ML18040A359

Responds to NRC Request to Submit Comment on NMP Petition to Extend Unit 1 Operating Period from 10,600 Hours to 14,500 Hours Before Insp of Core Shroud

ML18040A359
Person / Time
Site:	Nine Mile Point
Issue date:	12/03/1998
From:	Penn S SYRACUSE UNIV., SYRACUSE, NY
To:	Hood D NRC OFFICE OF NUCLEAR MATERIAL SAFETY & SAFEGUARDS (NMSS)
References
NUDOCS 9812240168
Download: ML18040A359 (24)

Text

~1 4 CATEGORY 1 REGULA '

INFORMATION DISTRIBUTIO SYSTEM (RIDS)

ACCESSION NBR:9812240168 DOC.DATE: 98/12/03 NOTARIZED: NO FACIL:50-'220 Nine Mile Point Nuclear Station, Unit 1, Niagara Powe AUTH.NAME, AUTHOR AFFILIATION PENN,S; Syracuse Univ., Syracuse, NY RECIP'.NAME'ECIPIENT AFF1LIATION HOOD,D.

Office of Nuclear Material Safety 8 Safeguards DOCKET 05000220

SUBJECT:

Responds to NRC request to submit comment on NMP petition to extend Unit 1 operating period from 10,600 hours0.00694 days <br />0.167 hours <br />9.920635e-4 weeks <br />2.283e-4 months <br /> to 14,500 hours0.00579 days <br />0.139 hours <br />8.267196e-4 weeks <br />1.9025e-4 months <br /> before insp of core shroud.

DISTRIBUTION CODE:

A001D COPIES RECEIVED:LTR ENCL SIZE:

TITLE: OR Submittal:

General Distribution NOTES:

E RECIPIENT ID CODE/NAME PD1-1 LA HOOD,D COPIES LTTR ENCL 1

1 1

1 RECIPIENT ID CODE/NAME PD1-1 PD COPIES LTTR ENCL 1

1 INTERNAL: ACRS NRR/DE/ECGB/A NRR/DRCH/HICB NRR/DSSA/SRXB NUDOCS-ABSTRACT EXTERNAL: NOAC 1

1 1

1 1

1 1

1 1

1 1

1 NRR/DE/EMCB NRR/DSSA/SPLB NRR/SPSB JUNG,I OGC/HDS3 NRC PDR 1

1 1

1 1

1 1

1 1

0 1

1 D

1 j

NOTE TO ALL "RZDS" RECIPIENTS:

PLEASE HELP US TO REDUCE WASTE. TO HAVE YOUR NAME OR ORGANIZATION REMOVED FROM DISTRIBUTION LISTS OR REDUCE THE NUMBER OF COPIES RECEIVED BY YOU OR YOUR ORGANIZATION, CONTACT THE DOCUMENT CONTROL DESK (DCD)

ON EXTENSION 415-2083 TOTAL NUMBER OF COPIES REQUIRED:

LTTR 15 ENCL 14

Dr. Steven Penn Syracuse University 201 Physics Building Syracuse, NY 13244 sdpennosyr.edu 315-443-5992 3 December 1998 Darl Hood Office ofNuclear Reactor Regulation Nuclear Regulatory Commission Washington, DC

Dear Mr. Hood,

I attended the public meeting held on 24 September at SUNY-Oswego to discuss the Niagara Mohawk petition to extend the Nine MilePoint One operating period from 10,600 hours0.00694 days <br />0.167 hours <br />9.920635e-4 weeks <br />2.283e-4 months <br /> to 14,500 hours0.00579 days <br />0.139 hours <br />8.267196e-4 weeks <br />1.9025e-4 months <br /> before inspection of the core shroud. Atthat meeting I raised several concerns and technical issues pertaining to the core shroud study performed by Niagara Mohawk and to the NRC safety evaluation ofNine MilePoint One. This letter is in response to your request to submit my comments in writing.

Iwillbegin the letter with a discussion of issues ofscientific integrity, ethics, and safety analysis. Iwillconclude with a review of several technical errors I have found in your report, "Safety Evaluation by the Office ofNuclear Reactor Regulation Regarding the Results of the Reinspection of the Core Shroud, Niagara Mohawk Power Corporation, Nine MilePoint Nuclear Station, Unit 1, Docket No. 50-220"-. ~ "" "

e8<22eoa68 981203 PDR ADQCK 05000220, I

P PDR

\\p IT

To provide some context to my comments Iwould like to begin by relating my educational background. I received both my SB and PhD in physics from MIT.My graduate work was in nuclear structure physics, specifically using electron scattering to probe nucleon-nucleon interactions at high momentum transfer. Igraduated in 1993 and am currently a research associate at Syracuse University.

