Report: Nuclear Plants Should Focus On Risks Posed By External Events

mdsolar (1045926) writes "Engineers at American nuclear plants have been much better at calculating the risk of an internal problem that would lead to an accident than they have at figuring the probability and consequences of accidents caused by events outside a plant, a report released Thursday by the National Academy of Science said. Accidents that American reactors are designed to withstand, like a major pipe break, are "stylized" and do not reflect the bigger source of risk, which is external, according to the study. That conclusion is one of the major lessons from the Fukushima Daiichi nuclear accident in Japan in 2011, which began after an earthquake at sea caused a tsunami.

Comments

already done

[Mr D from 63]

External events are considered in US plant design already, this author seems to be a bit ignorant on how the safety case for plants is built. Who cares if we refine the probability of an event is if the plant is already designed to withstand it? More total stupidity disguised as a serious study. Even highly unlikely events are designed against in our plants.

Now, Post-Fukushima, plants are adding response capabilities for apocalyptic type scenarios even though three is nobody that can provide an example of how such an event may happen for the particular site short of some major war type event. Fukushima was simple...don't put reactors that were not design to operate underwater where they can find themselves underwater. Given the situation, the outcome was quite easily predictable.

Re:already done

[tp1024]

It gets better, all the way back in 1975, the Wash-1400 report listed tsunamis as one of the potential ways to knock out the safety systems of a nuclear power plant, leading to the exact same outcome we have seen. All the way to the point of having to evacuate a few thousand square kilometers, given the BWR Mark I containment. (Actually, it was just one thousand, but the rest was off-shore.)

The main problem was that just about ALL the tsunami protection in Japan (both for cities and nuclear power plants) was based on the 1960 tsunami, that came all the way across the Pacific from Chile. The result was quite a disaster, but the worst part was the completely unprotected population and certainly not the nuclear power plants. Contamination is quite reversible, 18500 dead people not so much.

Re:already done

[Mr D from 63]

It will, in fact the reactors near Fukushima experienced major quakes beyond their design basis, remained intact and actually saw little or no structural damage. Only those plants that got flooded by the tsunami had problems, because they were not designed to be underwater.

If a major natural disaster hits, a nuclear plant is probably one of the safest places to be.

Re:already done

[Mr D from 63]

^you can make stuff up all you want, but there are no such thing as safety related electrical pillars. Offsite power supply is not credited in a safety analysis of the plant, and failure of those systems is just fine, as the safety related systems could more than handle the earthquake. The plant was doomed when it was inundated by water and the safety related systems became inoperable.

You should learn more about how a plant safety design basis is developed, and in particular the difference between safety related and non-safety related systems and components.

Re:already done

[brambus]

Gah, the stupid, it burns!

Natural radioactivity is mainly something that hits you from the 'outside'... it hits your skin

Except for the ~5kBq of K-40 in your nerves. And the C-14 in all of your tissues. Also, cosmic radiation doesn't stop on your skin - it's comprised of extremely high energy particles at 1 GeV or more. Those sort of energies make the radiation from nuclear reactors seem like child's play. That is not to say that you'd rather be inside of a nuclear reactor - most definitely not, the flux there is many orders of magnitude larger - but it does show that cosmic rays don't just "hit your skin", but instead fire right through you and irradiate your internals quite easily.

First of all a healthy person has no Uranium or Thorium in his body.

I'd be careful with throwing around superlatives like "none", but it's probably fair to say that the abundance of actinides in most humans would be classed as "trace" at best.

you are again mixing up external radiation by natural sources with radioactive elements incorporated into the body

Except that both K-40 and C-14 are both natural and inside your body. In fact, we use C-14 abundance in tissues to date when organisms died. Whether something is or isn't natural has no bearing on where it is harmful.

The fallout is measurable every where in north Japan.

This statement, while true, is misleading, or at the very least oversimplified. We have extremely sensitive measurement equipment, but the mere detection of the presence of a radionuclide does not in itself imply any danger from it. What needs to be assessed is the particular type of radionuclide, its abundance and sample distribution, in order to be able to at least roughly assess the potential biological impacts. In pretty much any scoop e.g. topsoil you'd be able to find all manner of toxic stuff, from mercury through arsenic, lead and even to uranium - this is simply a consequence of the magnitude of Avogadro's number.

I'll leave you with just one tiny factiod: long-haul flights are associated with elevated exposure to cosmic rays, easily 20-30x sea-level background and comparable to some of the hotter parts of the Fukushima exclusion zone. This has been repeatedly assessed and demonstrated. As such, one would expect to find radiation-related cancer clusters among airline crew, who spend a sizable amount of their lives in this elevated radiation environment. And yet, no reliable evidence for this has been found so far.

Re:already done

[Solandri]

External events are considered in US plant design already, this author seems to be a bit ignorant on how the safety case for plants is built. Who cares if we refine the probability of an event is if the plant is already designed to withstand it?

Technically, the Fukushima plant was also already designed to withstand this type of event. It had sufficient backup power systems necessary to continue operating the cooling pumps in the event of a catastrophic disaster of this type.

Where they screwed up was in the redundancy of the backups. This is unfortunately a fairly common failure mode in engineering designs. Say a single diesel generator has a 10% chance of failing to start up if you try to run it during an emergency. People then naively think that if you just put 6 diesel generators into the design, then that reduces the statistical probability of failure to 1 in a million. The chance of all six generators failing is (10%)^6 = 1 in a million.

That's the correct math for generator failures due to independent internal causes. But everything changes when you talk about external causes. Suddenly you have a cause like, oh, say, a tsunmai, which can affect all the generators simultaneously. The failure mode for each generator is no longer independent, and your redundancy does nothing to decrease the odds of a failure. All they had to avoid this effect was put the generators and diesel fuel tanks in different places. But no, the typical Japanese obsession with order and symmetry* mandated that they put all their generators in a row in the same place. And the tsunami took them out and contaminated their fuel all at once. Indeed the two newer Fukushima reactors where the generators and fuel were stored in a different location got through the earthquake and tsunami just fine.

* I rag on the Japanese, but the same thing happened with the Space Shuttle Challenger. They were having problems with poor O-ring seals in the solid rocket boosters. So to reduce the probability of a failure, they just added more O-rings. That worked to stop the independent failures (burn-through due to improper seating of an O-ring in one spot). But when an external factor popped up which caused all O-rings to fail simultaneously (cold weather), the safety of the redundant O-rings was negated.

Re:How would that be even helpful?

[Mr D from 63]

Earthquake probability and characterization is a 'continuously improving' science. Knowledge improvement is factored in by the regulator. Fortunately, plants are designed to withstand very large quakes with a large design margin added on top. In reality they will withstand quakes much larger than their stated design capacity.

Re:Stylized

[Anonymous Coward]

If the 1 in 10.000 years is per reactor, 18 years between accidents is "reasonable". With 400 reactors worldwide, that would mean approximately 25 years (~10000/400) between accidents.* Accounting for older designs, improving risk estimation, worse safety/quality standards in some parts of the world, etc. 18 years is close and not "far outside the claimed safety envelope".

Also, one "near miss" per year suggests luck, ten or more per year implies that there are enough safeties and checks in the systems to catch trouble before a catastrophe happens.

* I know this is not exact. It should be close enough. Fanatics can do the 1/(1 - ((10k-1)/10k)^400) stuff with a calculator.