ML20199H097
| ML20199H097 | |
| Person / Time | |
|---|---|
| Site: | Davis Besse  |
| Issue date: | 01/14/1999 |
| From: | Jeffery Wood CENTERIOR ENERGY |
| To: | NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM) |
| References | |
| 2581, GL-97-01, GL-97-1, TAC-M98561, NUDOCS 9901250116 | |
| Download: ML20199H097 (67) | |
Text
i e
y-Daws-Besse Nuclear Power Stanon 5501 North State Route 2 Oak Harbor, Ohio 43449 97(;0 John K.n6od Mc3 President-Nuclear y
Docket Number 50-346 License Number NPF-3 Serial Number 2581 January 14, 1999 United States Nuclear Regulatory Commission Document Control Desk Washington, D. C. 20555-0001
Subject:
NRC Request for Additional Information Regarding the Response to Generic Letter 97-01, Degradation of Control Rod Drbe 14echanism Nozzle and Other Vessel Closure Head Penetrations ', TAC No. M98561)
Ladies and Gentlemen, 1
On April 1,1997, the Nuclear Regulatory Commission (NRC) issued Generic Letter (GL) 97-01 (Toledo Edison Log Number 5027), requesting that licensees for the Davis-Besse Nuclear Power Station (DBNPS), Unit Number 1, provide a description of their program for ensuring the timely inspection of the reactor's control rod drive mechanism (CRDM) nozzles and other vessel closure head penetrations (VHP). The generic letter requested specific information regarding CRDM nozzle and other VHP inspection activities, and also a description of any ion exchange resin bead intrusions and their effect on the primary water chemistry. Toledo Edison (TE) initially responded to the generic letter by letter (Serial Number 2439a) dated April 23,1997 and committed to providing a final response to the NRC by July 29,1997.
]
The final response to GL 97-01 (Serial Number 2472), dated July 28,1997 provided TE's endorsement that the integrated response developed by the Babcock & Wilcox Owners Group (B&WOG) applies to the DBNPS. The B&WOG response is documented in B&WOG Topical Report, BAW-2301 titled "B&WOG Integrated Response to Generic Letter 97-01: Degradation of Control Rod Drive Mechanism Nozzle and Other Vessel L
Closure Head Penetrations." Topical Report BAW-2301, dated July 1997, has been submitted to the NRC.
By a letter addressed to the DBNPS dated September 8,1998 (Toledo Edison Log Number 5339), the NRC requested additional information (RAI) from TE within 90 days regarding the B&WOG integrated program and any deviations from the program specific to the DBNPS. Similar NRC requests for additional information were issued to other glT PWR licensees.
&$h j
P
Docket Number 50-346 Lic::nse Number NPF-3 Serial Number 2581 Page 2 On September 29,1998, the Alloy 600 Issue Task Group of the EPRI Materials Reliability Project met with the NRC staff to discuss proposed actions to respond to the NRC RAls on licensee responses to GL 97-01. The task group had developed a plan for i
a generic response for use by PWR licensees. The NRC staff indicated general agreement with industry's approach and agreed that licensees could extend their plant response submittal dates to January 15,1999. The DBNPS requested this extension in a letter (Serial Number 2569) dated November 19,1998, in order to participate in the industry's approach. Responses to the RAI questions were developed by the Owners Groups, EPRI, NSSS vendors, and contractors. These responses were integrated into a single response by NEI and submitted to the NRC on December 11,1998 (copy attached).
I to the NEIindustry response provides the response to the NRC RAI items applicable to the Babcock and Wilcox Owners Group members. The responses contained in Enclosure 3 for items I through 4 are applicable to the DBNPS with no exceptions.
Should you have any questions or require additional information, please contact Mr.
James L. Freels, Manager - Regulatory Affairs, at (419) 321-8466.
Very truly yours, i
FWK/laj Attachments cc:
J. E. Dyer, Regional Administrator, NRC Region III A. G. Hansen, NRC/NRR Project Manager K. S. Zellers, NRC Region III, DB-1 Senior Resident Inspector (Acting)
Utility Radiological Safety Board t
i 4
1
Docket Number 50-346 License Number NPF-3 Serial Number 2581 Page 3
RESPONSE
iO REQUEST FOR ADDITIONAL INFORMATION TO NRC GENERIC LETTER 97-01 FOR THE DAVIS-BESSE NUCLEAR POWER STATION UNIT 1 This letter is submitted pursuant to 10CFR50.54(f). Enclosed is additional information in response to the NRC's letter dated September 8,1998, regarding Generic Letter 97-01, Degradation of the Control Rod Drive Mechanism Nozzle and Other Vessel Closure Head Penetrations.
For:
John K. Wood, Vice President - Nuclear By:
Lonnie W. Worley, Directodavis-Besse Nuclear Assurance Affirmed and Subscribed before me this 14 day of January 1999.
WY
- "L-Notary Public, $ late of Ohio - Nora Lynn Flood My Commission expires September 4, 2002.
David J. Modeon Ouc T
ustoN I
December 11,1998 i
Mr. Gus C. Lainas, Acting Director Division of Engineering j
Office of Nuclear Reactor Regulation U.S. Nuclear Regulatory Commission Washington,DC 20555
SUBJECT:
Responses to NRC Requests for Additional Information on Generic Letter 97-01 l
PROJECT NUMBER: 689 The NRC has issued PWR licensees requests for information (RAls) on Generic Letter (GL) 97-01, Degradation ofControl RodDrive Mechanism No::le and Other Vessel Closure Head Penetrations. The six groupings of RAls are primarily categorized on the basis of PWR Owners Group membership and the model used by the licensee to assess head penetration crack initiation and gmwth. Due to the generic nature of the RAls, the Alloy 600 Issues Task Group of the EPRI Materials Reliability Project chose to develop generic RAI responses for voluntary use by licensees. These generic responses are provided to the NRC staff because licensees may reference portions of the enclosures in their RAI responses.
The Alloy 600 Issue Task Group developed the generic responses with input from the PWR Owners Groups and EPRI. Efforts were made to use consistent responses when common RAI questions were asked. The content of the seven enclosures attached are:
)
l
- The industry histogram that summarizes the predictive model results and i
identifies which plants have or will inspect their head penetrations : Responses to generic questions that apply to all PWR licensees (All six groupings of the RAls contain generic questions or a question that applies to all PWR licensees. Since the generic questions were not always asked using the same text, the RAI generic questions have been paraphrased.)
' : The responses for the RAI questions applicable to the Babcock and Wilcox Owners Group (B&WOG) members : The responses for the RAI questions applicable to the Combustion Engineering Owners Group (CEOG) members i
l J
Mr. Gus C. Lainas December 11,1998 l
Page 2 1 : The responses for the RAI questions applicable to the Westinghouse Owners Group (WOG) members
! : A discussion of the EPRI predictive model developed by Dominion Engineering, Inc. : A discussion of the Westinghouse predictive model If you have questions, please contact Kurt Cozens at (202) 739-8085 or koc@nei.org.
f Sincerely, l
David J. Modeen i
KOC/edb Enclosures c:
Mr. Jefferey F. Harold, U.S. Nuclear Regulatory Commission Mr. James Medoff, U.S. Nuclear Regulatory Commission Mr. Jack R. Strosnider, Jr., U.S. Nuclear Regulatory Commission Mr. Edmond J. Sullivan, Jr., U.S. Nuclear Regulatory Commission Mr. Keith R. Wichman, U.S. Nuclear Regulatory Commission i
l l
l 4
e l.
1 A
I HISTOGRAM OF RPV HEAD NOZZLE ASSESSMENTS AND PLANT INSPECTION PLANS i
f n
1 4
1 1'
f d
_m.,
Enclosure i Page1 HISTOGRAM OF RPV HEAD N0ZZLE ASSESSMENTS AND PLANTINSPECTION PLANS 1.1 Introduction The purpose of this document is to present a histogram (Figure 1) which compiles the PWR reactor pressure vessel (RPV) head penetration cracking assessments for all operating domestic PWRs (Combustion Engineering, Babcock & Wilcox, and Westinghouse designs).
The histogram groups all plants into three categories representing the amount of time remaining (in efTective full power years (EFPYs) from January 1,1997) until the plants are calculated to have the same probability of having a crack at the allowable depth as DC Cook 2 had at the time the 6.8 millimeter deep crack was discovered in 1994. The three categories are 1) less than 5 EFPYs,2) 5-15 EFPYs, and 3) more than 15 EFPYs.
1.2 Development ofIIistogram Based on Comparison to DC Cook 2 PWR RPV head penetrations were analyzed using models developed by the Electric Power Research Institute (EPRI) (see Enclosure 6) and by the Westinghouse Electric Company (see ). Both of these are probabilistic models which use the Monte Carlo method to handle uncertainties.
The histogram in Figure 1 is based on the cumulative probability of at least one penetration in the head of each plant having a crack at the allowable depth, typically 75% through-wall (see NRC Safety Evaluation Report addressed to NUMARC dated November 19, 1993). The analysis results are reported as the time (in EFPYs of operation from January 1,1997) for each subject plant to reach a reference probability level. This reference probability was established using the results of the DC Cook Unit 2 inspection in 1994 and calculations made for DC Cook Unit 2 using the same methodology as for the subject plant. It is the probability that a 75% through-wall crack existed at DC Cook 2 at the time ofits inspection. This probability is somewhat lower than that for the actual observed crack depth of 6.8 millimeters (43% through-wall), as illustrated in Figure 2.
By grouping the plants based on their relative probability of experiencing a flaw at the allowable depth, the results of both the EPRI and Westinghouse models were normalized and presented in the same histogram. The results plotted in Figure 1 show:
Seven plants calculated to have the same probability of having a crack at the allowable depth as DC Cook 2 in less than 5 EFPYs after January 1,1997.
Sixteen plants calculated to have the same probability of having a crack at the allowable depth as DC Cook 2 in 5 to 15 EFPYs after January 1,1997.
Forty-six plants calculated to have the same probability of having a crack at the allowable depth as DC Cook 2 in more than 15 EFPYs after January 1,1997.
Page 2 The plant names in each of the three histogram categories are provided in Table 1 in alphabetical order within each category and separated by inspection status.
1.3 Discussion of Plant Assessments and Inspection Status To date, six domestic plants have inspected some or all of the nozzles. These are DC Cook 2, North Anna 1, Oconee 2, Millstone 2, Point Beach 1, and Palisades. As reflected in the first row of Table 1, these six plants include plants in each of the three different histogram categories. DC Cook 2, North Anna 1, and Oconee 2 are in the group with highest calculated susceptibility, Millstone 2 and Point Beach I are in the group with intermediate calculated susceptibility, and Palisades is in the group of plams with lowest calculated susceptibility.
There are four other plants in the group with highest calculated susceptibility. These are Farley 2, Oconee 1, Oconee 3, and Surry 1. Farley 2 has plans to perform inspections. The remaining three plants in the highest susceptibility group are characterized by the licensee as
" sister" plants to similar plants owned by the same licensee which have already inspected:
Oconee 1 and Oconee 3 are lower calculated susceptibility " sister" plants to Oconee 2, while Surry 1 is a lower calculated susceptibility " sister" plant to North Anna 1. Consequently, all plants in the highest susceptibility group have inspected, have plans to inspect, or are lower susceptibility " sister" plants to units which have already inspected.
Four additional plants, which are in the intermediate susceptibility group, have planned inspections: Crystal River 3, Diablo Canyon 2, Ginna, and San Onofre 3. These inspections will provide additional assessment of the predictive models.
Table 2 summarizes the above information.
