M Paquette
Moderator Emeritus
I don't think you quite understand the problem. You're thinking of single disastrous events which are quite unlikely and consequently probably not worth worrying about. But real life disasters sometimes arise through the coincidence of events, each of which taken singly would be manageable, but which combine in a way nobody anticipated to produce a very bad result. If you don't know in advance what combinations can occur, how can you calculate the probabilities so you know you are within safe margins? (And by the way, Wikipedia has an entry Perfect storm - Wikipedia, the free encyclopedia .)
There's nothing unique to nuclear power regarding complex, or multiple point of failure, accidents. The Banqiao Dam accident certainly demonstrates that. In all cases the failure engineering folks try to make their best estimates of what could possibly go wrong, under the worst possible conditions, and design for that.
One of the things that has changed over the years is the accumulation of real world experience that informs engineering. The recent earthquake in Japan was, quite literally, a 'once in a thousand years' sort of event, which obviously was not considered in their original power plant siting and structural design.
In dealing with engineering in the face of complex events, the sorts of complex interactions possible can usually be reduced to a relatively small number of scenarios. Ways to address these can then be designed and incorporated into the finished product.
The gotcha, as always, is those pesky 'unknown unknowns.' Those are best addressed with a healthy dose of paranoia and what sounds like absurd hypothetical problems. "Suppose a double-ended shear of both main cooling loops occurs while operating the reactor at 100% power after four or more days. What happens?" "Suppose the control rods all disappear? What happens?"
The bad news is that we didn't know as much when the GE Mark I BWR was designed as we do now. The good news is that we know much more now than we knew back in the late 1960s.