June 22, 2023

Thoughts on the lost submersible

The news has recently been dominated by things of a nautical nature, thanks to the apparent loss of the Titan minisubmarine during a dive on a prominent wreck in the North Atlantic. I figured I should collect my thoughts (refined during discussions on the discord) and post them here.

First, a caveat. I am not an expert in deep-ocean operations, and I have no specific knowledge of OceanGate or this submersible beyond what I've read since this all started. That said, I've had some interest in this general field for a while, and have read several books on military deep-sea operations before all this blew up.

It's worth starting with a review of what we know about Titan. Most of this comes from a 2019 profile in Smithsonian. The sub's pressure hull is composed of a carbon-fiber cylinder with two titanium hemispheres on the ends. This is lighter than a conventional metal hull would have been, which in turn made it a lot easier to solve the basic problem of submarines: a fully-loaded submarine has to have the same density as water. With deep-diving submersibles, the hull is generally too heavy, and they have to attach additional flotation, something Titan was able to avoid.

That said, this is not the best use-case for carbon fiber.1 Carbon fiber's best properties are stiffness and extremely high tensile (tearing) strength, but a submarine's hull is in compression (crushing). There's also the problem that carbon fiber is a less reliable material than metal, with its properties depending heavily on exactly how it is built (it is much stronger with the grain of the fibers than across them, for instance). While there isn't any reason that you couldn't use carbon fiber for this role, there have been disquieting reports about OceanGate's practices with regards to safety and inspections. They also seem to have relied heavily on their acoustic monitoring system, which was supposed to hear fibers breaking well before any failures became catastrophic. I don't have the familiarity with carbon fiber as a material to have a strong opinion on this.2 The basic idea certainly doesn't sound crazy, but it's far from clear that it was tested and implemented well.

With that out of the way, we can turn to what we do know. Communication was lost with the submarine about 1 hour and 45 minutes into the mission, which is generally reported to have taken about 2 hours to reach the objective. This was not the first time comms had been lost during a dive, and Titan may have had only limited capability to navigate without outside help. Procedure was to wait for a while to reestablish communications, and if that couldn't be done, to drop ballast and abort. After that, we have had no definite contact with the submarine, despite extensive searches of both the sea surface and the bottom near the area.

That leaves three basic options for what happened to the submarine, and we can look at each of them in turn. First, there's the possibility of a catastrophic failure of the hull, killing everyone aboard almost instantly. Given the untested hull, this is fairly likely, and leaves a fairly large area where the remains of the submarine could be. The only external evidence would be a rather loud bang as water under hundreds of atmospheres of pressure came rushing in. I am sure that someone has asked the Navy to check the SOSUS tapes, but sound does weird things in the ocean, and the systems have been drawn down since the Cold War, so it may or may not have been captured.

The second option is that Titan is trapped on the bottom, and we need to free it before the crew runs out of air. I find this significantly harder to believe, assuming reports about the ballast system are correct. It appears that there are several options for jettisoning ballast in an emergency, including two (pipes which can be rolled off by getting everyone to the same side of the sub and a strap that would dissolve after 16 hours in the water) that do not require power. It doesn't appear that there are any large volumes which are normally buoyant but which could have flooded in an accident, as happened on Pieces III, leaving entanglement with an object on the ocean floor as the only way the submarine could be trapped intact. This seems unlikely given the general lack of obstructions on the floor of the deep ocean. The only real candidate is the wreckage they dove on, which would require the pilot to have gotten in too close after losing communications, which seems unlikely.

That leaves the possibility that they were able to surface after the loss of comms, and are drifting around the North Atlantic. I rate this one as more likely than being trapped on the bottom, for reasons described above, but less and less likely as time goes on, as we are quite good at finding things at sea. At one point, snorkel detection was a major ASW technique, and while Titan is not large, it's definitely larger than a snorkel. It is also worth pointing out that those aboard would still be trapped inside, as access comes through one of the titanium hemispheres, which is bolted on before descent. This isn't an inherently insane idea, as it removes a potential point of failure that a hatch would be, particularly in a carbon-fiber structure, but it would need to be matched with a reliable system for getting help once on the surface. Reports seem to indicate that they do not have such a system, one of many poor engineering decisions on the part of OceanGate.

