Hacker News Reader

Comment by sclarisse | Hacker News Reader

- Main hull of the craft is a tank filled with gasoline: lighter than water and only a little compressible. It is not a high pressure enclosure. A membrane keeps these two fluids apart mostly, but some maneuvers dump gasoline into the ocean.

- Ballast is held in place with magnets. You actually want to drop ballast as you descend, to make up for the added seawater.

- Sphere on the bottom holds observer and instruments. This is a high pressure part.

- There’s no real propulsion or navigation. You land more or less where you land.

View on HN

> The primary 24 volt electrical system of TRIESTE is powered by 56 twelve-volt, 48 ampere-hour, lead-acid batteries located in four aluminium saddle tanks fastened to the top of the float beneath the superstructure. Each of the four battery tanks or boxes is pressure- compensated through the use of transformer oil and, therefore, during submergence the batteries themselves are subjected to full depth pressure.’ Three of the boxes are allotted for powering the craft's five 3-horsepower electrical propulsion motors and for providing power to the lighting systems.

> The propulsion system consists of five special General Electric 3-hp de motors. These motors are designed to operate in inert fluid (silicone oil) and are subjected to full ambient pressure during diving operations.

> The motors themselves are in five different locations. Two of them provide horizontal motion forward and backward, two provide vertical motion either ascending or descending, and one is installed athwartships for turning. The motors all drive propellers through gear boxes.

It actually does have electric thrusters. Interesting that they were able to make the batteries neutrally pressured.

Ah! I was misled by the general comparison to an unpowered balloon that they make a few times. It’s not much propulsion and they take special care to differentiate it from a submarine that can actually travel the world.

> athwartships

What a phenomenal word!

Incidentally one of my favorite nautical terms. Just always liked the sound of it :-)

I recall reading about the craft using gasoline as an incompressible floatation fluid, but I hadn't realized the scale: 100 tons of gasoline, making up 2/3s of the craft's mass.

Basically the equivalent of a lighter than “air” craft but in a different fluid. I wonder if there’s a word for a craft that controls its buoyancy regardless of medium?

An Archimedes craft.

(I made this up just now).

Fluid is a Latin root, so why not "fluidostat" on the model of "aerostat"?

because -stat is a Greek root.

merriam-webster

   borrowed from French aérostat, probably back-formation from aérostatique "of aerostatics," with -stat (after héliostat heliostat) taken as the Greek agentive element -statēs "one who causes to stand" — more at -stat

I've never looked at it this way, but it really is literally an underwater blimp lol.

> I wonder if there’s a word for a craft that controls its buoyancy regardless of medium

sperm whale

I never knew the scale either. But the book says gasoline has a specific gravity of 0.7. So 100 tons will make roughly 30 tons of buoyancy. The negatively buoyant craft is 50 tons so combined with the air inside that is probably the displacement required to make it positive

When people say 'carbon fiber is weaker under compressive loads', or 'cylinders distribute compressive forces unevenly', these kinds of issues seem like engineering challenges not fatal flaws. I feel like the problem with the Titan submersible was the lack of respect for safety, not any specific design choice.

It could well be both.

Well, obviously going with a more conventional construction would make it easier to come up with a safe design. But I feel like people are going to take the wrong lesson from this accident.

The people whose lives depend on this sort of thing will definitely be learning the right lessons and the rest doesn't matter. In fact, plenty of them already had learned those lessons, the CEO of that company was a pretty special case (and not in a good sense).

The Boeing 787 is primarily made from carbon fiber, including the wings which experience a mix of tensile and compressive loads thousands of times over their design lives. Of course deep sea submersibles occupy a much smaller market where a comparable degree of rigor may not be justified, but I think it's silly that everyone suddenly seems to be a expert in composite and/or submersible design.

Airplanes deal with lesser atmospheric pressure though, not more. Flexibility is probably a good thing.

The 787 deals with a pressure differential of ~0.6 atmospheres.

A dive to the Titanic needs to deal with a pressure differential of ~340 atmospheres.

Commercial airplanes have only recently used composite wings, under large regulatory suspicion and are used in a far more benign environment than 4000 m of water head.

I really dont think ppl appreciate what 4000 m of water means! 1 atmosphere is only a 10 meter water column. 4000 m is 400 atm. Actually more because water can no longer be assumed incompressible.

