Technical Analysis on Titan Sub (Titanic Sub)

[JT Smith]

With regard to the self-limiting nature of rich thrill-seekers dropping a quarter million on a death ride: for sure that's true. But if there is strong evidence of deception, negligence, and/or violations of law I don't think it should matter. Fairness and justice shouldn't be about numbers.

 

[Vanadium 50]

Uranium would be best. You can find it by the nice blue glow.

I don't know what terminal velocity is in water - I'd guess 10 m/s for weights and 5 m/s for the vessel. So at 500m the weights would hit the bottom around a minute sooner. Plenty of time to introduce uncertainty and confusion.

 

[gmax137]

One thing that surprises me is the lack of telemetry involved. Add an optical fiber to the umbilical and you have all the bandwidth you need - a text every 15 minutes?

Why does that surprise you? There's no question that this project was done on a budget.

This is the vessel built with expired materials.
(WaPo, "Titan CEO spoke of ‘discount’ parts, journalist invited on submersible says")

Weissmann said Rush told him how he had gotten the carbon fiber used to make the Titan “at a big discount from Boeing.” Weissmann wrote in Travel Weekly that Rush said he was able to get the carbon fiber at a good rate “because it was past its shelf life for use in airplanes.”

It is controlled with a Nintendo console (BBC News, Titan sub: Cramped vessel is operated by video game controller). There may be nothing wrong with this per se, but I see Nintendo consoles at yard sales for 10 bucks.

There have been questions about the thickness of the CF (https://www.cnn.com/2023/06/21/us/titan-sub-safety-oceangate-employees/index.html)

The former employee became concerned when the carbon fiber hull of the Titan arrived, he said, echoing Lochridge’s concerns about its thickness and adhesion in his conversation with CNN. The hull had only been built to five inches thick, he said, telling CNN company engineers told him they had expected it to be seven inches thick.

 

[snorkack]

The outer shell was in compression. Was the initial failure mode likely to be buckling, or shear crack?

 

[hutchphd]

This material is anisotropic, not rigorously uniform, and subject to local hysteresis. I don't even know if the question makes any sense. The answer is don't get aboard.

 

[Astronuc]

The outer shell was in compression.

Until it wasn't. When a circular tube starts to creep down in one direction, on the perpendicular diameter, which is creeping out, the wall is bending, and the outer surface eventually goes into tension. If there is a flaw at that location, the stress concentrations are huge > 3x. The outer portions of the membrane pivot on the inner portions.

From experience, 2% ovality is about the limit before creep collapse rate starts to take off (increases rapidly) and time to failure decreases rapidly.

I've seen some examples and calculation, and I've some calculations/analyses myself.

 

[Vanadium 50]

Why does that surprise you? There's no question that this project was done on a budget.

Because there's money to be made by understanding how well the vessel performs.

The numbers that have been given are 2 hours down, 2 hours looking around and 2 hours going up, making in total 8. (I know...) If better understanding of the performance let you cut that to 4, you could do 2 dives in a day. Place the guide at the surface, and you can fill 4 seats instead of 3. And so on.

And it's no expensive - a few thousand dollars.

Nintendo console

I've seen this. My reaction is "so what?" It's not the part that failed. Lots of high-tech projects use low-tech parts. One of the Fermilab accelerators uses magnets from automatic car windows. A popular cable for experiments is ethernet, and HDMI is starting to gain popularity. One LHC experiment is shimmed with US dimes. One Fermilab experiment has detectors registered with tongue depressors.

There's plenty to complain about, but IMO, "Nintendo" is not one of them.

 

[Vanadium 50]

Until it wasn't.

An experiment any 10 year old can do. (Kept me amused when I was 10)

Take an empty can of Coke. Stand it up on its bottom and stand on the top of the can with one leg. It supports your weight fine. Now with your free foot genrly tap the side. The can buckles and collapses to a flat mess.

 

[gmax137]

Because there's money to be made by understanding how well the vessel performs.

The numbers that have been given are 2 hours down, 2 hours looking around and 2 hours going up, making in total 8. (I know...) If better understanding of the performance let you cut that to 4, you could do 2 dives in a day. Place the guide at the surface, and you can fill 4 seats instead of 3. Amd so on.