A last word on the banging noises reported on the 21st. Yes, they could have come from the submarine, and the timing is suggestive of human activity. But the oceans are very noisy, and sonobuoys are not optimized to listen for things 4 km down. Particularly given the rather limited capabilities of a sonobuoy relative to a full-scale passive sonar (none of which appear to be on the scene), this shouldn't be read too strongly. Particularly in light of the Coast Guard's announcement of the discovery of a debris field, which came out just as I was finishing up this post.

A year later, this post has held up quite well. The submarine imploded just as I predicted, and the Navy was able to pull the information off the SOSUS tapes. The investigation is still ongoing, but nothing seems to have come out which contradicts what we knew at the time I wrote this.

1 For those who are confused about what it is, "carbon fiber" is the generic term for what is more accurately called "carbon fiber reinforced polymer". Pure carbon is very strong but also very brittle, so the best way to use it is to create fibers by heating other material, then weave those into a fabric, which is then covered in plastic, formed into shape and cured. This allows the plastic and fiber to reinforce each other, giving the best of both worlds.

2 I learned a bit about it in college, but my materials professor was the worst I had in my entire time in school, and I haven't had to think about it since.


  1. June 22, 2023bean said...

    Just finished watching the press conference, and it looks like things went more or less as I predicted here. They found the debris field well away from the existing wreck and close to where communication was lost, with the failure having happened while it was some distance from the bottom.

  2. June 22, 2023Neal said...

    Thank you for this post Bean. I appreciate your explanation of carbon fiber.

    While we mourn the loss of five lives, the CEO had a strong aversion to having any external inspections performed on his projects and seemed, I do stress seemed, to have had a rather cavalier attitude toward calls to review his procedures and engineering.

    Additionally, any marine chandler can provide basic emergency signaling equipment. I copied the following from one of the pilot's forums: "Dukane beacon, 2” dia., 6” length, self contained power source, can transmit for 90 days, 8.8kHz, 160db, at depths up to 20,000’/6000m. Specifically designed for locating objects underwater, at ranges up to 7-12 NM, depending on environmental factors." I found this one on the Boeing site: https://shop.boeing.com/aviation-supply/p/DK120-90=7Q

    That said, I am impressed, as I most often am, by the search efforts that can be spun up when needed. Incident occurs on a Saturday afternoon and by Thursday morning the wreckage is found.

  3. June 22, 2023Brett said...

    Wall Street Journal is now reporting that the Navy heard the implosion sound days ago from where they lost contact with the Titan.

    Wish they could have saved the Coast Guard some trouble and tipped them off sooner.

  4. June 22, 2023EdH said...

    I was wondering if SOSUS (or whatever the current incarnation is called) would detect it. It is a very small vessel after all.

    Admitting detection gives possible opponents insight into its capabilities, and so is no doubt classified, and any release of information would require a review.

    The most reasonable course of action would be to communicate the info to the senior officers of the rescue effort, and for them to have their commands to continue on, but without further imperiling lives, an exercise as it were.

  5. June 23, 2023bean said...


    That report seems to indicate that they did in fact tell the Coast Guard. I haven't looked closely enough to be sure about how many ROVs were on scene yesterday, but I don't think it was a large number. And if I was the commander and had limited ROV resources, I might check the wreckage first, just in case the detection was wrong (remember, they have no a priori way of knowing that the SOSUS hit was right and the banging was spurious) because that has a chance of having someone trapped alive, whereas checking the possibility of implosion isn't similarly time-limited.


    That seems to be what happened. A lot of it is also that this was all done under blazing publicity, and by not revealing this (and it is worth noting that they were very non-specific about what system picked this up) they gave the Coast Guard the ability to do things without lots of awkward questions.

