Well for one submarines have parts that are over water, and can be stored in dry bays on the continent, for repair or construction (also the steel itself is stored on the continent prior to being used). For second the coldest wind chill recorded in the US was -100°F at McGrath, Alaska in 1989. So it's not too much of a stretch to take this value a minimal value for testing the material.

McGrath, AK is at 330ft of elevation in a river basin surrounded by mountains, 170mi from it's nearest shoreline (as the bird flies). Air temperatures at sea level never reach the extremes of coldest air temperatures on Earth, which are always at elevation. This is somewhat due to air temperature being a function of air pressure (the denser the air, the warmer it gets), and air pressure is greatest at sea level. But it is largely due to the freezing temperature of sea water, and that the warmest water will be at the surface due to convection. Warm water rises to the surface, and the coldest water temperatures are at the deepest bottom of the ocean, and never colder than the freezing temperature of sea water.

Wind chill temperatures only affect humans (and other living things). Materials cannot fail below the true temperature of the environment.

That is not true. We don’t necessarily call it wind chill, but evaporative cooling certainly can decrease the temperature of non-living things below the temperature of the atmosphere.

Wind chill only increases the rate of cooling, not the absolute value.

Evaporative cooling by definition cannot cool below the wet-bulb temperature.

But wet bulb temperature != true temperature of the environment, no?

> Wind chill only increases the rate of cooling, not the absolute value.

Right. The absolute value is related to the evaporation process, which is affected by wind.

> Evaporative cooling by definition cannot cool below the wet-bulb temperature.

Wet bulb temperature is not “the true temperature of the environment”. Besides, that is true for humans as well, so again not any different between living and non-living things.

The main point is that a thing left in a given environment can end up at temperatures lower than that of their local environment. This process depends on the atmosphere in contact with the thing, which depends on winds.

We're dealing with two misconceptions here. Wind chill is not affected by evaporative cooling. You can of course go below ambient with evaporative cooling, or air conditioning, or any number of methods.

Wind chill is explicitly and only the experienced subjective temperature caused by cold air being continuously replaced by the wind.

For example, take the US National Weather Service's wind chill formula, which only uses wind speed and tempurature as inputs and provides the wind chill in W/m².

Source: https://www.weather.gov/media/epz/wxcalc/windChill.pdf

> Wet bulb temperature is not “the true temperature of the environment”. Besides, that is true for humans as well, so again not any different between living and non-living things.

Wind chill and evaporative cooling are independent mechanisms, and the wind chill value and wet bulb temperature are separate values.

My reply was to your original comment, which said:

> For second the coldest wind chill recorded in the US was -100°F at McGrath, Alaska in 1989.

That value is not the temperature that an object could have reach at that time and place. It is the wind chill, an approximation of how cold it would have to have been on a windless day to make a person subjectively feel equally cold.

You always need to be pessimistic about operating conditions, because there is no way you’d be able to imagine everything that will happen. If you want to be sure that your structure copes with thermal contraction at -30°C, it makes sense to test at lower temperatures because even if it never gets that cold, other effects might have a detrimental effect that compounds the low temperatures (probably not for thermal expansion, but this is very important for fragile-ductile transitions). This becomes even more important in submarines, where the added cost is very small compared to the cost of losing a boat.

That’s just good engineering, and someone doing tests who does not understand that is clearly in the wrong industry.

That makes pretty good sense in the warmer end of temperatures, as even though sea water temperatures haven't been recorded above 130°F, friction may cause water temperatures against the hull to be much higher, and though probably never higher than the boiling point of water, it is possible for the temperature of steam to be much, much higher, though only at extreme pressures. Plus, air temperatures at sea level will be warmer than the water temperature in extremely warm areas.

But it is not possible for sea water, anywhere in the oceans, to be much colder than freezing, nor does the air temperature at sea level begin to approach the coldest temperatures recorded on Earth, which are always recorded at elevation. I can't seem to find the record lowest temperature recorded at sea level, but I can't imagine it could be much lower than a few dozen degrees below 0°F.

The question that comes to me is, why is that -100°F metal test even a standard? On the one hand, it is vanishingly unlikely a watercraft could ever experience temperatures that low or even 50°F above that, and on the other hand, temperature measure goes down all the way to -459.67°F. So why aren't they stress testing at -559.67°F? And if tests at those literally impossible temperatures were required, could we at least agree that that would be objectively stupid? And if so, there must be a way to determine what test temperatures are stupid to test at in a sliding scale diminishing towards test temperatures that are less stupid to test.