These are good points. Apparently too subtle for the business plan.

There's plenty to complain about, but IMO, "Nintendo" is notg one of them.

Yes, I agree. I almost didn't mention it, a cheap shot.

 

[snorkack]

Until it wasn't. When a circular tube starts to creep down in one direction, on the perpendicular diameter, which is creeping out, the wall is bending, and the outer surface eventually goes into tension. If there is a flaw at that location, the stress concentrations are huge > 3x. The outer portions of the membrane pivot on the inner portions.

From experience, 2% ovality is about the limit before creep collapse rate starts to take off (increases rapidly) and time to failure decreases rapidly.

But it is not yet in milliseconds at that point, because in that case it wouldn´t be "creep" collapse.

A pressure vessel could in general fail in several ways:
It might spring a limited leak and fill through that leak, with no collapse. But for a submarine, it would still be just as deadly! The wreck of Kursk could not surface, and water leaking in over several hours still killed everybody. Obviously, you don´t want weak spots in the pressure vessel, so you reinforce them... until the vessel fails in a random spot elsewhere than at seams.
Or it might buckle and undergo a purely elastic collapse. But this would still be just as deadly... if you could build a submarine metal shell that simply pops back to shape when lifted out of depth (no metal is actually that tough), it would be pointless because people would still be already dead inside.
I understand that the strongest vessels generally fail by implosion - they crack and shatter into a few large pieces. This means that a failure does propagate across the vessel once started, rather than stopping as a limited leak (and flooding the ship without imploding).
But how are the starting points of a future implosion detected and identified before actual implosion? What kind of flaw monitoring was it that the Titan operators should have done and neglected to do, or which could have been done on the conventional steel submarines (which have "test depth" and "collapse depth"), but which were somehow unworkable on titanium and carbon fibre submarine?

 

[Vanadium 50]

a cheap shot.

But it's not yours, it;s the media's.

I'm also curious about "expired" carbon fiber. Carbon fiber is not like tomatoes. Boeing has 12 year old 787s out there, and that carbon fiber is no youger than the plane. P can more easily believe "we thought we would use fiber type X but we switched to Y" or even "this is what's leftover on the spool when we wind a fuselage". At least more than going stale.

 

[Astronuc]

An experiment any 10 year old can do. (Kept me amused when I was 10)

Take an empty can of Coke. Stand it up on its bottom and stand on the top of the can with one leg. It supports your weight fine. Now with your free foot genrly tap the side. The can buckles and collapses to a flat mess.

I've done that too.

 

[DaveC426913]

The weights could be lead or depleted U

Lead is cheap. I don't see a lot of rationale for using anything denser, such as depleted U, since you don't have any constraint on space on the outside of the craft.

 

[gmax137]

I'm also curious about "expired" carbon fiber.

I read on another forum (for machinists) that the carbon fiber comes with a "sizing" to help the bonding between fiber and epoxy, and that this sizing is what has a shelf life. Whether this is accurate, I can't say.

 

[Astronuc]

purely elastic collapse

Collapse implies brittle fracture or permanent deformation - not elastic.

Note that submarines do not dive to the depth of the Titanic. The collapse depth of submarines is not in the public domain. The submarine Thresher apparently collapsed at about 2400 ft, but as far as I know, that is not official. Following the Thresher, new guidelines were issued.
https://www.usni.org/magazines/naval-history-magazine/2023/april/what-killed-thresher

In the case of the Thresher, the rear part telescoped into the forward section.

Buckling due to creep collapse starts as creep (below yield). As soon as the local stress state hits yield, we call that flow. As stress increases, flow rate increases. For brittle materials, it stays nearly elastic until rapid fracture, which occurs at about 1/3 the speed of sound in a solid (at least for metals). So a pressurized pipe (as in a pipeline) can propagate a fracture faster than it can depressurize. In metals the speed of sound is several 1000s m/s depending on the material.
https://www.olympus-ims.com/en/ndt-tutorials/thickness-gauge/appendices-velocities/

Results of tests on multiple samples indicate sound speeds of 10.763 km/s along the fibers (0°) and 3.042 km/s transverse to the fibers (90°).
Ref: https://www.osti.gov/servlets/purl/1117281

Supposedly, Titan had some acoustic monitoring system to detect crack (initiation?). Crack propagation at 3 km/s doesn't leave much time to response (i.e., ms). Brittle materials fail without much warning, which is why we prefer ductile materials - at least one gets some deformation in the elastic range. If Titan's systems warned them, it was too late, IMO.