Table 2. Summary of Plant Assessments and Inspection Status Status Assessment Groups
> 15 EFPYs 1
Inspected 3
2 1
j Plans to Inspect 1
4 0
l Sister Plants to Lead Units 3
Remaining Plants 0
10 45 Totals 7
16 46
)
Enclosure I l
Page 3 1.4 Summary The above results can be summarized as follows:
Six plants have inspected to date, including plants calculated to be in each of the three categories.
- There are seven plants in the group with highest calculated susceptibility. All seven of these plants have inspected, plan to inspect, or are " sister" plants to other lead units with higher computed susceptibility that have already inspected.
A total of five plants plan to inspect over the next four years, i
As additional inspection results become available, the industry Materials Reliability Project (MRP) will evaluate the results and take the necessary action to manage the issue.
)
1 l
Page 4 r O Plants That Have Already Performed Inspections EPlants That Have Announced Plans to Inspect a Sister Plants to Lead Units That Have Already inspected O emaining Plants R
50
~~
45
' fih;;l Bdh1 40 C;D m; ; MJ/
35
- % Mp
,3&yg,llJ.
/!V l
g st 30
) jf, ?
kWBNhl{;f 4-en 25 s-3 20 "4W1 E
N yM W @"
z 15 MM
- - g,m
- n y&t@
CEMI 2
10 M;
h L.
AQ i' ' *N 5
lJn.. ; t
^*
m WM
,gi i
0 ME
> 15 EFPYs Effective full power years (EFPYs) from 1/1/97 until probability of having a crack at the allowable depth matches DC Cook 2 probability of one 75%
through-wall crack at time of its 1994 inspection Figure 1. Industry IIistogram for RPV IIcad No721e PWSCC Table 1.
Identification of Plants in Indust y IIistogram for RPV IIcad Nozzle PWSCC
- --~-- -- - - - - - - - - -
)
i i
Page 5 g
Assessment Groups *
< 3 LFPYs 5-15 EFPYs
> 15 brPYs Plants That llave Already
'wm nunce e rausaces North Anna 1 Point Beach I Perfortned Inspections Oconee 2 i arley 2 trystal Kner J Plants That Have Diablo Canyon 2 Announced Plans to inspect Ginna San Onofre 3 t
Sister Plants to Lead Units Wonce 8 f,
,;+
O' "
W' That Have Already Inspected
'^
ri licaver Valley I ANUi 1
1 Beaver Valley 2 ANO2 DC Cook I Braidwood 1 Davis Besse Braidwood 2 j
North Anna 2 Byron 1 Robinson 2 Byron 2 Salem i Callaway Surry 2 Calvert Clifts 1 Turkey Point 4 Calvert Cliffs 2 I
Waterford 3 Catawba 1 4
Catawba 2 i
l Comanche Peak I Comanche Peak 2
)
Diablo Canyon 1 Farley I i
fort Calhoun j
Indian Point 2 Indian Point 3 1
Kewaunee 4
l McGuire 1 4
McGuire 2 j
Millstone 3 i
Remaining Plants Palo Verde 1 I
Palo Verde 2 Palo Verde 3 i
Point Beach 2 Prairie Island 1 Prairie Island 2 l
Salem 2 San Onofre 2 Seabrook SequoyahI Sequoyah 2 Shearon liarris l
South Texas 1 South Texas 2 St. Lucie 1 1
St. Lucie 2 Summer 1MI1 Turkey Point 3 Vogtle i Vogtle 2 Watts Bar i WolfCrcek
'Lt!ccuse i ull l'ower Years (Lt l'Ys) from lili97 until probatulity of haung a crack n' the allowable depth matches DC Cook 2 probability of one 75% through-wall crack at time ofits 1994 inspection.
____.m_
Page 6 l
t Re&cted nomMure of 6.ssana l
Deep (43%nsough!suli) Cock in DCCout 2 M
bU Reested hishibty of12 nun Dup (751Hrongh mull) b3 Crack in DCCook 2 lies eted firbedhty of g
j Oack al Allowsle Depth y5 la Subject 11ans t! "
g =l' 6
51m k-3
-l#u
'unn at Mich lista!ut M ich a
6 A rain Deep Subject Ilant has Qad mus Sanne Predicted Discoveredia ItchMEty of 75%
l DCCod 2 Thaowgh mullCsack asDCCuok 2 l
l Figure 2 Method Used to Normalize Model Assessments l
I
Enclosme 2 Page1 RESPONSES TO GENERICNRC REQUESTS FOR ADDITIONAL INFORhlA TION This enclosure provides responses to the questions that are common to the six forms of NRC Generic Letter 97-01 Requests for Additional Information (RAls). Descriptions of the EPRI and Westinghouse models are contained in Enclosures 6 and 7, respectively.
The RAI questions have been paraphrased to streamline and standardize the responses. The NRC has verbally agreed that these paraphrased questions are consistent with the original wording and may be used by licensees in their RAI responses. The order of the questions may be different than the order received by individual licensees.
OUESTION 1. " Describe how the variability in the productforms, material specifications, and heat treatments used tofabricate each CRDM/CEDMpenetration are addressed in the crack initiation andgrowth modeh. "
EPRI MODEL RESPONSE 1.a. The EPRI model time-to-crack-initiation predictions for a subject plant are based on the results ofinspections at plants which most closely resemble the subject plant in terms of material product form, material specification, material supplier, material heat treatment, and vessel head fabricator. This approach avoids the need for major corrections to reflect differences in material PWSCC susceptibility. Minor variations from nozzle to nozzle are accounted for statistically through the Weibull slope parameter and by applying a triangular distribution to the reference time to 10% probability of cracking. At the present time, EPRI considers that sufficient laboratory or field inspection data are not available to more precisely define the effect of product form, materiar specification, and heat treatment on the crack initiation rates. If proven correlations become available in the future, they will be included in the EPRI model. All EPRI model crack growth predictions are based on application of a log-triangular distribution to the available laboratory and field data corrected for temperature and stress intensity.
WESTINGIIOUSE MODEL RESPONSE 1.h. Since the Westinghouse probabilistic analysis models are mechanistically based, uncertainties are provided to directly account for the variability in such fabrication related input parameters as nozzle wall thickness, material grain boundary carbide coverage and monotonic yield strength. The Westinghouse mechanistic model also accounts for the variability in indirect fabrication related effects, such as the variation in surface roughness on crack initiation and the variation in the actual weld size on the local stress, where there is insufficient information to describe the causes and effects in a statistically significant manner. Specifically, the model input also includes the observed uncertainties on the coefficients used to calculate residual stress, initiation time and crack growth rate.
Page 2 OtissnON2. " Describe how the crack initiation andgrowth modelsfor assessingpostadated flaws in the no::les were benchmarked, andlist and discuss the standards that the models were benchmarked against."
EPRI MODEL RESPONSE 2.a. Benchmarking for crack initiation is performed using a reference nor:le concept.
After each plant inspection is completed, the vessel head and nozzles are analyzed using the EPRI model to determine the time to 10% probability of cracking for a reference no::le with a surface hoop stress level of 60 ksi and an operating temperature of 600 F which results in a 50%
cumulative probability of the observed inspection results when corrections for differences in stress and temperature between the reference no::le and the nozzles in the inspected pl:mt are included. This information is then evaluated relative to the results ofinspections for other plants to establish a time to 10% probability of crack initiation for each different group of nozzle materials.
Crack growth is benchmarked using reported crack growth rates obtained from controlled laboratory tests and field inspections corrected for differences in temperature and crack tip stress intensity. Please refer to the EPRI methodology description (Enclosme 6) for additional information on how the EPRI model is benchmarked.
WESTINGHOUSE MODEL RESPONSE 2.b. The Westinghouse models and software used for the probabilistic analysis of reactor vessel head penetration nozzles were developed using the structural reliability and risk assessment (SRRA) methodology. The application of this SRRA methodology to piping risk-informed ISI was extensively benchmarked against hand calculations, available failure data and alternative calculations as described in WCAP-14572, Revision 1, Supplement 1 (October 1997).
NRC is currently planning to issue a SER accepting this application of SRRA by the end of 1998.
As described in Table 4-2 of WCAP-14901 (July 1997), the SRRA probabilities for Alloy 600 PWSCC compare very well with inspection observations at four plants, where sufficient information existed to perform calculations for the worst head penetration nozzle at the time they were first inspected. While two of the plants (D. C. Cook 2 and Ringhals 2) with relatively high calculated probabilities had observed flaw indications, two other plants with lower calculated probabilities (Almaraz 1 and North Anna 1) did not. The initial WOG probabilistic model was revised as a result of the North Anna 1 inspection observations and an independent peer review by Alloy 600 PWSCC specialists (Jim Begley and Brian Woodman) at APTECH Engineering in the spring of 1997.
)
Orissnox 3. " Provide additional information regarding how the model will be refined to allow the input ofplant-specific inspection data into the model's analysis methodology "
Page 3 EPRI MODEL RESPONSE 3.a. Plant-specific inspection data are factored into the EPRI model predictions in two ways:
1.
As each plant inspection is completed, the vessel head and nozzles are analyzed using the EPRI model to determine the time to 10% probability of cracking for a reference nor:le with a surface hoop stress level of 60 ksi and an operating temperature of 600 F which results in a 50% cumulative probability of the observed j
inspection results. These data are updated periodically and provided to users of the i
EPRI model software. If an inspection indicates a significant change in reference nor:le conditions, users are notified.
2.
Once a plant has performed an inspection, the results of the plant-specific inspection, along with the results for other plants in the same nozzle material group, are used to establish a plant-specific reference for future predictions.
WESTINGIIOUSE MODEL RESPONSE 3.b. There are two kinds of variations that are considered in the Westinghouse probabilistic analysis: random and systematic. The random variation is that due to localized material variability and other effects with insufficient information available to completely characterize them. This could include the effect of the variation in surface roughness on crack initiation and the variation in the actual weld size on the local stress. For these types of uncertainties, a Bayesian updating process has been developed by Westinghouse that could be used to combine the prior distribution on time to failure, which gives the initial calculated probability of failure with time, with the observations from the inspection. The updated posterior distribution that is generated in this manner can then be used to generate an updated estimate of the probability of failure with time for each penetration that was inspected.
The systematic or mechanistic type variations, such as the time to crack initiation being inversely j
proportional to the stress to the 4th power, are included directly in the Westinghouse probabilistic model. If the observations from an inspection would differ significantly from what was calculated, then the basic model would need to be revised. This in fact has already occurred based upon the observations from the North Anna Unit 1 inspections. The revised model now provides calculated probabilities that are consistent with the current inspection observations (see previous response to question 1b).
Ol!ESUON 4. " Provide the latest model susceptibility ranking ofyour plant based on the model j
analysis results, including the basisfor establishing the ranking ofyour plant relative to the j
ranking ofother plants in your msners group analy:ed using your model. "
Page 4 EPRI AND WESTINGHOUSE MODELS RESPONSE 4. For industry planning purposes, plants have been grouped into three categories i
based on the predicted time to reach the allowable flaw depth limit. These results are provided in the industry histogram provided as Enclosure 1.
OUESUON 5. " Table 1-2 in WCAP-14901 provides a summary ofthe key tasks in the WEC vessel headpenetration no::le assessmentprogram. The table indicates that tasksfor 1) evaluation of mitigation methods, 2) crack growth data and testing, and 3) crack initiation characterization studies are still in progress. In light ofthefact that thepredictive models appear dependent in part on crack initiation andgrowth estimates, provide your best estimate ofwhen these tasks will be completed, and describe how these activities relate to and will be used to update the susceptibility assessments at your plant. "
}
EPRI AND WESTINGHOUSE MODELS l
RESPONSE 5. The programs on crack growth testing and crack initiation have been essentially completed, and the program on mitigation is now underway and targeted for completion in mid-2000. These programs have thus far served to confirm the assumptions used in the original safety evaluations and models. As additional information becomes available from the referenced testing, the models will be reviewed and updated as necessary. No major changes are anticipated.
