Third Starship prototype collapses during tank pressure test


Please consider donating to Behind the Black, by giving either a one-time contribution or a regular subscription, as outlined in the tip jar below. Your support will allow me to continue covering science and culture as I have for the past twenty years, independent and free from any outside influence.


 

Regular readers can support Behind The Black with a contribution via paypal:

Or with a subscription with regular donations from your Paypal or credit card account:


 

If Paypal doesn't work for you, you can support Behind The Black directly by sending your donation by check, payable to Robert Zimmerman, to
 
Behind The Black
c/o Robert Zimmerman
P.O.Box 1262
Cortaro, AZ 85652

During the final part of a tank pressure test the third SpaceX Starship prototype apparently collapsed, its outer welded hull failing.

Video below the fold. The prototype is on the right, and it appears it fall inward along its hull welds.

The SN3 (Serial Number 3) vehicle incorporated lessons learned from previous vehicles and test articles, and took advantage of improved manufacturing techniques and expanded facilities at SpaceX’s South Texas launch facility.

The next round of testing began this week with cryogenic proof testing. These tests saw the vehicle filled with liquid nitrogen at cryogenic temperatures and flight pressures. Proof testing began in Thursday and continued through to Friday morning when SN3 failed during what appeared to be the end of the test.

With Elon Musk noting “we will see what data review says in the morning, but this may have been a test configuration mistake,” on Twitter and the first-look observations, the fault may have been related to detanking, rather than another failure under pressure.

I’m no engineer, so I wonder how detanking could cause such a failure.

Share

19 comments

  • Scott M.

    There’s an unconfirmed statement floating around that one of the ground-support valves failed under cryogenic conditions. It might have resulted in ‘unplanned’ venting of the bottom tank, which then buckled under the weight of the still-full tank above.

    I’m sure we’ll find out the full story in a few days. SpaceX continues to impress me with their transparency.

    Oh, and BTW in case your blood pressure needs spiking here’s a bit of nonsense vomited up by The Verge:

    https://www.theverge.com/2020/4/2/21198272/blue-origin-coronavirus-leaked-audio-test-launch-workers-jeff-bezos

    I figure it’s good for a laugh if nothing else.

  • Max

    Another possible scenario is that when the pressure was released from the tank, vacuum sucked in the sides, in the way a water bottle or a pop bottle which is made to hold pressure, will collapse inward in the refrigerator.

  • Andrew_W

    I agree with Max, it looks like internal pressure dropped below external pressure

  • Ray Van Dune

    I am sure many of us have seen the video of a railway tank car crumpling when the cargo drain valves were opened, but the vent valves were accidentally left closed. The mass of the liquid cargo was sufficient to force it to partially drain it out, creating a partial vacuum in the tank above the liquid, eventually crushing the tank. Something similar could be what happened here.

  • Ray Van Dune

    Ps. It appears that it was the UPPER tank that depressurized and crumpled. Or perhaps it was the header tank inside it that depressurized, which might have had the same effect.

  • Micai

    I don’t work for SpaceX, but here’s my understanding:
    Because the walls of the tanks are made so thin, they rely on internal pressure to give it structural strength. If you look at the size of the vented plumes before the collapse, you will see the plumes from the bottom tank get very thin, while the top tank is in full blast. That says the LOX tank had very low pressure, while the CH4 tank was full, so the empty LOX tank couldn’t hold up the weight and collapsed. What I think they mean by detanking causing the problem is that they might have emptied the lower tank too fast given the top tank was still full/emptying.

  • Rose

    I sure does look like the LOX tank drew a partial vacuum, crumpled, then collapsed under the load of the partially full (of LN2) LCH4 tank, but it seems more likely that it simply lost the positive pressure (of GN2) necessary for its strength and collapsed under load. I think the smaller venting we see below the frost ring is from the LOX tank, and that implies some positive pressure in the tank.

