My (albeit short and 30+ years ago) experience in the nuclear industry as an apprentice would lead me to believe that it's about quality control and inspection as well as it probably being a long-winded task if the pressure vessel is 8 inches thick. I'd think it would need to be entirely welded at that depth, with each weld pass being X-ray inspected.

Everything I did when briefly working in a reactor (3 months of my 4 years) took a LOT of time because of safety checks and routine.

Sounds like my buddy who's a programmer for our social services. He said they spend about one day per week coding and the rest doing code reviews.

That sounds better than what my buddy who worked at Microsoft did, where he said he programed probably <10hrs a week and spent more time in meetings.

It's good that machine learning is taking over the coding part, so we can focus all our engineering efforts on meetings...

Personally, I'd love to use machine learning to automate the meetings instead.

And then all the meetings are about how you make time for resolving all the bugs introduced by copilot. But you can't cancel any meetings, so you need to create a new meeting because it is a new issue. And copilot is cheap so why not just buy more compute time? Surely that's cheaper because humans and machine solve problems in the amount of time and to the same degree of satisfaction, right? We should probably schedule a meeting so we can make a project to find out. We can probably find out if we measure the number of lines of code each can write. It'll be easy.

> Personally, I'd love to use machine learning to automate the meetings instead.

The old ‘could this meeting be an email?’. Rarely wrong, always controversial.

`Could this email be deleted?´

The luxury of having 10 hours a week to code

You call that luxury! When I was a lad we were put into a cardboard box on the middle of t’motorway, made to code an entire distributed database in our lunch hour, and if we had less than 100% coverage the project manager would cut us into little pieces and dance on our graves.

you had graves?! in my day management would sell our organs on the black market and feed the rest of our corpses to the HQ lobby piranha aquarium, though they'd still charge our estates for the tombstone retainer of course.

You had piranha tanks? Back in my day our managers just threw us in a empty concrete room with no windows and no outlets, tossed a computer at us and said "fix it". If we were lucky they'd say what "it" was. If we were extra lucky, they'd throw us a keyboard. Not that that mattered anyways.

You had keyboards? Luxury, back in my day we had to stamp punch cards with the tips of our fingers. If we made a mistake, the head operator would cut one of our fingers off with a bread knife. We didn't mind though, it made us tough, that and sleeping in the car park wrapped in tape from old backups just t'stay warm.

Did you sleep in cars in the car park or on the car park concrete?

You had concrete car parks? When I were a lad, we had to inherit from car park and implement parking spaces ourselves. And if you had an overflow you had to walk -2147483648 miles to work.

Ahhh, HN … thank you all :-)

That's my issue with articles like this. It sounds pretty great, but it provides barely any context. The quote at the end is a quote from the company's project lead, which sounds like it was taken from a company press release. So, yay?

I'm not a welding engineer.

I imagine half of it is checking the welds are to code and certifying it.

Might have be a foot thick wedge that was filled one bead weld at a time.

I once worked with a guy who had his nuclear ticket, on a team of welders working on high pressure steam and pressure vessels. Even many of the steam welds had to be x-rayed and he said the nuclear process was more intense, not just testing but technology and performance.

What's a "nuclear ticket"?

Means he had a certificate that showed he was considered (by the US navy in this case) competent in making welds suitable for building or repairing a nuclear reactor or related devices.

We didn't need that, of course, but because high pressure steam is so dangerous we used only union welders who were rated master and who already had experience in steam plumbing. I don't remember if there was a special ticket for that, but everybody we hired was interviewed and had to do some welding as a part of it (the equivalent of a leetcode test on the whiteboard I suppose). If it isn't obvious: I was not qualified to usefully interview any of them.

These guys were great to hang around with. go have a beer, horse around, etc with but wow, when they were on the clock all of them (machinists, etc too) were deadly serious. No fucking around.

"when they were on the clock all of them (machinists, etc too) were deadly serious. No fucking around."

Sounds like true professionals. There is no fucking around with serious work, when real lifes depend on it. I wish, more people had that attitude, then there would be less mess everywhere.

I used to work at a shipyard as a welder. The zero-bullshit professionalism when it was go time was genuinely satisfying to be working in: there was no way someone could fake their skill- literally everyone I worked with was competent.

If it weren't for the literal toxic working environment, I'd be tempted to quit tech and return!

Hah, an old family friend related a welding interview story on the other end of the spectrum.

He had been working as a machinist and welder for BART, on a mixture of road and rail car maintenance as well as other fixed infrastructure work. He interviewed for a job at Lawrence Berkeley Laboratory, but was intimidated when they asked him to "weld some gold tissue paper".

In America, they're all based on the American Society of Mechanical Engineers (ASME) Boiler and Pressure Vessel Code (BPVC). It covers all the different types of nuclear certifications for organizations, because normal welding certs don't cover the nuclear industry.