The nuclear power and nuclear regulatory community in the US is a rather small group, and their work is highly charged with political and economic forces. A group ofscientists and engineers should in principle adhere to the research goals of principled objectivity and of following the scientific method. But the unfortunate reality is that economic and political pressures can and have influenced the conclusions of scientific studies. This problem is particularly acute in areas such as the medical field where the outcome of a study can translate into large financial or career gains. I believe that the nuclear industry faces similar pressures and that any report issued or supported by that community should be reviewed with intense scrutiny to prevent any bias from coloring the results.

In regard to the recent petition, Niagara Mohawk has based their arguments for an inspection delay on the conclusions of a study performed by General Electric. General Electric built the Nine MilePoint One reactor and they are a manufacturer of reactor fuel rod assemblies. Their clear financial interest demonstrates an obvious conflict of interest. In addition General Electric's record for technical integrity is less than stellar the most notorious example being the 1989 report that GE was using substandard bolts when manufacturing aircraft engines for the DOD. GE initially denied these reports that were later found to be true. Curiously the news reports were aired on ABC and CBS but were absent from NBC, a subsidiary of GE.

While I do not claim that the report issued by GE includes any biased results, given GE's conflictof interest in this case, I believe that their report should only be utilized by the NRC ifa parallel study ofequal scope was performed as well.

Secondly, I would like to focus on the difficultyofperforming a proper safety analyses. As physicists and engineers we are trained how to calculate probabilities even for very unlikely events.

However, yourjob as a regulatory agency is complicated by the fact that you must weight each possible event by its impact or possible damage to the community. Simply from a mathematical

I, II

~

perspective this is a difficultcalculation since you are multiplying a small number (the probability of a rare event, for example the core shroud distorting to the point where itimpinges upon the fuel rod assemblies) times a large number (the damage caused ifthe operator lost control of the reaction). Each of these numbers has an associated error that results in a wide range ofpossible outcomes.

Finally we must define a level of acceptable risk and the method by which that level should be determined.

In the particular case of the core shroud cracking, the NRC faces a difficulttask in that the shroud cracking is not a well understood problem. Nor has a fullanalysis yet been done to determine the associated risk factor as a function of the extent of shroud cracking. This risk factor would of course have to include both'the direct impact from the cracking and the more indirect effects when the cracking is combined with other instrument degradation. Finally this risk factor must be compounded by the probable cost of a resulting accident. The limitingcases are trivial:if the shroud is at fullstructural integrity there is no risk and no cost, and ifthe shroud has lost its integrity then the risk ofit impairing the core operation and inducing a meltdown is high and the cost is enormous.

Unfortunately we are now in that uncertain and uncomfortable middle ground.

It is your role to perform the study to calculate that risk factor for all reasonable scenarios. This task is admittedly a daunting one. A reasonable person might even be tempted to neglect doing this work by assuming that we are still within the low risk regime. However the costs resulting from an accident are so extreme that we cannot delay in performing this evaluation, nor can we be so cavalier as to assume that this study would be unnecessary.

Before such an assessment is complete, we are proceeding either in ignorance or with limited intuition. One might reasonably argue that without knowing the risk involved we should shut down the reactors. However ifwe do choose to proceed we should do so with extreme caution and by assuming the most conservative model for the cracking. In the particular case being argued at the public meeting, debating over factors oftwo in the crack growth rate before that growth rate has even been measured, is an unwise and unsafe strategy. Only when the growth rate has been conclusively measured and well modeled should we feel justified and safe in assuming a less conservative value.

Third, I willreview issues regarding the use of theoretical versus empirical modeling. A complex system, by which I mean any system where there are several competing dynamics, and

lg r

Pl 5h 1

especially when these processes are nonlinear, is extremely difficultto model. Any good scientist can construct a model for a laboratory test where a single effect is measured and all other parameters are kept fixed. However combining several processes into a realistic model requires extensive knowledge about how these various processes interact. And that model, no matter how refined, is worthless unless it explains and predicts the data. Atpresent, despite extraordinary gains in science and engineering, the world is fullofcomplex systems that the scientific community can only model empirically.

