Hacker News Reader

Comment by redis_mlc | Hacker News Reader

Maybe somebody can explain this to me.

Welds are supposed to be tested at the weld level, not at the vehicle level. So what is wrong with their process or QA?

You are right. At the very basic levels we have PT. This is a test with dye sprayed on the weld and a developer sprayed on to show the defects. This is very commonly done.

Next we have UltraSonic(UT) testing. This requires expensive equipment and a good examiner to interpret.

Third we have X-Ray. This is used on welds that are very important as it is also hard to do and interpret. It also requires the weld to be ground smooth to distinguish flaws from ordinary weld.

I am sure there are a variety of other non destructive testing (NDT )techniques. These are the common ones used in industry.

I do want to mention that pressure vessels are hard to make. Especially something so large. This could actually be a design problem and not welds. Pressure vessels are often tested at 5/3 their working pressure. So failures at testing are not uncommon.

I really don't understand the build method they use. Ideally you would use a turntable to rotate the cylinder while welding the sections together with a sub-arc machine. https://www.youtube.com/watch?v=j-hfExEmGsE

Elon blaming "settings" seems like a silly explanation. There should be weld engineers and destructive testing done before the first beads are welded on an actual tank.

I've set up automated welding cells, and the "settings" excuse sounds totally reasonable to me haha. There are indeed a million settings (weld weave pitch and amplitude, wire feed speed, linear feed speed, gas flow rate, amperage, etc.) that are hard to tune 100% correctly and don't have "analytic solutions" (i.e. you tune some values like a monkey until it gives the result you want). Seems like this tank failure was their "destructive testing," although most people would probably have used smaller samples to save money. As someone else said: pressure vessels are hard!

> Ideally you would use a turntable to rotate the cylinder while welding the sections together with a sub-arc machine. https://www.youtube.com/watch?v=j-hfExEmGsE

Uh, at least in the video that steel is pretty thick. Starships skin is slightly less than 4mm thick... while in that video they have a 2cm deep groove, would the same technique actually work?

For what it's worth, ULA stainless steel (<1mm thickness) upper stages are resistance welded instead https://www.youtube.com/watch?v=o0fG_lnVhHw&t=41m10s

That said, I think a lot of the decisions right now are based on "what can we quickly experiment with" not "what is a good long term plan". Building a 9m turntable (diameter of the rocket) would take a long time.

I agree a 9m turntable would take some time to make. But once made you could crank these vessels out very fast. 4mm skin sounds OK and I am sure there are some structural supports inside to keep it from collapsing. I am not a weld engineer so I cannot say if Sub-Arc would be the best technique. I know that ships skins are being built with it.

You’re talking about production engineering. You don’t really do that until your design starts to be nailed down. It’s not.

Tory Bruno in an interview says ULA does between 1.1 and 1.25 safety factor for pressure vessels.

Boeing avoids welded seams on airplanes because they are very difficult to reliably inspect. Riveted seams, however, are straightforward to inspect.

I assume riveted seams don't work for pressure vessels? Would the air plane cabin be welded?

Rivets were the fastener of choice for pressure vessels in the steam age.

The pressure bulkheads in an airliner are riveted.

Dye penetrant is another NDT that highlights cracks for visual inspection too, but it's proven to be less effective than ultrasonic and x-ray (hence the changes some years ago in the inspection and rebuild manuals by AD's for many (it should be all) turbofan engines). I knew a German engineer consultant who did this on piping, welds and pressure vessels in the field for nuclear plants.

You can use all three of these NDT techniques, where are used in nuclear energy and turbofan engine applications, but still fail to detect weak welds or metal fatigue because of improper calculations or conditions in the real world.

There's no substitute for DT to validate the engineering and manufacturing, such as by overpressing a pressure vessel (which are also covered by ASME BPVC).

Oddly enough, I knew a civil engineer (PE) who was an expert witness and wrote code interpretations for BPVC to the tune of $40-60k per for a few hours of work (but having his experience on large industrial projects is an important prerequisite). IIRC, he had a massive house in Saratoga.

NDT = Non-destructive Testing

PE = Professional Engineer

ASME BPVC = American Society of Mechanical Engineers Boiler & Pressure Vessel Code.

Thank you

I'm not a welder, but you're talking about

A prototype that was manufactured in a tent and in a field, with welds being done in that field only a few days ago. I doubt much QA or weld level testing was done.

A prototype on which they knew welds were less than ideal...

And specifically a prototype of a rocket being built with new materials and new manufacturing methods. Rockets have the nearly unique distinction of having to be strong enough and no stronger because being stronger wastes mass which lowers the very important payload to orbit number. Bursting a few pressure vessels as they figure out where that point on the strength curve is seems pretty reasonable to me.

We don't (yet) know at what pressure this burst at, if it burst near the total pressure needed to achieve their safety margin I imagine they will be ecstatic. I wouldn't interpret this test failure as a process, or QA failure without more information.

> A prototype that was manufactured in a tent and in a field, with welds being done in that field only a few days ago. I doubt much QA or weld level testing was done.

What you're describing is how farm tractors are repaired, not how aerospace mfg. is conducted.

You can't "test out" bad metallurgy, welding or mfg. in rockets.

What I'm describing is literally how this rocket was made... literally the whole thing is being live streamed by people living nearby with both 24/7 cams on the facility and frequent higher quality closer films and pictures being taken. Occasionally there are tweets from Elon about the process too.

You are far from the only one who is skeptical that this is a valuable process, but there isn't any room to debate that it is what they are doing.

Edit: Here is a recent timelapse of some of that welding, as an example: https://www.youtube.com/watch?v=J7bIvLhI1hY

I keep thinking of the story "Have Space Suit—Will Travel", except it now it is "Have Welders—Will Construct Space Craft". What SpaceX is doing is both crazy complex rocket science, yet it's also quaintly pedestrian. It makes space feel more accessible to me.

What's a rocket but a tractor with a few tens of thousands of lbs of fuel and oxidizer strapped to it?

There's a reason SpaceX is eating "big aerospace's" lunch and tests like this are part of it.

> Rockets have the nearly unique distinction of having to be strong enough and no stronger

More precisely, airplanes are designed to being 50% stronger than the max load it would ever see. Rockets are designed to something like 10% stronger.

SpaceX is apparently aiming for a 1.4x safety factor: https://twitter.com/elonmusk/status/1215719463913345024

But my point here actually was the difference between "at least x times stronger than it needs to be", and "exactly x times stronger than it needs to be". Usually things are manufactured to the first spec, rockets are much closer to the latter spec because keeping weight down is so important.

I suppose airplanes must also be closer to the latter spec than most things, but they should still be more forgiving than rockets.

Isn’t the entire prototype part of the QA?

Reply on news.ycombinator.com