EDIT: to add another point - submarines are subject to implosion, which is a buckling failure. That sets up a vicious cycle where a small deformation gets worse, whereas a vessel at positive pressure will "straighten itself out"

Right material for the right application.

Just because it works under tension in an aircraft or even a cylinder of compressed gas, doesn’t mean it will work when the pressure is applied from the outside of the vessel.

It’s the opposite of plain concrete for instance, which has got compressive strength but doesn’t do as well for tension.

I understand just saying the words “carbon fiber” it feels like we live in the future but that’s the difference between marketing and engineering.

I feel like everyone either misunderstands your point or is deliberately misrepresenting it. It might very well be that composites turn out to be a viable material for deep submersibles. The fact that OceanGate managed multiple dives to the Titanic suggests that the idea is not completely far-fetched. Or it could be that it's a terrible idea. But it would require much more testing to properly understand how composites behave in these conditions.

It's a shame that it got tested out with deadly results by a company that was completely unserious about safety. Now the HN crowd has all watched the same bunch of Youtube videos over the last few days, so now everyone is a qualified submersible designer and has decided that composites are obviously a bad idea for a deep submersible.

One of carbon fiber’s failure modes involves damage due to repeated stresses. It is also fairly difficult to monitor how it degrades. That the sub survived a handful of dives isn’t indicative of solving the core issues with the material in this application. The failure indicates that they did not make much progress against those core issues.

Right, that's the point exactly. The company was clearly playing it fast and loose and not interested in understanding how their design behaved or how the material degraded.

Choosing carbon fiber for this application probably falls under 'playing it fast and loose'. Because it's a cost savings measure compared to creating the same part out of titanium (strong but very expensive) or steel (not that expensive but very heavy for the same strength) and it solves some ballast issues (because a similar part out of steel or titanium would be far heavier than carbon fiber and you can't discard the hull for obvious reasons if you want to ascend).

What is still a bit of a mystery to me is why the company that built the hull did so knowing what the application was. I'd love to see the correspondence between those two companies and whether or not there was an opportunity for the company that built the hull to flat out refuse to do so because they thought it was a bad idea or if they put an upper limit on the number of cycles.

https://www.compositesworld.com/articles/composite-submersib...

Is a really good article on this particular design, and there are some interesting details there that I know for a fact were different on the sub that imploded. The article clearly states: "Nothing will be mechanically attached to, or penetrate, the composite hull other than the titanium caps. "

A lot hinges on the use of that word penetrate, whether they meant all the way through or just from one side. There are multiple items screwed into the pressure hull on various photos including more than one monitor stand. That alone may have been enough to cause the failure.

What I also miss from the specification as quoted in the article is that there is no number of cycles specified. Finally the safety factor quoted in the article is 2.25, which is in excess of the 1.5 that you'd normally expect so theoretically the pressure hull should have survived this assuming no prior damage. On the first dive...

Finally, the whole 'we're monitoring things and will ascend if there are signs of impending failure' is something that sits wrong with me. Carbon fiber failures do announce themselves by the pulses of individual fibers breaking. But in a pressure vessel you get into a feed-forward loop that would cause such a failure to expand so fast that I don't think there would realistically be a chance to surface timely.

> [carbon fiber] solves some ballast issues (because a similar part out of steel or titanium would be far heavier and you can't discard the hull for obvious reasons if you want to ascend).

Incidentally there is another way they could have solved this problem; aluminum. With aluminum you can build a DSV that can dive deep enough to visit the Titanic, is light enough to not require additional buoyancy, and large enough to hold up to 7 people. This has been done: https://en.wikipedia.org/wiki/Aluminaut

I think they probably had a fatigue life issue with an aluminum hull (it was only in service for six years..) since aluminum infamously has no fatigue limit. But it should be easier to characterize, so you can retire the hull before it fails.

That is an amazing feat of engineering.

Interesting that they still have it on 'standby' just in case it is needed despite being out of service for so long. Unique capabilities.

It's a terrible idea. I've used carbon fiber since the 80's in various applications, and I know people that design deep sea ROVs for a living. I also work with/for a company that does this stuff. There are also a fair number of materials scientists that specialize in carbon fiber who have unambiguously stated that they wouldn't go on that sub, period. An engineer that worked for the company warned them about issues with the pressure vessel and got fired and sued.

Not a single person I'm aware of except for mr. Rush thought building the pressure hull of a passenger carrying submersible out of carbon fiber was a good idea.

