Read John Fuller (or Gil Scott-Heron) and you'd think We Almost Lost Detroit when the Fermi I fast breeder reactor went online.
It had serious problems; zirconium cladding came loose in the reactor vessel and clanged around in the liquid-sodium coolant flow, eventually leading to abnormal operation (and a partial melt-down) which caused the reactor to scram: it shut down.
The operators poked around, found the problem, spent nearly two years gimmicking up the tools and fixing it, and eventually got the thing more-or-less running again. It was never quite what the designers had hoped and in another couple of years, it was shut down.
You may have noticed that Detroit is still there.
So how come? It's not like a liquid-metal cooled fast breeder reactor is an especially forgiving design. But the thing had working safety systems and the operators paid attention to them. In the face of a design or construction problem, they didn't push the system past the point of no return, hoping to win brownie points or dodge a dressing-down, they shut 'er down and began to figure out what went wrong.
And we didn't lose Detroit. Or at least not to Fermi I.
4 months ago