OceanGate 2024: A Forensic Analysis of the Titan Submersible Implosion - SAMPE Presentation

[@hnrwagner]

Times codes: 0:00 - Intro 4:45 - Titan Digital Twin 7:38 - Manufacturing of the Titanium Aft Dome 8:50 - Analysis of the Titanium Hemisphere 14:02 - Manufacturing of the Titanium Front Dome 15:20 - The CFRP Cylinder and its manufacturing 17:48 - Cross-section details of the CFRP 18:59 - Analysis of the CFRP Cylinder 19:33 - CAD model of the TITAN 20:09 - Analysis of the CFRP Cylinder 2 21:52 - Design of the CFRP Cylinder 2 27:10 - How the TITAN looks like according to ASME Design Code 27:55 - This TITAN sub would never implode ;) 28:30 - Summary of Dives to the Titanic 29:15 - Fatigue failure estimate 30:36 - TITAN debris underwater cam 32:18 - Pressure gradient vs. Human Body 33:40 - How much energy the Implosion released 35:20 - Acoustic Sensors 37:19 - NTSB Reports 39:40 - Interface Rings to CFRP Cylinder 41:45 - Implosion Simulation 42:38 - Interface Ring failure 44:00 - Q&A Session 46:00 - Realistic implosion simulation with Imperfections 53:31 - Strain gages Analysis 55:30 - More questions regarding imperfections 58:31 - Q&A Session end 1:00:41 - updated Implosion Simulation - CFRP Cylinder Fracture 1:00:41 - updated Implosion Simulation - Acrylic Viewport Failure 1:00:41 - updated Implosion Simulation - Adhesive Interface Failure

[@PetesGuide]

Please pin this comment so that it is always the top comment!

[@vandecasa3795]

Maybe you should do a forensic analysis of your audio equipment.

[@hnrwagner]

It's a good enough

[@PetesGuide]

@@hnrwagner No, there is too much room reverberation. This is close to an echo and makes it difficult to understand you.

[@johnneyland3334]

Great Video and presentation analysis ! Thank you ! J

[@hnrwagner]

Thanks

[@billsteele495]

They used the wrong adhesive. Great work here ! 👍

[@hnrwagner]

or they used adhesive at all :)

[@ForensicCats]

That was interesting, thank you for sharing

[@hnrwagner]

Thanks, sry it was probably nothing new for you, more of a summary, I really like your comprehensive videos on the topic

[@ForensicCats]

@@hnrwagner thank you, and I just sent you the current link with new data from Coast Guard

[@Reach41]

Very interesting. As stated in the Q and A period, where the ends of the shell mated (by bonding) with the end hemispheres, there is a combination load that must be accounted for. In the center of the shell, hoop stress should be the driving load for design, but at either end a bending load is introduced when the shell and end caps don't strain identically. The longitudinal fibers pick up this load. In the design of the shell, hoop stress should be evaluated against compression allowables, and checked for buckling. Only the circumferential fibers can be used in those calculations. Buckling under hoop stress would be evaluated by assuming that the successive circumferential plies are all concentric. However, they are not concentric after the first few plies, they are budged outward, and some of those plies were actually sanded off during fabrication. Therefore, in the as-build condition, structural calculations would have to include post buckling analysis, with knockdowns inluded for the prosity, resin starved conditions, disbonds and contamination in the layup. Watching the video where the end rings were bonded to the shell made me want to pull my hair out. But, that said, I suspect that the sheared flanges on the rings were secondary to the implosion, which is consistent with your FE illustration. I agree with you that the strain gages warned the operators that the shell was degrading, which they rationalized away. They would have been of no use in alerting the operator that an implosion was imminent. I'm suggesting that the shell failure was due to buckling. When the shell wall started to buckle, total collapse would have progressed instantaneously. I haven't heard anyone suggest that "cracks" in the composite might have allowed water to seep in, which, upon expansion due to freezing, would have extended the defect with eact freeze/thaw cycle. With the state of the debris recovered being so poor, I'm not sure the NTSB guys would be able to find evidence of it. Love your work. Thanks.

