Hacker News Reader

Comment by tyingq | Hacker News Reader

"she thought it was "stupid" that the Navy required the tests to be conducted at -100F (-70C)"

I'd be curious to know what that requirement is for too, though I'd likely not just ignore it.

There are plenty of things to control for when designing and producing steel. An important consideration for some applications is the brittle to ductile transition temperature. Some steels have a DBTT around 0°C, which means that they loose most of their impact resistance below that temp. By adjusting the chemistry, processing methods, and post-processing methods, you can adjust the DBTT of steel down below -70°C. My guess is that they were being paid to supply steel with a DBTT below -70°C, and that the contract included proof testing. The standard testing method would be a Charpy impact test at various temperatures. It sounds like they were making complex castings, so they probably had some non-standard testing requirements.

More info:

https://www.punchlistzero.com/dbtt-ductile-brittle-transitio...

https://knifesteelnerds.com/2018/12/21/why-cold-steel-is-bri...

So sounds like designing for surface collisions (other vessels, icebergs) in a very cold area maybe?

The Navy is understandably cagey about how they surface through ice-sheets, but this Smarter Every Day video is the best summary I have seen. https://youtu.be/XFJnWp1tAdU

Basically, some part of the sub needs to be strong enough to reliably break ice at sub-zero temps. The rest needs to maintain its strength at those temps.

Yes, U.S. Navy nuclear submarines go under the polar ice caps somewhat frequently -- as did Soviet (and do Russian) nuclear submarines as well. It's a good place to hide from prying eyes, as not only can ships not (easily) follow you there, but the sound produced by all the ice moving and grinding makes passive sonar quite ineffective.

The YouTube channel Smarter Every Day recently had a video series on a nuclear submarine under the arctic ice.

Presumably it makes you invisible to any magnetic anomaly detection via the air or satellites as well.

Its interesting, but poor quality steel with sulphur contamination is one of the theories behind the failures of the titanic.

>Metallurgical and mechanical analyses were performed on steel and rivet samples recovered from the wreck of the RMS Titanic. It was found that the steel possessed a ductile-to-brittle transition temperature that was very high with respect to the service temperature, making the material brittle at ice-water temperatures. This has been attributed to both chemical and microstructural factors. It has also been found that the wrought iron rivets used in the construction of Titanic contained an elevated amount of incorporated slag, and that the orientation of the slag within the rivets may hold an explanation for how the ship accumulated damage during its encounter with the iceberg.

Source: https://www.govinfo.gov/content/pkg/GOVPUB-C13-17a17f71ae2f9...

Yeah, even a small % of sulfur can shave off 50% or more of your strength. A small amount of phosphorus (<1%) can raise the DBTT above 0°C. The average ocean temp at 200m is 4°C, so you could understand why the steel chemistry is important in subs. Luckily, these can be tested for non-destructively.

South pole mean winter low is -76F but -128F has been recorded. While the water isn't going to get that temperature, subs need to surface periodically. I bet they've looked at the probability of needing the ability to operate in extreme cold, and -100F is the compromise.

I had assumed the tests were related to pressure and depth, where the temp isn't much off of 0C. Not an area of expertise for me though.

A submarine cannot surface at the south pole.

Yet. Give it time. Though, i believe they just meant the waters surrounding Antarctica.

Yet? There's land underneath the ice at the south pole.

I believe yet refers to https://www.sciencealert.com/scientists-think-earth-s-next-s...

I used to design and build scientific equipment for the South Pole. We had to do all our own materials testing because manufacturers would only specify their products to the bottom of the MILSPEC temperature range and the South Pole goes below that.

I remember one spool of cabling in particular - it was beautiful stuff, highly flexible, high current carrying, just what we needed. It stayed awesome right to the bottom of the milspec temperature range, at which point the exterior sheathing went through some kind of phase transition and froze solid, followed by the interior insulation having a different coefficient of thermal expansion than the exterior sheathing resulting in the whole thing shattering itself into pieces just a few degrees below military operating temperatures.

That is a beautiful story of true engineering.

Likely it has to do with the scaling of the actual property being measured. For example with decreasing temperature, steels become less and less able to resist impact damage. Thus getting hit by something small at cryogenic temperatures is equivalent to being hit by something big at normal temperatures; so if you want to test whether a submarine can survive an impact that would be impractical to simulate in a lab, you could instead cool your test sample down a lot and test it with a smaller, cheaper device.

