ML20149E915
| ML20149E915 | |
| Person / Time | |
|---|---|
| Issue date: | 03/01/1994 |
| From: | Muscara J NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | Tim Reed, Strosnider J Office of Nuclear Reactor Regulation |
| References | |
| RTR-NUREG-1477 NUDOCS 9408080196 | |
| Download: ML20149E915 (8) | |
Text
.
=.
/<...,'o, UNITED STATES
!~
7.
NUCLEAR REGULATORY COMMISSION
{
WASHINOTON. D. C. 20SSE t
!.'.AR.' ! ISM MEMORANDUM FOR:
Jack Strosnider, Chief Materials and Chemical Engineering Branch Division of Engineering, NRR Timothy Reed, Steam Generator Task Project Manager Materials and Chemical Engineering Branch Division of Engineering, NRR FROM:
Joseph Muscara, Sr. Metallurgical Engineer I
Materials Engineering Branch
.y Division of Engineering, RES N
i
SUBJECT:
GENERIC LETTER ON VOLTAGE BASED REPAIR CRITERIA This memorandum is in response to Jack Strosnider's meno to Steam Generator Task Team Members of February 17, 1994. The meno requested documentation of concerns the task group members may have with the changes to the draft NUREG-1477 that are being proposed for incorporation into the generic letter on voltage based repair criteria. The memo indicates that the concerns will be included and addressed in the package for the generic letter that goes to CRGR for review.
Since input was requested from the task group members, I did not seek concurrence from management in this meno.
It, therefore, should not be interpreted as providing Office (RES) positions.
Significant departures in the proposed generic letter fr9m the task group
[
recommenpations, draft report NUREG-1477, are the relaxation in the interim olugging criteria (IPC) limit from I volt to 2 volts and a relaxation from plant specific tube pulls for plants using IPC to no tube pulls or an industry tube pull program. -
My memorandum of March 16, 1992 to C. Z. Serpan on the subject:
" Steam Generator Tube Inspection, Integrity and Plugging Issues" discusses my concerns with the use of voltage based plugging criteria and other issues.
If I were to rewrite that memo today, I might incorporate some refinements, but I believe those comments, concerns and issues are still valid. Therefore, I i
suggest that the March 16, 1992 memo be included in the CRGR package.
l Turning now to the IPC Task Group and its draft report,'NUREG-1477.
Page 1-2 of the report states that the NRC established "a special task group to review the acceptability of the voltage-based IPC. Specifically, the group was to review the technical bases for and outstanding issues related to voltage-based IPCs for 005CC at TSP intersections and to prepare conclusions and recommendations concerning implementation of these IPCs, including the technical bases, for presentation to the Committee for Review of Generic Requirements (CRGR)." The Task Group spent considerable time and effort in a) review and evaluation of information from submittals related to voltage based ce D
I W L u C..]
Oh r
9400000196 940301 1
I PDR
?
i Jack Strosnider, Timothy Reed 2
- fa, f i
plugging criteria, b) review and evaluation of other information from various studies and research results and in c) discussion, debate, evaluation and assessment of technical issues and concerns regarding voltage based plugging 1
criteria. The activities and deliberations of the task group resulted in a consensus regarding the issues and conclusions and recossendations for implementation of interim plugging critoria as reported in NUREG-1477. All IPC Task Group members concurred in the report. Jnaddition,thereportwas i
transmitted to the NRR and RES Office Directors by NRR Associate Director for i
Inspection and Technical Assessment and by RES Deputy Director for Research.
The transmittal received review and concurrence by the appropriate management chain in the two offices. The report was published for public comment in June 1993.
I have recently reviewed the public comments and again NUREG-1477. No substantial, relevant new information was presented in the public cosaments, nor developed or presented elsewhere that had not been reviewed and evaluated by the Task Group.
I conclude that the considered discussions, conclusions and recomunendations of the Task Group as reported in NUREG-1477 are still valid and that no basis exists for altering or relaxing the implementation of IPC as reported in NUREG-1477 especially with respect to the IPC limit of one volt and the need for plant specific tube pulls for plants using IPC.
My main concerns stem from application of voltage based criteria when there is a lack of a physical basis for the voltage to relate to the integrity and leak rate of cracked tubes. This lack of physical basis is discussed in my meno of March 16, 1992 and in several locations in NUREG-1477.
