Steel has a relatively low strength-to-weight ratio. The CF hull was probably close to neutrally buoyant, and had a large passenger cabin.
A steel hull would need to be much thicker, increasing the size of the vehicle if you wanted to maintain the same internal volume. And it would be nowhere close to neutrally buoyant, requiring extra buoyancy devices. Cameron’s vessel uses something called syntactic foam–hollow glass beads embedded in a matrix. The old submersible Trieste used a giant float filled with gasoline. In both of these, the passenger cabin was a tiny fraction of the entire vehicle size (granted, they also went much deeper).
The larger mass also means a much larger support vessel. Probably needing more crew to handle. Even more expense.
CF would be a fantastic material in this application if it could be inspected properly and if a suitable resin and winding process could be found. Those are huge ifs, though.
This. The primary problem with composite construction is that it is very hard to inspect. Stress can cause delamination, poor layups can create small air bubbles that reduce strength, etc. And once the layup is cured, it’s closed forever.
So you have to some fairly exotic nondestructive testing to really determine if the structure is okay, and in a high stress environment like this I would want it done after every trip.
If they wanted to come up woth a design lifetime for the sub such that they could feel safe for a certain number of trips, they would have had to build an assembly line, lock it down so it doesn’t change, build a handful of subs shells, then send them down (unmanned) to their maximum working depth +20% or more, and keep doing it until they either destruct or you hit some limit - say, 100 dives.
If you can build five subs on an assembly line that each survive 100 descents deeper than you are taking passengers, and they showed no damage to the composite structure afterwards, then maybe it would be safe to take passengers on the sixth one.
Right. Even if its inevitable that the material takes damage over time, that’s potentially acceptable if you very carefully qualify the material, and are able to continually test it to ensure the level of damage is within predictions, and that you leave in a generous safety margin to account for remaining uncertainty.
Really, this is the case for all materials. Aluminum is guaranteed to fatigue over time. But there’s an enormous amount of understanding as to how that works and how to detect upcoming failures.
I don’t think we’re near that level of understanding for CF even in aviation use (despite the tremendous amount of money put into it), and in this particular application there’s clearly even less. But it’s possible in principle.
As for the title question specifically, one way to turn compressive forces into tensile is via pretensioning. As you wind the hull, you apply tension equivalent to what it would experience at the rated depth. Then, the compressive forces serve just to relax that tensioning. You would have to gradually increase the tension from the inner layers (since they have to resist the tensioning itself) to the outer layers, among other things, and probably makes the structural analysis even harder. But it could help with the problem here (if it does turn out to be delamination).
Just to correct myself, this is probably false. Steel would have a much higher cross-sectional mass, but since it’s much denser than CF, it wouldn’t necessarily be thicker. The mass is still a problem, though. The extra buoyancy devices would increase the size of the vessel and increase the mass even further.
The composite version will need much more ballast. So starting out a composite version and steel version will be much closer in weight and need the same size support ship.
Why would CF need more ballast? Both need to achieve positive buoyancy when the ballast has dropped. The steel version would need to add buoyancy devices to achieve this, and then even more because the extra volume would add drag, which demands even more buoyancy to achieve a certain vertical speed. And likewise, it would need more ballast to achieve a certain sink rate.
You want it to be buoyant. The sub needs to go up when you drop the ballast. At worst (like if the hull had zero mass), you would need more ballast but the total mass would be less (since the hull didn’t weigh anything). So that would still be better than the heavy version.
CF isn’t that good, but in this particular application it seems to hit the sweet spot, where they achieve a decent lift rate without extra devices. The extra shroud part was flooded and wouldn’t contribute much.
Ballast is much denser than the equivalent buoyancy devices by a factor of 10x or more (probably more like 20x). So it doesn’t have the same impact on drag or vehicle size.
A simpler way of thinking about this is that a heavy hull is just ballast that you can’t drop. And that therefore you can’t use for anything useful. It’s always better to have 10 tons of hull and 10 tons of ballast than 20 tons of hull and no ballast (well, unless the 10 ton hull implodes after a handful of dives).
If you pretension the outer fibers, then the inner part of the cylinder, whatever it’s made of, will be under compression right from the get-go. That being the case, the inner part of this type of hull couldn’t be carbon fiber, because then those fibers would be under compression, which is exactly what you’re trying to prevent. To make this work, you would have to wind the pretensioned carbon fiber around some other kind of material that does tolerate compression well, like metal.
The end caps are compressing the hull cylinder axially, but water pressure is also compressing the hull cylinder radially. Pressurized pipes face a similar situation, in that they experience axial tension and radial tension. The radial tension is called hoop stress, and the math shows that this is always larger than the axial stress. This is why whenever you see a copper pipe that has ruptured, the tear is always parallel to the length of the pipe (do a Google image search for " burst copper pipe" and you’ll see what I’m talking about).
