Aging Nuclear Power Plants focusing in particular
on irradiation embrittlement of pressure vessels

Ino Hisamitsu *

Table 2: Results of monitoring tests on mother material of Genkai-1 reactor pressure vessel

(Continued from Nuke Info Tokyo No. 148)

    We must first understand the data on which this is based. Table 2 shows the results for the first to fourth monitoring tests. The amount of neutron irradiation is the amount for the specimens, not for the pressure vessel itself. The specimens were placed deeper inside the reactor than the reactor walls, so they were irradiated by more neutrons. Since the specimens have been irradiated by more neutrons than the reactor walls in the same time, operating years are converted to “effective operating years”.
    Effective operating years for the fourth monitoring test specimen was 66 years, meaning the reactor walls would be irradiated by the same amount of neutrons after 66 years. Since the reactors do not operate continuously, this amount of irradiation would not actually be reached until 85 years after the reactor began operating. How then are the present ductile-brittle transition temperature (DBTT) and the DBTT after 60-years estimated? Since DBTT is 98℃ after 85 years, bringing it back to 35 years and 60 years Kyushu Electric comes up with the lower temperatures of 80℃ and 91℃ respectively.
    The method used to derive this estimate is to redraw the prediction curve, adding a margin of error so that it passes through data point “×” in the top right corner of Figure 1 (see Nuke Info Tokyo No. 148), then to read off the DBTT corresponding to the amount of irradiation after 35 years and 60 years respectively. But for such a method to have a basis, the embrittlement prediction curve in Figure 1 must have some legitimacy. However, as discussed above, the formula used in the past has been pronounced invalid.

Figure 2: Genkai-1 Monitoring Data and JEAC-2007 Prediction

    So can the new 2007 prediction formula explain the DBTT of Genkai-1? The answer is no.
    Figure 2 shows the irradiation embrittlement prediction curve drawn by us on the basis of the 2007 prediction formula, and the observed DBTT. Like Figure 1, this diagram shows both the scale for DBTT and also for the increase in DBTT, the difference from the initial DBTT of minus 16℃.
    It can be seen that the observed data of 98℃ is 42℃ above the predicted curve. This cannot be explained in terms of margin of error. Compared to Figure 1, if anything the deviation is greater. Thus the 2007 prediction formula fails completely to reproduce the irradiation embrittlement behavior of Genkai-1. Hence, there is no explanation why a high DBTT was observed in Genkai-1. Given that such high DBTTs are observed when there is a high amount of copper impurity, or there is phosphorous grain boundary segregation, we cannot rule out the possibility that the Genkai-1 pressure vessel contains, depending on the location of the monitoring specimens, low quality steel with high levels of impurities. In regard to Genkai-1, both the 2004 formula (Figure 1) and the 2007 formula (Figure 2) have lost their predictive power. It is meaningless to estimate based on these formulas that the current DBTT is 80℃, or that after 60 years operation it will be 91℃.
    So what should we suppose the DBTT to be now? There is no sound method of estimating it. In that case, Kyushu Electric should respect the observed data of 98℃, assume that the pressure vessel itself has already reached this high DBTT (that being a true safety margin) and consider what response should be taken. The response should be to carry out the abovementioned PTS assessment based on a DBTT of 98℃, reconsider the operating sequence based on the 98℃ figure, and also carry out pressure tests based on 98℃.


NISA’s Response and Public Comments

    We were surprised at the observed high DBTT for Genkai-1. As soon as we found out about it we requested Social Democratic Party leader Mizuho Fukushima to arrange a hearing with officers of Nuclear Industrial and Safety Agency (NISA) to find out about the monitoring test methodology, etc. To our amazement, at that point in time (December 15, 2010) NISA had received no information about the results of the fourth monitoring test for Genkai-1. The first they heard of it was from the questions in our letter. Kyushu Electric had not informed NISA of the strikingly high DBTT and NISA said they did not know because they had no obligation to inquire. What a careless and lax safety monitoring system. At the hearing we demanded that NISA pay great attention to Genkai-1’s DBTT, and that it publish raw data for the Charpy test.
    It is a matter of great significance that the results of the fourth monitoring test for Genkai-1 cannot be accounted for by either the former prediction formula (JEAC 4201-1991), or the current formula (JEAC 4201-2007), and that the high DBTT is totally unpredictable. NISA called for opinions regarding the 2010 supplement to JEAC 4201-2007, so, in light of this serious situation, the Nuclear Aging Research Team submitted a public comment to NISA articulating fundamental questions about the monitoring test methodology.
    The essence of our public comment was as follows (abbreviated):