The fact that they managed multiple dives is probably predicated on your understanding that it was the same named vessel, but the carbon fiber cylinder had already been replaced at least once due to damage. Carbon fiber is a fascinating material, but for this application it isn't the best choice and for an application like this you don't take chances.

I've also looked at all of the photos of the interior of the sub and I've seen some things that anybody that understands this material would never ever have done. Such as to drill holes in it to mount stuff, rather than to use clamps or to integrate the mount into the fiber as you're laying it. The fact that it worked at all at those depths is a severe case of survivor bias, I'm fairly sure if you had tested the cylinder to destruction after a single dive that it would have failed at a much lower level of stress than prior to the first dive. And if you use a novel material and you're seeing unexpected behavior then you stress test until you drop rather than that you go sightseeing with paying passengers.

> Now the HN crowd has all watched the same bunch of Youtube videos over the last few days, so now everyone is a qualified submersible designer and has decided that composites are obviously a bad idea for a deep submersible.

Lots of people on HN are knowledgeable about various materials, you can discard them as well as all of the people who said prior to this accident that this was a bad idea from the start. But I sincerely hope that people who are in a position to actually act on all this knowledge (and the results of the investigation) will adhere to their previous mantra: when in doubt: test, test, and test again. Because no amount of knowledge or perceived understanding is going to replace that, it can help guide a design but it simply isn't the final word. The proof for any kind of pressure vessel is repeatedly cycling a design to the point where it fails. And if it doesn't fail when it has reached its design life, to check if the degradation is as predicted.

Note that such testing already starts when you receive a batch of fiber.

Are you an expert on carbon fiber or nondestructive testing and monitoring of carbon fiber? What is your personal experience building submersibles or aircraft?

Which people and what lesson?

Like who is it critical to not write off carbon fiber for extreme depth submersibles?

You can absolutely delude yourself with the wrong design and materials that have going for them that they're what you can afford and call it innovation and whatnot. It still is wrong design and materials.

I didn't pay attention to this until the day they were supposedly running out of oxygen. I searched for pictures to find a hollow long cylinder made of carbon fibre with end plates. That and reading their Wikipedia entry made me angry. It was obvious that the reason for losing contact days before was that it had imploded killing them instantly. My background is physics.

So is mine, and they made several successful trips which (among other things) makes me think that carbon fiber could be an appropriate material for a submarine. Metallic hulls would also implode catastrophically if not well designed.

The reasonable and ethical engineering approach is to do one's best with models, simulations, first principles analysis and design something that works on paper. And then built prototypes, test the hell out of them. Without people inside. Standardize everything they can about manufacturing. If all looks good, they can operate. Never stop measuring and testing.

The operators decided this was too much work. They gambled and they lost.

They did try a model but it failed at the euqivalent of 2000m I believe

Move fast and break people. How many entrepreneurs learned from the Titan disaster that they should really hire test pilots?

Composite failure is very hard to predict and hard to evaluate. By comparison metal fatigue is very well understood.

Compression loads on submarines are extreme and cyclical. This causes the fibers to delaminate loosing strength. Furthermore, De-lamination is difficult to measure non-destructively (I think only xrays work?)

Manned conposite subs should always have their chief designer or financier aboard

ultrasound works quite well as well - but no NRE technique is perfect

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7503374/

Didn't they specifically refuse to do any testing because they trusted that proprietary acoustic system?

The paper is from 2020. My understanding is that composite testing is still in its infancy.

Also note that the hull was 5 inches thick (safety factor > 2.5). Depending on the dispersion properties of the medium ultrasound might not penetrate and get a good result. In Ti (some alloys at least), for example, you cant get much signal past a couple of inches due to the scattering (from the grains?)

That was supposed to hear the hill cracking/creaking.

Ultrasound for delamination is like medical ultrasound and looks for a discontinuity in the sheet from the delamination.

I think someone wanted to do the delamination hull scan and they skipped it presumably because they trusted the creak listener

Just to add, isnt the reason why metal fatigue is well understood because it has killed a lot of people?

No, it is well understood because the failures of metals are much more reproducible than of other materials.

Both for ceramics and for composites the failures are caused by microscopic defects that are never the same for different samples, so the conditions for failure are very variable from sample to sample. Therefore the results of experiments with such materials have lower predictive ability for the behavior in a real application.

To some extent. "Mistakes were made" as they say. But its a lot more nuanced than that.