The "nuclear ticket" for most is probably:

> NPT - Fabrication of parts, appurtenances, welded tubular products and piping subassemblies

From ToughNickel [2]:

> BPVC Section IX outlines the brazing and welding qualification procedures for nuclear pressure vessels. Section IX also gives the welding and brazing qualifications required to work on equipment for use around nuclear reactors. Welders must be qualified per the procedures listed in paragraph NCA-5250.

Also, appurtenances up above, not actually a typo.

[1] https://www.asme.org/getmedia/eaacef3f-e8b5-44a4-b0c6-451e1e...

[2] https://toughnickel.com/business/Nuclear-Welding-Certificati...

I'm not in this field, but in trades tickets are technical certifications (e.g. journeyman ticket, fall arrest ticket)

Presumably its a certification to weld on or around nuclear equipment.

Welding stuff at this scale/depth takes time. It isn't just inspections either. There is lots of prep work. Sometimes the material itself has to be pre-heated/cooled in order to prevent warps/cracks in huge blocks of metal. But, for the same reasons, that preheating itself cannot be done too quickly. Multiply this by hundreds or thousands of weld lines and 12 months doesn't sound so long.

Why does there need to be 'hundreds or thousands of weld lines'?

Typical welding can only safely deposit a certain volume of metal at a time due to physics constraints.

Either the puddle gets too large and sags, or flux needs to be removed between passes, or the work piece needs to be moved with a certain speed/force that doesn’t scale, or the heat can only be deposited on a single surface at a limited rate, which causes problematic distortion and limits penetration.

Welding a 6” thick piece of anything with stick/tig/mig may take dozens or even hundreds of passes depending on geometry, material, available welding positions, etc.

Welding the same thing with stir welding may take 100+ or even 1000+ tons of pressure in a moving weld fixture.

And because of that, will also cause major distortion to the part unless done very carefully.

Electron beam welding doesn’t have many of these issues because the electron beam doesn’t get absorbed in a point at the surface (and the beam can be very narrow), allowing very deep penetration compared to something like an electrical arc, also allowing a lot less total heat input to get the welded spots to fuse, and that also means weld joint shapes can be used that won’t need filler metal being added, or movement, it doesn’t have puddle size issues or flux removal problems (necessarily). Due to the lower total heat input, it also wouldn’t have as significant distortion effects.

The internal tension/distortion created from welding heat can be really mind blowing - we’re talking enough to twist automobile frames into pretzels if not done correctly, or rip apart armor plating and heavy machinery joints just from the act of cooling.

It also doesn’t have force scaling issues like stir welding (generally).

It would require completely, perfectly clean surfaces and a non-reactive environment, and clearance for the beam emission equipment though.

Pretty neat overall!

I would like to subscribe to your newsletter.

Thanks for that fascinating explanation!

I am thinking that this technology will apply nicely to tank armour?

The UK has lost it's capability to make tanks from scratch, maybe this approach could be used to substitute in some way?

If it was homogenous steel plate, definitely. My understanding is that modern tank armor is actually a composite with many layers of materials like hardened steel, rubber, ceramic, etc. Those allow disruption of kinetic penetrators and shaped charges more effectively than a homogenous piece of steel, so effectively more armor 'thickness' for a given weight/thickness.

No idea how they weld those things together!

There are some neat videos on youtube showing some pretty cool simulations of the how and why.

Yup good points. I now rememeber "dorchester" and "chobham" armours. I must say that I too am at a loss to imagine how they are fabricated into solid hulls!

Not sure if this applies, but IIRC some welds are embedded in complex structures that require only very skilled workers... and sometimes only a left handed guy can access the surface.

Does the left-handed welder get to charge a premium? Or is a rightie expected to make it work?

Yeah, I'm wondering about that as well. Did this exact design take 12 months previously? Or did a 30ft-wide BWR vessel take 12 months? That would be a pretty significant difference.

I did a small amount of nuclear-grade welding myself and it was always one hour of welding, one or more days of testing. So if they could replace many small passes with one all-the-way-through pass, that would vastly decrease production time. But even then, testing a weld that spans the length or circumference of a vessel is still going to take a very long time. Since they don't mention any massive, profound, fundamental improvements in post-weld testing methods, I'm guessing they just left that out of their "24hr" claim, which makes it very misleading.

It could be that the final testing would also have to happen after the 1 year of conventional weld-test cycle. But yeah it's not clear.

The heavy forged sections of a pressure vessel are really thick and there is a limit to how thick a single weld can be using conventional techniques. So you have to do many passes. Each one of those requires bringing the temperature of the massive forged sections that are being welded up to temperature, cooling it, and then x-raying the partial weld before proceeding with the next cycle. The actual time spent doing the weld is cumulatively not more than a week.

gold dust of an answer - thank you.

The materials they are welding can have lead times of months/years from the forgeries and the quality requirements are insane because of the safety/enviromental/economic considerations.

So therefore the work has to be inspected[0] regularly and everything has to pass before work continues.