Talk:Core damage frequency

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Mathematics (Rated B-class, Low-importance)
WikiProject Mathematics
This article is within the scope of WikiProject Mathematics, a collaborative effort to improve the coverage of Mathematics on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Mathematics rating:
B Class
Low Importance
 Field: Probability and statistics
WikiProject Statistics (Rated B-class, Low-importance)
WikiProject icon

This article is within the scope of the WikiProject Statistics, a collaborative effort to improve the coverage of statistics on Wikipedia. If you would like to participate, please visit the project page or join the discussion.

B-Class article B  This article has been rated as B-Class on the quality scale.
 Low  This article has been rated as Low-importance on the importance scale.

removal of accident statistic and classification of incident[edit]

User:Nailedtooth removed my additions

Core damage is the worst thing that can happen to a nuclear reactor, as due to damage of this structure, the operator might loose control of the nuclear reaction, possibly leading to a meltdown.


In a study commissioned by the European Commission, the core damage frequency of reactors in Europe is estimated to be once in 20.000 reactor years (5*10-5/a[1]).

after an edit conflict on List of civilian nuclear accidents, which he reported here.

I have now restored it. -- Eiland (talk) 10:15, 6 June 2008 (UTC)

The information you have posted is false and the phrase "Core damage is the worst thing that can happen to a nuclear reactor" is particularly meaningless. Why you insist on altering nuclear related articles with such poor information baffles me.Nailedtooth (talk) 14:29, 6 June 2008 (UTC)
Im sorry, but I believe to give the implication of a core damage does add usefull information to the article. I wonder what other problems you see to core damage than the one I wrote? As per the references, check the meltdown article, which - accidentally? - says the same. And while your at it, why did you remove the title of the article as "correcting link"? -- Eiland (talk) 16:03, 16 June 2008 (UTC)
The problem is you're using wiggle words "might" which makes that information useless. Core damage - might - do a lot of things, not just what you mentioned. If you're going to talk about the implications of core damage, talk about them all. Another implication of core damage is the reactor continues to function correctly. But you didn't include this, why?
Also, there is no reference for that information, so I'd like it removed. References to other Wikipedia articles do not count.Nailedtooth (talk) 22:23, 16 June 2008 (UTC)
Yes, core damage might also save the world, but the reason that this figure is used is that it is the critical issue of nuclear reactors. If the core gets damaged, reactor moderation MIGHT be impossible, cooling MIGHT become inadequate, and meltdown MIGHT occur. We're talking statistics here, so of course there's no guarantee. I think school kids making a project on nuclear reactors MIGHT want to know the possible implication of the figure under discussion. Oh, and here's a reference for core damage leading to meltdown. -- Eiland (talk) 23:10, 16 June 2008 (UTC)
Well, since we have to come to some sort of consensus we obviously need solutions other than keep/remove. Also, TMI had a meltdown due to failure of a pressure release valve, not core damage. The core was damaged, but that was a result of the accident, not a cause. Nailedtooth (talk) 00:07, 17 June 2008 (UTC)
  • I've come over due to the ANI posting. Are there sources that say core damage is the worst thing for a nuclear plant? ·:· Will Beback ·:· 00:21, 17 June 2008 (UTC)
Hi. Are you the terminator? :) The worst thing to happen to an NPP is to loose control over the reactor as then potentially the contents of the reactor might come in the environment. Either plane impact, war, earthquake, inattendence, ageing, material fatigue, design error, you name it, can lead to this. If the reactor gets damaged, by either which reason, the chances are rapidly increasing that it cannot be properly controlled anymore. Maybe to say " the worst thing" is too absolute, because maybe by some cosmic quirk, a reactor might spawn a black hole, gulping the entire universe, after which this wikipedia entry wasn't correct. But I think an indication of the risk carried in core damage accidents, including meltdowns, should be included in this page. What about "Core damage is dangerous, as due to ..." -- Eiland (talk) 10:44, 17 June 2008 (UTC)
Core damage doesn't mean a loss of control over the reaction. There are quite a number of reactors that have suffered core damage and continued to function for many years.Nailedtooth (talk) 14:23, 17 June 2008 (UTC)