B&WOG RESPONSES To NRC REQUESTS FOR ADDITIONA.L INFORMATION
1 Page1 B& WOG RESPONSES TO NRC REQUESTS FOR ADDITIONAL INFORnfA TION The B&W Owners Group (B&WOG) developed a cooperative, integrated program in response to the Nuclear Regulatory Commission (NRC) Generic Letter 97-01 titled " Degradation of Control Rod Drive Mechanism Nozzle and Other Vessel Closure Head Penetrations." This program is documented in the B&WOG Topical Report, BAW-2301, dated July 1997 (reference 1). The NRC subsequently issued to member utilities of the B&WOG requests for additional information (RAls), which are summarized in reference 2, and the responses are provided below.
Ouestions and Responses Applicable to All B&WOG Member Utilities Q1. Provide a description ofhow the various productforms, material specifications, and heat treatments used tofabricate each CRDMpenetration no :le at the B& WOG member utilities are handledin the CIRSE model.
The basic B&WOG response to this question is provided in Response 1.a of Enclosure 2. For clarification, the Electric Power Research Institute (EPRI) CHECWORKS Reactor Pressure Vessel llead Nozzle Module (RHNM) model described in Enclosure 2 is an industry adaptation of the original CIRSE model software and uses the same inputs and assumptions as the CIRSE model cited in reference 1. A more detailed description of the EPRI model, than that provided in Appendix B of reference 1 for the CIRSE model, is provided in Section 3 of Enclosure 6.
The EPRI modeling approach, which is used by the B&WOG, predicts time to crack initiation for a particular plant based on the results ofinspections at plant (s) that most closely resemble the particular plant in terms of material product form, material specification, material supplier, material heat treatment, and vessel head fabricator. For the B&WOG, the reference plant is currently Oconee Unit 2. As additional inspections are performed, these data will be incorporated into the reference plant (s) analysis (e.g., Crystal River Unit 3).
As stated in Section 2.2.1 of the B&WOG Generic Letter 97-01 response (reference 1), all of the existing 483 CRDM nozzles at the operating B&WOG plants were fabricated from the same product form (i.e., hot-finished seamless tubing) and from only 13 individual heats, which comprise 17 heat treatment lots, of Alloy 600 material. The Alloy 600 materials were either supplied by the Babcock & Wilcox Tubular Products Division (B&W-TPD) or by the International Nickel Company, Inc. Huntington Alloys Product Division (INCO). B&W-TPD manufactured 11 of the 13 heats of material (472 of the 483 CRDM nozzles), and B&W fabricated and installed all 483 CRDM nozzles. Therefore, relatively consistent material properties, such as microstructure, mechanical properties, material chemical composition, and as-fabricated surface finish were maintained for the CRDM nozzles installed at B&WOG plants.
Also, only 11 CRDM nozzles were fabricated from the two heats of material supplied by the other material supplier (INCO). However, the differences in yield strength among the 17 heat treatment lots are used as input to the EPRI model to ascertain the inside surface hoop stress for each CRDM nozzle.
Page 2 Obviously, there are heat-to-heat material variations in the B&W-design plants, as shown in Table 1 of reference 1. The triangular distributions and statistical treatment used in the EPRI model software were judged to be adequate for development of the B&WOG integrated inspection plan. When additional laboratory test and field inspection data become available, they will be evaluated and incorporated into the EPRI long-range planning model, as appropriate.
Q2. Provide any additional information, ifavailable, regarding how the model will be refined to allow the input ofplant-specific inspection data into the model's analysis methodology.
The basic B&WOG response to this question is provided in Response 3.a of Enclosure 2. In this manner, plant-specific inspection data applicable to the B&WOG (e.g., the Oconee Unit 2 re-inspection results and the Crystal River Unit 3 inspection results) is then factored into the model predictions.
These results, in addition to other planned U.S. industry inspections, worldwide industry experience, and economic factors, will be used to evaluate the time frame and need for future B&WOG plant inspections.
Additional details are provided in Section 3 of Enclosure 6.
Q3. Describe how FTI's crack initiation and crack growth modelsfor assessingpostulatedflaws in vessel headpenetration no::les were benchmarked, andprovide a listing and discussion ofthe standards the models were benchmarked against.
The B&WOG response to this question is provided in Response 2.a of Enclosure 2. Framatome Technologies, Incorporated (FTI) uses the EPRI crack initiation and crack growth models for assessing postulated flaws in vessel head penetration nozzles at B&WOG member plants.
Additional details are provided in Section 3 of Enclosure 6.
Q4. Provide the latest CIRSE model susceptibility rankings ofB& ll'designedfacilities based on the CIRSE model analysis results compiledfrom the analyses ofthe CRDAf and instrumentation no::les at thefacilities.
The basic B&WOG response to this question is provided in Response 4 of Enclosure 2.
Since there are only two B&WOG plants (Oconee Unit I and Three Mile Island Unit 1) that each have eight thermocouple nozzles, the thermocouple nozzles ' nave been evaluated independently and are addressed in the response to B&WOG member utility RAI question 5 below.
Questions and Responses Applicable to Duke Power Company for Oconec Units 1. 2. and 3 and GPU Nuclear. Incornorated for Three Mile Island Unit 1 Q5. Compare the overallsusceptibility rankings ofthe thermocouple no::les at ONS-1 and TAff-I to that ofthe plants with the most susceptibly ranked CRDAipenetration no::les. Based on Page 3 this assessment, indicate whether the thermocouple no::les at ONS-1 and TMI-1 will be inspected during the year 2001 refueling outage. Ifit is determined that the thermocouple no::les will not be inspected, provide the basisfor omitting the inspections ofthe thermocouple noz:les in the year 2001.
As mentioned in reference 1, the RV head thermocouple nozzles (ONS-1 and TMI-l only) are included in the integrated B&WOG inspection program. When the Generic Letter response was prepared in 1997, preliminary results of stress analysis indicated that these vessel head penetrations potentially ranked among the higher susceptibility nozzles at the B&WOG plants.
Based on those preliminary results, inspections of the thermocouple nozzles were tentatively scheduled for 2001 at both ONS-1 and TMI-1.
Since that time, additional analytical evaluations have been performed, and it has been concluded that the thermocouple nozzles are still ranked among the higher susceptibility nozzles for the reactor vessel head nozzle population (e.g., ONS-2 and CR-3). However, the predicted integrated probability of at least one nozzle cracking of the eight thermocouple nozzles at ONS-1 is less than that predicted for the high susceptibility CRDM nozzles at ONS-2 and CR-3. For the thermocouple nozzles at TMl-1, the integrated probability of cracking is notably less. In addition, the following items provide furtherjustification for concluding that the thermocouple nozzles are not of significant concern in the short term:
- 1. The thermocouple nozzles are attached to the reactor vessel head using a partial penetration weld. The predominant mode of cracking with this type of weld has been shown, both by finite element analysis and PWSCC experience, to be axially oriented.
Circumferential cracking has occasionally been observed to initiate on the surface of some nozzles (e.g., pressurizer nozzles) and shown not to propagate beyond a very shallow depth.
- 2. The thermocouple nozzles are not shrunk fit into the reactor vessel head penetration prior to welding, and therefore a diametrical gap (approximately 5 mils) exists between the thermocouple nozzle and the reactor vessel head penetration. Therefore, should any amount ofleakage occur, it would be readily observable.
- 3. No known reaming or grinding operations were performed before welding, which would increase the PWSCC susceptibility of the thermocouple nozzles.
Therefore, it is concluded that no safety concern exists with the thermocouple nozzles.
Preliminary inspection plan activities are underway by Duke Power Company to inspect the thermocouple nozzles on the reactor vessel head of ONS-1; however, this inspection is planned for the end-of-cycle 20 outage in the spring of 2002, as there currently is no outage scheduled in 2001. Since TMI-l has significantly less operating time than ONS-1, GPU Nuclear, Incorporated has decided to evaluate the thermocouple nozzle inspection results from ONS-1 before scheduling an inspection.
i Page 4 i
i Ouestions and Responses Applicable to GPU Nuclear. Incorporated for Three Mile Island 3
Unit 'l Q6. Given that the TAff-1 facility experienced an extendedintrusion ofthiosulfate ions into the TA11-1 RCS, andsince the degradation ofAlloy 600 steam generator tubes at TAff-1 has inpart been attributed to this event, justify why the Alloy 600 CRDAfpenetration nozzles at TAff-1 are not being scheduledfor volumetric inspection in the near term.
l-Response will be submitted directly by GPU Nuclear.
References i
i
- 1. "B&WOG Integrated Response to Generic Letter 97-01: ' Degradation of Control Rod Drive 1
Mechanism Nozzle and Other Vessel Closure Head Penetrations'," BAW-2301, July 1997, available from Framatome Technologies, Inc., Lynchburg, Va.
- 2. " Request for Additional Information Regarding the Generic Responses from the Pressurized Water Reactor (PWR) Owners Groups to Generic Letter (GL) 97-01, ' Degradation of CRDM/CEDM Nozzles and Other Vessel Closure Head Penetrations'," Memorandum from J.F. Harold through S.S. Bajwa to NRR Project Directors, August 11,1998, available in the l
Nuclear Regulatory Commission Public Document Room, Washington, D.C.
Y a
3 E
d 3
4 1
1 1
1 J
CEOG RESPONSES TO NRC REQUESTS FOR ADDITIONAL INFORMATION Page1 CEOG RESPONSES TO NRC REQUESTS FOR ADDITIONAL INFORhfA TION The Combustion Engineering Owners Group (CEOG) provided a generic response (CE NPSD-1085,"CEOG Response to NRC Generic Letter 97-01, Degradation of CEDM Nozzle and Other Vessel Closure Penetrations", July 1997) to NRC Generic Letter 97-01.
The NRC subsequently issued to CEOG members requests for additional information (RAls) on the CEOG methodology for determining the susceptibility of vessel head nozzles to PWSCC.
The RAls noted that the NRC had informal information indicating that the CEOG had decided to change the methodology for evaluation of vessel head penetration susceptibility to PWSCC from the model described in detail in CE NPSD-1085 to a new EPRI model. The CEOG has decided that the most appropriate manner to respond to the RAls is through an integrated response, as suggested by the NRC. The CEOG responses to the RAls addressed to its member utilities are as follows (questions are paraphrased):
Question 1.
Specify whether the ABB-CE PITM model or the EPRI model is being endorsed for the assessment ofCEOGplants.
Response
The CEOG now endorses the use of the EPRI model for the assessment of PWSCC in Alloy 600 reactor vessel head penetrations. The use of this model and the results generated from its application supersede those of the probabilistic inspection timing model (PITM) described in CE NPSD-1085.
CE NPSD-1085, in responding to GL 97-01, noted that the PITM was used in selecting which CEOG plants to inspect and the timing associated with the inspections. The PITM was developed in 1993 to evaluate the susceptibility to PWSCC of all CEDM and ICI nozzles in all CEOG plants. The PITM addressed crack initiation and crack propagation to determine the time for a crack to grow to a detectable size and to propagate completely i
through wall. The PITM used (a) a two parameter Weibull model to predict time-to-crack initiation for individual nozzles with estimates for the scale and shape parameters being derived from operating experience with small diameter Alloy 600 nozzles in CE plants.
(b) a power law fracture mechanics model for crack propagation.