Though other comments are intolerant of the metallurgists' actions, fundamentally, she was correct, although it is clearly unethical she faked tests results, her reasons for doing so are not entirely irrational. The more intelligent and the more expert an individual is, the less tolerant they will be when asked to do things they know could never matter. Though the military is famous for requiring irrational actions, such as constantly digging a ditch to only fill it in again (which I believe has something to do with proving compliance and following orders, maybe testing the command structure), the woman was not in the military. While I do not condone faking tests, I have some empathy for her, as it is difficult for intelligent individuals to intentionally do stupid things, especially when they are required to do so by less intelligent individuals. I could only guess that most of HN members are a lot like her, and less tolerant of objectively stupid requirements.

> metallurgists' actions, fundamentally, she was correct, although it is clearly unethical she faked tests results, her reasons for doing so are not entirely irrational.

She falsified tests and produced parts that were not up to the specs required by the customer. That is very much wrong, regardless of how justified the specs are.

If the navy is overly pessimistic, the discussion needs to happen with them. Second guessing and assuming that they won’t really need that anyway is what I’d expect from a cut-throat Chinese factory selling in AliExpress, not a supplier for a major western power’s navy.

You test at the extremes so you know it'll perform in the usual.

I'm sure that isn't always true, that if something performs at the extreme it necessarily means it performs in the usual. I find it difficult to believe that, if normal operating temperatures are going to be somewhere between the upper and lower limit temperatures of sea water and the upper and lower limits of air temperature at sea level, that a metal batch that failed at -100°F would need to be rejected. It's one thing for the purposes of science, to know, but my understanding is that engineering and material specifications for an actual thing needs to be more practical, especially for something that already costs $3.5B.

The big reason for these kind of "extreme" tests is that they help you verify the models and simulations you've done. If your model breaks down at -40f, you want to know that, and figure out why, what that means for the actual material, what that means for the specs you promise, and what that means for the product.

It also can be useful if these extreme test are better at revealing manufacturing defects. Maybe you literally only want the most perfect 1% of steel coming out of the foundry to actually be used, because you believe that will make a better submarine. In that case, sure maybe -100f is overkill for the specific application but you have to remember these objects are literally designed to have explosives near them. Being EXTREMELY GOOD at resisting brittleness literally will save sailors lives when it comes to close calls from depth charges.

> I'm sure that isn't always true, that if something performs at the extreme it necessarily means it performs in the usual.

That’s why we test over the operating range for the different parameters, not just at the limits.

> It's one thing for the purposes of science, to know, but my understanding is that engineering and material specifications for an actual thing needs to be more practical, especially for something that already costs $3.5B.

You need to be pessimistic about operating conditions if you don’t want it to fail unexpectedly. That is absolutely critical in engineering. Now, you set the boundary differently depending on the field and the importance of the widget you engineer (tolerances are quite different for a phone than for an aircraft), but the concept itself is part of engineering.

There are cases where practicality is less important than reliability. Things like space probes and a lot of things related to national security. That is one of the reason why they cost billions in the first place.

My best guess is that it was a proof designed to verify the success of the casting process, not resistance to theoretical real-world exposure.

All metals have a ductile-to-brittle transition temperature. Ductile failures are generally preferred to brittle (think bend/dent vs shatter). Some steels have a D2B temp around 0°C, which makes them basically untrustworthy in colder environments.

The D2B Temp can be lowered by altering the elemental makeup of the metal, controlling the crystal structure, and other various metallurgical techniques. These same techniques are utilized to produce metals with other desirable properties, like increased ductility, higher tensile strength, corrosion resistance, etc.

Casting, especially large complex castings of custom tailored alloys, are temperamental. Controlling the elemental makeup is relatively easy, but controlling the crystal structure is difficult.

My guess is that they were paying for a specific alloy & temper of metal because it had their desired properties, and successfully cast samples would have a D2B below -70°C.

It is a standard & easy test to perform, but I could see how the tech running it would not connect the dots above. That being said, they don't need to know why the test was ordered & paid for, they just need to run it.

I believe it would look something like this: https://www.youtube.com/watch?v=9990wY1DEhk

The extreme cold temperatures of -55°F and colder are never recorded at sea level. You only get those kinds of temperatures at elevation. The ice in that video is no colder than 32°F, as the freezing temperature of water is the same as the melting temperature of ice. So a submarine breaking through the ice would never experience temperatures 32° below the freezing temperatures of water, let alone 132° below freezing temperatures.