corrosion, embrittlement, hydrogen attack, lying with statistics, methodological assumptions, nrc embrittlement, nuclear dangers, nuclear reactor, nuclear reactor pressure vessel, nuclear reactors, nuclear safety, NUREG, Pressurized Thermal Shock, PTS, PWR, reactor pressure vessel, statistics
[We added to this post on early 12 May UTC. See update at bottom.]
“NRC-2014-0137 Technical Basis for Regulatory Guidance on the Alternate PTS Rule (NUREG-2163)”
Comment Deadline May 12 2015, Eastern Time http://www.regulations.gov/#!docketDetail;D=NRC-2014-0137
Documents to comment: http://www.regulations.gov/contentStreamer?documentId=NRC-2014-0137-0003&disposition=attachment&contentType=pdf http://www.regulations.gov/contentStreamer?documentId=NRC-2014-0137-0002&disposition=attachment&contentType=pdf
NUREG-2163 (document to comment) states: “The United States Nuclear Regulatory Commission (NRC) promulgated Section 50.61a, “Alternate Fracture Toughness Requirements for Protection against Pressurized Thermal Shock Events,”….” (p.1)
This already suggests that there is a problem. Why are they promulgating “Alternate” requirements? And, why is the “stakeholder” whose opinion appears to matter the utility funded EPRI?
“This action became desirable because the existing requirements, as contained in 10 CFR 50.61, “Fracture Toughness Requirements for Protection against Pressurized Thermal Shock Events” , are based on unnecessarily conservative assumptions. The Alternate PTS Rule reduces regulatory burden for those PWR licensees who expect to exceed the 10 CFR 50.61 embrittlement requirements…“(p.1) The utility funded Electric Power Research Institute (EPRI) is apparently the only “stakeholder” which they consider. (See p. 11) “Unnecessarily conservative” means that they might have forced these old reactors to be shut down, as originally planned.
It’s important to recall that it’s not just pressurized thermal shock which can cause the nuclear reactor pressure vessel to fail. And, it’s not just neutrons which do the damage. There is also hydrogen attack, corrosion, and possibly substandard materials or manufacture. Any, a combination, or all of these things together can lead to failure at any temperature. Although pressurized nuclear reactors (PWR) are more at risk, because of the higher pressures, all are at risk. In the US there are around 100 nuclear reactors still in operation and 2/3rds are PWR.
Interestingly enough, despite this NUREG draft’s weak methodology and lack of statistical power, the above problems are so overwhelmingly true, that even this NUREG notices that there is a problem leading to reactor pressure vessel damage being greater than they expected in their models:
“Type A, B, and D deviations all give rise to concerns that the embrittlement trends predicted by the ETC may produce non-conservative estimates of the embrittlement experienced by materials used to construct the RPV that is being evaluated.” (p. 27) (They appear to mean conservative in regards to public safety here, though the NRC also uses conservative to mean protecting the financial interests of the nuclear industry-utilities). However the NUREG then goes on to explain how the utility can screw around with the stats to get what they want.
Let’s return to methodology. People are generally familiar with political campaign polls. Many things can go wrong with these polls. Recently there were concerns of a hung parliament with no clear majority in the UK, and yet, a clear majority came. This may have been faulty polling or perhaps the poll predictions influenced vote and voter turnout.
While incorrect polling methods might cause a pollster to lose his or her jobs, what are the repercussions of reactor pressure vessel failure? Non-nuclear pressure vessel failures can and have been deadly. This was a major problem for steam boats.
However, the repercussions of a failed, exploding nuclear reactor pressure vessel may be a Chernobyl style disaster or worse. Although most hear little about it, the repercussions of Chernobyl continue to be a major problem in Europe, even after 29 years and this is only counting Cesium 137, which has gone through almost one half-life since Chernobyl. It is not counting the long-lived ones.
Scandinavian Reindeer have continued to sometimes have Cesium 137 levels which exceed even the high levels allowed in US food – themselves over 10 x greater than allowed in Japan. UK sheep have continued to be contaminated at over 1,000 Bq/kg, which while it would be acceptable in US food, is not acceptable for Europe which allows only 600 Bq/kg – an amount still exceeded sometimes in German Wild Boar. A couple of years ago they quit testing UK sheep even though some still exceeded 1,000 Bq/kg allowed for consumption within the UK.
So, is any error acceptable? No, it is not. Yet this NUREG also allows for some cracking of the reactor pressure vessels. What’s more it allows for screwing around with the crack measuring and sizes. Vast parts of the earth are already contaminated due to Fukushima and Chernobyl and other disasters, all while the world population itself has exploded. Insurance does not cover these accidents and people will be forced to stay in contaminated zones deemed acceptable by the US government which has repeatedly shown the disdain which it has for its own people by running around the world, and asking other countries to send their high level nuclear waste to America, in the name of non-proliferation; allowing other countries to burn nuclear waste and to literally dump it in America, and by allowing more radiation in food than any other country (of which we are aware) and certainly more than Europe, and Japan, hence opening itself up to become the dumping ground for radiative food. What to say about such a country? And, little wonder that people are alarmed at unnecessary war games being played out within communities, when the US government has huge areas of land upon which to play its war games and model towns to practice door to door exercises (as does the UK who took control of some villages during World War II or before to practice urban or village combat).