Integrity 1.e. the failure pressures, failure modes and leak rates, of cracked steam generator tubes is governed by the lengths, depths snd morphology of the dominant cracks. The integrity of complex cracks such as infant 005CC is also affected by the presence and size of ligaments between cracks. The voltage response from cracks is related to the volume of the crack. Various combinations of crack width (tightness), lengths and depths may produce.the same voltage but
~
different burst pressures.
Further, tight cracks regardfess of size tend to produce low voltages, similarly clusters of cracks with ligaments in between l
tend to produce low voltages. As the cracks grow and ligaments shorten, the.
voltage response may not change substantially, but eventually the ligaments become small enough so that the tube would fail under the pressures of i
interest. Therefore, cracks may be present that are of low voltage but would also fail under relatively low burst pressures.
In addition there ist a saturation of voltage as a function of crack length for part-through wall cracks and to a lesser extent for through-wall cracks. Voltages for part-through-wall cracks are relatively low, while through-wall cracks produce
)
higher voltages. The crack length at which the voltage saturation occurs varies between approximately 0.2 to 0.5-inches. As an example, an 80 percent through-wall flaw 0.5-inch long would produce a given voltage; if this crack were tight and included ligaments, the voltage would be low. The failure pressure (in the burst mode, i.e., primary to secondary) for a tube with this crack would be approximately 4,200 psi (data / correlations from previous NRC research). A flaw of similar depth, (80 percent through-wall), tightness, and with short ligaments but approximately 0.8-inch long would produce a similarly J
low voltage (because of part through-wall, tightness, ligaments, voltage saturation effect) but a tube with this flaw would rupture under a 2600 PSI (MSLB differential) pressure.
In the first case, the tube would not fail
.m.-
,.m-__
-_m,.-
.m
)
1 Jack Strosnider, Timothy Reed 3
D6 / ;g4 under MSL8 pressure, while in the second case, failure by rupture would occur l
under MSLB conditions. A given voltage criterion (even low V) would have l
accepted these two tubes, which produced similar voltages but the voltage l
response was not able to characterize (predict) the gross differences in the failure pressures, failure modes and consequences. Similar arguments as above (length, depth, tightness, ligaments, saturation) indicate that the voltage cannot predict the leak rate.
In order to be able to do so, the voltage must uniquely relate to through-wall crack lengths and to the crack lengths and depths for part-through-wall cracks (since these govern the through-wall crack lengths when cracke pop through). A leak rate correlation with voltage is even more difficult to envision for operating plants because of the effects on i
leak rate of the varying tortuous leak paths through different cracks, varying crack tightnesses, foullag and blockage of cracks by corrosion products, deposits and presence of support structures.
Another characteristic of cracks and voltage response is that there is a discontinuous behavior in the voltage increase when a crack progresses from deep part-through wall to through wall. There may be several volts increase when a crack progresses for example from go percent through-wall to all the way through-wall. This discontinuous large increase in voltage as cracks propagate through-wall is not adequately accounted for in the development of voltage limits, or when predicting the leak rate or burst pressure at the end of a given cycle.
For example, the voltage limit is based on the voltage that corresponds to a burst pressure of three times the nomal operating pressure differential minus a voltage allowance to account for flaw growth between inspections and voltage measurement error. The voltage allowance for flaw growth is the averane voltage change ot' served from previous operating cycles.
These average voltage growths typically range from a fraction of a volt to somewhat over one volt. The average voltage growth may be appropriate for flaws that don't propagate through-wall but it may not account for the several volt increases that may be experienced for those flaws that propagate through-wall during the operating cycle. The probability of leak, the leak rate and the probability of burst would be greater for these flaws than would be predicted when the average flaw growth allowances are used. Further, the burst pressures would be 1cwor than predicted. The numbers of cracks growing through-wall during one cycle of operation could be considerable depending on the voltage limit used and the particular cracks left in service.
For example, NUREG-1477 page 3-7 indicates that the average crack growth rate inferred from eddy current tests and tube pulls from Trojan was approximately 0.0051 inches /yearwhileapproximatelyIgpercentofthecracksweregrowing at rates higher than 0.021 inches / year. At the average growth rate, most cracks would not grow through-wall during a typical fuel cycle, only those that may have been 85-g0 percent through-wall at the beginning of the cycle would grow through. However, the 19 percent of cracks growing at the higher rates would probably all grow through-wall during one fuel cycle if left in service because at this rate the cracks need only be 40 percent through-wall at the beginning of the cycle, and since they were detected they were at least
' Note: There was a typographical error in NUREG-1477 in the decimal point.
i,, -
IME '
- Q Jack Strosnider, Timothy Reed 4
40 percent through-wall.