  • Are there any sources that talk about this? I don't doubt you're right, but our opinions count less than the opinions expressed in reliable sources. ·:· Will Beback ·:· 10:50, 17 June 2008 (UTC)
Dunno, what would count as a credible source? [1]? [2]? —Preceding unsigned comment added by Eiland (talkcontribs) 11:22, 17 June 2008 (UTC)
  • I checked the first one and I didn't see any mention of core damage. I couldn't open the second one - can you paste here the material covering core damage being the worst thing for a nuclear plant? ·:· Will Beback ·:· 19:06, 17 June 2008 (UTC)
NRC Says Ohio Reactor Damage 'Significant'
Associated Press Writer
Damage to the reactor head of the Davis-Besse power plant in Ohio ranks among the top five most serious nuclear plant accidents or near-accidents since Three Mile Island, the Nuclear Regulatory Commission said Monday.
Davis-Besse, along Lake Erie in northwest Ohio, was closed for two years after inspectors found corrosion on the reactor in March 2002. Leaking boric acid almost had eaten through a 6-inch-thick steel cap; repairs cost $600 million.
While the plant was shut down, engineers found that its undersized sump could have become clogged with debris during an accident, which choked off the flow of water to cooling pumps, said an NRC analysis released Monday.
Federal regulators estimated there were six chances in a 1,000 that the plant could have experienced a meltdown during the year before it was shut down for routine maintenance in February 2002.
Normally, the risk of an accident happening at Davis-Besse is about six in 100,000, NRC spokesman Scott Burnell said. The NRC considers the risk "significant" when circumstances at a plant bring the possibility of core damage within one chance in 1,000. [3]
The primary concern is the fuel within the nuclear reactor and the spent fuel stored onsite after its removal from the nuclear reactor. The fuel, whether inside the nuclear reactor or not, must be cooled to prevent damage from overheating. If the fuel is damaged, government studies report that the radioactive material released from either the reactor[3] or the onsite spent fuel[4] can kill and injure tens of thousands of people living within 500 miles and render large regions uninhabitable for long periods.'[4] —Preceding unsigned comment added by Eiland (talkcontribs) 23:10, 17 June 2008 (UTC)
The problem with Davis-Besse wasn't core damage, it was with a faulty seal in the pressure vessel and - as the article stated - an undersized sump. Core damage would have, like TMI, been a result of the accident, not a cause. You still haven't shown that core damage is 'the worst thing that can happen' to a nuclear power plant.
I think with the input of Will Beback we've come to the conclusion that the phrase "Core damage is the worst thing that can happen..." needs to be removed. Nailedtooth (talk) 02:22, 18 June 2008 (UTC)
it doesn't matter what's the cause, but if core damage occurs (remember we're talking "core damage frequency" here)- for which the chance significantly increased during the DB incident, chances for meltdown increase. As I wrote before, I proposed the alternate wording, "Core damage is dangerous, as due to ..." You're always quick at proposing removals. I believe its better to try to reach consensus through compromise. -- 11:15, 18 June 2008 (UTC)
I propose the phrase "Core damage is potentially dangerous due to an increased chance of other accidents, though no single outcome is a certain result." This phrase is factually correct in every way. Nailedtooth (talk) 14:24, 18 June 2008 (UTC)
I don't agree, thats a truism, and you yourself had something against that. The big risk of core damage is that it paves the way to a melt down, and I will reject any phrasing which not clearly states that. If regulators would have argued like you, they would have come up with the phrase 'meltdown frequency'. Lets see if Will Be Back will be back, because you and I tend to never agree -- Eiland (talk) 14:50, 18 June 2008 (UTC)
Making absolute demands is an attempt to dictate the article's content without discussion or consensus.Nailedtooth (talk) 21:53, 18 June 2008 (UTC)
The problem at davis-besse was not seal failure, it was CRDM housing circumferential cracking, compounded with gross negligence.VmZH88AZQnCjhT40 (talk) 19:11, 12 October 2013 (UTC)

NOTE: 'Core damage' damage in a nuclear reactor is analagous to individual personality 'core damage' as psychological effects. Stress and corruption affect individual personality core, including the undermining effects of impersonation such as the "I'm calling for...[person]" voice-contact type used in business interactions. (talk) 02:05, 22 June 2008 (UTC)beadtot