[@hnrwagner]

Great comment thanks

[@hnrwagner]

I would love to know how the proper interface of the titanium rings with the cfrp cylinder would look like, unfortunately I have no expertise in this area

[@nian60]

I assume it will be impossible to exactly replicate the Titan implosion due to the fibers being sanded down, and thereby changing the strength properties of the carbon fiber hull. We can't know in hindsight exactly where those fibers were sanded down, and exactly by how much. And then the resin or glue was apparently unevenly applied by hand, which will also be impossible to replicate. It's like trying to replicate a kid's carved bark boat. The piece of bark will never be exactly the same, and the carving by hand will never be exactly the same.

[@hnrwagner]

It may be impossible to replicate exactly but it can be approximated good enough

[@nian60]

Interesting analysis. I don't know if you said it but apparently the titanium domes were downgraded from titanium grade 5 to titanium grade 3. Stockton Rush did that to save money.

[@hnrwagner]

I will check the consequences of it, thanks for the advice

[@kouider76]

Very Interesting, is it possible to share the CAD or at least the datasheet you used to construct the model as I am intending to simulate it in Ansys and why not compare the results with the one you got

[@hnrwagner]

I will upload the step file , would that be good enough?

[@kouider76]

@@hnrwagner yeah enough as all the data regarding the composite layup you have already explained and discussed in your presentation. Many thanks 🙏

[@imhusker]

The fiber placed hull looks better but is it “obviously” better quality. The fiber is placed without tension. It is autoclave cured 5 times thru the thickness. During the autoclave cycle the thickness is compacted, reducing the diameter and causing wrinkles.The wrinkles on the outside of the hull are numerous as is seen in a picture in this video. Recently it was shown the wrinkles go deep into each 1” layer. The wrinkles were ground smooth before the next layer, cutting several plies in the process. This not only reduces the compressive strength, it severely affects the local stiffness. Whereas the geometric middle surface is nearly exactly cylindrical, the local stiffness deviations causes the stiffness middle surface to be wavy in the wrinkle regions, which are very numerous. So the buckling capability will be reduced.I would have scrapped this hull after the first autoclave cure. Without the autoclave, it would have looked like Swiss cheese. The filament wound hull is wound with tension in the hoop direction and develops about 1 psi interface pressure with the mandrel per ply. That adds up to 400 psi, but won’t be that high due to compaction. If the mandrel is stiff enough, the positive hoop tension will be maintained to the end. They don’t autoclave composite pressure vessels. I saw someone try it and it did just what this one did and failed prematurely. I’ll put my money on this one to out perform the pretty one. There’s not much difference in the hand and machine placed axial plies.

[@hnrwagner]

It's crazy that they cut those layers of and didn't repeat the manufacturing, that is unbelievable negligent

[@imhusker]

It would be interesting to model the wrinkles in your buckling model. Geometrically, the hull was made on a nearly perfect cylinder, so the hull deviates very little from a circular cylinder. The thickness is also constant. The wrinkles could significantly warp the structural middle surface and degrade the buckling load. I don’t think even delams would have much affect because of high inter-laminar compression and friction.

[@hnrwagner]

@@imhusker I am on it and try to model it

[@baronhelmut2701]

I love how you recognize the crash site, but the failure mode from the animation doesnt match it at all. The window didnt collapse inward. It got thrown outwards by the hull tearing on its join to the front titanium ring.

[@hnrwagner]

The behavior you described can't be simulated by me at the moment but I will try in the future, thanks for suggesting

[@imhusker]

At 54:20 you show depth/ strain plots but the strain is mV. Is the gage factor known to convert to actual strain. The change is about 700 mV from 0 to 3800 m. By my calcs, the cylinder hoop strain should be .002 to .0025.

[@nian60]

It might be a good idea to put the language of the video to English. KZbin's auto translation thinks the language is German, so the subtitles are not working.

[@hnrwagner]

I will

[@wackyvorlon]

What do you think of the loud bang that was heard near the end of dive 80?

[@hnrwagner]

I will look into it

[@Matthew-z9c]

Have you modelled this with a solid laminate to account for transverse shear effects, given this is not a thin-walled structure?

