As an example, the initial safety assessment may assume that a) 5mm is the limit for crack width, and b) by the time cracks are 1mm wide, they will be 2mm wide a month later, and, because of that growth rate, pick 2mm as the “take out of commission now” point (nb: numbers 100% made up; for illustration only)

If, in real life, it turns out that it takes a year for a crack to grow from 1mm to 2mm, the model used can be improved. Result _could_ be that the “take out of commission now” point gets moved to 3mm.

They also may remove some stuff with 2mm cracks and test it for strength, and learn more about how the stuff behaves after being irradiated for decades. That, too, can change (in either direction) the life expectancy of the reactor.

Perhaps, but surely you realize that there is currently a lot of pressure (and therefore bias) for keeping the plant operational: Jobs, energy supply and the operator's profits. And the thing is, those biases affect both the regulator (the government), the local population (who probably want to keep their jobs) and the operator (EDF Energy).

So at this point in time I would be skeptical of new experiments and model updates performed by the operator and approved by the regulator. IMHO: Don't approve the model change, do some lab experiments / examine more empirical evidence over time and then possibly change the model for a _future_ reactor.

I do, but there also is a lot of pressure to close nuclear reactors because the perceived danger of nuclear power is a lot higher than its real danger.

Also, would you be in favor of _not_ changing the model for current reactors if updated models predict a shorter life time? If you think the model should be updated in those cases, that introduces bias. It also could mean reactor owners would stop making measurements, and that, in turn, would mean we would learn less about making safer reactors.

If such conditions exist, they can be determined while the reactor is shut down. Otherwise that's recklessness. The original safety margins are there for good reason. Setting them aside without thoroughly testing new limits with justifications that include reasons why the original limits were wrong is reckless. "Nothing bad happened" is not a good reason.

Safety limits are very conservative on nuclear power for fairly well established reasons though. If your worst case scenario makes the two largest metropolitan areas in a country uninhabitable for a while, for the current power equivalent of around 6 months new build in offshore wind (Hunterston is 1 GW and we are currently building out around 2 GW a year in offshore wind), then perhaps the caution is worth it.

The strategic industrial behavior is to accept conservative safety limits initially and then apply for exceptions as needed to keep the site running. You'll see the same thing from nuclear power to various chemical plants to NASA.