Stumbled upon this page, doing some cleanup. The math is wrong and some underlying assumptions in the article aren't totally correct. Hopefully my edits will resolve the above debate too. Text from [5] pages 136-137 is summarized here for your reference:

Severe accident sequences in nuclear reactors are failures that result in a loss of containment of the radioactive inventory... The core subsequently melts and releases part of the radioactive inventory to the containment structure. If the containment breaches... part of the released inventory will be released to the environment. Subsequent dispersion of the radionuclides will result in damages to public health and the economy.
The probability of severe accidents, i.e. accidents leading to a core meltdown, has been studied for a variety of reactor types and in different countries... Studies investigating the sequence of technical events occurring in such scenarios are called Probabilistic Safety Assessments (PSA).
‘Core meltdown’ or ’core damage’ means the melting of much or all of the reactor core material (oxide fuel and metal cladding) due to a loss of capacity to remove the heat of radioactive decay. Damaging of single fuel elements or exceeding technical limits of operation for fuel elements are not considered a core damage in a PSA. The given probabilities are those for damages of the core as a whole, not for small parts of the core or single fuel pins.
Model, Country, Location, Type, and CDF
NUREG-1150 USA Surry/PWR - 4*10-5
NUREG-1150 USA Peach Bottom 2 / PWR - 4.5*10-6
WASH1400 USA PWR - 2.6*10-5
WASH1400 USA BWR - 4.6*10-5
Sequoyah USA Sequoyah/PWR - 5*10-5
EPS900 F CP2/PWR Yes 4.95*10-5
EPS1300 F 1300MW/PWR Yes 1*10-5
Hinkley Point GB 610MW/AGR - 1*10-6
Japan JA 1100MW/PWR - 1*10-7
DRS-B D Biblis-B/PWR yes 3*10-5
SWR Phase II D - yes 2.7*10-6
Ringhals 3/4 S 915MW/PWR - 3*10-6
The reviewed studies, for different countries and types of facility and reactor, yield core damage frequencies between 10-4/a and 10-6/a... Core damage frequencies of 5*10-5/a are a common result, a figure often adopted in further risk studies...

--Tollsjo (talk) 09:41, 29 November 2008 (UTC)

Phew, also found an informative article at [6] that, after a lot of cross-referencing, suggested that splitting hairs over damage/meltdown/catastrophe and US vs The World stats was missing the point. In fact I'd even recommend moving this out of Statistics as it's a risk-management metric not a measure of overall reactor safety. Let's say Ford Pinto cars have a 1% death rate during normal driving and a 10% death rate during a rear-end collision, what does that tell you about Ford or American automobiles? Nothing, just avoid rear-end collisions in a Pinto. If CDF can be used as an industry performance stat, that's secondary to its main purpose in risk analysis.

Tollsjo (talk) 11:56, 29 November 2008 (UTC)

In fact the maths was correct: 5 × 10−5 is equal to 0.00005, which is equal to 1/20,000, not 1/50,000. Johnsul (talk) 00:55, 7 December 2008 (UTC)

Of course, what I'd be interested in is a discussion of how the (large) given figures relate to the reality of reactor failures: if there are 439 reactors in the world, at 1 per 20,000 reactor years you might expect (using a whole bunch of simplifying assumptions) one event per 45 years, which is clearly not the case. Is the explanation that the figures are accurate for all reactors except the ones that failed which individually had low CDFs, or that known melt-downs have not counted as the right sort of event for CDF calculation purposes? Johnsul (talk) 01:06, 7 December 2008 (UTC)

Mediation Accepted[edit]

Hi All, I've accepted the chance to mediate this case. Come join the discussion linked to in the template at the top of this talk page. HatlessAtless (talk) 17:01, 11 July 2008 (UTC)

This mediation has been closed due to lack of participation. —Preceding unsigned comment added by HatlessAtlas (talkcontribs) 18:30, 26 August 2008 (UTC)

Incident of Core Damage Statement[edit]

I would like to take issue with the last statement of the subject article - "Assuming there are 500 reactors in use in the world, the above numbers mean that, statistically, one core damage incident would be expected to occur somewhere in the world every 40 or 100 years, respectively."