My point here is that it is the data from a complex system which allows us to test our theoretical models or develop an empirical model. Data from laboratory tests which examine a single factor help us in developing a theoretical model, but taken alone that data does not replace direct measurement on a complex system. In regard to the core shroud at Nine MilePoint One, there have been two measurements done on the core: the measurement of crack length done in the Spring of 1997 and the analysis performed on the two boat samples which was submitted to the NRC in January 1998. This data provides a snapshot of the level of cracking in the core.

However, we are interested in knowing the crack growth rate. To determine the crack growth rate requires several measurements of the crack size at various times. We should not assume that the growth rate would be linear." The fundamental data necessary to test our theories about the core shroud cracking does not exist. That data is only obtained by performing repeated measurements of the core shroud crack size over the next several fuel cycles. Without that data our models are only conjectures. With the scant data that currently exists I would, as an experimental physicist, say that we know very little about the core shroud crack growth rate.

Atthe 24 September meeting a Niagara Mohawk representative made the statement that there are actually two measurements of the crack size since we can take as given that the cracks were of length zero when the shroud was installed. This data is only relevant in establishing an average growth rate over a long time period. The instantaneous growth rate can deviate significantly from the average rate. Moreover since one expects the initialcrack growth rate to be nonlinear, then this "initialmeasurement" ofzero crack length, is practically irrelevant. To illustrate this point we consider the data on core shroud cracks at Nine MilePoint 2 (NM2). The NM2 core shroud was inspected in October 1993 and found to have no perceptible cracks. An inspection on 2 June 1998

P V

found cracks ranging from 0.25 0.65 inches deep over about 60% of the vertical welds. This data would indicate an average growth rate of0.6-1.6 x10 inches/hour for 100% operating time. The water conductivity was less than 0.20 pS/cm for the whole first 3 cycles. I should note that this average growth rate at the given water conductivity is not very different from the instantaneous growth rates predicted by Niagara Mohawk at this level of water conductivity. Nine MilePoint One (NM1) had water with higher conductivity for much of its operating life and should have thus sustained higher crack growth rates, yet even ifits crack growth rate had been as low as the average rate forNM2, the NM1 welds should have cracked through entirely by now. They have not. Therefore the average and instantaneous rates must be different.

My final comment on modeling concerns the Niagara Mohawk conclusions regarding the method of cracking. The NMPC stu'dy concludes that the core shroud cracking is basically intergranular stress corrosion cracking (IGSCC). Since I am not a material scientist Iwilltake as given that the boat samples have the fracture pattern indicative of this method of cracking. Iwill also assume, for now, that we understand that the crack growth rate is primarily a function of the stress intensity in the material and the water chemistry. Given this knowledge itshould be possible for NMPC to use their history of water chemistry and current level ofcracking to model the stress intensity in the shroud and see ifitprovides a self-consistent picture. A more in-depth model is truly what is required. Simply quoting crack growth rates does not do justice to the complexity of the problem. For example, let us assume a crack growth rate of 2.2 x10 inches/hour which is the value cited in the NMPC report. The crack depth along vertical weld 9 (V9) has been measured to be 82% of the wall thickness or 1.23 inches. At the assumed growth rate that crack should have formed in 6.4 years. That crack, which extends for at least 70 inches along the weld, should have grown to this length in 363 years.

These numbers are seemingly ridiculous because the model for crack growth is a great deal more complex than a simple linear rate calculation. One has to model the nonlinear processes of crack formation and crack growth when the material ligature is small compared to the crack dimensions.

To truly prove that they understand the problem, NMPC should make a predictive model that can match several crack size measurements over a range of time. Without a realistic model and with such scant data, I believe it prudent in the near future to assume the NRC's upper bound crack growth rate of 5 x10 inches/hour.

In addition a more

f.

II L

i~L '

I

~

g N

f A

,t 4

detailed model should be required within the next few years to assess the associated risk factor of the core shroud cracking and to determine when the shroud's structural integrity is. likelyfall below,

't the ASME standards.

I would like to conclude my letter by focusing on the NRC report entitled "Safety Evaluation by the Office ofNuclear Reactor Regulation Regarding the Results of the Reinspection of the Core Shroud, Niagara Mohawk Power Corporation, Nine MilePoint Nuclear Station, Unit 1, Docket No. 50-220". This report contained several basic math or conceptual errors which left me seriously questioning the technical ability of the authors of the study. Let me focus on a few of the errors.