(c) Monte Carlo simulation to calculate the risks of cracks or leaks for all nozzles in a specific head considering uncertainties in the input parameters and varying susceptibilities of nozzles to PWSCC.
After issuance of CE NPSD-1085, the CEOG compared the PITM with newer industry models and determined that the PITM needed to be upgraded or replaced with a newer model. The CEOG contracted EPRI to provide a new model which is described in
Page 2. This is the PWSCC susceptibility model that the CEOG is now endorsing and was the model used to develop the CEOG input to the industry histogram that is being submitted to the NRC along with this document.
The EPRI model general characteristics are the same as those of the CE PITM model described above - i.e.,it uses l
(a) a two parameter Weibull distribution to predict time to initiation.
(b) a power law fracture mechanics relationship to predict crack growth.
l (c) a Monte Carlo analysis to calculate the integrated risk of cracks for an entire nozzle population considering the uncertainties in input parameters and varying PWSCC susceptibility ofindividual nozzles.
Unlike the CE model, the EPRI model permits utilities to evaluate the risk of cracking i
and the costs for a number of alternative strategies to address the cracking issue. Aside from an economic modeling capability, the EPRI model includes many enhancements not j
available in the PITM. Some of these include:
I use of Weibull distributions of head nozzle cracking data from " reference plants",
which have performed inspections, to predict time to cracking in plants which have not inspected.
improved estimates of head nozzle temperatures, obtained from a recent study to define the temperature of each CEOG plant.
improved estimates of nozzle ID surface stresses for all CEDM and ICI nozzles based on 57 finite element analyses as compared to four FEA in the PITM.
crack growth relationship suppoded by data from several laboratory test programs, including the EPRI program, and data from several plants in France.
a correction to account for product form (bar forgings versus pipe) and fabrication practice, using the reference plant approach.
predictions far other nozzles, such as the small diameter head vent nozzles in CEOG plants.
i improved bench marking of the model capability to incorporate plant specific inspection data i
flexibility to easily incorporate additional correction factors (such as for material I
microstructures, water chemistry efTects, additional fabrication effects, etc. if appropriate correlations for these effects are developed) into the predictions.
i
Page 3 Question 2.
Indicate how the model being endorsed relates to the CEOG integrated program and whether the modelis consistent with Topical Report CE NPSD-1085.
Response
The EPRI model is being used by the CEOG as part of the integrated inspection program and it supports the previous selection of CEOG plants for head nozzle inspections.
CE NPSD-1085 indicated that the plants with the highest susceptibility to PWSCC based on PITM predictions, were Millstone-2, San Onofre-3 and Palisades. These plants had the fewest EFPH to predicted nozzle through-wall cracking of all CEOG plants. CE NPSD-1085 indicated that these plants had inspected their head nozzles (Palisades, a partial inspection) or were planning to inspect at the next refueling outage. Millstone-2 did complete an inspection of all CEDM and ICI nozzles in August 1997 and San Onofre-3 has an inspection of all head nozzles planned for the next refueling outage (Spring, 1999).
When the 14 CEOG plants were re-analyzed in 1998 with the EPRI model, the top four plants in susceptibility to PWSCC from the prior analysis remained the top four plants.
Three of these (Millstone-2, San Onofre-3 and Waterford-3) fall into Category 2 (5 to 15 EFPY from 1/1/97 to reach Section XI depths) of the industry histogram and the fourth (Palisades) falls into Category 3 (greater than 15 EFPY to reach Section XI depth). Thus, the results from the new model are consistent with the prior model results as described in CE NPSD-1085 and support the selection of plants for inspection as described in that document. In summary, the EPRI model supports the previously announced CEOG integrated inspection program.
Additional inspections beyond those described in CE NPSD-1085 are not currently scheduled. The CEOG number utilities will monitor reactor vessel inspections at CEOG plants and at other domestic PWRs and use results from these inspections to evaluate the need for future inspections of additional plants or re-inspections of plants that have already inspected. The EPRI model will be a tool in these evaluations.
Question 3.
Describe how the variability in the product form, material specifications, and heat treatment are addressed in the models.
Response
The CEOG response to the question on PWSCC model treatment of variability in product forms, specifications and heat treatments is provided by the Response 1.a to Question 1 of Enclosure 2.
Question 4.
Describe how the crack initiation andgrowth models were benchmarked.
Response
i Page 4 The CEOG response to the question on bench marking of the PWSCC initiation and growth model is provided by the Response 2.a to Question 2 of Enclosure 2.
Question 5.
Provide additional information regarding how each model will be refined to consider inspection data.
Response
l The CEOG response to the question on PWSCC model refinement to include plant specific inspection data is provided by the Response 3.a to Question 3 of Enclosure 2.
Question 6.
Provide the latest model susceptibility rankings, including the basis for comparingplants within each group.
Resnouse i
The CEOG response to the question on the latest model susceptibility rankings for the CEOG plants is included.in the response to Question 4 of Enclosure 2.
l WOG RESPONSES TO NRC REQUESTS FOR ADDITIONAL,INFORMATION l
l
Page1 WOG RESPONSES To NRC REQUESTS FOR ADDITIONAL INFORMA TION I. Responses to Generie RAI Questions This section provides responses to questions in the NRC Generic Letter 97-01 Requests for Additional Infonnation. These include responses to four generic questions posed to all plants and one question regarding the effect of ongoing laboratory tests posed to the WOG plants. This sectica also includes responses to owners group specific questions related to plant ranking, thermocouple nozzles, thiosulphate intrusions, etc. Additional supporting technical information regarding each of the models is presented in Enclosures 6 and 7 of this submittal, depending on the model used.
Q1.
" Describe how the variability in the productforms, material specifications, and heat treatments used tofabricate each CRDM/CEDMpenetration are addressed in the crack initiation and growth models. "
RI.
See Enclosure 2 Response to Question 1.
Q2.
" Describe how the crack initiation and growth modelsfor assessingpostulatedflaws in the no::les were benchmarked, andlist anddiscuss the standards that the models were benchmarked against. "
R2.
See Enclosure 2 Response to Question 2.
Q3.
" Provide additional information regarding how the model will be refined to allow the input ofplant-specific inspection data into the model's analysis methoaology. "
R3.
See Enclosure 2 Response to Question 3.
Q4.
" Provide the latest model susceptibility ranking ofyourplant based on the model analysis results, including the basisfor establishing the ranking ofyourplant relative to the ranking ofotherplants in your owners group analy:ed usingyour model. "
R4.
See Enclosure 2 Response to Question 4.
QS.
" Table 1-2 in WCAP-14901provides a summary ofthe key tasks in the WEC vessel headpenetration no::le assessmentprogram. The table indicates that tasksfor
- 1) evaluation ofmitigation methods, 2) crack growth data and testing. and 3) crack initiation characteri:ation studies are.stillin progress. In light ofthefact that the predictive models appear dependent inpart on crack initiation andgrowth estimates, provide your best estimate ofwhen these tasks will be completed, and describe how these activities relate to and will be used to update the susceptibility assessments at yourplant. "
Page 2 R5.
See Enclosure 2 Response to Question 5.
II.
Responses to RAI Questions Specific to WOG Member Plants Westinghouse Owners Group members each made separate submittals in response to Generic Letter 97-01. Some of the utilities used the Westinghouse methodology, and others chose to use the EPRI methodology. Those who used the Westinghouse methodology either submitted or referred to WCAP 14901 to describe the methodology.
The requests for additional information which have been issued for the Westinghouse plants fell into two categories: those that apply to all plants, and those that apply to Westinghouse plants only. The generic questions have been answered separately; this section contains the answers to the those questions which apply to the Westinghouse plants.
Q1.
In the NEIletters ofJanuary 29,1998 (Ref 1), and April 1,1998 (Ref 2), NEI indicatedthat inspectionplans have been developedfor the VHP no :les at the Farley Unit 2 plant in the year 2002, and the Diablo Canyon Unit 2 plant in the year 2001, respectively. The staffhas noted that althoughyou have endorsedtheprobabilistic susceptibility model describedin IVCAP-14901, Revision 0, other WOG member licensees have endorsed a probabilistic susceptibility model developed by an alternate vendor ofchoice. The IVOG'sproposal to inspect the VHP no::les at the Farley Unit 2 andDiablo Canyon Unit 2 plants appears to be based on a composite assessment ofthe VHP no::les at all IVOG memberplants. Verify that such a composite ranking assessment has been applied to the evaluation of VHP no::les at yourplant(s). Ifcomposite ranking ofthe VHP no::les at IVOG memberplants have been obtainedfrom the composite results ofthe two models,jusufy u hy application of the probabilistic susceptibility model described in IYCAP-14901, Revision 0, would yield the same comparable rankings ofthe VHP no::leforyourplant(s) as would application ofthe alternate probabilistic model used by the TVOG memberplants not subscribing to IVCAP-14901, Revision 0. Comment on the susceptibility rankings of the VHP no::les atyourplant(s) relative to the susceptibility rankings ofthe VHP no::les at the Farley Unit 2 and Diablo Canyon Unit 2 plants.
RI.
The announcement ofinspection plans by individual WOG plants is the result of each individual plant's economic situation, along with their future operational plans. The individual plant results are all compared in the histogram in Enclosure 1. An individual plant's category in the histogram is one of the many considerations which must be evaluated in making inspection decisions.
Q2.
Some IVOG members have chosen to evaluate the vessel headpenetrationsfor their facilities according to aprobabilistic methodology that was developed by another
Pcge 3 vendor ofchoice. WEC and WOG did not provide a description ofthe crack initiation andgrowth susceptibility model usedfor the assessment of WEC vessel headpenetration (VHP) no::les in plants endorsing WCAP-14902, Revision 0.
Provide a description ofthe crack initiation andgrowth susceptibility model usedfor assessment ofthe VHP no :les atyourplant(s).
R2.
The WOG members which chose not to use the Westinghouse model all used the EPRI model, whose methodology is described in Enclosure 6.
I l
1 l
1 4
i 1
i i
i-i DESCRIPTION OF THE EPRI RPV HEAD NOZZLE PWSCC l
PREDICTIVE MODEL i
j (BASED UPON EPRI TR-103198-P8) i l
l i
I e
j
RESPONSES TO GENERIC NRC REQUESTS FOR ADDITIONAL INFORMATION
\\
i l'
Page1 DESCRIPTION OF THE EPRI RPV HEAD N0ZZLE PWSCC PREDICTIVE MODEL i
i i
Table of Contents' l
1.
Introduction 2.
Summary EPRI Model Description t
3.
Methodology and Modeling Assumptions i
.j
.i 1
Page 2 Section 1 Introduction The purpose of this document is to describe the EPRI (Electric Power Research Institute) RPV (Reactor Pressure Vessel) head nozzle PWSCC (Primary Water Stress Corrosion Cracking) i predictive model and to provide information supplemental to the response <, to generic requests for additional information (located in Enclosure 2 elsewhere within this industry submittal).
The model described in this report was originally developed by Dominion Engineering, Inc., for several individual utilities and was known as the CRDM Nozzle PWSCC Inspection and Repair Strategic Evaluation (CIRSE) program. This model was later adopted by the B&W Owners Group and partially formed the basis for the B&WOG head nozzle inspection plan.
In 1997, the CIRSE model was selected by EPRI to be included in the EPRI CHECWORKS suite of software under the name RPV Head Nozzle Module (RHNM). The EPRI RHNM model is an extension of the previous CIRSE model using the same basic input and modeling assumptions. The model was extended to Combustion Engineering Owners Group plants in early 1998. This EPRI model has been used to provide input to the integrated industry histogram for the B&W Owners Group, the Combustion Engineering Owners Group, and 14 individual Westinghouse plants. All histogram assessments reported for the RHNM/CIRSE model were generated using a single, censistent basis. The completed software package, including strategic planning features, will be issued by EPRI to its member utilities in December 1998.