    This public comment calls for a fundamental review of JEAC-4201, which stipulates the monitoring test methodology for steel in pressure vessels, and for an explicit statement in the rule that there are cases where the option of permanent shutdown should be selected.
    NISA’s response to our public comment was published on its web site on May 6, 2011. There was no direct response to the points we made. The response made no reference to the striking deviation in the Genkai-1 data. It simply stated that where there is a deviation the margin for error should be reset and that there was no problem. NISA’s reply was an insult to our intelligence. What needs to be corrected is the thinking behind the monitoring test methodology that uses margin for error to paper over problems.

Discussion and Issues in the “NISA Advisory Committee on the Technological Assessment of Aging in Nuclear Reactors”
Launch of the “NISA Advisory Committee on the Technological Assessment of Aging in Nuclear Reactors”

    Last November the Nuclear Industrial and Safety Agency (NISA) initiated the NISA Advisory Committee on the Technological Assessment of Aging in Nuclear Reactors. As it turned out, I was invited to become a member of the committee. Hitherto, NISA has ignored our ideas. I decided to participate in the Hearings because I believed it was necessary to have a forum in which to communicate our thoughts about the issue of aging nuclear power plants, in particular concerning the extraordinary embrittlement of the Genkai-1 plant. However, these Hearings are, as their name implies, a forum in which committee members’ views are heard and debate takes place, but in the end NISA takes responsibility for writing the report. I was aware of this limitation when I decided to become a committee member.
    The following three issues have been considered during the Hearings:

