Submersible Implosion - fiber-composite issue

[DaveW]

The main weakness in fiber composites is that they do not do well in compression along the fibers. Any load or condition sufficient to buckle (severely bend) the fibers can lead to failure. This is much more the case with "brittle" fibers such as Kevlar or carbon - Nylon, polyester, etc. tolerate this better. When these brittle fibers bend sharply, they lose much of their strength, and when a large enough % of the fibers are damaged, the composite can fail.

This phenomenon was evident in analyzing air spring failures when I worked at Bridgestone/Firestone. That's why tires that have been run severely under-inflated totally come apart and the tread section can completely come off. This also applies in racing to CF tubs or suspension components, and that's why they can eventually lose stiffness or break.

So, IMO, too many severe compression cycles on the sub primed it for the resulting implosion failure.

[Pi_guy]

I remember some arguments when we were developing a car. Several of us complained that 1/4 inch bolt needed to be upsized it was failing too many times.

The response from the designer-engineer was the numbers show that is the proper size, I don't care what crew chiefs have to say they are not engineers.

Further down the time line the bolts were upgraded to larger diameters by the designer.


Sometimes you wonder if the choices are made by engineering decisions or ego?

[E1pix]

What a sad deal.

Like with racing, explorations of most types are risky — but there’s sure lots of living to be had.

The media’s been focused mostly on fault, as if going to 13,000 feet could be 100% safe.

It does sound like this craft mostly skipped Tech, and improvements are likely moving forward. But I wouldn’t hesitate a minute in boarding this thing, life’s too short to not witness dazzling moments.

I can’t or don’t want to imagine the terror in those last seconds. Was it worth it? I’d like to think “Yes.”

[DaveW]

I remember some arguments when we were developing a car. Several of us complained that 1/4 inch bolt needed to be upsized it was failing too many times.

The response from the designer-engineer was the numbers show that is the proper size, I don't care what crew chiefs have to say they are not engineers.

Further down the time line the bolts were upgraded to larger diameters by the designer.


Sometimes you wonder if the choices are made by engineering decisions or ego?

Engineers, unfortunately, miss things in their design process (btw, I am an engineer and I know I am not immune to that). I try to not let my ego make me ignore physical evidence or suggestions. A cursory analysis can show the component to be strong enough, but if ANY load condition beyond the obvious was missed or ignored, you have a recipe for failure.

[DaveW]

...I can’t or don’t want to imagine the terror in those last seconds. Was it worth it? I’d like to think “Yes.”

The implosion likely happened in an instant and they were gone before they knew what was happening. So they probably didn't suffer. Still, it's unfortunate that it happened, since with sufficient preventive/predictive practices it might have been avoided. Like some racecar components, a CF sub should "time-out."

[Jerry Kehoe]

My guess most engineers base their specs on a calculated assumption of the loads encountered, unfortunately then the specified component based on the calculated loads is designed and proclaimed right. The key issue is the "assumption" that the loads are correct and if not accurate failures can occur. As I remember the first Tyrrell designed by a brilliant guy broke in half on the first test and certainly his assumed sums had it right but in reality the "assumptions" were not. Probably happens more often than anyone wants to think! think that's why we see things like "modification one, two or three" pop up as reality is found out!

[Pi_guy]

Engineers, unfortunately, miss things in their design process (btw, I am an engineer and I know I am not immune to that). I try to not let my ego make me ignore physical evidence or suggestions. A cursory analysis can show the component to be strong enough, but if ANY load condition beyond the obvious was missed or ignored, you have a recipe for failure.

But from my years of knowing you you always approached items with an open mind and accepted criticism and evaluated the information given to you, without already having made your decision.

[DaveW]

My guess most engineers base their specs on a calculated assumption of the loads encountered, unfortunately then the specified component based on the calculated loads is designed and proclaimed right. The key issue is the "assumption" that the loads are correct and if not accurate failures can occur. As I remember the first Tyrrell designed by a brilliant guy broke in half on the first test and certainly his assumed sums had it right but in reality the "assumptions" were not. Probably happens more often than anyone wants to think! think that's why we see things like "modification one, two or three" pop up as reality is found out!

As the computer guys say, "garbage in, garbage out."

[R. Pare]

I have found that the biggest issue with too many engineers today is that they rely on what the computer is telling them, and that, combined with little to no real world experience, leads to bad input into the computer, resulting in a failure like this one.

Quite a few years ago I was commissioned to design and build a power transfer box for one of the "car that can fly" companies. They were a bunch of extremely bright MIT engineers, but with no where near enough real world experience. Way too many of my phone and email conversations with them involved my explaining the material properties that they knew nothing about since it was never discussed in their test books.

I suspect that this is what happened with this sub.

[E1pix]

That’s a fine analysis, Richard.

We see this issue all the time lately; be it sub tech, solar power theory, or seemingly anything quoting theoretical, best-case performance.

This is why whenever we ask “Is this problem solved?” the standard reply is “It *should* work” — and why I tend to trust a MIT grad less than one from UHK (University of Hard Knocks).