So, say someone wants to do a campaign poll in the US. They call around 2,000 people randomly and ask how they will vote. This always shocks people as it is so few. An early methodological flaw was when this was done and few people had telephones, leading to a false prediction that Dewey won. Maybe you send around a survey, and then there are issues of who responded and why or why not. Maybe you will go door to door asking to get the surveys back. Then there is the problem of questions asked and of open ended questions, which may not fit in a neat box.
For quantitative analysis, statistics, you need to start with at least 50 of whatever you are examining, in order to start speaking of statistical significance, whether hard science or social science. And, really they should be chosen randomly.
So, here we have around 100 US nuclear reactors and 2/3rds are PWRs (Pressurized Water Reactors). It should be clear that to study PWRs, the entire population of PWRs would have to be properly studied. And, really all nuclear reactors would have to be studied.
This, in fact, is what Digby McDonald recently suggested should be done due to major flaws found in Belgium nuclear reactors. He said that all nuclear reactors, all over the world, need to be tested, all need ultrasonic survey of their reactor pressure vessels, because the degradation was worse than predicted by their models: https://miningawareness.wordpress.com/2015/03/03/nuclear-reactor-cracks-widespread-disease-scourge-warns-nobel-in-chemistry-nominee/ This is because the world population of nuclear reactors is so small (approx 435 to 443) and the repercussions of a nuclear accident so great.
In qualitative methods we might look at a case or two or three, like the Belgium nuclear reactors. Studying them in depth allows everyone to see that they have serious problems and are a public danger. From this we may extrapolate, as Digby McDonald does, the need to test more or all. However, one cannot extrapolate, as the US NRC appears to do, outwards from these few to all reactors, unless it is to say that all may be dangerous and should be properly tested or immediately shut down. But, one cannot assume that all are ok because one is. It’s like asking your neighbor who they vote for and assuming that the one person will tell you the winner. Even three neighbors may not tell you the winner. These old nuclear reactors have different makers, different life histories.
And, even in this in-depth study in Belgium they had to get it right; they had to redo both their methodological assumptions to make them opt on the side of caution and increase sensitivity of measuring instruments: “In addition, Electrabel has increased the sensitivity of the ultrasonic detection device MIS-B to ensure that all flaw indications were detected. In May and June 2014, the operator conducted new ultrasonic inspections in Doel 3 and Tihange 2 with these adapted detection parameters. The results of these inspections were then interpreted using the aforementioned new measurement method.
The combination of these two improvements has resulted in the detection of about 60% more flaw indications in the steel walls of the reactor pressure vessels of Doel 3 and Tihange 2. The new method implies that signals with a very low amplitude, which is characteristic for small flaw indications that wouldn’t be detected with the detection threshold of 2012, are being regrouped. This regroupment leads to an increase of the average and maximum length of the flaw indications.” http://www.fanc.be/fr/news/doel-3/tihange-2-clarifications-regarding-the-detection-the-position-and-the-size-of-the-flaw-indications/753.aspx
Even as it is clear that these nuclear reactors are too dangerous to start up again, they may decide to “prove” that they are safe anyway. After trying to do things right, they appear to be now quibbling about whether the cracks are that dangerous or not, in true US NRC style.
The US appears to have worked with a few brand new reactors to come up with an original theory of damage and now a few used reactor pressure vessels. From this they seem to have extrapolated to all nuclear reactors and even in their comparison they have only 3 to 8 points, whereas they would need in the range of 50 to 120 test points per reactor, if what they were testing was correct and it does not appear to be. Their methodology and statistical power appears hopelessly flawed and fucked up.
The US NRC under the influence of the utility funded EPRI, has chosen an option which reduces safety margin: “because NDE techniques tend to oversize smaller flaws, thereby distributing detected flaws into larger bins where the allowed number of flaws is smaller. In such cases, adjustment for NDE flaw-detection and -sizing uncertainties may result in a less conservative distribution of flaw sizes, possibly allowing a comparison of the adjusted NDE data to the flaw tables to be successful.” (p.63)
Furthermore, it seems that they have only tested a few reactors. “The Alternate PTS Rule is based, in part, on analysis of information from three currently operating PWRs.” (p.7) The loophole which allows this appears to be here: “III. Surveillance Program Criteria
A. No material surveillance program is required for reactor vessels for which it can be conservatively demonstrated by analytical methods applied to experimental data and tests performed on comparable vessels…”
Nonetheless, the old nuclear reactors are so messed up and the optimism that they could go on and on so ridiculous that even the statistically weak NUREG methodology points to DANGER.