~ Page 5, final paragraph:

The report discusses how the systematic error is calculated for the measurement of the crack length. The crack length is measured by determining the position of each end of the crack using a detector. The detector is positioned using a delivery system. Let us declare, a, the position of the detector according to the delivery system, and, b, the position of the crack end as measured by the detector. In this case the crack length is given by i = (a, + bt) (as + ha} where the subscripts refer to the measurements for a given end of the crack. In this case the resul'ting error in the length is given by O'I =

2 0'+CUBI, where cr and o'> are the errors in a and b respectively. Using the errors cited in the report, a', =1.106 and 0'> 0.364 inches, yields a length error cr, = 1.647 inches which is about half the figure given in the report. The author assumed that o, = 2(o ytv l which is incorrect.

~ Page 6, first paragraph:

The EVT (enhanced visual test) error is given at 1.2 inches. Iwill assume that this error was calculated correctly although I am left wondering ifit does not suffer from the mistake noted on page 5. Nonetheless, while the error is first listed as being the error in the length, in the next sentence the author then calculates the total measurement uncertainty as 2.4 inches. I willassume that the author means for the firsterror to be the error in determining the end of the crack. In this case the crack length error is cJI 1.7 inches. In any case cr< o 2.4 inches. The author then makes the egregious mistake of adding the error to the value of the measured crack length. Values and errors are two entirely separate types of objects

that obey completely different mathematics. For those unfamiliar with the basics oferror I

analysis let me recommend that you read Data Reduction and Error Analysisfor the Physica1 Sciences by P. R. Bevington. This book teaches the well founded and proven method for propagating errors through a calculation. When one simply adds the error to the value, as was done in your report, then that value is wrong and all further calculations done using the value are wrong as well. And, even though the error may have been added to the crack length to yield an upper bound, that does not imply that all further calculations done using this length value willyield the most conservative value (i.e. consider the crack growth rate in this case, or any case where errors are not constant for all data).

~ Page 6, Section 3.3, Paragraph 4: The authors note that NMPC was unable to detect welds V5 and V6 using their current detection methods.

They also point out that the tie rod repair loses its effectiveness ifthese welds are completely cracked and welds H2 and H3 are completely cracked. Ifind itsurprising that this point is passes over so lightly. One of the two major points that the NRC report addresses is the failed tie rod repair. Ifthese welds are significantly cracked then the tie rods lose thermal preload. In addition since the ring segment is mounted normal to the direction of tie rod stress, these welds willbe stressed due to the torque from the tie rods and willthus be subjected to higher crack growth rates. NMPC should model how the stress from the tie rods willeffect crack growth rates in the vertical welds. The authors conclude this discussion by stating that inspections reveal that the cracks in weld H3 were not extensive enough to inhibitsafe operation. However the report on the weld inspection, given in App* d'.

h ldH3 HaL'mal

(

h*

• I d

h report).

~ Page 8, Paragraph 2: The author is discussing the crack growth data from the Brunswick BWR and comparing the crack size measured using UT during two successive refueling outages. The author notes "that there was no change in length or depth" which is a meaningless statement in the context of a measurement where the error'in the detector reading is twice as large as the expected signal. Let me explain. Ifwe assume a typical fuel cycle of 14,000 hours0 days <br />0 hours <br />0 weeks <br />0 months <br /> and a crack growth rate of 5 x 10 inches/hour, then we might expect a crack length increase of about 0.7

h I

II ~

H

inches. However as noted above the length error is cr< = 1.647 inches for the method used by k

NMPC. Iwillassume that the method used for the Brunswick measurement was ofsimilar accuracy to the method used by NMPC (ifnot then questions arise as to why NMPC would use a less accurate method). Since the error is large compared to the expected signal one MUST cite the error to understand the measurement.

Even ifthe exact same length and depth were measured the crack may still have grown. In fact by assuming the values for the length error and the fuel cycle duration the error in crack growth rate would be 16.64 x 10 'nches/hour or 3.3 times the NRC limitof 5 x10 inches/hour.

~ Page 8, Paragraph 2: "AtNMP1, no field data on crack growth are available for vertical welds.

Of the horizontal welds, field data are available only for the H8 weld." This fact alone should be the cause of great concern to the NRC for without data from the system in question we can not truly say we understand the system. Gathering more data on NMP1 core shroud crack size should be ofparamount importance to both the NRC and NMPC.

~ Page 8 Paragraph 2: The single measurement on crack growth rate was performed on weld H8.

