[lispm]

The safety requirements are much higher than in the past.

There are lots of problems. For example the control systems are now all digital. Now find THREE INDEPENDTLY designed and programmed control systems, so that these are available redundant and don't fail at the same errors. How many companies are there that can get their stuff certified at the required levels?

[hga]

Plenty I'm sure, after all aren't there firms doing this work right now (in that the world never stopped building nuclear power plants? I've read that this approach is used by Airbus for dual redundancy.

Given my very limited understanding of the requirements, isn't nuclear reactor control a lot easier than fly-by-wire avionics? And somewhat safer, in that good designs (e.g. not the RMBK) are designed to safely passively fail as Three Mile Island Babcock & Wilcox design quite nicely did in a near worst case accident (about half the core melted with 20 tons of uranium flowing to the bottom).

[lispm]

For Finland there was ONE company capable of doing that, IIRC. They had a hard time finding other companies willing to invest in that area.

Airbus sells many airplanes. Boeing, too. It took a long time to go to fully digital control systems on airplanes.

I don't think any one wants a reactor to passively fail. That's a worst case scenario. If your reactor control system has to fall back to that, then it has no chance. The control system has a) to keep a reactor at all times in safe operation conditions and b) it has to work under failure conditions. If the core melts, then this is an economic loss of billions. The reactor in Finland will cost upwards of 6 billion Euros (last estimates I read were at 5.7 billion Euro). No one would want to have a core melting at such an expensive machine.

Read more here:

http://news.bbc.co.uk/2/hi/europe/8138869.stm

http://www.guardian.co.uk/environment/2009/nov/27/nuclear-po...

This sounds even worse:

http://www.electricenergyonline.com/?page=show_news&id=1...

[hga]

Agreed that no one wants a passive fail in these type designs; Three Mile Island Unit 2 was in operation for only three months before the accident took it off-line for all time.

Hmmm, reading those links and the most relevant of the items linked by the last, it sounds like:

The French may have done an inadequate job in the control system design; at the very least multiple other country's regulatory bodies are concerned about the same and very basic thing (module independence) and that sounds bad to me.

The other conclusion is that it was insane to start building 3? of these EPRs without getting all the way through the design process (!!!) let alone getting the first up and running. That's just appalling bad project management, which I think the U.K. regulator implied in its complaint (the "experts" knew of the control system architecture problems but management wasn't listening to them).

France has been technically quite successful in their nuclear power program (ignoring the experimental Superphénix) and was finishing construction of their last operating plant as late as the last month of 1999. I wonder what happened (well, we can guess well enough).