And, this danger should not be an excuse to continue to massage and manipulate the data, but rather to test all of the nuclear reactors thoroughly, as recommended by Dr. Digby McDonald. Or, more logically to just shut them all down permanently. Some parts of some nuclear reactor pressure vessels apparently cannot be reached for examination, unless new technology allows.
Whereas Belgium opted to test a second time with greater sensitivity to be certain and found more flaws and has opted to group together small flaws to count as larger ones, the US seems to have reduced flaw sizes on the assumption that they were overestimated.
Here is what the NUREG-2163 (to comment on) points out as far as methodological errors:
Type A Deviations”errors in the chemical composition values or errors in the unirradiated ΔT30 value associated with the surveillance sample…”
Type B Deviations “the existence in the surveillance sample of an embrittlement mechanism that is not represented in the ETC calibration dataset…”
Type C Deviations “errors made in testing, labeling, or controlling the notch orientation of surveillance specimens…”
“Potential origins of Type D deviations may include, but 2 are not limited to, (1) a large measurement error in the single datum or (2) the rapid emergence in the surveillance sample of an embrittlement mechanism that is not represented in the calibration dataset (e.g., rapid emergence of a Type B deviation).” (pp.26-27)
“If they are statistically significant” [they appear not], “Type A, B, and D deviations all give rise to concerns that the embrittlement trends predicted by the ETC may produce non-conservative estimates of the embrittlement experienced by materials used to construct the RPV that is being evaluated. For this reason, methods to assess the statistical significance of these deviations are described in Sections 5.4.1 through 5.4.3. Type C deviations, if they are statistically significant, suggest that the surveillance program for the material in question may not provide a reliable indication of embrittlement trends for that material. Because Appendix H to 10 CFR 50 requires the performance of surveillance on the “limiting” (meaning “most irradiation-sensitive”) materials used to construct the RPV beltline, the existence of a Type C deviation is important from a regulatory viewpoint, but not in the context of indicating a potential non-conservatism in the predictions of the DT30 ETC adopted in the Alternate PTS Rule. For this reason, statistical procedures to detect Type C deviations were not included in the Alternate PTS Rule (see Reference ) and, therefore, are not described in Section 5.4.” (p. 27)
“The Alternate PTS Rule is based, in part, on analysis of information from three currently operating PWRs.” p. 7
“Criteria relating to evaluation of plant-specific surveillance data: The Alternate PTS Rule includes statistical tests that must be performed on RPV surveillance data to determine whether the surveillance data are sufficiently close to the predictions of an embrittlement trend curve (ETC) that the predictions of the ETC are valid for use.” (p.viii)
“2. Data-Quantity Requirements: To perform these statistical tests, at least three plant-specific ΔT30 values measured at three different fluence levels should be available. If this condition is not met, the ETC described in the Alternate PTS Rule should be used to estimate ΔT30.
3. Data-Binning Requirements 31 a) As discussed in Item 1, data obtained for the heat of material in question as part of a 32 10 CFR 50 Appendix H surveillance program conducted for any plant that is operating, or has operated, under a license issued by the NRC should be binned together and considered in these statistical evaluations. b) For plates and forgings, Charpy data is often obtained for different orientations of the notch relative to the primary working direction of the plate or forging. For the purpose of these statistical tests, ΔT30 values for a particular plate or forging should be computed from unirradiated specimens having the same notch orientation. Once ΔT30 values are calculated, different notch orientation data from the same heat of material should be binned together for the purpose of the statistical tests described in Sections 5.3 and 5.4. This is appropriate because the differences in unirradiated transition temperature caused by notch orientation will be subtracted out when ΔT30 values are calculated.” ( p.24)
Updated Tuesday, 12 May 2015, UTC:
This diagram from p. 55 of the NUREG-2163 gives a glimpse of how the NRC screws around-cheat with “standards”.
For some things they are allowing outdated 1998 or earlier ASME standards. Then there is the “regulatory relief” from standards. Based on this diagram they are applying half of a 2013 ASME standard (but it seems only to 3 reactors and extrapolating to the rest) and then they cut off ASME and have made a new standard, which uses a non-statistically significant number of data points per reactor (ranging from around 2 to 20 with around 8 appearing most frequently based on looking at it, when 50 to 120 or more date points per reactor would be needed). These data points themselves seem to be based on methodologically flawed extrapolations based on several new and/or several old reactors. Or, they are based on experiments on materials believed similar to the reactor in question. Standards are not a mix and match affair. And, how can someone teaching at MIT have no basic knowledge of methodology or statistics? They need to be sent back to take classes on methods. Despite all of their screw-ups, they still find that these reactor pressure vessels are aging more quickly than their models suggest! So, why are the reactors being given operating extensions?
[This post represents our current understanding of the NUREG to be commented upon. It may or may not be edited-added on to prior to the deadline on 12 May, 1159 pm. Please see related posts which should prove useful. ]