[Marty Nygard]

My favorite part of this whole fiasco is the submersible was controlled by a modified(?) hand held video game controller! There is no indication that there was even a spare controller duct taped to the bulkhead. Did they even put new batteries in prior to the dive?
It is reported that the carbon composite pressure vessel was built in six weeks within the company facility. Maybe next time they will subcontract with Dauntless.

[R. Pare]

There is a video showing the CF filament being spun around the hull form - I think that was actually at their shop.

Most likely they took some equations from somewhere that gave the CF compression performance and simply went to a thickness that the computer told them was "safe" for those depths - I have no idea, of course, what safety margin they used. Personally, I would be rather skeptical about any degree of safety margin for the simple following reason:

The problem with that is that during each dive, some of the fibers will break each time, weakening the hull, and after enough dives, the hull can no longer withstand at those pressures. The non-spherical shape of the hull was also a red flag, as the pressure stresses are WAY harder to predict and control.

The hubris that this guy had about safety is rather unbelievable.

[Pi_guy]

That’s a fine analysis, Richard.

We see this issue all the time lately; be it sub tech, solar power theory, or seemingly anything quoting theoretical, best-case performance.

This is why whenever we ask “Is this problem solved?” the standard reply is “It *should* work” — and why I tend to trust a MIT grad less than one from UHK (University of Hard Knocks).

My 2 cents is you can't just go by labels. You need to ask the questions and based on the response make your decision. Been involved with many developmental things and no matter what your level is you should be able to explain it. Answers like you wouldn't understand or you really don't need to know just raise very large flags. If you can't explain it to me how well do you know it?

When the owner of the project makes a statement we are not seeking certification every one should have run away. If you are an explorer you can cut corners and do your thing when you start seeking passengers for a commercial venture you need rules.

[DanW]

There is a video showing the CF filament being spun around the hull form - I think that was actually at their shop.

Most likely they took some equations from somewhere that gave the CF compression performance and simply went to a thickness that the computer told them was "safe" for those depths - I have no idea, of course, what safety margin they used. Personally, I would be rather skeptical about any degree of safety margin for the simple following reason:

The problem with that is that during each dive, some of the fibers will break each time, weakening the hull, and after enough dives, the hull can no longer withstand at those pressures. The non-spherical shape of the hull was also a red flag, as the pressure stresses are WAY harder to predict and control.

The hubris that this guy had about safety is rather unbelievable.

As a kid growing up in SoCal, we lived near Aerojet in Azusa. Very early '60's they were winding fiberglass tanks, both cylinders with bolus ends and spherical tanks for the Gemini and Apollo programs. They quickly discovered during pressure testing their assumptions about pressure integrity on wound filament tanks were wrong. Cracking noises of filament failures were the canary in the coal mine. Thankfully the engineers insisted on hydro testing instead of using air....

[Hardingfv32]

There are a lot of conclusion being jumped to about this event. The fact is we do not know why there was an implosion. Maybe the window failed.

We have a lot of knowledge about how carbon fiber fabrication function. There is no reason for an engineer to not rely on this data when putting a design together. Just because CF is not at its best in compression does not mean you have to avoid this condition. It can be accounted for in the design.

You guys are all afraid of CF because you do not have detailed knowledge of its properties. Until the material experts examine the wreckage everything being stated is just mindless speculation.

Brian

[Steve Demeter]

Reading up on the history, one of the main engineers quit because ha had doubts about the design and construction of it almost from the very beginning. Additionally, their chief Exec decided to NOT have it tested to recognized international standards for deep diving vessels. They were warned numerous times in advance by numerous experts that they were playing with fire plain and simple.

And also remember with any composite lay up ANY air not removed becomes an instant weak spot.

I read that the implosion took about 1 millisecond and that the human brain takes about 25 milliseconds to respond to stimuli. So I doubt that they knew what hit them

[DaveW]

There are a lot of conclusion being jumped to about this event. The fact is we do not know why there was an implosion. Maybe the window failed.

We have a lot of knowledge about how carbon fiber fabrication function. There is no reason for an engineer to not rely on this data when putting a design together. Just because CF is not at its best in compression does not mean you have to avoid this condition. It can be accounted for in the design.

You guys are all afraid of CF because you do not have detailed knowledge of its properties. Until the material experts examine the wreckage everything being stated is just mindless speculation.

Brian

Of course, it did implode. So it is, IMO, NOT mindless speculation. The fact that proper testing was not done before they took paying customers on a very risky adventure and failure and death resulted says safety was mindlessly disregarded.

Of course I would trust CF in a proper application. IMO, this was just not a situation where one could afford to do that.

[Marty Nygard]

"I read that the implosion took about 1 millisecond and that the human brain takes about 25 milliseconds to respond to stimuli. So I doubt that they knew what hit them"

The reports tend to indicate that the submersible was attempting to resurface; probably because of cracking noises from the hull. I think they were all aware of what could hit them.

[R. Pare]

That may have been from an earlier dive - one of the passengers - a submersible expert - said afterwards that the hull was making cracking noises as they went down.

[DaveW]

Some experts, however, weren't surprised — including film director and deep-sea explorer James Cameron.