The basic equations for calculating hoop stress and axial stress assume that the wall thickness is small compared to the diameter of the pressure vessel. No doubt the Titan’s hull is quite thick compared to its diameter, so those thin wall equations would not be applicable. There is a whole separate set of equations for thick walled pressure vessels that would be applicable here. I’ve only ever seen equations that assumed high pressure inside the vessel, but even if those aren’t applicable to a submarine hull, I’m sure there exist an analogous set of equations that assume higher pressure on the outside of the vessel. Similar to the situation with hoop stress and axial stress in a thin-walled pipe, I have no doubt that those equations would show that the hoop compressive stress in the Titan’s hull would have been substantially greater than the axial compression stress. And both of those stresses would be compressive, not tensile.
So when you imagine the catastrophic failure, don’t imagine a long cylinder being compressed into a short cylinder or pancake. Instead, imagine a fat cylinder being compressed into a narrow cylinder.
Thanks. But as I look at I don’t have a clear idea on why the fibers are in tension when the tube is under external pressure. It is obvious why when the pressure is internal.
For the submarine hull, the fibers are not in tension. The material on the outer surface of the hull is under axial compression, radial compression, and tangential compression. The material on the inside of the hull is under axial compression and tangential compression, but no significant radial compression (just a tiny amount, about 14.7 psi, due to the atmospheric pressure inside the cabin).
I heard someone (an important taling head expert) elucidate the problem as such. Imagine having many layers of composite. Physically, on the ends or seams snd layers of material where they come together, under great pressure the water penetrates and weakens the bond of the layers and for all practical ends rips apart at th3seams shearing. Imagine taking a multilayer
wafercookie and pressure washing it from the “side”.
In theory, the exposed ends of the composite tube were protected by having the endcap mated to them, but I’ve seen footage of the assembly process (9:38 in the video below) and it’s just a bunch of guys enthusiastically daubing epoxy on the mating surfaces with no control over evenness of application, then putting it together in an open warehouse space with no control over contamination from foreign bodies such as dust, hair, insects, etc.
Honestly, the whole thing looks pretty slapdash to me. I’m sure there is an extent to which being under pressure would just push that joint together more, which should strengthen it, right up to the point where it really matters, and at that point, irregularities become places where stress will be unevenly distributed, inducing point failure.
Of course, I am not an expert on assembly of carbon fibre pressure vessels, but that’s sort of how I can recognise the shortage of controls in the footage above. That process looks like how I would do it. Me- an amateur.
Ok, thank you. I don’t recall why but I had the misimpression hoop stress would put the structure in tension under external pressure. That didn’t make sense for uniform external pressure and turns out not to be so. I don’t know the winding pattern though, for bodies with internal pressure the radius and pressure combine to winding fibers helically (forget what they call it) so the fibers have some axial direction too. Except for the uneven distortion by pressure towards the center or the end caps I don’t see that helping much in this case. This design makes even less sense than ever.
The endcaps are just relying on the glue seam between the composite tube and the end caps. That’s the part that seems dicey to me. (BTW: Is the hatch in an end cap?). I suppose it would be much more expensive to make a prolate spheroid that has fibers both radially and axially aligned. But why waste money on a stronger more reliable structure?
I took a materials science class for one quarter in college. I remember the professor going through an equation to determine the yield strength of a material under tension. As you apply tension, the material will stretch. (Most materials have some degree of elasticity to them.) Stretching relieves the tension, but as it stretches it also gets narrower. The thinner material is weaker than the thicker piece had been before the tension was applied. The yield strength was when the weakness of the material getting thinner was more than the stress relieved by stretching.
If I have any of that wrong, I hope someone will let us all know.
The thing is, one of the tests in that class had a question about a perfect sphere under water, subject to force pressing on it from all directions. Apparently, we were supposed to use the same equation for yield strength. That didn’t make any sense to me. I figured a material under compression would get thicker, and therefore stronger. I mentioned it to the professor after the test, but I don’t remember how that got worked out.
Titanium is exceptionally strong for its weight, but working with the metal is difficult and massively expensive. Much rarer than iron, just sourcing titanium is costly. Russia is a major supplier of titanium to nations all over the globe and has much easier access to the metal (the US had to secretly import titanium from the Soviet Union to build the SR-71 Blackbird).
Actually building the titanium hull requires special argon-infused facilities and workers who are trained to work with the metal. Bending and shaping titanium panels is difficult, and there is an uncomfortably high risk of imperfections within the metal that could result in a catastrophic failure of a submarine subject to extremely high pressures.
All of these problems culminate in a mind-bogglingly expensive production process.