The depth of the crack measured during the 14th fueling outage was less than that measured during the 13th fueling outage. NMPC drew no conclusions about this data and the NRC simply went along with the industry's erroneous thinking. Once again the NRC neglected the importance of the relevant error analysis in this case. When the measurement errors were considered was the data within the error. Ifnot then how different was the measurement in terms of the error (e.g. lo? 2a?) Ifthere was a large deviation then the NRC should be concerned that the detector error being reported by the NMPC is too small. Conversely ifthe data was within the error then the data should not be disregarded for it is useful data which implies a range of possible growth rates. One is always left to wonder when data is disregarded whether such exclusion is done justly or whether it is excluded because it does not conform to the experimenter's preformed conclusions. The NRC need to be especially watchful of such tainted science when dealing with scientists who may be under pressure to achieve a conclusion which benefits their employer.

Ih

~L

%7 II

"'tS' II

~

l h

~ Page 11, last paragraph:

The allowable vertical weld flaw size is being modeled assuming that the horizontal welds are cracked through wall. The last sentence reads, "Typically, the allowable crack sizes are large and approach or exceed the length of the weld itself." What exactly does it mean when the allowable crack size is larger than the weld itself. Ifthe model truly allows for such unphysical results then the relevance of the results of that model should be seriously questioned.

~ Page 13, Section 4.2.4: This section discusses the NRC staffs independent calculation for the largest acceptable axial flaw. According to the authors the NRC performed a calculation for the bounding final crack length and found it to be slightly larger than would be allowed assuming both LEFM failure mechanism and the ASME criteria. Since the authors do not cite any numerical results the reader has no way of assessing the validity oftheir conclusions.

Nevertheless the NRC notes that NMPC also obtained a similar result when they performed their calculation assuming that the cracks were through wall and that the horizontal welds were fullycracked. The authors then noted that a more detailed NMPC calculation, which included the additional strength from partially cracked welds, was within the ASME safety standards.

They then jump to the conclusion that ifthey performed a similarly more detailed calculation that their results would also fall within the safety requirements. The authors inay be right, but it is the responsibility of the NRC to do these calculations and not speculate what the results might be. It is the NRC's job to perform the necessary safety calculations to ensure that the public's safety is being protected.

~ Page 14, Section 4.3.1: There is a math error in calculating the crack area. It states that a crack has an area of 3 square inches for a 0.003 inch wide and 90 inch long crack. Assuming the crack to be 0.003 inches wide, the correct area is 0.27 square inches. The leakage rates should be reduced by a similar factor although this willnot effect the conclusions drawn in this section of the report.

~ Page 17, Section 5.2.1: In this section the authors review the condition of the tie rod assembly which had been in place for a single fuel cycle. A great deal of interest is placed on the tie rod at

r 0

II

,ui,'

270'hich had vibrated loose and was thus not providing the fulldesign tension. Indeed the updated design was intended to remedy this defect. However I was alarmed to learn that the clasp at 90'ad snapped and that part of the clasp was thrown over to 330'. In any engineering design, one expects that the hardware willbe designed to withstand stresses at least double the maximum expected stress. I would assume that in a nuclear reactor the safety margin should be even higher. Why then did that clasp snap? Was the material ofpoor quality? Was the calculated stress too low? How did this flaw get past the initialNRC review? These are serious questions which are not addressed and leaves one in doubt concerning the design's revision.

~ Page 17, Section 5.2.5: NMPC reported that the tie rod at 270'ad loosened because the base of the rod was not correctly seated in the tie rod anchor on the apron of the core shroud. As part of the design revision they recommended a new installation procedure which, they said, would insure that the rods were properly seated. However two paragraphs later the NRC notes that upon inspection after installation NMPC found that the tie rod middle support at 90'nd 166'as no longer in contact with the RPV. They attributed this loss of contact to "movement of the lower support assembly up the cone toward the shroud." It is just such movement that the new installation procedure was supposed to eliminate. This description suggests that the new design does not correct for the loosening problem.

I am deeply concerned in reviewing your report that you have not implemented enough oversight to prevent conflict of interest from tainting your technical conclusions. I am also disturbed at the numerous errors in basic experimental analysis that exist in your report. Ifyou have made such frequent mistakes in the most basic error calculations, what then should I reasonably assume about your more complex calculations?

And how should I feel about your competence in your role as the public agency charged with performing technical oversight of the nation's nuclear industry? Ifthis oversight is compromised, then would not reactor safety and the

~<

O

11 public health both be at risk? I await your response and do sincerely hope that you willdemonstrate that my conclusions regarding your competence and the associated risk to public health are in error.

Sincerely, Dr. Steven Penn Cc: David Lockbaum, Union of Concerned Scientists

0 J

'<, (c