 

[Astronuc]

Rapid collapse, but ductile (just 1 atm, or 1 bar)

Slow collapse, and still ductile ( slow decrease to <1 atm, or <11.01325 bar)

https://www.progressiverailroading....tank-car-legend-on-MythBusters-TV-show--47621Consider with 350 or 380 atm differential pressure.

 

[gmax137]

Here's an interesting article from 2017, Composites World magazine.
https://www.compositesworld.com/articles/composite-submersibles-under-pressure-in-deep-deep-waters

The magazine goes into details of the OceanGate "Cyclops 2" submersible, it isn't clear whether this is the Titan vessel.

The rationale for the carbon fiber design is identified:

OceanGate CEO Stockton Rush says the company had been evaluating the potential of using a carbon fiber composite hull since 2010, primarily because it permits creation of a pressure vessel that is naturally buoyant and, therefore, would enable OceanGate to forgo the use — and the significant expense — of syntactic foam on its exterior.

There's info in the article that makes me question the "out of date" CF and the 7 vs. 5 inch stories.

Spencer opted for a layup strategy that combines alternating placement of prepreg carbon fiber/epoxy unidirectional fabrics in the axial direction, with wet winding of carbon fiber/epoxy in the hoop direction, for a total of 480 plies. The carbon fiber is standard-modulus Grafil 37-800 (30K tow), supplied by Mitsubishi Chemical Carbon Fiber & Composites Inc. (Irvine, CA, US). Prepreg was supplied by Irvine-based Newport Composites, now part of Mitsubishi Chemical Carbon Fiber & Composites Inc. The wet-winding epoxy is Epon Resin 682 from Hexion Inc. (Columbus, OH, US). The curing agent is Lindride LS-81K frLindau Chemicals Inc.cals (Columbia, SC, US).

Initial design work indicated that the hull, to be rated for 4,000m depth with a 2.25 safety factor, should be 114 mm thick or 4.5 inches, which OceanGate opted to round up to 5 inches (127 mm) to build in an additional safety margin.

 

[snorkack]

Collapse implies brittle fracture or permanent deformation - not elastic.

Consider something like a plastic drinking straw failing in excessive suction/external pressure, with the result that it is no longer round - it is compressed to a flat strip with only open parts at the edges of the strip where the bending rigidity stops it from folding against itself? It is elastic deformation (remove suction and the straw reopens), but do you have another term for it besides "collapse"?

Supposedly, Titan had some acoustic monitoring system to detect crack (initiation?). Crack propagation at 3 km/s doesn't leave much time to response (i.e., ms). Brittle materials fail without much warning, which is why we prefer ductile materials - at least one gets some deformation in the elastic range. If Titan's systems warned them, it was too late, IMO.

True, responding to final freely propagating crack is too late.
Do brittle materials have deformation through formation of stopped cracks? Cracks which propagate only a small distance to obstructions, cause deformation, redistribution of loads, generate acoustic effects (both in creation and while sitting there as cracks) - but do not initially propagate through, nor connect to leak water through? Until there is enough pre-failure brittle deformation accumulated that the cracks propagate by linking up to stopped cracks?

 

[Vanadium 50]

Note that submarines do not dive to the depth of the Titanic.

That anybody admits to, anyway.

Seriously,, you can look at a submarine and see that it's not going to go anywhere near that deep. I don't know what the deepest diving military submarine is - if you disregard the one-offs like the Papa and the Dolphin, probably the Oscars. But there's not a whole lot of benefit to going deeper and deeper and deeper. And what benefit there is can be countered by building more, cheaper subs.

There's am old Navy joke about going too deep.

Old salty chief: "Son, if this boat exceeds its maximum depth, what will be going through your mind?"
New sailor: "Gosh, Chief, I don't know."
Old salty chief: "The hull."

I can't speak to other navies, but the US Navy takes sub safety very, very seriously. It only takes one slip up.