1
Page 3 Section 2 Summary EPRI Model Description l
The EPRI RHNM model assists utilities in managing the RPV Alloy 600 head nozzle PWSCC issue through prediction of crack initiation and growth. The fundamental methodology and modeling assumptions of the model are as follows:
The EPRI model uses a Weibull distribution of RPV head nozzle cracking data for reference plants which have performed inspections in order to predict the time to crack initiation for subject plants which have not performed inspections. The crack initiation predictions include the capability to correct for differences in temperature, stress, materials, and fabrication practices between the subject and reference plants.
i A key factor in this approach is the selection of reference plants for the predictions which are as similar as possible to each subject plant with regard to materials and I
fabrication methods. This eliminates the need to make large corrections for significant differences in materials and fabrication variables.
The EPRI model uses a power-law stress intensity equation to predict crack growth.
This equation is based on results of field inspections and controlled laboratory tests corrected for differences in temperature and crack tip stress intensity.
The EPRI model uses the Monte Carlo analysis method to calculate the integrated risk of cracks for the entire nozzle population on a vessel head taking into account uncertainties in input parameters and varying PWSCC susceptibilities among i
individual nozzles.
j The EPRI model allows utilities to evaluate the risk of cracking and cost for a number of alternative strategic plans to address this issue.
A more detailed description of the EPRI model follows in Section 3 of this document.
Responses relative to the EPRI model to questions in the NRC Generic Letter 97-01 Requests for Additional Information are located in Enclosure 2. These include responses to four generic questions posed to all plants and one question regarding the effect of ongoing laboratory tests posed to the WOG plants.
I i
)
Page 4 F
Section 3 Methodology and Modeling Assumptions The following outlines the methodology and modeling assumptions of the EPRI RHNM model.
3.1 Monte Carlo Method The model uses the Monte Carlo method of analysis. This method was selected since prediction of PWSCC must be treated as a statistical process, and it is desirable to assess the effect of a significant number of variables and attemative strategic plans.
l The time to crack initiation, crack growth rate, and many other variables are defined by distributions of values rather than by singular values in order to reflect uncertainties. The Monte Carlo method allows each input parameter to have distributed values and provides a framework l
which allows calculations required for the risk assessment and economic analyses to be performed in a straightforward manner.
The Monte Carlo method establishes input values for each Monte Carlo trial by random sampling of the distributions for each variable being considered. While most random samples fall near the mean value of a given parameter, small percentages of the samples are extracted near the upper and lower bounds of the distribution.
The model analysis uses 12 distributed variables which are listed in Table 3-1. These variables j
include the time to 10% probability of cracking for the reference conditions, the Weibull slope, j
the stress level in the nozzles, the crack growth rate under standard conditions, etc.
The predictions of cracking for each scenario being evaluated are saved for each Monte Carlo trial so that they can be sorted or counted to produce cumulative output distributions. Sufficient trials are run so that each input distribution is well sampled. When a sufficient number of trials are chosen for the Monte Carlo method, successive runs of the same case, using a different set of random numbers, will yield essentially the same results.
Because the Monte Carlo method is a statistical approach, the output results are also in the form of statistical distributions. The EPRI model reports median, lower bound, and upper bound values from the output distributions. For example, one of the bounds can be set to produce the input data to the industry histogram, i.e., the probability of DC Cook 2 having had a 75%
through-wall crack at the time that inspections were performed.
1 Page5 3.2 Crack Initiation Predictions L
i The EPRI model predicts the time when a nozzle develops a crack and the maximum crack depth within a nozzle as a function of time by combining crack initiation and crack growth models.
This section discusses the model for time to crack initiation, and Section 3.3 describes the model for crack growth.
3.2.1 Statistical Modelfor Crack Initiation The model predicts the time to crack initiation for a subject plant based on statistical results from inspections performed at plants with similar materials, materials processing methods, and fabrication methods. This avoids the need to make major corrections for differences in material processing, microstructure, and fabrication methods.
However, corrections are made for I
differences in operating temperature and nozzle stresses, which can be quantified.
For example, the time to crack initiation in a plant with Huntington Ah%ys extruded nozzle material would be based on the results ofinspections at other similarly conssucted plants with similar Huntington Alloys material. These predictions would not be based on insoection results i
at EdF plants, which have forged nozzles and machined tapers or counterbores on the nozzle inside diameter.
i 3.2.2 Weibull Distribution to Describe Time to Crack Initiation The EPRI model uses a two-parameter Weibull distribution to calculate the probability of crack initiation as a function of time:
F(t)=1-e (eg. 3-1) where:
F the probability of a particular nozzle cracking by time I
=
degradation time (operating time corrected for temperature variations) t
=
" characteristic time" to 63.2% probability of cracking 6
=
Weibull" slope" which represents scatter b
=
The two-parameter Weibull distribution is an industry standard approach for this type of phenomenon and is frequently used to predict PWSCC in steam generator tubing [3.1]. Rather than using 0, the model uses a corresponding quantity, the time to 10% probability of a nozzle cracking t. A typical Weibull distribution is shown in Figure 3-1.
3.2.3 Relative Susceptibility Factor Because the time to crack initiation is a function of material susceptibility (microstructure),
amount of cold work during machining, surface stress, operating temperature, water chemistry environment, etc., the time to 10% probability of nozzle cracking for any reference plant must be l
Page 6 corrected to account for differences in these variables between the subject plant and the reference plant. This is accomplished using a Relative Susceptibility Factor (RSF) which relates the PWSCC susceptibility of a subject nozzle to that of a reference no=le (see Section 3.6 for a discussion of how the time to 10% probability of crack initiation is determined for reference no=les):
Im = RSF (eg. 3-2) v
- O' <
'h" RSF =
exp (eq. 3-3) s [ch.rer j ([ tab.r.r j ([ mat.rer j s sur.r.r,
, R J T,.rj_
S where:
time to 10% probability of crack initiation (used to calculate probability F in to
=
eg. 3-1)
RSF relative susceptibility factor for scaling im
=
f;,,
water chemistry factor (constant for all nozzles in a unit)
=
fa nozzle fabrication factor (to account for undesirable surface conditions
=
caused during fabrication) material factor (constant for all nozzles of a given heat) f,
=
s, maximum inside surface hoop stress
=
stress exponent x
=
activation energy for crack initiation (kcal/ mole)
Gj
=
4 R
gas constant = 1.103x10 kcal/ mole R
=
absolute nozzle operating temperature ( R)
T
=
Note: the "ref" subscripts denote the reference plant values 3.2.4 Correctionsfor Differences in Material PWSCC Susceptibility Differences in material susceptibility are addressed by selecting a reference Weibull distribution from an inspected plant which has nozzles produced by the same supplier at about the same point in time and using the same material processing and fabrication parameters as the subject plant.
This avoids the need to make large adjustments for differences in materials (e.g., chemical composition, microstructure, etc.) and fabrication practices. Accordingly, the material and fabrication factors for the subject and reference nozzles are currently taken as 1.00. If future plant operating experience and laboratory testing confirm the need for different material and fabrication factors, then these factors can be directly input into the EPRI model.
3.2.S Correctionsfor Differences in Operating Temperature It is known that temperature is a key variable which affects time to crack initiation and crack growth. For example, many steam generators have significant amounts of cracking at expansion Page 7 transitions on the hot leg side but essentially no cracking ofidentical transitions on the cold leg side. Similarly, several plants have experienced PWSCC of Alloy 600 instrument nozzles in hot leg piping without having any reported PWSCC in essentially identical nozzles in cold leg piping.
The difference in operating temperature between the subject and reference plants is taken into account by an activation energy model. Experimental and field data suggest that the thermal activation energy for crack initiation is in the range of 44-53 kcal/ mole [3.2]. A recent summary paper by Staehle [3.3] suggests 50 kcal/ mole. A nominal value of 50 kcal/ mole has been used for the EPRI model calculations.
In many cases, plants have operated at essentially a constant hot leg temperature over the plant life such that the effective PWSCC degradation time can be taken as the actual plant running time in effective full power years (EFPYs). Ilowever, if a unit has operated for extended periods at different hot leg temperatures, then the effective full power years must be corrected for these temperature changes. For these ca::es, the model assumes the current operating temperature, and the EFPYs are adjusted as necessary to reflect the temperature changes.
3.2.6 Correctionsfor Differences in Nozzle Stresses It is known that inside surface stress (residual plus operating) is a key variable in establishing time to crack initiation. This has been demonstrated by researchers [3.4,3.5,3.6, and 3.7] and confirmed by several field observations:
cracking tends to occur first in outer row nozzles which typically have the highest calculated stresses, cracking has been axial and located on the uphill and downhill sides of the nozzles as predicted by stress analyses, and cracking has tended to occur most rapidly in nozzles with large welds.
i EPRI has performed elastic-plastic finite element stress analyses of a wide range of CRDM, CEDM, and ICI (in-core instrumentation) nozzles using identical modeling assumptions with parametric-type material yield strength and geometry inputs [3.8]. The analysis of each nozzle geometry has been perfonned for several different material yield strengths.
The EPRI RHNM model corrects for differences in stresses between the subject and reference nozzles in the following manner.
For cases where there are a limited number of nozzles of each basic configuration (e.g., ICI nozzles in Combustior Engineering plants), the EPRI model uses the finite element analysis results for the particular nozzle design and weld geometry in the analysis. The software interpolates results for intermediate yield strengths not explicitly analyzed.
i Page 8 i
For cases where there are many similar but not identical nozzles designs (e.g., CRDM nozzles in Westinghouse plants), the EPRI model uses algorithms based on the results of the finite element analysis work on representative nozzles which predict the peak j
nozzle inside surface hoop stresses as a function of key variables such as nozzle yield i
strength, incidence angle in the vessel head, weld size, weld distribution (geometry),
l etc.
Differences in predicted stress level between a subject nozzle and the reference nor:le are accounted for by taking the ratio of stresses between the subject and reference nozzles to a power with the mean value of four. This exponent is based on the work reported by van Rooyen, 1
Yonezawa, and Seman [3.5,3.6, and 3.7].
3.2.7 Correctionsfor Differences in Fabrication Methods The EPRI model is based on using reference plants for PWSCC predictions which are as similar i
to the subject plant as possible. This includes having been fabricated by the same vendor at approximately the same point in time. Accordingly, the fabrication factors for the subject and l
reference nozzles are currently taken as 1.00. As mentioned previously, if future plant operating experience and laboratory testing confirm the need for different material and fabrication factors, then these factors can be directly input into the model.
3.2.8 Correctionsfor Differences in Plant Water Chemistry At present, the industry consensus is that there is no significant effect of primary water chemistry on Alloy 600 PWSCC within normal industry allowable water chemistry limits. Therefore, i
normal variations in water chemistry between the subject plant and the reference plant (s) do not affect PWSCC susceptibility significantly, and the water chemistry factors for the subject and reference nozzles are taken as 1.00.
t' 3.2.9 FinalCrack initiation Model i
The following equation is used to establish the Weibull distribution of time to 10% probability of j
crack initiation in a subject nozzle. The key factor I,m,,, represents the reference no::le for plants l
with the same basic material and fabrication parameters.
I'"*'
,1 (eg. 3-4)
'tm=7 y
G, 1
s, R
T T,,c
<s,,,,,,,
g
i Enclosurc 6 Page 9 3.3 Crack Growth Predictmns The EPRI model predicts crack growth using the power-law fracture mechanics model which has become the accepted industry standard for RPV head nozzle PWSCC [3.9].