 (1) Assessment of the aging of individual plants:
 (2) Relation between aging and the Fukushima Daiichi accident:
 (3) Cause of the greater than predicted embrittlement of Genkai-1:
    Consideration of how to interpret the results of the monitoring tests of the Genkai-1 DBTT, which exceeded the predicted 98℃, and whether the equation for predicting embrittlement is appropriate.
    Theme (3), which relates to irradiation embrittlement in Genkai-1 and whether the existing prediction equation is appropriate, is the issue that interests me most. Debate about the cause of the high DBTT (98℃) observed in the Genkai-1 pressure vessel monitoring tests revolved around two theories: [i] was it caused by poor quality pressure vessel material or a bad manufacturing method, or [ii] was it because the embrittlement prediction equation does not accurately reflect reality in the high irradiation range?
    Kyushu Electric claimed that the results of a chemical analysis of the steel materials showed that there were no irregularities and that uniformity was maintained. They also claimed that examinations carried out by the Central Research Institute of Electric Power Industry (CRIEPI) and others into micro-organization in the monitoring samples showed a good correlation between embrittlement and the formation of impurity clusters, so there was no abnormal embrittlement. However, to confirm the accuracy of this judgment and form a conclusion about whether or not the material of the pressure vessel is sound, instead of getting a research organization like CRIEPI, which is part of the nuclear industry, to assess the samples, they should be given to fair and trustworthy university researchers to examine their micro-organization.
    To support Kyushu Electric’s claim, a report entitled “Preliminary Consideration towards Improvement of the Accuracy of the Embrittlement Prediction Method ” jointly produced by CRIEPI and the Federation of Electric Power Companies (FEPC) was submitted to the eighth meeting (February 22, 2012, document 10). It concluded that it is not necessary to change the thinking behind the embrittlement model and the reaction rate equation, which form the basis of the current prediction equation, and that the variation from reality arose due to the lack of data in the high irradiation range. Further, by giving importance to the high irradiation range data (applying a weighting) and resetting the parameters of the equation (impurity cluster formation rate equation coefficient) the Genkai-1 data fit was improved. In fact, however, the fourth data point of 98℃ is still above the standard deviation margin and the second and third data points drop below, making the curve look very suspicious. In other words, they were unable to draw a meaningful curve connecting the third (56℃) and fourth (98℃) data points.
    It is problematic that in order to improve the fit in the high irradiation range the coefficients for the reaction rate equation, etc. were greatly changed. These reaction rate equations are the master equations that determine the whole method, so for the parameters to change greatly depending on the data sets that are used indicates the brittleness of the model itself. The reliability of the embrittlement prediction equation model, which is the basis of JEAC4201-2007, is therefore called into question. The problem goes beyond the Genkai Nuclear Power Plant. It extends to all aging nuclear power plants.
    Looking at the diagram in which NISA compared the prediction equation for aging nuclear power plants with the observed data (Hearing number 5, 23 January 2012, document 2), a large gap between the predicted figure and the observed figure can be seen in the high irradiation region. It is a fact that the prediction equation is unable to predict reality. However, the inaccuracy for Genkai-1 is particularly striking. The inaccuracy for other reactors is within 20℃, but the data from the fourth monitoring sample for Genkai-1 is out by 42℃. Besides the fact that the embrittlement prediction equation does not match the pressure vessel of Genkai-1 (see [ii] above), we must consider that the extraordinary embrittlement is due to the materials or the manufacturing method ([i]).
    Another surprising thing was that when we investigated CRIEPI’s embrittlement prediction equation, we discovered an elementary but important error in the equation itself. This prediction equation expresses changes in the micro-organization, namely the formation of impurity clusters and lattice defect clusters, which are the cause of irradiation embrittlement, as a reaction equation set, by tracing impurity atoms (copper atoms, etc.) and point flaw reaction (combination and disappearance) processes, and relating this to the rise in DBTT. This can be said to be an epoch-making change, compared to the rough and ready 2004 equation that just tried to fit the data, ignoring the rate of irradiation. However, there was a vital error in the reaction rate equation.
    The main cause of irradiation embrittlement is the formation of copper clusters (or impurity clusters in general). In the model there are two types, irradiation induced clusters and irradiation promoted clusters. Irradiation induced clusters are accumulations of copper atoms in lattice defects caused by neutron irradiation. The rate of formation is proportional to the concentration of copper atoms and the rate of diffusion of copper atoms (the speed at which they move). Physically this is an appropriate assumption. However, CRIEPI’s report says, “Because the formation of irradiation-enhanced clusters is a process in which copper atoms that exceed the solid solubility limit form a nucleus together, it is described by the square  of the quantity of copper above the solid solubility limit and also the square of the diffusion coefficient.” It must be said that this is a mistake. Because two (or more) copper atoms come together to form a cluster, it is appropriate to the think that it is proportional to the square of the concentration of copper atoms, but it is a mistake to say that it is proportional to the square of the dispersion coefficient*. Because two atoms move, at first sight it might seem that it would be proportional to the square of the speed, but that is not the case. Whether one atom is moving or stationary at one point, the rate at which they come together is the same. This can be proved mathematically. For example, the chance of two people meeting in a crowd in a stadium is the same whether one of the two is moving or stationary.
    As stated above, there is an error in the basic model of CRIEPI’s prediction equation. Naturally, any arithmetical calculation using this equation will produce the wrong result. Since the JEAC4201-2007 embrittlement prediction equation includes this fundamental error, it is a useless equation for predictions.
    In addition to the abovementioned brittleness of the embrittlement prediction equation, a mistake in the derivation of the equation itself was discovered. The JEAC4201-2007 embrittlement prediction equation must be discarded. The current situation is that there is no reliable prediction equation.