 

[Astronuc]

Initial design work indicated that the hull, to be rated for 4,000m depth with a 2.25 safety factor, should be 114 mm thick or 4.5 inches, which OceanGate opted to round up to 5 inches (127 mm) to build in an additional safety margin.

My guess is that folks will be rethinking that. Actually, that might apply to the first dive. Maybe the subsequent dive is 0.99-0.97 of that, and each subsequent dive would be 0.99-097 of that from the previous dive; and maybe it's nonlinear, e.g., 0.99*0.98*0.97*. . . . And that may be from pristine material.
One could put numbers in a spreadsheet and test different scenarios for 24 or 25 successive dives.

Environmental degradation and phenomena like cyclic fatigue must be considered.

They could have taken a scale model built to their design requirements and run it up and down, and measured after each cycle. It might take a month to do that - and it does cost money.

But if one is putting passengers in such an experimental vessel and exposing it to extreme conditions, then one needs to assure vessel integrity to guarantee 'safety' to some extent.

It's still not clear to me how they designed for 4000 m, nor is it clear how OceanGate validated the design. Hopefully, the investigations will shed light on what was done, and what wasn't.

I noticed that not only have the OceanGate and OceanGate Expeditions been deactivated, but Spencer Composites removed their page for Marine applications.

From a Google search

Titan Submersible
https://oceangate.com › our-subs › titan-submersible

Titan is a Cyclops-class manned submersible designed to take five people to depths of 4,000 meters (13,123 feet) for site survey and inspection, research and ...

Well clearly, the catastrophic failure would call into question their claim about 'designed' . . . 'to depths of 4,000 meters (13,123 feet)'.

 

[russ_watters]

And if you ban this, do you also ban space tourism?

I don't see that anyone suggested banning it (at least before you brought it up -- I didn't read through all the rest of the posts). But no, I wouldn't ban either. Nor would I allow a free-for-all if possible. The fact it happened international waters and the perpetrator died in it may make this particular incident impossible to prosecute, but it sure looks like murder to me. And there's no jurisdictional issue for criticism.

Which one did you have in mind?

The lack of safety factor is the eye-popping one to me. Having an appropriate safety factor (which is generally higher for larger risks/life safety) is a pretty basic engineering principle. Heck, I design air conditioning systems with safety factors and the main risk there is someone might be a touch warm for a couple of hours on the hottest day every couple of years (also, money). Life safety engineering scares the hell out of me as it is.

More specific and not necessarily inviolable: spheres are stronger than cylinders.

Similar: you can't push a rope (that might be Navy as much as engineering....).

Related but bigger than just engineering: if everyone else does something successfully one way, you should consider why and be careful doing it another way. This applies to both being cylindrical and using carbon fiber.

This operation had a tech-bro hubris feel to it to me. Tech-bros don't care about things going badly because there's no personal risk. It's just other peoples' money, and even the ones who eventually go to jail don't seem to see it coming (nor did this one see death coming).

I don't believe diving to the Titanic can be done safely - and I am defining "safely" as a 99% survival rate. There seem to have been about fifty manned dives, and one failure.

Several things I take issue with here:

1. I don't think it's correct to use the claimed outlier in the risk analysis. In other words, it sounds like you are saying that prior to this incident nobody had died visiting the Titanic. That sample size is small though.
2. Because of #1 I don't believe the occupants of the sub expected that they were undertaking about the riskiest adventure trip there is. 99% survival rate is on par with attempting Mt Everest or space travel. Not Blue Origin either -- actual space travel.
3. Nor do I think visiting the Titanic inherently is as risky. Pressure is what I'd call a "passive" risk. It's always there and it's big, but it never changes. It's always exactly the same. And because of that, it's relatively simple/straightforward; once you've engineered for it, there's not much else to do. Space travel on the other hand involves massive complexity. Everest involves massive uncertainty/variability.
4. Stories are coming out of the woodwork now about people in the know who raised alarms. This particular outfit was known to be much, much more dangerous than the baseline risk.

Late edit:

Now, if you want to argue it should be safER, OK, how much safer? Factor of 2? Factor of 10? And how do you verify this number?