]
3.3.1 Power-Law Fracture Mechanics Crack Growth Model The power-law fracture mechanics crack growth model is given by the following expression:
d = A(K-K )"
(eq. 3-5) where-i rate ofcrack depth increase (m/s) d
=
A growth rate constant (amplitude) (m, s, MPa)
=
nozzle maximum stress intensity (MPa 6)
K
=
1.ls.a 6
=
where:
crack depth a
=
s,a mid-wall hoop stress
=
l Kh stress intensity threshold
=
i power-law exponent n
=
The effect of temperature on crack growth is considered by scaling the growth rate constant A using an activation energy model:
. G, ' 1 1 '.
[
A = A,.r exP (eg. 3-6)
T,,,7 R
T t
where:
~
reference value ofA selected based on field and laboratory experience A,,
=
activation energy for growth (kcal/ mole)
Gg gas constant = 1.103x10-2 kcal/ mole R R
=
absolute nozzle operating temperature ( R)
T
=
i g
reference temperature used to normalize field and laboratory experience T,,,
=
3.3.2 Reference Temperaturefor Plotting Crack Growth Rates Since the crack growth rate is a function of the material temperature, laboratory and field crack growth rate data must be plotted at a comrnon reference temperature for comparison purposes.
For purposes of the EPRI model, the reference temperature has been taken as 325 C (617 F).
This temperature is widely used for crack growth rate tests.
Page 10 The most recent activation energy for crack growth in Alloy 600 materials in primary water environments reported by EPRI is 32.4 kcal/ mole [3.10]. The nominal activation energy value used for the EPRI model is O = 33 kcal/ mole.
3.3.3 Modsfied Peter Scott Growth Equation Early in the evaluation of cracking in Alloy 600 nozzles, Peter Scott of Framatome proposed an equation for crack growth based on laboratory tests conducted on steam generator tubing specimens [3.11]. The crack growth rates measured for the steam generator tubing specimens were later corrected to account for cold working of the specimens in preparation for the tests.
The Scott model has been used by several organizations performing crack growth analyses. The most recent modified form of Scott's equation at 325 C [3.10] is:
2.23x10a2 (g_9)u' m/s (where Kis in MPaM)
(eg. 3-7) 4
=
A more recent paper by Amzallag and Scott [3.12] reports the crack growth model at 323 C to be:
5.6x 10a2 (g_9)u' m/s (where Kis in MPa M)
(eq. 3-8) d
=
This is about twice the amplitude reported in EPRI TR-109136 [3.10]. Both of these curves are plotted in Figure 3-2.
3.3.4 ThresholdStressIntensity The Scott curve is computed using a threshold stress intensity of 9 MPa6. However, it is known that cracks which initiate and grow through the cold worked region must eventually propagate into the base material. Assuming a 0.004 inch (0.1 mm) cold worked surface layer and a 30 ksi (207 MPa) stress in the base metal below the cold worked region, the threshold stress intensity for the crack to propagate would have to be as low as K,3=1.lcr6
=1.l(207 MPa)fr(0.0001 m)
(eq. 3-9)
= 4.0 MPaM The EPRI model uses an assumed stress threshold of 4 MPa6. As shown in Figure 3-2, this has the effect of producing a more conservative crack growth rate for lower stress intensities relative to the modified Scott curve.
Page1I 3.3.5 Laboratory Crack Growth Rate Measurements Tests have been performed by several laboratories to determine crack growth rates in Alloy 600 materials as a function of crack tip stress intensity. The results of these tests, corrected to 325 C, are plotted in Figure 3-2. The tests included are the following:
Westinghouse laboratory tests [3.10]
EdF laboratory tests [3.13]
CIEMAT laboratory tests [3.14]
3.3.6 Field Crack Growth Rate Measurements Results of field inspections have also been used to determine crack growth rates. This has involved 1) measurements of crack growth based on repeat inspections and 2) estimation of crack growth based on a predicted time to crack initiation and the growth rate required to achieve the final measured crack depth.
EdF has reported on repeat inspections of a number of cracked nozzles [3.12]. The temperature corrected results of the repeat measurements on hot head plants are plotted in Figure 3-2. These data show four points at or below and two points somewhat above the modified Scott curve (eq.
3-7). There were also several cracks in hot head plants which did not grow from one cycle to the next. These points were conservatively not plotted in the figure.
During refueling outage 9, nozzle no. 75 in the DC Cook 2 plant was found to have three flaws.
The deepest flaw was reported to have a depth of 6.8 mm (0.268 inches) [3.15]. During refueling outage 10, the deepest flaw in nozzle no. 75 was inspected and reported to have grown to 7.3 mm (0.287 inches) depth. The crack growth rate for the deep crack in the DC Cook nozzle was estimated two ways.
First, the measured crack growth between the two inspections was plotted relative to the crack tip stress intensity computed using the same model as for other EPRI model calculations. As shown in Figure 3-2, the resultant crack growth is well below the other data. While the depth was reported to have increased by 0.5 mm, it should be noted that this increase is within the typical inspection accuracy and might actually have been a little higher or lower.
Second, it was assumed that the crack initiated at 6.0 EFPYs and grew to the measured 6.8 mm at outage 9. The crack growth curve, corrected for temperature differences, which would produce this assumed growth is also plotted in Figure 3-2.
This growth is much higher than the measured growth rate between outages 9 and 10 suggesting that the rate of growth has decreased rather than increased with crack depth.
.. _ _.. _ - _. _ _ _ _ _.. ~
Page 12 Cracks were discovered in several Ringhals 2 nozzles during an inspection in 1992. Cracks in i
nozzles located above the bottom of the weld were removed while cracks below the bottom of the weld were left in place. The cracks left in place were inspected in 1993 and again in 1994.
Vattenfall has reported that there was no significant increase in the size of the flaws left in place
[3.16].
3.3.7 High Reported Crack Growth Rates in EdF Cold Head Piants The major reported exception to the activation energy model discussed in the preceding paragraphs is the rapid crack growth rate measured in EdF cold head plants. Amzallag, et al.,
have reported that there is essentially no difference in the measured crack growth rate between nozzles in cold head plants at 290 C (554 F) and hot head plants at 315 C (599 F) [3.12]. The activation energy model would suggest a factor of 3.5 times lower crack growth rate at the cold head temperature using an activation energy G = 33 kcal/ mole.
The cause of the significantly different crack growth behavior reported for materials in the EdF cold head plants is not known at present. Possible sources of the difference could include materials processing method (forging at low material temperatures without subsequent annealing) or uncertainty of the actual nozzle operating temperature in plants with small bypass holes through the internals flange.
Because the reported behavior of these materials is inconsistent with other laboratory and field experience, including EdF's own crack growth testing, and because the EPRI model does not address EdF plants, the crack growth measurements for the EdF cold head plants are not plotted in Figure 3-2 or included in the subsequent evaluation.
3.3.8 Curve Fitting to Laboratory and Field Data The scatter in the data in Figure 3-2 clearly suggests that crack growth should be considered to be a statistical process with a distribution of reference values. Figure 3-3 shows the empirical distribution of power-law constants A (thin line) for the 36 data points in Figure 3-2. (The DC Cook 2 reinspection point is an outlier and is conservatively excluded.)
A log-triangular distribution (see paragraph 3.5.3 below) with lower bound a, upper bound b, and mode (nominal) value c A ~ exp{triang[In(a),In(b),In(c))
(eg. 3-10) was fitted to the empirical distribution such that the two distributions had equivalent bounds and intersected at a cumulative fraction of 0.500 (median). The log-triangular distribution shown in Figure 3-3 fits the data well and is implemented in the model. A plain triangular distribution was found to fit the data poorly.
Figure 3-2 shows the median, upper bound, and lower bound crack growth rate curves based on the set of reference data. Note that the upper bound growth curve is conservative with respect to
Page 13 the EPRl-reported modified Scott curve, eq. 3-7, and essentially identical to the " upper bound" curve reported in Scott's most recent paper, eq. 3-8. The figure also shows crack growth rates of I and 3 mm/ year for reference purposes.
3.4 Effect of Previous Inspections on Predictions The continuity of the Weibull curve and the Monte Carlo analysis always results in a small but finite probability of early cracking which will be inconsistent with the absence of observed leakage during the standard Generic Letter 88-05 required leak detection walkdowns or with the absence of observed cracking during past NDE inspections of nozzle inside surfaces. Therefore, after the crack initiation and crack growth predictions are integrated on a statistical basis for each nozzle, the predictions are modified (" truncated") as necessary to be consistent with the results of past leak detection walkdowns and NDE inspections ofindividual nozzles such that there are no predicted leaks at the most recent outage where visual inspections of the head did not show a leak, and there are no predicted cracks at the most recent outage where nondestructive examinations of the nozzle inside surfaces showed that no cracks existed.
" Truncation" lowers the predicted probability of cracking and leakage for the few cycles following the " truncation" inspection because the predictions must have continuity with respect to time. Because through-wall cracking below the bottom of the J-groove weld does not result in leakage, the statistical " truncation" for leak detection walkdowns only affects above-the-weld cracking. On the other hand, NDE inspections may verify the absence of cracking above or below the J-groove weld. The EPRI model provides inspection result inputs on a nozzle-by-nozzle basis for the two regions.
Truncation has a very small impact on predictions of time to cracking at the probability level used for the industry histogram.
3.5 Distributed Parameters The EPRI model treats many of the variables as statistically distributed parameters to capture the effect of uncertainties. For example, the reference time to 10% probability of cracking is allowed to vary within a defined range because the reference time to cracking data available today are j
sparse. Similarly, there is some amount of uncertainty in the inside surface hoop stress.
Table 3-1 gives a list of the model inputs which are treated as statistically distributed parameters.
i The sampling frequency heading in Table 3-1 indicates how often the distributed parameters are sampled. The appropriate frequency is generally once at the start of each Monte Carlo trial.
However, crack growth curves vary from nozzle to nozzle, so sampling for A is performed for each nozzle (above and below the weld) and trial.
Page 14 3.5.1 WeibullDistributionsfor Crack Initiation Above and Below the Weld Field experience has shown that cracking tends to initiate either above the J-groove weld or below the J-groove weld. Cracks located below the bottom of the J-groove weld are not as significant as cracks located above the J-groove weld since they cannot result in leaks or loss of structural margin even if they grow through-wall. Accordingly, the EPRI model statistically samples two initiation times for each nozzle, one for the region above the bottom of the J-groove weld and one for the region below the bottom of the J-groove weld. The Weibull sampling and relative susceptibility factor calculations are performed independently for each region in each nozzle. The EPRI model input to the industry histogram is based on the predictions for the region above the bottom of the J-groove weld.
3.5.2 Triangular Distributionfor Most Other Distributed Parameters Table 3-1 shows that most of the distributed inputs are assumed to have triangular distributions.
The triangular distribution is the preferred distribution when the available data do not justify a more elaborate choice. A variable with a triangular distribution can take on any value between a lower bound (a) and an upper bound (b). A third parameter (c) defines the mode of the distribution (value for which the density function is maximum). The mode can take on any value between a and b. The three parameters are chosen based on available industry and field data, controlled laboratory tests, and engineering judgment. A variable x has a triangular statistical distribution ifits cumulative distribution function F(x) has the form
'O if x < a (x-a)2
,nfasxsc F(x) =, (b-aXc-a (eq. 3-11)
(b-x) i 1-(b-a)(b-c) ifc<xsb 4
1 1
if b < x and is written x ~ triang(a,b,c).