Is Genkai-1 Pressure Vessel Sound? NISA’s Predictable Assessment

    At the 12th Hearing, held on March 29, NISA submitted a draft report entitled “Concerning Neutron Irradiation Embrittlement of Reactor Pressure Vessels (Draft)” (referred to hereon as “Draft Report”). The purpose was to bring to a close the debate since January this year about “the cause of embrittlement in excess of predictions in the Genkai-1 reactor.” I strongly opposed the Draft Report and listed the problems. In the end the report was not finalized in March as planned and debate continued.
    I strongly opposed the report because even though the reason why a high DBTT of 98℃ was observed was hardly explained, the conclusion was drawn that the pressure vessel of Genkai-1 was sound, and the fact that the DBTT failed to agree with predictions was blamed on flaws in the prediction equation. Furthermore, NISA concluded that the pressurized thermal shock (PTS) assessment carried out by Kyushu Electric was appropriate and that the pressure vessel was in sound condition. However this type of assessment is totally inappropriate.

Figure 3: Kyushu Electric’s Pressurized Thermal Shock (PTS) Assessment for Genkai-1 Pressure Vessel.

    Figure 3 shows the results of Kyushu Electric’s PTS assessment. The curve that looks like a mountain in the bottom right hand corner is called the PTS state transition curve (K1 curve). In the case of a sudden large loss of coolant (Loss of Coolant Accident = LOCA), the Emergency Core Cooling System (ECCS) kicks in and coolant is fed into the reactor. The K1 curve shows the change over time in the force (strictly speaking the stress intensity factor K1) applied under those circumstances to the leading end of cracks that are presumed to exist in the inner surface of the pressure vessel. As a result of inserting cooling water, the temperature of the internal surface drops. At the same time, a temperature difference arises across the thickness of the pressure vessel and tensile stress is applied to the inner wall. Eventually the temperature difference of the pressure vessel becomes smaller and the value of the K1 curve decreases towards the bottom left.
    On the other hand, the curve rising to the right from the bottom left of Figure 3 is called the fracture toughness transition curve (K1C curve). It shows how the fracture toughness K1C changes depending on the temperature. If the material becomes brittle the curve shifts to the right. How is this curve derived? Besides Charpy shock test specimens, specimens are placed inside the pressure vessel to measure fracture toughness. These are extracted and the fracture toughness is measured at various temperatures. A curve is drawn as an envelope around the bottom limit of the measurements, in other words below which there is no data. In the Japan Electric Association’s standard JEAC4206-2007 this curve is derived using the following equation:

    K1C=20.16+129.9exp[0.0161 (T-Tp)]…(C8)

Parameter Tp is determined so as to draw an envelope around the measured data (i.e. so that all the data falls above the curve).
    As the amount of neutron irradiation increases, the fracture toughness is reduced and breakage due to embrittlement occurs at higher temperatures. In order to derive a fracture toughness transition curve that corresponds to amounts of irradiation embrittlement other than those given by the measurement test specimens, with the measurement data on the horizontal axis the curve is shifted an amount ΔTK1C parallel to this axis in the higher temperature direction. In that case, ΔTK1C is said to hold. ΔRTNDT is the difference in the DBTT (the amount by which DBTT shifts). In other words, it is assumed that if the temperature at which the fracture toughness value was measured is shifted by the same amount that the DBTT increased, the same fracture toughness value will be obtained. There is no theoretical basis for this relationship, but since it more or less works experimentally, JEAC4206 used this assumption.
    Theoretically, an enveloping curve can therefore be drawn using all the observed test data from the first to the fourth test at Genkai-1, as well as data measured before irradiation. Also, for an arbitrary amount of neutron irradiation, a fracture toughness transition curve (C8) can be drawn. In this way the two curves in Figure 3 show the current K1C curve and the K1C curve 60 years after commencement of operation for estimated amounts of irradiation of the inner surface of the pressure vessel.
    According to NISA’s draft, “The fracture toughness measurement for accumulated irradiation equivalent to that in 22 years from now (60 years from commencement of operations) was approximately double (over 50℃ in terms of temperature) the critical stress intensity factor. This fracture toughness measurement is a directly measured value not related to the accuracy and correlation equations of the prediction method. Even bearing in mind that in general there is a variation of ±25% in fracture toughness for materials within the transition temperature range, it was confirmed that at this point in time there is sufficient margin for operation of Genkai-1.” (p. 11)
    Is this true?