Metal fatigue is understood because we have a mechanistic understanding of what it is, how fatigue accumulates, and how different metal crystal strictures respond to fatigue.

We know, for example, that metals like aluminum will always fail from fatigue given enough loading cycles, no matter how small the applied stress.

We also know that other metals, like iron and titanium, have a "fatigue limit" below which fatigue doesn't accumulate and these metals can endure infinite loading cycles.

We have, to some extent, the ability to repair metal fatigue.

We build airplanes from aluminum knowing their aluminum hulls and wings will fail (whereas if it were built with Ti fatigue failure could be eliminated) because metal fatigue is very predictable and we can withdraw a hull from service after a regulatory determined number of landings.

So yes, until we developed our current understanding of fatigue, people died. But, often, this was from a callous disregard to traditionally accepted safety factors by cowboy "innovators".

(Im a materials eng. PhD in polymers w/ background in Eng. Phys. Im not a metallurgist for what its worth)

I'm just an armchair observer here, but from what I can tell from the discussions, carbon fiber certainly can be an appropriate material for a submarine, as long as it's only used for one dive.

The reason it's a useless terrible material for a sub is that until it survives the first dive, you don't know if it has a hidden flaw.

And after the first dive, even if it survives, you know that 100% that it does.

Weight is not nearly the problem it is for a sub as it is for spacecraft or aircraft. The problem wasn't safety protocols or XBox controllers or any of that. It was conflation of "big idea people"[1] with real engineers.

[1] see Jobs, Musk, etc.

As I said, I'm an armchair observer here, not a materials scientist.

However, there are a lot of things that are absolutely designed and built to be single-use, especially in fields like space, where rocket engines have been single-use until very recently.

As for weight, I think the reason that was important was because they didn't have a big enough ship that could operate a crane and haul the submersible out of the water. Really, the whole operation was a bad idea: this kind of exploration is expensive as hell, and the only way to cut costs (with current tech) is to do really dangerous stuff.

The problem with carbon fiber is that it's fiber. Think of a rope. It's very strong in tension; you can hang from it, climb it, hoist things with it, etc. Carbon fiber is useful in an aircraft fuselage because the inside pressure is a tension load on the fuselage.

Rope is useless under compression. You can't push a rope, or climb a rope that's only attached to the floor. So under compressive loads, you're relying on the glue mainly. The carbon fiber gives the glue something to stick to. A submarine is the opposite of an aircraft: the high pressure is on the outside, so the hull is under compression not tension.

It's a terrible choice for a submarine.

Some good points from first principles. But there are ways. If you manage to make the composite with a very good process so the fibers are straight and closely packed, then it can have good compression strength as well.

This can be achieved for example in pultruded carbon rods, where the carbon fiber is under tension when it passes through an epoxy bath. They have been used in aircraft wing top spars which receive compression loads (when taking positive gees).

How to make a cylindrical vessel that can take compressive loads? The creation process certainly needs some thinking and attention.

Thinking about this more from first principles, if I got it right, the cylinder hoop stress is twice the longitudinal stress. So you could pultrude a unidirectional layer of hoops, or very shallow angle spiral to handle half of the the hoop stress only on the outside. Have normal cross wrapping layer on the inside probably.

This is misleading. Only raw CF is like a rope, once made into a composite with resin it becomes a stiff member with entirely different properties. It's more analogous to a wood beam - very stiff and resistant to bending but once you get past it's limit it cracks.

I suspect the failure mode in a submersible like this isn't so much the carbon fibers themselves but either the carbon > metal interface at the ends or gradual delamination between layers of fibers due to the cyclical pressure loads.

Under outside pressure, the fibers in the matrix take close to no load, while under tension they take a lot. That alone is the reason why carbon fibre composites are the least suited materials for submarines and submersibles.

You should watch the initial interviews with James Cameron. He explains 1) composite hulls have a finite number of compression cycles before they fail, typically less than 10 (this was dive #7 give or take, I forget the exact numbers) and 2) no composite material is perfect, the water pressure will find imperfections and work it's way into the material, causing weak points

Steel is a significantly better material for this task. I'm stunned they even considered carbon fiber.

Metal hulls do not degrade on every dive. If you start off with a safety factor of 2.5 (according to the design specs as they are known today) then after a number of dives you'll be below '1' and failure is a given. Without the ability to test the degradation it's a roulette.

