California's last nuclear plant turns to generative AI for filing and finding the fine print A California startup is deploying what it says is the first commercial installation of generative AI at a US nuclear power plant, but don't get too excited (or worried) about what it's going to be doing quite yet – it's a pretty run-of-the-mill use in an enterprise environment. Atomic Canyon and Pacific Gas and Electric (PG&E …
Just please don't let the dolts that run the now defunct San Onofre Nuclear Generation Station (Yes, SONGS) any ideas.
In case you're not aware, San Onofre shut down permanently after "upgrading" the generator system in such a way as to cause the thing to leak like a radioactive tea kettle. I believe that's what football (aka soccer here in the States) fans call an "own goal."
"In case you're not aware, San Onofre shut down permanently after "upgrading" the generator system in such a way as to cause the thing to leak like a radioactive tea kettle. "
Hmm, the report I remember reading says that after the main steam turbine had been re-installed after a maintenance at the manufacturer, it turned out to have an imbalance when operated above 80% of rated capacity. Regulations don't allow for a permit if the plant can't run at full nameplate capacity. The cost to have a new turbine manufactured after determining that repair wasn't an option, was more than the anticipated profit from operation before the Hot side would time out. They had no choice but to shut down completely. The nuclear side was right and tight. It was the power conversion part that was November Foxtrot Golf.
...to do the job that could be handled by a couple of relatively low paid full time archivists to me.
You remember those, the people we used to pay to know where all the technical documentation was and keep it properly indexed and up to date?
Step one should be to design and implement a complete record keeping system.
The only place I have organized in a way that takes a computer to find things is a rack of drawer cabinets with small parts. It made no sense to put one transistor in a drawer, label it and keep them in order. it made more sense to fill a drawer with transistors and make a catalog entry so I can find that part again. The really common transistors I use all of the time do have their own labeled drawers, but I also buy estate lots of electronics and wind up with plenty of vintage parts that are nice to have around but only if I can find them. With a big budget and a staff, I would have walls of drawer cabinets and all those parts in their own drawers all nicely labeled. I might need GPS coordinates and differential corrections at that point.
tag em, bag 'em and catalog em.
And by that I mean "stick a bunch of 'em to a sheet of paper" then bag the sheet or drop it in a folder or drop it in a paper drawer"
You can index by sheet number and X-Y coordinates with the device number written on the paper
it's worked for me for decades (non static sensitive devices - it's fractionally more complicated for stat senstitive and involves foil tape)
"You can index by sheet number and X-Y coordinates with the device number written on the paper"
I do a rough sort so there's a plastic storage container with assorted TO-3 package transistors which I have a lot of in zip lock bags with the part number on them. Another box has small transistors up to the TO-220 package (ish) format. Another box has nearly all of my surface mount components of all sorts with heavy paper dividers. At one point I was trying too hard to make everything instantly findable but I noticed that the majority of the time the stuff I didn't need that often took more time to organize than to find again if it was just roughly sorted. Space is also an issue. If I was buying new as I needed something, I had a majority of the parts on hand the things I go to again and again. By buying estate lots of parts, I wind up with loads of things I don't. I could just throw those things away but from time to time I do make some fair money on them by selling them or having them to use on a repair to keep to the original BOM of something. A lot of what I hunt for now are carbon comp resistors since I've been playing a lot with tube circuits. Carbon film resistors aren't rated to a high enough voltage and the ones that are can be expensive. If I need a 470k 1W carbon composition resistor, I have it and it cost me pennies to buy but it likely came in a good size box with loads of other stuff.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
