what we're betting SpaceX will call an "anomaly" during a fuel tank test today. Viewers of a internet live-stream of the test were able to make out the top of the tank protruding from the Starsh** stage as SpaceX filled the thing with fuel. There was the usual venting at first …
It probably wasn't exothermic. They were loading liquid nitrogen to test pressurisation. The pressure got too high and a weld failed. There was no flame, and no chemical reaction; it was more like a balloon bursting. The cold nitrogen condensed water vapour from the air.
That's one reason to call it an anomaly rather than an explosion. "Explosion" would be misleading as to what happened.
Also, phrases like "Somehow he managed to do it without killing himself." and "They usually detonated on contact with the oxidizer, as several possessors of piles of junk that had originally been
ignition delay equipment could testify, and did." add a certain ... zest to an already powerful story.
well, any animosity aside, a catastrophic fail in the test phase is part of the process. It's just that SpaceX can't do this without people seeing it and laughing a bit.
Who out there has NEVER let the blue smoke out of a component? I most recently had that happen to me when someone handed me a 12V power supply that REALLY turned out to be a 24V power supply, with the same connector, and I plugged it in without reading the @#$% label... blue smoke and arcs and OH CRAP and I repaired the board but the fried regulator managed to take the CPU with it...
(fortunately the CPU was TQFP, unfortunately had trouble soldering it with $CLIENT's tools, nearly had to bring it home to fix it, then managed to blob-and-wick some solder onto a questionable-looking CPU pin, then all good)
This post has been deleted by its author
This post has been deleted by its author
Sometimes a test is done to find the weak spot and where the effort should be going. This may not have been their intention, but I bet they learnt a lot.
I've been told many times that any fool can make kit unbreakable, but if you don't break something then it was overengineered :-)
Yes the lab coat please with the copy of "Ignition!" in the pocket.
And sometimes you actually want a weak spot, but you find that it is somewhere you don't want it.
Like HPN adding stiffening braces to the BMW R80GS and R100GS frames, but finding that now the frames were breaking instead of the (original) welds. Fix was to pre-weaken the braces.
"Like HPN adding stiffening braces to the BMW R80GS and R100GS frames, but finding that now the frames were breaking instead of the (original) welds. Fix was to pre-weaken the braces."
In WW1 a German fighter plane was breaking its wings. The main structural beam was strengthened - and it still broke. After several iterations someone applied some science. The beam needed to be weaker - presumably so it would flex. The book "Structures: Or Why Things Don't Fall Down" may be the one I recall explaining how engineers and architects have made mistakes.
The book has a nice quote in its preface.
Among the innumerable mortifications that way-lay human arrogance on every side may well be reckoned our ignorance of the most common objects and effects, a defect of which we become more sensible by every attempt to supply it. Vulgar and inactive minds confound familiaritv with knowledge, and conceive themselves informed of the whole nature of things when they are shown their form or told their use; but the Speculatist, who is not content with superficial views, harrasses himself with fruitless curiosity, and still as he enquires more perceives only that he knows less.
"The Idler is naturally censorious; those who attempt nothing themselves, think every thing easily performed, and consider the unsuccessful always as criminal."
Samuel Johnson 1758
SpaceX say they were testing "to the max" and that the result was "not entirely unexpected".
good point. back in the day we called this sort of thing 'hydrostatic testing'. Normally weren't expecting it to blow up but just in case, better during the test than during operations. So you pressurize with water and run it up to maximum expected levels. In the case of a catastrophic fail, something cracks and pressure drops significantly due to it being water.
In THEIR case, however, they had to use fuel, probably due to cryogenic temperatures. At cold temperatures of cryo-fuels (like LOX) metals become EXTREMELY brittle. And if their fracture toughness isn't quite right, you end up with, uh, an "anomoly".
For all we know, it was the welding process or something like it, that was responsible. It can happen. In WW2 a liberty ship broke in half during construction due to bad welds and brittle fracture.
Along similar lines on the behaviour of metal when welding, the story of Salesforce Transit Centre in SF:
TL;DR: opening of new building delayed after structural cracks found in two beams. Entire structure checked and no further issues found - cause was incorrect post-weld treatment resulting in micro-cracks that resulted in brittle cracking when placed under load.
"At cold temperatures of cryo-fuels (like LOX) metals become EXTREMELY brittle"
Some alloys are stronger at cryogenic temps.
The usual process is to test multiple samples. Some will be tested to destruction by pressurizing them with water until they fail and further tanks tested to 1.5x or so of nominal working pressure as part of the qualification for use. "Maximum working pressure" is a bit of a misnomer. That maximum is the expected working pressure. If it fails at that pressure, there is zero safety margin.
Somewhere I have the photos of the burst tanks we did at work to test a new design. We had some issues with welding on the proof tanks, but got it all sorted before ever filling up with LOx and Helium (we used a pressure fed system on the small rockets rather than a turbopump).
To have an anomaly like SX did is not a good sign. If anything, the test article should have been beefier than needed. Most aerospace stuff is like that. You overbuild the test article and then refine down rather than shaving stuff down on the first go and seeing what blows up and then bulking it up a bit to see if that helps. Is it the FEA software that all the new kids have complete faith in? You know, those fresh out of college engineers that have never actually built something with real materials shaped by the lowest bidder according to some idealized computer model. It's like learning electronics at university with "ideal" opamps, transistors and capacitors. In my ME courses we started by ignoring weld inconsistencies and assumed the weld was just as good as having contiguous base metal. Good luck with that on designs with critical joins.