That's not how it works. Safety factor is an exact percentage of a design rating, not a failure/accident probability. They vary based on things like risk and predictability:
https://en.wikipedia.org/wiki/Factor_of_safety

I can't say if it should be 25% or 100% here, but note that the decision on the value includes mitigation; lower safety factors require more testing, quality control, etc. to provide "safety" in a tighter design tolerance. The point is: if it isn't a cookie-cutter scenario they require a detailed risk analysis to establish the value. And the value is never zero because that basically guarantees failure.

 

[Astronuc]

Consider something like a plastic drinking straw failing in excessive suction/external pressure, with the result that it is no longer round - it is compressed to a flat strip with only open parts at the edges of the strip where the bending rigidity stops it from folding against itself? It is elastic deformation (remove suction and the straw reopens), but do you have another term for it besides "collapse"?

One could do a test. I'm not sure I can pull a complete (or near vacuum) with my mouth, but if one has a vacuum pump that might work. With what I could draw, there is some permanent deformation and the straw cross section is oval with a sharp profile where bending was the greatest. But that's a thin wall plastic, not a metal or brittle composite.

Pinch the straw between one's figures as see a difference. Then repeat several time.

One could also freeze the straw and perform the same experiment.

Do brittle materials have deformation through formation of stopped cracks?

It depends on the bulk and local stress states and microstructure. It's possible a crack once initiated becomes arrested - depending on the stress level.

Cracks which propagate only a small distance to obstructions, cause deformation, redistribution of loads, generate acoustic effects (both in creation and while sitting there as cracks) - but do not initially propagate through, nor connect to leak water through?

In brittle materials, the deformation is very little. I don't know about carbon fiber composites, since I have not researched them.

Until there is enough pre-failure brittle deformation accumulated that the cracks propagate by linking up to stopped cracks?

In the case of Titan, there may have been little evidence externally (by sight) that any cracking had occurred. Apparently, there was not testing performed. I'm still trying to understand the titanium-CFC combination.

Some information on the design of Titan's hull.
https://web.archive.org/web/2021080...bmersibles-under-pressure-in-deep-deep-waters
I haven't verified.

The Wikipedia article (unverified information) has the following:

The entire pressure vessel consisted of two titanium hemispheres with matching titanium interface rings bonded to the 142 cm (56 in) internal diameter, 2.4-metre-long (7.9 ft) carbon fibre-wound cylinder.

I don't know how reliable the information is.
https://en.wikipedia.org/wiki/2023_Titan_submersible_incident

I do not find the hull dimensions in the cited reference [12].
https://www.autoevolution.com/news/...g-on-its-way-to-the-titanic-wreck-216772.html
https://web.archive.org/web/2023062...g-on-its-way-to-the-titanic-wreck-216772.html

I would not trust Wikipedia with respect to information concerning the specifics of Titan. Official data seems to have disappeared from public view. We will have to wait for the published investigation reports.

OceanGate has removed their material. The following is archived.
https://web.archive.org/web/20230619233914/https://oceangate.com/our-subs/titan-submersible.html

 

[Vanadium 50]

Maybe the subsequent dive is 0.99-0.97 of that,

Maybe.

I have very little experience with pressure vessels, mostly watch other people buy them. (Had an application where one would be filled with lead perchlorate - you don't want that leaking!). I do however, know a little bit about superconducting magnets and magnetic forces are like a pressure (outward, not inward).

Some magnets "train" - i.e. they become stronger with use. The phenomenon is not well-understood. Others degrade over time, but eventually hit a plateau. And rarely you get a dog that just gets worse and worse.

As far as scale model testing, the problem is that various effects scale differently. Taking the magnet as an example, if I scale it up by x, the pressure is constant, but the pressure per unit mass is not. If you are driven by material flaws a smaller device will perform ,uch better than a full-sized one. And so on.

FWIW, I think cycling fatigue was the culprit (I may have been the first to have brought it up) but probably at the junction between bell and cylinder. I also suspect a cascading failure - a bad spot developed, the loads transferred away from the spot causing other bad spots to develop until the whole thing "unzipped".

 

[DaveC426913]

So ... they would very likely have survived if he had built and used a new vessel instead of recycling a used one to save time and/or money.

...is what I'm hearin'...

 

[russ_watters]

In the case of Titan, there may have been little evidence externally (by sight) that any cracking had occurred. Apparently, there was not testing performed.