  6. June 24, 2023Lambert said...

    I assume the effects of anisotropy and other quirks of CFRP are amplified by the fact we're talking about buckling failure. Unlike tensile failure, the maths behind buckling doesn't tend to give you nice solutions. You either run a numerical analysis or use rules of thumb full of empirically-derived magic numbers. It'd be easy to miss an assumption that is valid for solid titanium or alu but not for composites. The other thing that comes to mind is delamination. Even a small section of hull losing its ability to resist shear could lead to local buckling (like how a can wrinkles as you start to crush it) followed by catastrophic failure.

  7. June 24, 2023Doctorpat said...

    @bean: They also seem to have relied heavily on their acoustic monitoring system, which was supposed to hear fibers breaking well before any failures became catastrophic. I don’t have the familiarity with carbon fiber as a material to have a strong opinion on this.

    I do have some familiarity with carbon fibre, enough to have a medium level opinion on this.

    • Yes, a good acoustic monitoring system will give you warning of impending failure.
    • Days warning? Hours? Minutes? Seconds? That's an important question. And something much harder to determine. Especially in a buckling failure where seconds warning is certainly on the cards.
    • And it's something that would be very difficult to test. The first 3 tests give you hours warning, number 4 (real life) gives 2 seconds.
    • As mentions, oceans have all sorts of weird sound effects, which means lots of false positives on your acoustic monitoring, which means a high threshold for hitting the panic button.
    • Delamination and crack growth are your other fears, and this requires regular inspection using systems like ultrasound or xrays. And is much more difficult in carbon fibre where all the layers already look like cracks and discontinuities.
  8. June 27, 2023Bernd said...

    Do you think there'll be enough intact debris to find out how it failed? I'm assuming there isn't much publicly available info on submarine implosion investigations to go on, given how rare and classified they are.

  9. June 27, 2023bean said...

    Depends on how precise you want to be. I'm sure that there's enough debris to know if it was, say, a carbon-fiber failure vs a failure of one of the end spheres. I'm not sure they'll be able to know much more than that.

  10. June 28, 2023Doctorpat said...

    It's certainly also a possibility that a failure starts somewhere, and then as the collapse happens you get huge forces suddenly applied in all sorts of unplanned ways and now dozens of fractures occur throughout the structure.

    Which one was first? Who knows?

    There is a fair amount of experience built up about such things when it comes to air crashes and road crashes. But carbon fibre in the deep sea ie. huge external water pressure? You'll be pioneering new ground.

  11. June 28, 2023Anonymous said...

    Also what regulatory changes are likely to result?

    At a minimum I'd expect carrying paying passengers on an uncertified submarine to become just as legal as doing it with an uncertified airliner and probably even more skepticism of composites but is anything else likely to change.

  12. June 28, 2023bean said...

    I'm not sure if we'll see much in the way of regulatory changes, because this was in international waters, which makes figuring out how to stop people from doing things like this at least mildly difficult. They'll probably figure out how to restrict it more now, which I'm not sure is a good thing. But it won't be airliner levels of illegal because very few things are.

  13. June 28, 2023Garrett said...

    Regulatory changes are going to be a challenge under a common-law situation. From snippets I've read, participants were notified in-advance that the sub was experimental, that it wasn't certified, and that there was a risk of death. It wasn't passage booked on Priceline or some such thing.

    The countervailing example in the US is the whole aviation regime, where I don't think there's any amount of paperwork which would allow you to charge for a flight in an experimental aircraft. Sometimes you can do cost-sharing as a non-commercial pilot, but this gets into areas I haven't read up on much.

  14. June 28, 2023Ian Argent said...

    The thing about airliners is they take off and land in the USA. Same with cruise ships (dock and return to).

    If nothing else, the Jones Act is going to make basing something like this outside US law much more attractive ANYWAY, so...

  15. June 28, 2023Ian Argent said...

    Seen on Twitter - photos allegedly of wreckage of the Titan submersible recovered from ocean floor


  16. June 28, 2023Basil Marte said...

    Based on the number of passengers per year, the bathyscaph tourism industry needs about as much regulation as the suborbital space tourism industry. (Notable comparison is Spaceship One with an in-flight breakup killing the two pilots.) Even ignoring the fact that regulation would be premature (hindering engineering exploration), legislators' time is expensive, and regulation is costly infrastructure that the economic size of the field cannot support (benefit from). It's akin to setting up a Fordist assembly line when one wants to manufacture ten items.