    I don’t believe that these are quite the balloon tanks of the old Atlas days (see https://www.thespacereview.com/article/1326/1 “Not a bang, but a whimper” Dwayne A. Day, March 2009 — Video of collapse: https://www.youtube.com/watch?v=imkdz63agHY ) which needed to be pressurized to support their own weight, but I don’t think they are far from it, and do need to be pressurized to support much additional load.

    The detanking procedure presumably calls for the LCH4 tank to be fully drained before the LOX tank below it is depressurized, and either a faulty valve or an incorrect or mistimed command depressurized the LOX tank early.

  • Rose

    Here is a video with the audio synchronized to compensate for the delay: https://www.youtube.com/watch?v=D8-RZmTOtXQ&t=110

  • MDN

    What impresses me about SpaceX is that they understand this stuff is complicated and takes lots of incremental steps and tests to really get right. It is not easy to design a light and efficient tank, and the method of using the tank (and the control systems and software that will manage it in production use) are just as important as the design when the tank is a stressed member (i.e. structural element) of the ship.

    Now is the time when you can afford and want to uncover problems because now is when they can be quickly and inexpensively fixed. The way NASA runs things we might see this with a flight article instead of a prototype because they are scared of anything that might fail. That means they inherently over-design and always need a ridiculous schedule and budget to even have a chance of success, and then risk BIG failures late in the program anyway because in the end there simply isn’t enough testing.

    Kudos to SpaceX for moving fast and not being afraid, or freaking out, over failure. If your don’t fail a few times during the design you aren’t pushing hard enough.

  • Lee S

    @Scott M….. The most telling line for me from the article is “Blue Origin’s primary competitor at the moment is Virgin Galactic”
    That one sentence speaks volumes!

  • Lee S

    I gotta say, SpaceX impresses the crap out of me…. Build it, try it, it blows up, work out why, fix the problem, build it…etc etc…
    They are hand riveting the hull together right now, with wooden scaffolding inside… But SpaceX is not testing the integrity of the hull right now…. ( Fortunately!). NASA would be friction welding custom panels together using custom alloy steel before they even considered doing a test on on the lighting system…. Respect to SpaceX for failing proudly in their development!! It’s true you learn so much more from a failure than you do from a successful test!

  • Ray Van Dune

    Is my video player playing upside down? Various commenters seem to be saying that the BOTTOM tank lost pressure, crumpled, and could not support the UPPER tank. What I see pretty clearly is the UPPER tank collapsing, and in effect TOPPLING OFF the lower tank.

  • David

    The upper tank is the one that is frosted over. The crumple is directly below that, at the top of the lower tank.

    While I agree with everyone else that it’s amazing and wonderful that SpaceX is willing to try, fail, pick up the pieces and try again. it’s somewhat surprising that they’ve had this many of the same basic failure in a row. I bet there are some heated meetings happening that we don’t hear about.

  • pzatchok

    That crumple zone looks like it is in the lower tank but above the liquid level inside the tank.

    The simple fluid pressure could be whats helping it keep its shape.

    When it falls you can see even the lower area eventually crushes and rips.

    Either they messed up and emptied the lower tank first. And forgot to open the lower vent valve. Or it got frozen and stuck.
    Or they wanted to empty both at once (which I find dangerous) and the upper drain valve did not open.

    Or this was a what if test to destruction. What if the lower tank went empty by mistake or failure. Broken fill line for instance.

  • Edward

    This looks more like column failure than a vacuum failure, and Scott M’s explanation sounds reasonable. David’s description of the event is correct; the upper tank is smaller than the lower and the failure occurred at the top of the lower tank. Rose’s comparison with the early Atlases sounds reasonable, but I suspect that they would want a stronger (heavier) vehicle for the manned version(s) of Starliner.

    SpaceX may be trying to save weight by using a weaker skin, but I think that they may not need to try so hard and could take a little hit on the overall payload to orbit. I suspect that SpaceX is trying to get to Mars as much mass as it can with each launch and is trying for more performance than they tried for while designing their Falcons. The goal of the Falcons was to reduce launch costs, and compromises on efficiency and performance did not seem to bother them. With Super Heavy-Starliner, they would be the price leader even if they could get only half their expected mass to orbit.