This is a simplistic use of math that portrays core damage frequency in an unnecessarily poor light. The facts from Reference 5 of the article is that the EPRI report states as of 2005, the average core damage frequency for the US Industry is 2 X 10-5. This means, the risk of having a core damage incident for the 104 nuclear plants in the US is 1 in 50,000 reactor years. Further, the graph depicting the core damage frequency clearly states that ONE data point was used from 2001 to 2005 (hardly an average) and goes further to state that CDFs closer to 10 -7 have neglible effect on the average and aren't considered. This shows that a lot US plants could have a much lower CDF number than 2 X 10-5. Dividing this number by 500 and stating a core damage incident would be expected once every 100 years is incorrect. I suspect the European figure has also been misrepresented in deriving the once every 40 years figure.

This statement appears to be anti-nuclear propoganda and clearly shows how figures and charts can be twisted to say anything the writer wants. I would prefer to see this sentence deleted. [[[User:Orionfree|Orionfree]] (talk) 12:07, 24 September 2010 (UTC)]

It is most definitely not anti-nuclear propaganda. I would know, since I wrote it. It is simply another way to put the numbers in the article in perspective. If there is a problem, the problem is with the estimated average CDF being calculated badly (which I have no trouble believing, since the CDP is so low already). europrobe (talk) 19:21, 24 September 2010 (UTC)

Well…….I won’t take issue with your statement that average CDFs are too low since I’m familiar with the process but do not know all the failure and fault tree assumptions they employ and also since they are very plant specific. I won’t even argue the “anti-nuclear propoganda” statement since it’s subjective. I do, however, take issue with putting the numbers in the article in a perspective that is incorrect. If the average CDF for US plants is 2 X 10-5, then dividing them by any number of plants is wrong. It is correct in saying that since the average CDF for the US nuclear industry is 2 X 10-5 per reactor year, then the US nuclear industry has a risk of an incidence of core damage of once every 50,000 years. Dividing 50,000 by 104 (the US plant population, since this IS a US industry value not the 500 world number you used) and coming up with 480 years would be wrong. The EPRI report states (on the bottom of the first page)…..”the CDF value represents the number of occurrences per reactor year” (in this case 2 X 10-5) and further, “but can also be stated as occurrences per number of years” (again in this case 2 every 100,000 years for the entire US industry). Further manipulation by dividing by the number of plants is misrepresenting the calculation. [[[User:Orionfree|Orionfree]] (talk) 15:10, 25 September 2010 (UTC)]

The CDF is, as you know, stated in incidents per reactor-years. If the world had had 500 reactors running at 100%, 500 reactor-years would be generated each year. After 100 years, 50000 reactor-years would have been generated, and -statistically- one incident would have happened. This is using the lower CDP of the US estimate, which resonably should be better (lower) than the world average. europrobe (talk) 20:20, 25 September 2010 (UTC)
Just to clarify as to how to calculate the probability of core damage, the mean time to core damage for a 104 reactor fleet operating continuously at a CDF of 2e-5, the formula would be MTTF=1/[1-(1-CDF)^(Number of reactors)]. This is where 480 years comes from.VmZH88AZQnCjhT40 (talk) 18:55, 12 October 2013 (UTC)


I just want to add that I found this article informative in the light of recent events in Japan. I'm still an advocate of nuclear power, just that we should only use plants with passive savety mechanisms. But if this article needs anything, is that apparently the numbers that have been quoted are much too low, given that in the past 50 years we've had three major accidents already.

I do agree that the 1 X 10^-5 statistic doesn't mean that you have an accident every 100 years if you have 500 operating reactors. It would be about every 140 years (0.99999^70000 = 0.497). For 2 X 10^-5 it's less, about 70 years for 500 operating reactors. User:AdriaanRenting, 18 March 2011 —Preceding undated comment added 03:22, 18 March 2011 (UTC).