In addition, since voltage does not relate uniquely to crack length and depth, it does not relate to crack growth rate. Applying an average voltage growth rate to flaws that are growing at lower rates does -
not compensate for the growth and consequences of those cracks that are growing at higher rates than the average.
In the discussion above, for the Trojan example, applying an average voltage growth rate to all the cracks would still produce a low voltage for most cracks at the end of the cycle, indicative of cracks not growing through wall and reflecting low probability of leakage, low leak rate and high burst pressures; while in fact, if voltage growth rates related to crack growth rates, higher voltage growth rates should be applied to the lg parcent of the cracks that are growing at higher rates than the average (greater than 4 times higher) so that at the end of the cycle these cracks would show higher voltages to indicate their progression through the tube wall resulting in higher probability of leakage, higher leak rates and lower bur:t pressures. To predict the voltage at the end of an operating cycle, higher volts should be applied to these cracks to account for higher i
growth rates and additional volts to account for the discontinuous jump in volts as the cracks penetrate the tube wall.
It would appear that in order to set voltage limits (even if a basis existed for it) for acceptance of flaws, infomation is needed from each flaw beyond just the voltage response at the time of inspection and average values from previous inspections; for the particular flaw, one needs to also know the size (depth) of the flaw and its growth rate since the voltage at the end of the cycle is determined by the particular voltage growth rate and whether the crack breaks through the wall of the tube.
The above discussions are intended to show that there are a number of cases where the voltage can not be expected to relate to burst pressure, failure mode, or leak rate of variously degraded tubes.
For certain classes of flaws, 4
the. voltage could be low while the burst pressure also would be low and most importantly for some of these low voltage flaws, the voltage could not discriminate between stable cracks and unstable cracks which would rupture and produce high leakage rates under nomal or accident pressure conditions.
Further, the voltage limits can not be set by simply applying average voltage growth values from previous inspections and not accounting for the voltage increases for those flaws growing at considerably higher rates and not accounting for the voltage jumps when the cracks break through wall. The IPC Task Group recognized and discussed the limitations of the voltage based approach in several sections of NUREG-1471.
It recognized the lack of a physical basis, the lack of a unique correlation between voltage and crack size, the saturation of voltage with increasing crack length for part-through-wall cracks, the lack of a unique correlation between crack growth rate and voltage growth rate and the lack of correlation between voltage and leak rate.
The Task Group also recognized that although a strong statistical relationship exists between burst pressure and voltage, the data.on which the correlation is based are limited and also exhibit significant scatter. Furthemore, in discussing the leak rate versus voltage data, the Task Group indicated that there is a lack of data in the low voltage range 0-to-3 volt that is of great interest.
In relation to the burst pressure / voltage date, the current writer notes that, although the limited data contains some low vdtage flaws, the low voltage tubes were not necessarily selected for their characteristics of e
Jack Strosnider. Timothy Reed S
$8 ' r g interest. The major aim in pulling tubes was to obtain cracked sections that produced higher voltages; in removing these higher voltage tube intersections other cracked intersections on the same tube were removed that were of lower voltage. What is needed in the data base are those tubes of low voltage (below the plugging limit) that might at the same time represent long (or effectively long, i.e. with very short ligaments in between) and deep flaws.
)
In pulling tubes one must search for low voltage response cracks and utilize the other eddy current parameters to select the longest and deepest ones for testing and evaluation.
I believe that when these particular flaws are added to the data base, lower burst pressures for these tubes would be registered than have been observed previously at similar voltages, thus increasing the data scatter and decreasing the correlation factor in the burst pressure versus voltage relationship. The task group identified (NUREG-1477 pg 3-20) six areas of weaknesses related to the voltage-based approach, the sixth is as follows:
"the need for additional data to further evaluate the possibility that flaws with low voltages (morphologies with ligaments) could produce low burst pressures." The Task Group report also indicated that 1) although it is possible to have cracks of low voltage and low burst pressure, this does not seem to be consistent with empirical data regarding the development of 00 SCC l
and with the examination of 00 SCC defects removed from service and 2) the voltage-based approach provides a better description of operating experience to date than the NRC mechanics-based approach. This empirical data regarding the development of 00 SCC and the observations on operating experience must be analyzed carefully. Firstly, with respect to development of ODSCC, the Task Group states (NUREG-1477 pg 3-14) " Examination of 00 SCC defects removed from service and developed in model boilers indicate that in early stages of
. development 00 SCC is characterized by numerous ligaments between linking microcracks.