  17. June 28, 2023Philistine said...

    I don't know what regulation would even be needed. One guy decreed that fetishizing safety was a waste of money and he and his company weren't going to stand for it; now that one guy has died for his trouble. And of course it's tragic that he took four other people down with him - but does anybody really think his business model is going be widely emulated in the future? Really? Does anybody think his business model even could be emulated - that the already-limited customer base for this kind of expedition would stand for it?

  18. June 29, 2023Anonymous said...

    Where could you base it from that wouldn't enact any safety regulations? If you wanted to see the Titanic you'd be limited to countries that would enact any SOLAS updates.


    Does anybody think his business model even could be emulated - that the already-limited customer base for this kind of expedition would stand for it?

    I don't think any potential customers would want to stand for it, would they have the knowledge to be able to tell a repeat is another matter.

    Because the people he took with him certainly didn't.

    Also https://oceangateexpeditions.com/ is still offering Titanic expeditions.

  19. July 02, 2023Philistine said...

    The difference between now and last month is that now we've had a spectacular example of why you don't dive with the guy whose explicit, publicly-stated mantra is "Safety Is For Suckers."

  20. July 03, 2023Anonymous said...

    Yes, but a lot of people don't exactly have long memories.

  21. July 04, 2023Fionn said...

    A long-ish form article on the submersible:


    Confirmed a bunch of rumours I had heard floating around while the rescue op was still underway.

    • Passengers were signed on as crew/"mission specialists", thus dodging all passenger regulations re: safety, classification, etc.

    • There was essentially zero non-destructive testing of the carbon fibre hull done, instead he went all-in on his unproven realtime monitoring system.

    Can only quote and weep:

    ... The meeting ended in Lochridge’s firing.

    Soon afterward, Rush asked OceanGate’s director of finance and administration whether she’d like to take over as chief submersible pilot. “It freaked me out that he would want me to be head pilot, since my background is in accounting,” she told me ... As soon as she was able to line up a new job, she quit.

    Whatever about the other passengers, who were able to exercise some degree of due diligence about the OceanGate & the sub, the 19 year got a particularly raw deal, being dragged along whether he wanted it or not. What a waste.

  22. July 05, 2023Anonymous said...


    Passengers were signed on as crew/"mission specialists", thus dodging all passenger regulations re: safety, classification, etc.

    If you pay to be on it you're a passenger.

  23. July 05, 2023Fionn said...

    If you pay to be on it you’re a passenger.

    Well from a common-sense POV you are completely correct, which is why Rush structured OceanGate with several corporate entities scattered across multiple countries so his activities would fall between the cracks and he could operate as he wished. For all we currently know the passengers transferred their fare to a corporation in one country and signed on as crew to another corporation in another country. Without a whistleblower or passengers giving details to some authority who could do a deep investigation it would be very difficult to determine what was actually going on inside OceanGate.

    I also read that this "paying to be crew" business did happen in the days of sailing ships, sometimes junior crew would pay for their passage to gain experience in order to advance their careers, and perhaps this was one of the loopholes that was used. I would imagine a court would not look very well upon this arrangement though, for obvious reasons. Perhaps somebody here knows more about this one.

  24. July 05, 2023bean said...

    I also read that this “paying to be crew” business did happen in the days of sailing ships, sometimes junior crew would pay for their passage to gain experience in order to advance their careers,

    This is sort of true. What happened was that until surprisingly late (WWII-ish) a lot of countries required sail experience to get a license as a merchant officer. As a result, it was common for people trying to get that certificate to sign on for free or even pay with the few remaining sailing ships to check that off their list. This helped keep sail viable on a few routes into the 30s, IIRC. The book The Last Grain Race talks about this and is highly regarded, although I haven't read it myself.

  25. July 07, 2023Anonymous said...

    That was also pretty obviously training which is basically the only exception to fare payers always being passengers.

Comments from SlateStarCodex:

Leave a comment

All comments are reviewed before being displayed.

Name (required):

E-mail (required, will not be published):


You can use Markdown in comments!

Enter value: Captcha