My knowledge of shipbuilding is limited but I believe the navy suffered quite a bit during WW2 from certain steels undergoing a transition in the low temperatures of the north atlantic that caused brittle fracture. I would imagine testing at ultra low temperatures accelerates the rate at which such brittle fractures can be detected.

Brittle fracture toughness for carbon steels (resistance to cracking) gets really bad below about -20 °F. The colder you get, the more things you have to do to keep that toughness up, or else steel will crack like cast iron instead of bending and stretching like it's supposed to.

I come from the pressure vessel industry where impact testing is routinely required at temperatures below -20°F, sometimes at -320 °F for cryogenic applications.

Fracture mechanics and low temperature behavior of steels is a relatively "new" field in engineering, only emerging sometime in the last 80 years, so I can see why the metallurgist may not have known too much about it. Widespread study was kicked off in the US with the Liberty ship disasters:

https://metallurgyandmaterials.wordpress.com/2015/12/25/libe...

If you want to know more about stuff like that, ask away; it was far and away the most interesting part of my mechanical engineering education and career before I switched over to tech :)

Which is sort of the same thing. "I don't want to run this expensive or complicated test because it's stupid" is certainly a kind of cost cutting.

Maybe, but not necessarily. I bet many of us paper over requirements we think don't add value, where we can get away with it. Solely because we think it adds no value.

While both might be illegal, there's a massive difference between papering over a requirement for a business application and papering over a requirement for a mechanical specification.

The former might cost a customer money; the latter puts lives at risk.

>the latter puts lives at risk

Maybe. There are certainly past cases of the military creating unreasonable/unhelpful specifications.

Edit: Not justification for what she did. Just curiosity, having spent a fair amount of time in the military.

It is not the job of an individual to simply just ignore a spec instead of bringing up a complaint.

Yes, I'm not justifying what she did. I am, however, curious if the requirement was "stupid" or not.

There is always a chance it is, however without some deep knowledge on the reasons for the spec, its impossible to say why a spec exists in the first place.

Many of these large scale engineering specs are "take the worst possible situation you can ever think of and design it so that it can survive 1.3x of that". Yes it sounds stupid to the layman, but this is how you make shit that survives everything you toss at it.

Ignoring "dumb specifications" is the reason why SpaceX lost a payload. They used non aeronautical grade metal, which did not conform to the requirements for spaceflight, causing a mission failure.

Anyone who would paper over requirements that involve lives is a terrible engineer and should be blacklisted from anything resembling engineering.

It probably significantly improved her return on brain damage.

According to other coverage it was not a failure to run the tests, it was changing failed grades to passes.

https://www.wavy.com/news/military/navy/metallurgist-faked-s...

It would be interesting to know how cold the ballast tanks (and connected pipes) gets when the compressed air get released. Air going from high pressure to low can quickly drop the temperature, especially if its already at a fairly low point to start with.

I don't know about how cold deep artic waters may be (I doubt that much), but strength-stress curves are likely why. The easiest way to make something last long and not have unexpected accidents is to overrate it big time - such that any wear would need to become blatant.

To take an ergonomically silly exampke shovel handle could be a three inch diameter of solid steel that can take five metric tons per square centimeter as pressure without bending. If it wears out enough that it could snap by human body weight it would have to be worn quite a bit.

Salt water won't go much below zero. At oceanic salinity levels I doubt there's any place with liquid water below -2C.

Submarines don't stay all the time under water.

They're designed to spend as much time under water as possible (i.e., months, basically). That's the whole reason for nuclear power on submarines.

The official requirement of the US Navy seems to disagree and I tend to trust they know better than us what they design their gadgets for. The metallurgist also thought they knew better, because they're operating submarines for their lives... oh wait.

We have no knowledge of what those castings were used for. It's impossible to judge whether the requirement for these parts was excessive/copy-pasted, or not, and if it was, then by how much.

Okay I think I understand now where you're coming from and it was answered somewhere else in this discussion: when we receive a weird requirement we can ask for confirmation and clarifications and even express disagreement, but we will implement it after all that. Anything else would be at least disingenuous, if not worse.

I'm curious what the requirement was for, and why she thought it was stupid. Nothing justifies how she just skipped it, but it is interesting that she did, and how long it went undiscovered.

IT is entirely possible the requirement was stupid, over-specified, and irrelevant, but that is beside the point.

Reply on news.ycombinator.com