    The first problem is the qualification, “Even bearing in mind that … there is a variation of ±25% in fracture toughness.” Is not the variation in the fracture toughness larger within the transition temperature range? Is it not said that it is from double to half? If there is a variation of 50% in the 80℃ measurement of the fourth monitoring test, what will happen to the K1C curve? I drew this in Figure 4. The result is that the K1C curve approaches much closer to the K1 curve.
    The second problem follows on from the above quote, “In regard to the variation in the monitoring measurement values, although the measurements each time are few in number, they are carried out continuously for fracture toughness for temperatures which take into account the increase in temperature (which can be thought of as the DBTT) for each monitoring test and it is considered rational to take the overall lower limit.” This is also on p. 11 of NISA's draft report. This sentence refers to a shift in the fracture toughness ΔRTNDT based on the abovementioned assumption that ΔTK1C=ΔRTNDT.  However, I submitted an opinion to the Hearings with an analysis that specifically showed that for Genkai-1, at least, this assumption does not hold. It is unacceptable that NISA compiled this draft with no reference to my analysis.
    If this assumption does not hold, the shifted data point is not valid and the only two data points that can be used to draw the K1C curve are those from the fourth monitoring test. With such limited data it is hard to claim that a reliable value for fracture toughness can be derived. I therefore presented the curve in Figure 4 taking into account a variation of 50%.
    However, in appendix A to JEAC4206-2007 there is a rule about what should be done “in the case where the value for fracture toughness is not derived.” This is an instruction to use the following equation to derive the K1C curve from the DBTT values.

    K1C=36.48+22.78exp[0.036(T-RTNDT)]…(A7)

Figure 4: Results of Authors’ Examination of Genkai-1 Pressurized Thermal Shock (PTS) Assessment. JEAC4206-2007 Appendix C and Appendix A, using references (a) and (b).

    Figure 4 shows the curve derived by inserting the fourth monitoring test values for DBTT RTNDT = 98℃ into equation A7. This curve approaches almost to the point of touching the stress curve K1. If the curves were to cross that would mean the pressure vessel would break.
    Next I would like to consider the PTS state transition curve (K1 curve), which shows the size of the stress arising. Are Kyushu Electric’s calculations sufficiently conservative? The assumption in JEAC4206 is for a semi-elliptical 10mm deep and 60mm long crack in the inner surface. It calculates the stress applied to the leading edge of this crack (stress intensity factor K1). Figure 3 shows the PTS state transition curve derived by Kyushu Electric for Genkai-1. According to document 20 presented to the Hearings by Kyushu Electric, for the PTS assessment the most severe large rupture LOCA (loss of coolant accident) is assumed. Kyushu Electric said that it is a conservative assessment in which, without considering the temperature conditions of the inner surface or mixing with cooling water, the temperature would fall in steps from 291℃ to 27℃. (Kyushu Electric gave a confusing explanation implying that the temperature of the inner surface also falls in steps.)
    On the other hand, in Figure 4 the K1 curve referred to as ‘Matsubara and Okamura’ shows the results of a PTS assessment for a pressure vessel of the same dimensions as Genkai-1 (plate thickness 168mm, diameter 3.37). It is a diagram showing the case of a 10mm deep crack (a ratio of crack depth to plate thickness of 0.06). This curve gives a much larger K1 curve than the curve in Kyushu Electric’s assessment. Matsubara and Okamura’s paper assumes a sufficiently long crack, so compared to assuming a crack of 60mm length the values are rather large, but that variation is about 15% based on stress calculations (personal correspondence from Dr. Aono). Even if that amount is subtracted it is above Kyushu Electric’s K1 curve. There is therefore a possibility that Kyushu Electric’s assessment is not sufficiently conservative in regard to pressure conditions, etc.
    On this point, committee member Meshii said that the K1 curve changes greatly depending on the heat transfer coefficient h of the inner surface. If the equation is taken as h=1kW/m2K the result is close to Kyushu Electric’s analysis, but if it is taken as h=2kW/m2K the result is about the same as the Matsubara and Okamura analysis, and for h=∞ it crosses the K1C curve. From this result, Meshii concluded, “The PTS assessment carried out by Kyushu Electric was judged to be close to realistic, but not so conservative that it was not necessary for variation in the fracture toughness value to be taken into account.” He is saying that the curve in the assessment is at the limit and that Kyushu Electric’s analysis does not have sufficient leeway .
    Seen in this light, the conclusion in NISA’s draft report that it has been confirmed that Genkai-1 is “sound enough” in regard to pressurized thermal shock must be seen as lacking foundation. At the sixteenth meeting of the Hearings NISA submitted a new draft which to some extent took into account the various critical views expressed. Debate on this draft is set to begin. However, even though the wording is slightly changed and the data reinforced, the arguments and the conclusion in this draft are the same as before. The conclusion that the Genkai-1 pressure vessel is sound was there from the beginning. The new draft does no more than add all sorts of considerations.
    For reactors with such extreme irradiation embrittlement that the conclusion concerning whether or not they are safe varies depending on the analytical method and point of view, there is no other way to ensure people’s sense of security than to make a decision to shut these reactors down.