These ligaments result in lower eddy current response, but at the same time testing of these tubes indicates that the ligaments result in
~
high burst pressure and low leakage rates. The lowest burst pressure observed for any tube removed from service in a U.S. plant to date' was 5400 psi, well above normal operating or postulated pressure differentials." The discussion is correct, however, the observations relate to a fixed point in time in the early stages of development of the 00 SCC cracking process.
It stands to reason that in this early stage of development as the cracking progresses, the ligaments get smaller and smaller and that before the ligaments are completely corroded away, there still is a conductivity path for the eddy currents and therefore the voltage does not change sopreciably. However, as the ligaments get smaller and smaller the burst pressures get lower and lower (and the leak ratas higher and higher).
In fact, the Task Group completes its discussion by stating (continuation of above paragraph in quotes from pg 3-14) "However, this does not preclude the possibility of the existence of cracks whose morphologies may result in lower burst pressures." Secondly, with respect to observations regarding the different approaches and operating experience, the mechanics-based approach predicts cracks growing through wall and higher leak rates.
The voltage-based approach predicts low leak rates.
If 00 SCC cracks were present in operating steam generators that are through wall or that would propagate through-wall ur. der accident pressures, how would they manifest themselves? 00SCCs that develop at the support plates are tight, tortuous, fouled with corrosion products and deposits and surrounded by packed crevices and support structures. Therefore, through-wall cracks in these locations,
)
1 1
%.' f g Jack Strosnider, Timothy Reed 6
l whether short or long, would produce no or little observable leakage.
Furthermore, how would those cracks where mechanics-based models predict failure or rupture under accident conditions manifest themselves in service in l
the absence of accident pressures? Even if accident pressures were l
experienced in service and the cracks propagated through-wall, little or no leakage would likely be observed due to packed crevices and presence of support structures unless the support plate were to move during the accident.
So, the fact that we do not experience leakage in service with steam generator tubes containing 005CC at the support plates, does not mean that potentially dangerous cracks that could produce high leak rates are not present nor that the voltage is a good predictor of tube integrity - the cracks simply do not manifest themselves at the tube support plates. Other observations from operating experience come from the evaluations (burst and leak rate tests, etc.) of pulled tubes.
Tube specimens from 44 tube support plate (TSP) intersections have been removed from five reactor plants that use 7/8-inch-diameter tubing. How do these 44 samples from five plants represent the conditions at any of the approximately 150,000 TSP intersections at any given plant? Furthermore, those samples lack the low voltage, low burst pressure cracks that are of interest since no attempt was made to search for them in the pulled tubes as discussed previously. Even if only one 005CC in one thousand were of the low voltage, low burst pressure morphology in operating steam generators, what are the chances that it is represented in the 44 samples? When conducting inservice inspections of steam generator tubes at a given plant and several thousand indications are reported, how many potentially dangerous (low burst pressure, low voltage, high leak rate or rupture) cracks are left behind even if a low voltage plugging criteria is used and even if only one in a thousand cracks are of the low voltage-low burst pressure morphology?
Is it 2, 3, several? These potentially dangerous cracks can dominate the calculated leak rate.
I believe that the operating experience and the pulled tube data available can not bt used to predict the presence or numbers of potentially dangerous flaws that may be left inservice after inspections even when low voltage plugging criteria are used. Further, the absence of detected leakage in service does not mean the absence of potentially dangerous cracks in the steam generator.
The Task Group recognized that 1) the industry proposed voltage-based approach is an empirical approach based on statistical evaluations of limited tube burst and leakage data which exhibit considerable scatter, 2) there is a lack of a unique correlation between voltage and crack size and voltage growth and crack growth and 3) there is the potential for developing 005CC cracks that generate small voltage response while at the same time having low burst pressures. Therefore, the Task Group concluded that, there is some l
probability that greater leakages and lower burst pressures than those predicted by the voltage based approach could occur. The Task Group further ricognized that the overall conservatism of the approach had not been quantified.
The industry was encouraged "to quantify ths overall conservatism of the analysis by examining the uncertainties and conservatisms being propagated through the analysis from start to finish, beginning with the inspection results and ending with the corresponding estimates of offsite dose for a postulated accident occurring at the end of the next operating cycle."
Since the IPC Task Group report (NUREG-1477) was published, no such analysis
N 1 i$
Jack Strosnider, Timothy Reed 7
has been prepared and reviewed by the staff. When one considers all the uncertainties in every phase of the voltage based approach and, for example, a value of I gpa leakage is calculated at the end of the next cycle, is the real value between 0 and 10, 0 and 100, 0 and 1000 gps?