Fukushima clearly shows that existing CDF figures for BWRs should be taken with a grain of Iodine-131. Three cores have been damaged, which is clearly a far higher rate than the literature predicted. There is nothing exceptional about the circumstances either - a repetition could easily happen in Central Japan, where a megathrust earthquake is also due.Theeurocrat (talk) 12:01, 5 April 2011 (UTC)

As far as I understood from some media reporting there is a serious problem with the risk-assessment to the extent that failures of backup systems are considered to occur completely independently from each other. This means no account is taken of the possibility that a single cause (say a Tsunami or Earthquake) may damage both primary, secondary and tertiary systems at the same time. Additionally, explosive failure of one system is assumed to never ever damage a backup system. Fukushima shows this assumption makes no sense at all. I have no sources though as I heard this on the radio somewhere. Arnoutf (talk) 11:36, 1 July 2011 (UTC)
Fukushima may well be understated as well, while 3 reactors had Core Damage that we know about, both Fukushima Dai-chi 5,6 and Dai-ini 1-4 may well have had core damage as well. TEPCO and JP Govt have been notably mute on many a problem instead of being forthcoming. -- (talk) 14:28, 10 May 2013 (UTC)
MTTF for a CDF of 1e-5 and 500 reactors is 200 years. 1/(1-(1-1e-5)^500)=200.5a. For 2e-5 and 500 reactors it is 100a. The CDF for Fukushima was bad because of the possibility for confluence of tsunami energy from multiple emission points at the site, which has zilch to do with Central Japan.VmZH88AZQnCjhT40 (talk) 19:05, 12 October 2013 (UTC)

Fukushima and Design Basis[edit]

Currently, this article says the reactors at Fukushima "failed due to the extreme beyond design basis conditions." This could not refer to the earthquake, which was about 6.5 to 7.0 at the plant and within design basis there. But it does not really describe the tsunami either. The owners of the plant were warned on several occasions, including in 1991 and 2007, that the seawall of 5.7 meters was far too low. Earlier earthquakes in 869 and 1896 were of magnitude similar to that of 2011, and the one in 1933 may also have been able to top the seawall at Fukushima. Another term for a design basis accident is a "maximum credible accident." To say something is much beyond a maximum credible accident is itself incredible, when the possibility both follows historic precedent and has been repeatedly warned about by professionals in the field. I am not sure whether this article should be reverted or altered. The discussion of design basis quite possibly belongs here. But I do think it needs to be corrected. ghh 19:15, 3 March 2012 (UTC) — Preceding unsigned comment added by George H. Harvey (talkcontribs)

The fact that the engineering design basis was wrong, and that TEPCO may have known that it was wrong, is independent of the fact that the accident was beyond the design basis. The design basis for the reactor is what is in their approved safety assessment, not what you or anyone else thinks is credible.VmZH88AZQnCjhT40 (talk) 19:08, 12 October 2013 (UTC)

There may be an error in a source reference.[edit]

I took a look at the source reference, "Reassessing the frequency of partial core melt accidents." While the author has clearly done a fair amount of careful homework for this article, he seems to have missed the meltdown in the type "KS 150" A1 reactor at Jaslovske-Bohunice, Czechoslovakia, in 1977. This means the number of meltdowns relevant to this article is 12, not 11.ghh 16:28, 4 March 2012 (UTC) — Preceding unsigned comment added by George H. Harvey (talkcontribs)

What factors are included in a CDFcalculation?[edit]

How does one arrive at a CDF calculation? Presumably its something like:

CDF = (p1)(p2)(p3)...(pn)

that is, the product of the probabilities of each event required to cause core damage.

CDF might be useful as an index for comparing safety of reactor types. However, since the math predicts between 0 and 1 core damage incidents since 1950, something is missing.

I think it's safe to assume that earthquake damage and other natural disasters aren't included and neither is deliberate or even accidental human action. At any rate, you can't predict at all what systems will be caused to fail in an externally caused disaster (although you can guess at which are more vulnerable to certain types of damage).

I'd like to see some explanation by a knowledgeable person about how external damage to the plant is treated and how CDFs are derived.

Is it possible that CDF only applies to a reactor that is immune to external damage? That is, is the engineering problem of plant safety so well worked out that we build reactors that should virtually never fail because of random failure of components? (And yet they do fail because they're located on Earth and operated by humans.)

Roches (talk) 15:53, 26 February 2014 (UTC)

  1. ^ CE (2003) Environmentally harmful support measures in EU Member States, J6, p. 137