    The dangers of nuclear power plants are not limited to earthquakes and tsunamis. Aging is another big problem. In this context, the irradiation embrittlement discussed in this paper is the most fundamental problem requiring attention. Operating for 60 years nuclear power plants which were assumed to have a life expectancy of 40 years is just increasing the danger.
    Destruction of the pressure vessel due to embrittlement is an accident that must not be allowed to happen. If the pressure vessel is destroyed the nuclear fuel will be spread over a wide area and there will be no way of cooling the nuclear fuel to remove the decay heat. Emergency response fire trucks and power supply trucks will all become ineffective. Reactors with even a small risk of being destroyed due to embrittlement should be shut down.

    Of the seven nuclear power plants identified in this paper as having striking irradiation embrittlement, the Fukushima Daiichi Unit 1 reactor has been transformed into a hideous mess and will not operate again. We believe the other six aging reactors should be permanently shut down forthwith.
    A bill to wind up NISA and NSC and establish a new Nuclear Regulatory Commission is now being debated in the Diet. The bill proposed by the government contains a clause saying, “The life of nuclear power plants will in principle be 40 years.” This condition allows a life extension of 20 years in exceptional circumstances, so there is the possibility that the 40-year condition will be gutted of meaning. It should state that nuclear power plants will, without exception, be decommissioned after 40 years.
    All nuclear power plants that began operations in the 1970s will be over 40 years old by 2019. All these early reactors have numerous problems with manufacturing technology and quality of materials, and they are deteriorating. Of course Tsuruga-1 and Mihama-1&2, which are already over 40 years old, should be closed down, and Genkai-1 and Takahama-1, which have extreme irradiation embrittlement, should be closed down without waiting for them to turn 40.

    Acknowledgement: Much of this paper is based on discussions of the Nuclear Aging Research Team. I express my thanks to Chihiro Kamisawa, Yuuta Aono and all the members of the Team.


* The equation is: Formation rate of irradiation-enhanced clusters =  A × (quantity of copper above the solid solubility limit × its diffusion coefficient)^2

(a) Nuclear Industrial and Safety Agency, “Concerning Neutron Irradiation Embrittlement of Reactor Pressure Vessels (Draft),” Hearings on Technological Assessment of the Aging meeting 12 document 5, March 29, 2012.
(b) Kyushu Electric Power Company, “Responses to Committee Member Comments”, Hearings on Technological Assessment of the Aging meeting 8 document 6, February 22, 2012, pp. 3-5.

Return to NIT 149 contents

Return to Nuclear Safety page

Back to first part of this article