    Now they are trying to settle and colonize other worlds (Mars and maybe the Moon), so they may be going for more efficiency and performance than in the past. The development cost may be much more, but the payoff would be worth the price.

  • Dick Eagleson

    Lee S,

    Starship has no rivets anywhere. It’s an all-welded construction. The engine bay at the bottom has many columns of spot welds where stiffeners are attached to the skin there. In photos, these often seem to get mistaken for rivets. You’re not the only one to have done this.

    David,

    The three failures have not been basically the same. The Mk1 failure was of a peripheral weld seam on the uppermost tank dome. The SN2 failure was of a weld on the “thrust puck” structure at the bottom of the vehicle. The SN3 failure wasn’t weld-related at all. It was a buckling failure that occurred when pressure in the lower tank was somehow allowed to get below ambient atmospheric pressure.

    I doubt that internal meetings got heated for two reasons:

    1) By SpaceX standards, this is not a particularly remarkable series of failures. SpaceX had a lot more failures learning to land F9 1st stages.

    2) SpaceX has almost no meetings involving more than a handful of people at a time and these are short and very tightly focused as to agenda. This is just one of many ways SpaceX differs very significantly from both NASA and its legacy contractors.

  • Edward

    Dick Eagleson wrote: “By SpaceX standards, this is not a particularly remarkable series of failures.

    I’ll add that we tend to only see the spectacular events. Successful static tests have a tendency to look like nothing happened, so there is nothing interesting to get everyone’s attention.

    Since the article says that the first two failures were tests in which loads were increased until the structure failed (test to destruction), then those two were also successful tests. I am surprised that we were not told at the time that those two tests had been intended to go to destruction.

  • Lee S

    @Dick Eagleson, Thanks for the info! It actually reassures me that it is fully welded rather than riveted. Do you have any idea of what system of welding they are using? I need at some point to do a deep dive into NASA’s friction welding process, which I can kinda get my head around the theory, but have no idea how it works in practice, only that it is very, very expensive. ( No surprise really, given its NASA)

  • Max

    Watching it several times more using roses link, You can see the crumbling start at a dimple/dent that already existed. (The weakest spot of course) what is most informative as the venting of the top tank is less noticeable than the one on the bottom. The ice formation on the top take is severe where it has not yet begun on the bottom.
    My best guess is this system was designed to pump the liquid back into the holding tanks rather than vent the tanks to atmosphere. (Most engineers would never waste so much material unnecessary)
    The unintended consequence of venting to atmosphere is it takes time for the liquid to boil. The more it boils, the colder it becomes until it freezes. Slowing down gas off.
    I know this because I had a summer job, when I was 18, of roofing houses. The large propane tank, that would heat the tar, would freeze up because we consume the gas faster than tank could warm. One of my duties entailed building a fire with 2 x 4 wood scraps under the propane tank for extra heat to maintain a vigorous flame.
    I’m sure most of you seen the experiment of dipping a hotdog in an open container of liquid nitrogen. The gas off is extremely slow because of the cold temperature. This equilibrium has prevented the upper tank from draining.
    The top tank gas vapor plume is definitely slower than the bottom tank. It is also covered in a thick layer of ice, adding more weight from the warm humid air, is evidence that it has reached a state of near frozen and has ceased to drain.
    The bottom tank, having more volume, was venting nicely with the warmth of its extra mass, but being open to atmospheric pressure, did not have enough internal pressure to maintain the stability/rigidity.

    Conclusion; my guess is that after the pressure test was concluded, someone made the mistake, when they were told to drain the tanks, of opening the drain valves to atmosphere, rather than pumping the fluid back into the storage tanks. Simple miscommunication.

Leave a Reply

Your email address will not be published. Required fields are marked *