No new bases, calculations or information have been provMed to justify I
changing the Task Group assessments and recommendations for implementation of a limited voltage-based plugging criteria.
The Task Group beltaved "that use of a 1-volt limit on an interin basis constitutes an appropriately conservative and cautious approach pending additional data and experience, and will provide adequate assurance that nature stress corrosion cracks will not remain in service. The task group also concludes that it is appropriate to limit approval of the interia 1-volt limit to one operating cycle at a time to 1
ensure that future applications of voltage-based plugging criteria properly reflect the latest data and experience."
I concur with the recommendation of the 1-volt limit in that it may provide a low enough probability of leaving in service dangerous cracks that might produce low burst pressures and high leak rates, although I have no quantitative information regarding this probability or the numbers of dangerous cracks.
I can not concur in a 2-volt limit in i
that I do not believe an adequate basis exists for voltage criteria in general and that at this level, many more dangerous cracks could be left in service that would produce low burst pressures and large cumulative leakage. The Task Group also recossended that tubes should be removed at each outage at each plant implementing the IPC to enhance and validate the empirical burst and leakage correlations, to provida data for assessing the reliability of inspection and to confirm that axial 00 SCC continues to be the dominant degradation mechanism.
It is extremely important to confirm that 00 SCC with morphologies providing low voltage and low burst pressure and/or nature 005CC are not left in service. Since the voltage criteria can not provide this confirmation, tube pulls are necessary.
Furthermore, since the SCC phenomenon is highly complex and variable, the confimation must be obtained from the plant under consideration. One needs to know that, at a particular point in time at a give plant, the progression of the 005CC is as expected and bounded by the generic data. Although an industry wide program for tube pulls will provide additional data, it must be assured that the particular plant at the particular time of interest falls within the generic data group And that no low voltage / low burst pressure or nature cracks are being left in service.
Therefore, I concur with the task group recommendation that plant specific tube pulls are necessary to validate the acceptability of the 1-volt criteria.
The Task Group also has stated that "the use of voltage-based criteria for 005CC at TSPs is not intended to preclude the use of length-and depth-based criteria. Over the long tem, the use of length-and/o.r depth-based criteria for plugging and repair would be a preferable approach, and the industry should pursue improved nondestructive testing technology to support such criteria." How does the use of voltage criteria provide incentive for use of the preferable length-and/or depth-based approach and the development of improved nondestructive testing technology? Furthemore, why would there be i
such incentive when increasing the voltage limit as more and more cracks (and more and more potentially dangerous cracks) can be left in service?
j l
l
I ISS4 Jack Strosnider Timothy Reed 8
To conclude, I support all the IPC Task Group recommendations for implementing voltage-based criteria as stated in Draft NUREG-1477 especially with respect to the limit of 1-volt and the need for plant-specific tube pulls.
I do not concur i,ith any relaxation of these parameters.
Finally, based on my knowledge, experience and judgment, and I understand and accept that we all have limits in these areas, I would like to conclude as follows:
1.
I derive ng comfort in the notion that voltage and voltage-based criteria can predict, or allow to maintain, steam generator tube integrity with respect to burst pressures, rupture and leakage, 2.
I derive a little comfort that a maximum of 1-volt plugging limit may acceptably limit the number of potentially dangerous cracks remaining in service, and 3.
I derive ash comfort in realizing that the cracks in the steam generator tubes at the tube support plates are surrounded by a packed crevice, deposits, corrosion products and the support structures which act as the " glue" that keeps the tube constricted, the cracks from excessively leaking or from rupturing and the tube support plate from moving, under normal or accident.
operating conditions, whether the cracks are shallow, deep, through-wall, short, long or of critical size.
N Joseph Muscara, Sr. Metallurgical Engineer Materials Engineering Branch Division of Engineer,ing, RES P
9 a
t/g b 1..i4 DISTRIBUTION:
E. Beckjord W. Russell T. Speis A. Tadani M. Virgilio L. Shao OGC ACRS (10)
L. Plisco, EDO E. Murphy T. Sullivan K. Karwoski J. Hcpenfeld A. Busiik K. Eccleston J. Donoghue S. Long A. Drozd R. Jones E. Butcher J. Craig P. Xadambi J. Cunningham C. Serpan E. Hackett L. Abramson R. Landry C. Liang W. Long 6
.. -, - -.. - -...,,,, - -,