3.5.3 Log-Triangular Distributionfor Crack Growth Rate The only exceptions to the Weibull and triangular distributions are the crack growth rate constant A,r and exponent n.
Because A,,, may vary over more than an order of magnitude, a log-triangular distribut. ion was chosen.
The log-triangular distribution is calculated from the triangular distribution as x ~ exp{triang[In(a),In(b),In(c)).
Currently, not enough data exist to support a distribution for n, and Scott's recommended value of 1.16 is taken as the unique value.
Page 15 3.6 Predictive Model Calibration A key aspect of the EPRI model development was establishing a method to calibrate the analytical model with the results ofinspections performed on actual reactor vessel heads. The calibration procedure consists of the following main steps:
1.
A reference no::le was established for each unique combination of material supplier and fabricator. The reference nor:le is assumed to have a 60.0 ksi inside surface hoop stress and to operate at 600 F.
i 2.
Trial and error analyses were perfomied for each plant which has already performed inspections to determine the time to 10% probability of crack initiation for the reference no :le which would produce a 50% cumulative probability (best estimate) of the observed inspection results (number of cracked nozzles and crack depth) in the reference plant at the time the inspection was performed. These calculations are based on the Weibull curve for the reference nor:le with corrections for differences in inside surface stress level and operating temperature between the reference nor:le and the reference plant being evaluated.
3.
If the inspection did not show any cracks, then it was conservatively assumed that a crack initiated in a single nozzle immediately after the inspection.
l 4.
If the reference inspected plant has nozzle materials of different types, such as i
extruded SB-167 ICI nozzles and rolled SB-166 CEDM nozzles, then each material group is evaluated separately.
For each subject plant to be evaluated, the EPRI model selects the reference no::le for the previously inspected plant which is most similar in terms of material supplier and fabricator. If several similar plants have been inspected, the model uses the available information to establish upper bound, lower bound, and nominal reference values. If no plant with similar nozzle materials has been inspected, then an estimated time to 10% probability of cracking is established based on knowledge of the material being considered relative to materials at plants which have inspected and engineering judgment.
3.7 Risk Assessment An important objective of modeling PWSCC in RPV head nozzles is to determine the risk of various depths of cracking at future refueling outages. The probability of cracking versus time is calculated statistically by the EPRI model as the ratio of the number of Monte Carlo trials with at least one cracked nozzle of a given depth at a particular refueling outage to the total number of trials run.
As previously noted, the EPRI model performs crack initiation and growth calculations for locations above the bottorn of the J-groove weld, which could lead to a leak, and below the bottom of the J-groove weld, which would not lead to a leak. The model risk assessment includes results for both regions.
Page 16 3.8 Note on Model Quality Assurance The models and software for the probabilistic analysis of reactor vessel head penetration nozzles were developed, verified, validated, and controlled in accordance with the Production Grade standards of the EPRI " Software Development Standards," EPRI TR-105061. Analysis work to compute operating stresses in nozzles, including the effects of welding residual stresses, were performed using the same computer software code and program listing, which Dominion Engineering, Inc., uses for nuclear safety related work perfornied to requirements of 10CFR50, Appendix B. The DEI 10CFR50, Appendix B, QA program is accepted by nuclear utilities through audits by NUPIC.
3.9 References 3.1 Electric Power Research Institute. Statistical Analysis ofSteam Generator Tube Degradation, EPRI NP-7493, September 1991.
3.2 PlVSCC Predictive Modelsfor Alloy 600 Vessel Head Penetrations, Minutes of April 27 and 28,1995, meeting at EPRI, Palo Alto, California.
3.3 Staehle, R. " Initiation and Propagation of LPSCC," Proceedings: 1997 EPRI IVorkshop on PlVSCC ofAlloy 600 in PTVRs, EPRI TR-109138-P2, November 1997.
3.4 Berge, P., Buisine, D., and Gelpi, A. "PWSCC Effect ofInitial Surface Preparation,"
Proceedings: 1997 EPRI IVorkshop on PlVSCC ofAlloy 600 in PlVRs, EPRI TR-109138-5 P2, November 1997.
7'i '
3.5 Bandy, R. and van Rooyen, D. " Quantitative Examination of Stress Corrosion Crackit r Alloy 600 in High Temperature Water - Work During 1983," Proceedings: 1983 Y
IVorkshop on Primary-Side Stress Corrosion Cracking ofPIVR Steam Generator Tv W;;
EPRI NP-5498, November 1987.
3.6 Yonezawa, T., et al. "Effect of Cold Working on the Stress Corrosion Cracking Re y
ma af Nickel-Chromium-Iron Alloys," Materialsfor Nuclear Reactor Core Applicatiu-a
- 2. BNES, London, October 1987.
((
k' 3.7 Seman, D. J., et al. " Primary Water Stress Corrosion Cracking of Alloy 600 - Effec:
Processing Parameters," Proceedings: 1991 EPRI IVorkshop on PlVSCC ofAlloy 600 in PTVRs, EPRI TR-100852, July 1992.
3.8 Hunt, E. S., Gross, D. J., and White, G. A. " Stress Predictive Algorithms for CRDM Nozzles," Proceedings: 1997 EPRI IVorkshop on PTVSCC ofAlloy 600 in PlVRs, EPRI TR-109138-P2, November 1997.
Page 17 l
3.9 Electric Power Research Institute. PWSCC ofAlloy 600 Afaterials in PlVR Primary System Penetrations, EPRI TR-103696, July 1994.
3.10 Electric Power Research Institute. Crack Grmvth and Aficrostructural Characterization of Alloy 600 PlVR Vessel Head Penetration Afaterials, EPRI TR-109136, December 1997.
3.11 Scott, P. M. " An Analysis of Primary Water Stress Corrosion Cracking in PWR Steam Generators," Proceedings ofNEA/CSNI UNIPEDE Specialist Afeeting on Operating Experience with Steam Generators, Paper 5.6. Brussels, Belgium, September 16-20,1991.
3.12 Amzallag, C., et al. " Analysis of the Cracking Behavior of Alloy 600 RVH Penetrations:
Part II-Derivation of a Crack Growth Model and Analysis of Crack Growth Rate Field Data," PVP-Volume 324, Fatigue and Fracture, Volume 2, ASME,1996.
3.13 Le Hong, S. and Vaillant, F. " Crack Growth Rate Measurements of Vessel Head Penetration Alloy 600 in PWR Environment," Proceedings: 1997 EPRI IVorkshop on PJVSCC ofAlloy 600 in PWRs, EPRI TR-109138-P2, November 1997.
3.14 Briceno, D. G. and Lapena, J. " Crack Growth Rates in Vessel Head Penetration Materials," Proceedings: 1994 EPRI Workshop on PWSCC ofAlloy 600 in PlVRs, EPRI TR-105406, Part 2, August 1995.
3.15 Mickatavage, B. C. " Donald C. Cook Nuclear Plant Unit 2: RVH Penetration Inspection,"
Presented at EPRI/NEI Ad-hoc Advisory Committee Meeting, Tampa, Florida, November 18,1994.
3.16 Sjustrand, H. "Ringhals Unit #2, #3 and #4 RVHP Examination Results from NDE and Boat Sample Analysis," Proceedings: 1994 EPRI Workshop on PWSCC ofAlloy 600 in PWRs, EPRI TR-105406, Part 1, August 1995.
i
l Page 18 4
Table 3-1. EPRI Model Distributed Inputs Assumed bampling Distributed Inputs Units Distribution Frequency Nozzle Inputs Nozzle Temperature F
Triangular Trial Nozzle Stress Tolerance
% from NOM Triangular Trial Crack Initiation Reference Reference Time to 10%
EFPY Triangular Trial Weibull Slope Triangular Trial Activation Energy (Initiation) kcal/ mole Triangular Trial Stress Exponent Triangular Trial 4
Crack Initiation Susceptibility Factors Water Chemistry Factor Triangular Trial Material Factors Tolerance
% from NOM Triangular Trial Fabrication Factors Tolerance
% from NOM Triangular Trial Crack Growth Reference 2
Reference Growth Curve Constant m, s, MPa Log-Triangular Nozzle 2
Reference Growth Curve Exponent Dependent Nozzle Activation Energy (Growth) kcal/ mole Triangular Trial I
NOTES:
'The reference growth curve exponent values are assigned based on the sampled growth curve constant value.
2Sampled independently for the above and below the bottom of the weld regions for each Monte Carlo trial.
4 Page 19 i
s.
Weibull Plot
.90 5
.. 0.63 0.5
../.
/
l...
.. 0.2 1
(
.10 6
Fraction y
. Parameter Fit...
. 0.05
\\
of
}f
,_,, 3,4,00,,o,0,0,,,,,
Tubes "01
., /.....
. 0.00s t
LTMA tutWng
- 0.002
.001 Thot.609'F (320.5'c)
...... Plant B....
. 0.0005
~
0.0002
.0001.,
l 1
10 100 Service Time (EFPY) i i
I 1
b.
Same Data and Fitted Line with Linear Scales O.9 --
1 O.8 --
1 0.7 --
/
0.6 --
Fraction 0.5 --
/
Failed
/
Tubes 0.4 --
/N Fitted Weibull Distribution 0.3 --
Data Point
./
0.2 --
g
+.
- 0.1 --
y
.e
... +. e4 0
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 Service Time (EFPY)
Figure 31. Typical Weibull Distribution
Page 20 1.E-09
~AlfDaia'Correctellii32Pd Using G'i33 liial!niile I
~
l l
l l
4 '~l f-i[310]
l i
\\
4-i /O
[3.12] j \\
_.$ r. -l l *
.......j
- ....*T....-
3
}
- 4 m
1.E-10.,,
g; -
- 1 '. / ~}, s n'
,a l
r w
_j 4
3 _,-
,>i:
!i :
- . sA = x 3
/
a l ^.. -V
- 1 f
i i
b,,i.,. T _
,it>'lj,l O~,< %'
X71
!)
l l
l t
l f,-
,s' l'
l 8
IL y<a:- p<;
}
l e
g 1.E.11.
u in ___ /
E
! '-fe/
i 1
U 4f i
i
~N i
i i
i i
_g i
i i
i i
i I
I
_j/.;
l l
l l
.n v
\\
I 1.E-12. m f
- 0 10 20 30 40 50 60 70 80 Stress Intensity K (MPaVm)
-~~-I mm/yr ~
3 mm/yr
...... Scott Curve [Ref. 3.10]
...... Scott Curve [Ref. 3.12] -. -. RilNM. Upper RIINM. Median
-.. RilNM Lower A
West. Lab (GBC>34%)
o West. Lab (ileat 69) l 0
EdF-Lab X CIEMAT. Lab
+
DC Cook 2 (Nozzle #75)
[
. DC Cook 2 (Nozzle #75) g EdF - Field Figure 3 2. Summary of Field and Laboratory Crack Growth Rate Data
Page 21 l
i i
l 1.0 All Data Corrected to 325'C Using Q = 33 kcal/ mole Power-Law Exponent = 1.16 3 0.8 i
O 4
a 0.7
.2 i
0.6
. Log-Triangular _
Distribution J
Fit to Data g 0.5 r.
.c
.'3 0.4 C
C U
'g 0.3 Median Rankin I
of 36 Field and z 0.2 Lab Values of U
0.1 0.0 Crack Growth Power-Law Constant A (log scale) i i
l i
Figure 3-3.
