dpb wrote:

There are some differences but the really significant ones only
come into play over a longer time frame than an initial response
to an accident. And, of course, to date no CANDU has been tested
to see if the implementation holds for them in practice as well
as the theory...

When you look at the over-head pictures of the Japanese reactor sites,
they really don't look like they were swamped by a tsumai wave.

So I don't understand how or why their coolant circulation systems
failed. Seemingly not from mechanical / structural breakage.

On-site diesel electrical generation can be housed in water-tight
buildings and can operate for days or weeks - given a competent fuel
supply.