Cameron told ABC News that he believes the Titan's hull began to crack under pressure and that its inside sensors gave the passengers a warning to that effect.

"We understand from inside the community that they had dropped their ascent (I think he meant descent) weights and they were coming up, trying to manage an emergency," he said.

[holmberg]

There's a lot of information on Wikipedia about the lack of testing and inspection, use of old CF, and general attitudes about safety. CEO said that meeting safety requirements stifled innovation.

OceanGate Titan

Greg

[bob darcey]

They lost communication about 1 3/4 hours into a 2 1/2 hour descent to 12,500 feet. That's probably when the problem began; they may have been down about 8800 ft or so, and the pressure would be about 3850 psig. That's a lot less than the 5500 psig they'd have at 12,500.

[Rick Kirchner]

As far as the implosion's effects on the people....

Where I worked we developed the Fuel-air explosive. Basically a coal-mine explosion in the open air (they call them thermobaric weapons now because FAEs got a bad rap as "the poor man's nuke").

On the day that the air war for Desert Storm kicked off they tested one to see if they could put out a fire in an oil-filled trench, as the Iraquis had planned to use those to slow our advance. Ground zero for the test was 14 miles from my house - and it cracked the stucco around all of my windows.

I was told by the old-timers at work that despite the impressive visual display of a FAE detonation, it only produces overpressures of under 10 psi. Doesn't sound like much right? Multiply that by the surface area of your body. It's like being driven over by a LARGE vehicle. The sheep and pigs they used as test subjects looked like they'd been stepped on by the Jolly Green Giant. Your insides exit via any available orifice.

Now consider how much pressure the was at that depth 5800 psi....

The whole bit about using old Boeing CF - well, I don't think this was a QC problem.

The game controller. We use game controllers to fly UAVs half the world away. They are amazingly reliable for the cost, and the human factors have all been worked out on a population of millions.

Here's the thing though. There was only one way to develop this system, and that is to build one, instrument it, and drop it to crush depth. Then you design one for the operational depth with sufficient margin, and then drop that thing down and bring it up hundreds of times. By the time you get to that point your simulation should begin to match the experimental data. There's no test chamber for this environment, no simulation. It's something you have to do over and over again to build the required databases. Don't forget, it's not just pressure, but temperature as well. What would have happened had they decided to go down to a pacific trench and investigate black smokers?

The problem with simulation is that it's like masturbation - do it enough and pretty soon you think its the real thing.

Can't tell you how many times there's been an experimental failure because the simulation didn't represent the actual environment.

One of my mentors worked at the Navy lab in Hawaii with Ballard. He said that the reason Ballard was so successful compared to his peers, was because his stuff came back. It was all massively overbuilt. So while everybody else was building and re-designing their lost stuff, Ballard was back out collecting data.

I still have acquaintances that work at that lab. One of the things they do for people is to take a styrofoam coffee cup and put it in one of the baskets on a deep submersible. When it comes back to the surface you have a souvenir - a coffee cup that's now solid styrene the size of a thimble.

[pacratt]

And, despite the possibility of bad engineering, let's not forget possible material & manufacturing defects.
As a machinist, how many times did I turn or mill a piece of "certified" material only to find porosity or a crack deep inside the piece?
Just another piece of a complex puzzle that can either lead to immense success or devastating failure.

Personally, I still lay blame on the company for what seems to be an incredible lack of testing and disregard for safety.
(I said 'personally')
Glenn

[R. Pare]

This article is obviously speculation, but it does point out a real problem in the materials used that I also suspected - the interface between the titanium bulkheads and the CF hull, where the stiffness is vastly different:

https://www.realclearscience.com/art...le_942957.html

Hopefully they will sometime soon release the pictures of the pieces, or actually bring the pieces up, so that the failure can be analyzed.

[R. Pare]

Just read that they are indeed retrieving the parts, so we can expect an analysis report sometime in the future.

[DaveW]

This article is obviously speculation, but it does point out a real problem in the materials used that I also suspected - the interface between the titanium bulkheads and the CF hull, where the stiffness is vastly different:

https://www.realclearscience.com/art...le_942957.html

Hopefully they will sometime soon release the pictures of the pieces, or actually bring the pieces up, so that the failure can be analyzed.

IMO, that thick CF hull was so rigid that the relatively flexible titanium had almost no role in the structure's strength either positively or negatively. The real issue was the integrity of the CF hull which obviously was not sufficient.

[Rick Kirchner]

you think about all the stuff that has to be done "differently" with composites - like designing for stress around holes, inserts for fasteners, etc.

I can see them thinking "well, they wind continuous fiber high pressure cylinders and those are about half the pressure we'll face, so we'll just make it twice as thick...."

[DaveW]

you think about all the stuff that has to be done "differently" with composites - like designing for stress around holes, inserts for fasteners, etc.

I can see them thinking "well, they wind continuous fiber high pressure cylinders and those are about half the pressure we'll face, so we'll just make it twice as thick...."

And of course, compression vs tension loading...

Tension loading (pressure inside a container) is an inherently stable condition whereas compression loading (pressure outside) is not.