[@hnrwagner]

Yes, I will upload the step file to github later this day, then you can check it out

[@Matthew-z9c]

@@hnrwagner Sorry, I didn't mean the CAD as a solid, I meant the laminate definition, i.e, is it modelled with solid or shell elements? Shell elements can't account for transverse deformation/shear with Kirchoff-Love, and given the thickness of the cylinder, these will have been significant. Typically, in Ansys, I would model this using either solsh190 elements to account for Mindlin-Reissner, or with solid elements with full 3d elasticity.

[@ForensicCats]

@dr-ing, Ronald, can you please run your numbers with just the hemispheres without the interface rings? My thinking is, the interface rings and bolted connection gave rigidity to each other ... What would each behavior like if the bolts were not a factor? - would they oval in shape??? For context, in one of the reports, they mentioned "ovaling" and I think this is more yiur wheelhouse than mine... Could each section withstand the psi without the bolted connectors ... I have considered the "cylinder-piston pressure from the ends, trying to buckling the carbon fiber hull and therefore the bolts are "useless "... But, ovaling is mention in the report and i am thinking, some point during the implosion (the carbon fiber buckled) and this sudden "release of pressure " created a huge "over pressure " and the interface rings and the end domes "ovaled (during that event)... Because the report identifies the "oval", this means the section that ovaled is now plastic (revealed by the fact that it is in the report as oval shape)...

[@hnrwagner]

You mean the titan analysis but without interface rings? So only hemisphere and composite cylinder?

[@ForensicCats]

@@hnrwagner allow me to make a video reply on my channel and that way people can see where to come for the answer and hopefully subscribe to your channel also. I have to find the section in the report where it describes ''oval" and also discuss my position for and against this in reference to before or after the implosion... I have a mechanical engineer who I will mention that is leaning towards before... I am not leaning that direction and simply, I think the bolts and interface would have failed many trips prior (unless they have more data to the contrary - for example your modeling)... Again, I will do a video within a few minutes and post in about two hours. Thank you and goodnight (Germany time now is around 9:30p.m.)

[@ForensicCats]

@@hnrwagner I made a video response and shout-out to your channel ... kzbin.info/www/bejne/aXuUaHiQlMuUmpY (keep in mind, the example at the end is only ''modeling'' the Rhino liner ...

[@uncoolmartin460]

Interesting. I wonder about your hexagonal structure and if it was inside of a cfrp cylinder, providing some reinforcing of the cylinder, how it might change the modelling. Keeping the internal volume the same as the Titan, with the same wall thickness of the cfrp cylinder or better. My intuition says it should help considerably, but I realise intuition can be wrong.

[@hnrwagner]

The hexagonal sub design stands on its on, it is mo support structure for the cfrp

[@uncoolmartin460]

@@hnrwagner Ok, I get that, sorry I didn't mean to demean or belittle your design. It was just a suggestion, If I could do the modelling myself I wouldn't have bothered you.

[@metal--babble346]

the CEO of Oceangate refused to build a successful test model. Regardless of known dangers, Oceangate built the full scale vehicle, and immediately began selling expensive tickets for an "Explorer's Club" adventure.

[@hnrwagner]

Greed before safety

[@ForensicCats]

Dr-Ing, Ronald, here is a video that I hope you take interest in... ''snap buckling" kzbin.info/www/bejne/h367ep53eLCngrs and in my private room, I have a model (I will release it after our elections, November 6 or so). But, I think you can work with this video... The report I used is from fiberglass and is from the 70's

[@NielNBob]

youtube hid this from me til today. god will punish......

[@hnrwagner]

we need more interaction for the algorithm to push the video !

[@chris11980]

Most german english ever heard.

[@reginawagnerbiolife7867]

Do your ears bleed?

[@chris11980]

@@reginawagnerbiolife7867 Ja verdammt, meine Ohren bluten.

[@hnrwagner]

It is not that bad :P

[@laibachsdoku-kanal3330]

Könnten Sie Knallkopp bitte Deutsch reden?

[@hnrwagner]

Immerhin gesitzt

[@laibachsdoku-kanal3330]

@@hnrwagner Immerhin!