The Windscale Fire

Image credit

In 1957 the fuel in an air-cooled tritium production reactor in England caught fire in what remains the worst nuclear disaster in history outside of the former Soviet Union or Japan. You can read more about it here:
The Windscale fire

So much went wrong at the Windscale pile it's hard to find anywhere in the chain that's blameless. The reactor never should've been making tritium in the first place; it was designed to produce plutonium and was reconfigured to run at the higher temperature needed to convert lithium into helium and tritium without an adequate design review. The fuel powering the reactor was metallic uranium, not the chemically inert uranium ceramic used in modern reactors, and hot metallic uranium in an air-cooled reactor is an obvious fire hazard in retrospect. To increase their confidence that they could run the reactor beyond its original redline, technicians installed a temperature-monitoring system that didn't adequately monitor temperature. There was no containment building around the pile. Nobody knew what the reactor was doing when it caught fire, except that it was running hotter than its designers intended and that there was no easy way to put it out. In the event, the fire raged for days, releasing large amounts of radioactive dust, before fire hoses were able to drown it.

There was at least one good decision made in this entire kerfuffle: Physicist John Cockcroft's insistence on putting a filter atop the reactor's exhaust chimney. As it was, the amount of radioactive iodine released resulted in about 240 additional cancer cases in the Cumbrian countryside. Had those filters not been in place, the damage would surely have been much worse. When designing something complex and dangerous, the lesson seems to be, listen to the worryworts.