"It can happen. In WW2 a liberty ship broke in half during construction due to bad welds and brittle fracture."
Many of the early Liberty ships sank due to a structural failure. The problem was that the cargo hatches had sharp right angle corners which concentrated fracture stresses. The solution was to round the corners.
Surprisingly a few years later - the first Comet passenger jet planes disintegrated from the same design error by having square windows.
This is probably one of the most annoying things about the internet. 30 years ago this kind of testing would have been done without anyone in the world knowing about it, and unless someone was killed, people would be presented months or years later with the finished product.
Now everything is streamed to the internet. Joe Public and headline seeking journalists then froth at the mouth screaming down the intertubes how unsafe Product-X is. This then goes viral, and suddenly every self qualified expert commentard offers their factual opinion, resulting in everyone "knowing" that Company-X produces unsafe products. Company-X then goes bust as investors pull out because of a bad reputation.
I'm currently working on a marine project at the moment where destructive testing is considered a good thing, and even more so when it's unexpected, but still within the testing period. Contrary to what expertards believe, science and engineering do not know everything, which is why we test. Best the engineers discover then mitigate issues during production, and not when the ship is sinking.
"30 years ago this kind of testing would have been done without anyone in the world knowing about it, and unless someone was killed, people would be presented months or years later with the finished product."
30 years ago, the tank design would have gone through physical tests rather than just FEA sims on a computer. It wouldn't have blow while being tested at it's working pressure but only if something were to have gone very wrong.
If you are going to make your testing live and public, it's a good idea to spend the time to do things properly. Yes, it takes time and costs money but so does repairing test facilities and metaphorically cleaning the egg from your face.
"I'm with you there, wouldn't want to be near any rocket during fuelling, It's probably the most dangerous event apart from ignition during the whole launch cycle.
In other stories, NASA is not happy with SpaceX wanting to have astronauts in the capsule while the the rocket is being fueled on the pad. SpaceX claims that waiting until after fueling is complete allows the system to warm up and doesn't let them get the fuel density as high as they need.
SOP for Crew Dragon will be to fit the astronauts into the capsule AND THEN start fueling. Should just call them organic payload. I can't see how they will be able to do anything with the rocket engines lit on a touch display.
Look closely at all of the crewed NASA vehicles and you will see that there are physical switches with guards and places to anchor the hand. They learned early on that the vibration was too much for astronauts to accurately actuate controls and lots of opportunity to flip the wrong switch by accident.
The counter-argument is that even approaching a fuelled rocket is dangerous. It wouldn't be allowed for an uncrewed launch, for example. So loading the astronauts into a ready-fuelled rocket presents a danger both to the astronauts and to the ground crew who help them into the rocket. They are all very exposed as they are wandering around the pad. If you load the astronauts first, by the time you start fuelling the ground crew are well away, and the astronauts themselves are largely protected by the Dragon 2 and its launch abort capability.
As for the Dragon 2 touch screens, those are mostly informational during launch. SpaceX vehicles are autonomous. They don't have human pilots controlling them.
Musk's optimism for timescales is pretty notorious (recall the 2011 video showing the F9 US landing? Great CGI)
So they learned something they shouldn't do next time.
I'm pretty sure SS is going to go through at least a couple more iterations in design before this thing flies to orbit.
I'm also pretty sure it will happen barring Musk keeling over.
I'm just not sure when.
Not entirely unexpected. Mk 1 has been hastily fabbed up and was already entirely obsolete. For instance, it was fabricated from panels of stainless because that's what they could get. The Mk 2 (in Florida) and Mk 3 (in Texas with the Mk1) are fabricated from rings of steel straight off a roll - far fewer welds, far less complex assembly.
In normal SpaceX style they're iterating quickly and finding problems early and fast - compared to the more traditional direction Blue Origin have taken of spending 19years trying to design the perfect rocket before cutting steel.
Might delay the first flight of a Starship prototype (which will now be a Mk2 or Mk3!) but it's unlikely to put much of a hold on the project overall.
The camera was set up to record one frame every millisecond. When the nuke blew, the lid was caught in the first frame and then disappeared from view. Judging from the yield and the pressure, Dr Brownlee estimated that it left the ground at more than 60 kilometres per second, or more than five times the escape velocity of our planet.
It's called rocket science for a reason.
As I recall, the US Air Force in the 1950s had a good number of spectacular launch failures while developing rockets before they managed to get one off the pad in once piece. (Not counting the German NAZI rockets they'd carted out of Peenemunde in '45, of course.) If it hadn't been for the cold war and the need to get spy sats over Russia the boffins probably never would have stuck with it. And later rocketry wasn't without its problems, either. Think Apollo 1 and two lost Space Shuttles... Or the Russian space program's encounters with failed descent parachutes, explosive decompression and goodly number of explosions on the pad as well. So I'm all for testing the heck out of everything, and I'm not even expecting to ever fly in one of these things.
This is how technology gets developed. You do the best you know how; you test; you fail; you change what you borked and you test again. If everyone knew how to get it 100% right the first time... Well, where's the fun in that?
So here's a pint raised to the guys who test stuff!
Biting the hand that feeds IT © 1998–2021