There was obvious (if qualitative and not visual) evidence: loud cracking noises.

During a trip on board the Titan off the coast of the Bahamas in April 2019, Karl Stanley, an expert in submersibles, knew immediately that something was off: He heard a cracking noise that got only louder over the two hours it took for the submersible to plunge more than 12,000 feet.

https://www.nytimes.com/2023/06/23/us/titan-safety-warnings-titanic.html

They were hearing the vessel destroy itself with every dive!

 

[russ_watters]

To expand - I understand the "there outta be a law" reaction to tragedy. But how much in the way of resources should we spend saving stupid rich people from themselves?

Is that all about regulation or is there an implication for search and rescue too? Because SAR can't discriminate on the basis of wealth or stupidity. That would not be morally acceptable even if we dislike the people we abandon.

 

[russ_watters]

Alvin Dive Statistics
https://www.whoi.edu/what-we-do/explore/underwater-vehicles/hov-alvin/dive-statistics/

I was hoping to find how many dives they have performed at 4000 m and deeper.

I'm not sure how much depth per dive matters. I'm a bit thin on this for metals much less composites, but I don't think there's a "no damage" threshold for fatigue. I feel like it would be quadratic with depth; 5,000 dives at an average of 2,000m = 1,250 at 4,000m (this would align with the noise report I quoted/linked). Really impressive career either way.

 

[Astronuc]

There was obvious (if qualitative and not visual) evidence: loud cracking noises.

https://www.nytimes.com/2023/06/23/us/titan-safety-warnings-titanic.html

They were hearing the vessel destroy itself with every dive!

I had heard that there were sounds, but I didn't know the details.

Looking at one comment in that article:

In the April 2019 email to Mr. Rush, Mr. Stanley said the loud cracking sounds that they had heard during their dive “sounded like a flaw/defect in one area being acted on by the tremendous pressures and being crushed/damaged.” He wrote that the loud, cracking noise signaled there was “an area of the hull that is breaking down.”

Apparently, after that dive or season, "he [Rush] made some changes to the Titan, including building a new hull, and called off the planned dives for that year." So it was a new hull (new materials?, new design?, ??)

Another statement in the article:

Saltwater that had been trapped in between different materials in the vessel from dives in 2021 and 2022 worked its way through fibers and softened it up, making it more susceptible to a leak, experts said.

It occurred to me regarding Titanium and Carbon mating could be problematic in seawater. Carbon is more noble than Ti, so the Ti could begin to corrode, and possibly, the carbon composite reacts with elements in seawater, including dissolved oxygen. AND, if the infiltration occurred at depth at 370-380 atm, could the seawater expand at 1 atm?

Lots of aspects to investigate. This will be one of the classic cases in failure/forensic analysis - unfortunately.

 

[russ_watters]

I think this is the crux here. The passengers in the sub were technically trained people capable of decision (maybe not the 19 yr old?).

I'm seeing the CEO and and a deep sea diver/explorer, but the other three were a kid and two businessmen with no relevant expertise beyond "adult". So one perpetrator, one accomplice and three victims.

That being said I believe that any expectaton of extraordinary emergency response should be predicated upon some robust and sufficient pre-certification, and that should be a societal norm. If you want to just hang your hindquarters way over the edge then go for it, but don't expect me to help pay the piper.

And how many bodies are adrift in the Mediterranean today?

Ehh?? Were those bodies in the Med previously on certified craft?

Besides being incorrect/illogical, such emergency response priority grading is wholly immoral. Western societies, at least, do not operate that way.

 

[hutchphd]

Besides being incorrect/illogical, such emergency response priority grading is wholly immoral. Western societies, at least, do not operate that way.

I think I was unclear. The priority I was referencing has taken place. Witness the resources expended to save the refugees off Greece (~Nada) relative to the huge response for 5 people who should have known better and were under no duress to be in that foolish submersible.

I'm seeing the CEO and and a deep sea diver/explorer, but the other three were a kid and two businessmen with no relevant expertise beyond "adult". So one perpetrator, one accomplice and three victims.

Unless there was true malfeasence, every one of them had the resources to seek out and the intelligence to recognize the need for expert analysis. This craft had no certification or record. They were not victims, they were gamblers.