As I said, I don't know the safety systems of CANDU :)
Large heat sink buys you time to get the cooling working again. Ultimately the heat must be transferred out somehow. In modern reactor designs these systems are usually designed to passively for very long times. With CANDU's it is very likely to be the case too.
In nuclear engineering one must always consider also the chance, that not everything is working as designed (like the destroyed diesel generators and external power sources at Fukushima). PRA (Probabilistic Risk Analysis) is used for that. PRA analysis are used for detecting most vulnerable systems in a nuclear power plant and this information is used to design new safer reactor types and to update the old ones little by little to be still safer. Harsh weather, seismic activity etc. is also considered in these analysis, but sadly in Fukushima even detected vulnerabilities didn't lead to improvements in time.
Even, if everything is done as well as possible, there is still a chance (although with modern designs almost arbitrarily small) that under certain conditions all the safety systems will fail. An ultimate example of such an event, would be a 100 km meteorite smashing the plant to atoms.
We agree on older plants being not as safe as new ones. My original sentence was about things you just can't prepare for (with any reasonable means). After that I just wanted to say, that it is possible to build plane-safe NPP.
Large heat sink buys you time to get the cooling working again. Ultimately the heat must be transferred out somehow. In modern reactor designs these systems are usually designed to passively for very long times. With CANDU's it is very likely to be the case too.
In nuclear engineering one must always consider also the chance, that not everything is working as designed (like the destroyed diesel generators and external power sources at Fukushima). PRA (Probabilistic Risk Analysis) is used for that. PRA analysis are used for detecting most vulnerable systems in a nuclear power plant and this information is used to design new safer reactor types and to update the old ones little by little to be still safer. Harsh weather, seismic activity etc. is also considered in these analysis, but sadly in Fukushima even detected vulnerabilities didn't lead to improvements in time.
Even, if everything is done as well as possible, there is still a chance (although with modern designs almost arbitrarily small) that under certain conditions all the safety systems will fail. An ultimate example of such an event, would be a 100 km meteorite smashing the plant to atoms.