This keeps coming up and I keep wondering how in the world are large transformers that complicated to build. And I don't mean that in a dismissive way but a "boy do I have a gap in my knowledge of the world" way :-)
I think that is a pretty common reaction when it comes to physical/mechanical objects. Another way of thinking about it which might be helpful is. Hundreds or thousands of very talented engineers have been working non-stop for 150+ years in a transformer arms race.
This has resulted in incredibly high performance designs and incredibly high requirements.
No doubt. But same could be said of, say, automobiles, or microchips, and many other things. Even larger objects like power plant turbines seem relatively frequently manufactured. Large transformers on the other hand appear to be closer to nuclear power plants in rarity/difficulty of replacement, and as a result for me they have a very high difficulty-of-replacement-to-perceived-complexity ratio.
Large transformers are built to a very specific custom specification. Each one is a custom engineered unit. A transformer is basically a sealed vessel full of volatile oil that you pass an extremely large voltage and current through. The machine has to be perfect. The smallest clearance issue, insulation problem, etc and you have a bomb. And that's just making it safe. We haven't even talked about making it effective.
Transformers are the sort of think that are not rocket science to build, but every single detail is very very important. It requires a team of people to design, construct, and validate and all these people need to have the skills from doing it many times before. These people and facilities simply don't exist in large enough numbers because they are very expensive.
There are also supply chain issues to consider - the steel used in transformer magnetic circuits for example is a very specific metallurgy and construction. During COVID it was quite difficult to find. Poor steel means poor efficiency which at the scale of these machines can mean millions of wasted mwh of power over time.
Thanks for the detail, this is very neat. I wonder if transformers like this are in an "uncanny valley" of manufacturing where there is barely not enough demand to bootstrap mass production like for example turbines. Turbines have very exact tolerances as well, but it seems that with higher unit counts orgs like GE managed to churn them out.
Perhaps the liquid piece is also making things more difficult; I'm sure a dry object is a lot easier to warehouse until needed.
They are dry until delivered, because they already max out the transport vehicle's capability (often highly specialized transformer carrier train cars with like 16 axles at each end to spread the weight).
