> Since then, actual experience showed that about every 20 years
That's not really a good metric. What you need to do is a time varying analysis and consider the total number of reactors (globally).
Eg. if we hold danger of reactors constant but we increase the number of reactors over time then you'd see disasters being more frequent (but the opposite is true).
Alternatively if we hold the number of reactors constant and pretend that all new reactors replace old reactors you can determine if new reactors are more, the same, or less dangerous.
But to do a real analysis you have to combine both of these (and some more stuff). Otherwise you're pretty much just saying that the average of the set {1,2,3} is 2. That 20 year thing is a bit contrived and I'm surprised it has popped up a few times.
That's not really a good metric. What you need to do is a time varying analysis and consider the total number of reactors (globally).
Eg. if we hold danger of reactors constant but we increase the number of reactors over time then you'd see disasters being more frequent (but the opposite is true).
Alternatively if we hold the number of reactors constant and pretend that all new reactors replace old reactors you can determine if new reactors are more, the same, or less dangerous.
But to do a real analysis you have to combine both of these (and some more stuff). Otherwise you're pretty much just saying that the average of the set {1,2,3} is 2. That 20 year thing is a bit contrived and I'm surprised it has popped up a few times.