General Electric boiling water reactor stress corrosion cracking

Cited as the exemplar of attacking a problem across all three of Tom's circles (metallurgy / mechanical / corrosion) rather than just the metallurgist's circle.

I've never forgotten my three interacting circles — the metallurgy circle, which is microstructure, is one, and there's only so much I can do after I've made my material. Then there's the stress circle. If I relieve the residual stresses or shrink the strain, if I keep it nice and clean — my environment, don't let hydrogen get in, don't let boric acid get up in there — if I keep the system clean by my process or by putting hydrazine in on shutdown to get the oxygen out of the water, I shrink that circle. You can't solve all the problems in the metallurgy circle. Some you have to solve in the mechanical behavior circle, which is residual stresses. Some you have to solve in the corrosion circle, which is the environment. You want to think of it that way. When General Electric had their big stress corrosion cracking problem, they worked on all three circles. One of my problems with most of the philosophy of welding in this country right now is that everything's in the metallurgist's circle. You don't solve it just looking in one area. You attack every area. So that's your question.

Sensitization of stainless steel welds in BWR safe ends cost GE "a couple billion dollars" in the 1970s. Westinghouse PWRs avoided it by using Inconel (10× the cost) for higher temperatures. The teaching consequence of the John Wulff sensitization patent and the broader carbon-control rationale.

In fact, John Wulff, my academic grandfather, had a patent in the 1940s to sensitize stainless steel, heat it in this region, get something like this, throw it in nitric acid overnight to make stainless steel powder. The nitric acid doesn't attack the 18% chrome, it does attack the 10% chrome, eats away the grain boundaries, and you get typical intergranular cracking. This cost General Electric a couple billion dollars in the nuclear reactor industry in the '70s from the welds getting sensitization cracking. If you were Westinghouse, you used Inconel — about ten times the price of stainless. If you're building a pressurized water reactor, because of the higher temperatures, they needed Inconel. General Electric had a boiling water reactor, they could use stainless — less expensive material but more susceptible to stress corrosion cracking. You get cracks through the reactor safe end. That means that you're not going to be able to flood the reactor with water if you have a near meltdown. And you will not have a near meltdown — you'll have a real meltdown.

GE's response to in-reactor sensitization-driven SCC, via the 304→304L→ultra-low-carbon→LN-grade (low-carbon, nitrogen-strengthened) sequence. Tom uses it to motivate the AOD-driven cost collapse for low-carbon stainless production.

Tom's marquee case for sensitization in service. GE built 1970s boiling water reactors using 304 stainless steel and incurred a two-billion-dollar worldwide problem from stress corrosion cracking in ultra-pure water with parts-per-million chlorine and oxygen. Motivated industry-wide move to 304L and 304LN.

Is this a problem? Yes. General Electric built nuclear reactors in the 1970s, and what happened to them — they were surprised — they're using the world's cleanest water. If you take absolutely pure hydrogen and absolutely pure oxygen gas, react them to make water, and measure the electrical resistance, it will be 18 megohms. Whenever you put any mineral in there, some sulfate or chloride, the resistance goes down because the ions conduct electricity. When we talk about nuclear-reactor-grade water, the goal is 18 megohms — that's absolutely pure water with no ions in it. Distilled water — you don't even talk about resistance; it has lots of conductivity because there are still enough ions in distilled water. But when you're down to parts per trillion or less of salt in the water, you can start measuring water purity in megohms.