> engineering challenges not fatal flaws

These challenges are on the order of "non-flammable paper" or "non-magnetic iron"

> "non-flammable paper"

These exist, but are not common.

> "non-magnetic iron"

Austenitic stainless steels are a thing.

https://en.wikipedia.org/wiki/Austenite

So I'm not sure if I like those examples. It's more along the lines of using baked clay under tension, or a rope under compression.

As long as you have two metal end caps connected by metal poles, and you lay the carbon fiber in the direction that connects the end caps, then it seems that the hull would become concave/"bulge inward" and the fiber would hold pressure by tension.

Not sure whether this would actually work in practice though or whether it would be better than the simple metal sphere design.

That will surely fail because the load isn't spread evenly.

Seriously, this isn't a tent.

OK they weren't the best example, but for the record non-flammable paper is not actually non-flammable, it just doesn't sustain combustion because of a chemical in the paper. Austenitic steel is not iron, it's iron + carbon.

It might be a matter of it being weak enough that by the time you added enough to make it work, it would negate any advantage of using the carbon fiber while also having worse properties in some other way?

There's nothing wrong with using carbon fiber for a compressive load like this. Like most engineering the devil is in the details - how do you interface the carbon to the metal end caps for example? They likely are a weak point as the compressibility will be different for both materials making them want to shift (or fracture if they are glued together) during pressure cycles. I'd love to know how they solved this (or if they didn't and that's why we're reading this).

Contrary to, I don't know, tze selection of a stack, the selection of the correct material for a given application is an integral part of engineering. Finding means to get away with the wrong material is not an engineering challenge, it is simply bad engineering. Not even that at the extreme ends of the spectrum.

> You actually want to drop ballast as you descend, to make up for the added seawater.

The ballast is dropped during the ascension, not the descent. This makes the ascension-step impossible to fail and requires no electricity [1]. Incidentally, I first came to know about the bathyscaphe while reading Peter Watts's Rifters trilogy (which is an amazing hard SF series set in the deep sea).

[1] https://en.wikipedia.org/wiki/Bathyscaphe#Mode_of_operation

The document says otherwise. As gasoline is slightly compressible, seawater is admitted in the float during the dive, increasing negative buoyancy, and increasing the rate of descent. So, to control the rate of decent, some ballast is released during diving too.

Ballast is dropped during both descent and assent. The linked document explains how the gasoline used for buoyancy is significantly compressible and becomes denser as the bathyscaphe descends. If no ballast was dropped during descent the rate of descent would continue to increase due to the feedback mechanism of increased pressure, leading to increased compression, leading to increased density.

FWIW you don’t need a bathyscape to experiment with this. When Scuba diving in temperate locations (like Tobermory, Ontario), people with 7mm wet suits and extra vests or jackets experience this as they descend: The neoprene compresses, and if you don’t add small amounts of buoyancy to your BCD, you’ll start to free-fall.

It's Tobermory. Absolutely beautiful area, so many birds. I went camping there several times. It's also pretty much the end of the world, the only way through is by ferry.

You’re correct, no “e.” Thanks!

The diving in Five Fathoms Marine Park is exquisite, and it’s a short drive from Lion’s Head, which has absolutely stellar rock climbing.[1] We’ve taken the kids on the ferry to Manitoulin Island[2], and from there driven over to Sudbury.

I recommend trying that ferry at least once, it was a lovely experience.

[1]: https://www.thecrag.com/en/climbing/canada/lions-head

[2]: https://en.wikipedia.org/wiki/Manitoulin_Island

Isn't the neoprene air-containing foam, and it's the air that compresses? (Not that that negates your point...)

Absolutely, but that air is still an integral part of the foam, right? I think it's fair to say they both compress.

Neoprene is a closed-cell foam. An open-cell foam (like a dish sponge) would not compress as air/water can just flow through.

You can experienced this in pools! 12ft of water adds an extra 1/3 atmosphere of pressure, compressing the gas in your body by 25%. (That's why your ears pop!)

By varying the amount of air in your lungs, you can choose which depth to be neutrally buoyant at. Weirdly fun!

Or at least, you can if you don't sink instantly at any depth!

Some steel shot ballast was jettisoned to slow the descent. They didn't want to crash into the floor and damage the pressure hull.

It says at the link, on page 5, "The compressional loss of buoyancy is of such a magnitude that approximately 1 ton of ballast must be dropped for every 3000 feet of descent to maintain equilibrium."

Reply on news.ycombinator.com