Statistical Distribution of Crack Growth Rate Power-Law Constant l
l
i 1
l METHODOLOGY FOR EVALUATION OF REACTOR VESSEL CLOSURE HEAD PENETRATION INTEGRITY FOR THE WESTINGHOUSE OWNERS GROUP (BASED UPON WCAP-14901) i
l Page1 MET 110DOLOGY FOR EVALUA TION OFREACTOR VESSEL CLOSURE ffEAD PENETRA TION INTEGRITYFOR Tile WESTINGil0USE OWNERS GROUP 1.0 Westinghouse Crack Initiation Model Development and Crack Initiation Testing 1.1 Crack Initiation Model Westinghouse advanced an Alloy 600 PWSCC initiation model for primary components in Pressurized Water Reactors [10]. Briefly, the model incorporates three contributing factors for the prediction of crack initiation time; namely, material condition, stress, and temperature. These i
are discussed below.
Material Condition and Microstructure I
As reported by several authors [17,18,19,20, and 21], the Alloy 600 microstructure is a function of the thermomechanical history of the material heat as well as its carbon content.
Alloy 600 material heats subjected to mill annealing at low temperatures exhibit a fine grained microstructure with heavy transgranular carbide precipitation and little or no carbides precipitate on the grain boundaries. Such a microstructure is reported to be more susceptible to PWSCC.
On the other hand, a high temperature mill-anneal (>1000 C) tends to put more carbon into solution, increases grain size, produces grain boundary chromium carbide precipitation and renders the material more resistant to PWSCC. Norring, et al. [22], did not find a correlation J
between the total content of carbon and the crack initiation time, but they observed good correlation between the amount of grain boundary carbides and crack initiation time. The fact that grain boundary precipitation is beneficial to PWSCC has been reported by many researchers
[23]. Norring, et al., [22], showed that the crack initiation time varied directly (linearly) vith grain boundary carbides. Their data suggested that when the grain boundary carbide coverage is increased by a factor of 3, the crack initiation time also increased by a similar factor (from 4,000 hours0 days <br />0 hours <br />0 weeks <br />0 months <br /> to 12,000 hours0 days <br />0 hours <br />0 weeks <br />0 months <br />). Bandy and Van Rooyen [24], pointed out that in addition to grain boundary carbide coverage, other features relating to processing history variables such as carbon concentration gradients, substructural features, grain size distribution, cold work, intragranular carbide distribution and the grain boundary segregates all play an important role in the cracking behavior of the Alloy 600 material.
When considering the influence of microstructure on the PWSCC susceptibility for the purpose of the current evaluation, to enable comparison of heats fabricated at different vendor shops, the thermomechanical processing history effect is separated from the grain boundary carbide coverage effects. In general, the influence of the grain boundary carbides is known and the coverage (G) can be easily measured directly from the microstructure. The influence of other structural features due to processing history cannot be assessed directly. These processing effects are represented in the current treatment by a single parameter (A) characteristic of the fabrication shop (vendor). This approach provides a means of comparing the PWSCC susceptibilities of Alloy 600 material heats from different vendor shops although they may contain similar grain boundary carbide contents.
Page 2 Influence ofStress Steady state tensile stress in the component, either due to residual and/or applied loads, has a strong influence on the PWSCC.
Bandy and Van Rooyen [24], reported that the time to failure varied inversely as the fourth power of applied stress in both annealed and cold-worked specimens. They also reported data to support that cold work reduces the resistance to PWSCC. The effective stress at a given Alloy 600 location is a function of the fabrication steps and their sequence, the yield stress of the material, and the service stress. In general, the local residual stresses resulting from fabrication can play a more significant role than the service stresses themselves.
Temperature Effects Several investigators [17,24], examined the role of temperature on PWSCC. It is well established from these results that the crack initiation time decreases exponentially with temperature and that they are related through an Arrhenius equation expressed as a function of the activation energy of the process. The experimental results confirm that Alloy 600 PWSCC is a thermally activated process and the activation energy for the process varies approximately between 50 to 55 kcal per mole. An activation energy value of 55 kcal/ mole is consistently applied throughout the current assessments, for crack initiation. A different value,32.4, applies for crack growth as was discussed in Section 2 of WCAP-14901.
1.2 The Westinghouse Crack Initiation Model Consistent with tche contributing factors discussed above, the crack initiation time (tj) or the rate of crack initiation (1/t ) is proportional:
i 1/I, x (Stress )"
~QIRT xinverse of the grain boundary carbide coverage factor, (1/G) n -QlRT so that 1/t, x G
Since the nature of the vendor thermomechanical processing is also a significant contributing factor, one can say that for a given fabrication process n -QIRT gg 1/t, = A (1-1)
G i
The proportionality constant "A" can be chosen to represent the processing conditions representative of a given manufacturing process or manufacturer, and could include parameters such as yield strength as part of the expression.
Page 3 "A" can be assessed for a given heat by substituting the parameters of a service component with a known cracking history for the heat of material. "A" will then represent the processing condition (or the ver. dor) by the definition we have just established.
The parameters in the above rate equation (1-1) are described below:
A is a constant, relating to the processing, and fabrication conditions of the material G is the grain boundary carbide coverage factor F
is the effective tensile stress (resulting from applied and residual stresses) is the stress exponent having a value ranging from 3.5 to 4.5 for Alloy 600 in primary water n
G is the activation energy for the crack initiation process and has an approximate value of 55 kcal/ mole R
is the gas constant (1.987 cal / mole degrees K)
T is the absolute temperature in degrees K, and tf is the time to initiate cracking.
I
Page 4 2.0 Technical Description of Probabilistic Models To calculate the probability of failure of the Alloy 600 vessel head penetration as a function of operating time I, Pr(t 5 If, structural reliability models were used with Monte-Carlo simulation methods. This section describes these structural reliability models and their basis for the primary failure mode of crack initiation and growth due to primary water stress corrosion cracking (PWSCC). The models used for the evaluation of head penetration nozzles are based upon the probabilistic and economic decision tools developed previously for the Westinghouse Owners Group (WOG). The capabilities of this software have already been verified in the following ways:
1.
Calculated stresses compare well with measured stresses, 2.
Crack growth rates agree with measured field data.
Recent improvements have also been made to the model in order to maximize its use for individual plant predictions. The changes include the following:
1 1.
The model accepts measured microstructure (replication) and also has the capability to ignore its effects, if desired.
2.
The relationship ofinitiation time to material microstructural effects and yield strength has been improved to more closely match the observations from the recent inspection at North Anna Unit 1.
3.
Statistically based Bayesian updating of probabilities due to initial inspection results has j
been added (e.g. the lack of any indications at any given plant).
4.
The uncertainty en crack growth rate after initiation has been updated to reflect the findings observed in the recent Westinghouse test data and the recent in-reactor measurement data to be published by EdF [16] (see Figure 4-2 of WCAP-14901).
5.
All models have been independently reviewed by APTECH Engineering (Begley and Woodman)[25], and an improved model was developed for the effect of monotonic yield strength on time to initiation.
6.
A wide range (both high and low values) of calculated probabilities are consistent with actual plant observations as discussed below.
The most important parameter for estimating the failure probability is the time to failure, if n i
hours. It is dermed as follows:
f = t, +(a -ad/(da/dt)
(2-1)
I f
where:
Page5 ff time to initiation in hours,
=
af failure crack depth in inches,
=
crack depth at initiation in inches and a
=
o da/dt =
crack growth rate in inch / hour.
In equation (2-1), both the crack depths at failure and initiation may be specified as a fraction of the penetration wall thickness, (w). The failure depth af epends upon the failure mode being d
calculated. Since the failure mode ofconcern is axial cracks in the penetration that are deeper than the structural limit of 75% of the penetration wall thickness (w), it would be specified as:
j = 0.75w (2-2) a The time to PWSCC crack initiation, t in hours, is consistent with the previous equation (1-1) by i
Rao [3] and is defined by:
C + (1 + C,P,yr )exp'RT G, '
i t'. =
(2-3) a "' S,">
where:
a log-normal distribution on the initiation coeflicient, which was based upon the data C,
=
of Hall and others [26] for forged Alloy 600 pressurizer nozzles, with only the uncertainty based upon the data of Gold and others [27],
coefficient for the effect of grain boundary carbide coverage, which is based upon the C,
=
data ofNorring and others [22],
the maximum residual and operating stress level derived from the detailed elastic-o
=
plastic finite-element analysis from the WOG study of Ball and others [28] as shown in Figure 2-1, with its normally distributed uncertainty being derived from the variation in ovality from Duran and others [29] (see Figure 2-3), which is a trigonometric function of the penetration diameter and setup angle (local angle between the head and longitudinal axis of penetration).
S,,
yield strength of the penetration material,
=
n, n, exponents on stress and yield strength, respectively (n, = 4, n = 2.5)
=
G, the activation energy for crack initiation, which is normally distributed,
=
universal gas constant, and R
=
the penetration absolute temperature, which is uniformly distributed based upon the T
=
calculated variation of the nominal head operating temperature.
Page 6 Equation 2-3 is equivalent to the initiation equation by Rao (3), where G / A = C, + (1 + C 17;r )/S"'
2 r
Either data from field replication [30] or the correlation model by Rao [31] can be used to determine the percent grain boundary carbide coverage, Pase in equation (2-3). The model {31]
is a statistical correlation of measured values with the following materials certification parameters:
- Carbon content,
- Nickel content,
- Manganese content,
- Ultimate tensile strength and
- Yield strength.
The uncertainty on this model applies equally well to both the predicted and measured values.
The hours at temperature per operating cycle, which is normally distributed, is used to check if crack initiation has occurred. Once the crack has initiated, it is assumed to have a depth of a and o
its growth rate, da/dt, is calculated by the Peter Scott model, which matches the latest Westinghouse and EdF data and the previous data given in the WOG report on the industry Alloy 600 PWSCC growth rate testing results [32]. The crack growth model is:
- = C (K, - K, )
exp (2-4) 3 n
dt
- rRT, where:
C a log-normally distributed crack growth rate coefficient,
=
3 Kt the stress intensity factor conservatively calculated assuming a constant stress through
=
the penetration wall for an axial flaw at the inside surface with a length 6 times its depth using the following form of the Raju and Newman equations [33]:
K, = 0.982 +1.006(a/ w)2y[xy)u (2-5)
G, activation energy for PWSCC crack growth, which is also normally distributed, and
=
Kru =
threshold stress intensity factor for crack growth The probability of failure of the Alloy 600 vessel head penetration as a function of operating time
- t. Pr(t.s t), is calculated directly for each set ofinput values using Monte-Carlo simulation.
Monte Carlo simulation is an analytical method that provides a histogram of failures with time in a given number of trials (simulated life tests). The area under the simulated histogram increases
Page 7 with time due to PWSCC. The ratio of this area to the total number of trials is approximately equal to the probability of failure at any given time. In each trial, the values of the specified set of random variables is selected according to the specified distribution. A mechanistic analysis is l
performed using these values to calculate if the penetration will fail at any time during its lifetime (e.g. 60 years). This process is repeated many times (e.g. 6000) until a sufficient number of failures is achieved (e.g.10 per year) to define a meaningful histogram, which is an approximation of the lower tail of the true statistical distribution in time to failure. The shape of I
the distribution depends upon the input median values and specified distributions of the random variables. It is not forced to be an assumed type of distribution (e.g. Weibull) as is done for other non-mechanistic probabilistic methods. For the worst penetration in one plant, the mean time to failure was greater than 160 years but its uncertainty was so large that the normalized area under the histogram (estimated probability) at 60 years was 8 percent.
To apply the Monte Carlo simulation method for vessel head penetration nozzle (VIIPN) failure, the existing PROF (probability of failure) object library in the Westinghouse Structural Reliability and Risk Assessment (SRRA) software system was combined with the PWSCC structural reliability models described previously. This system provides standard input and I
output, including plotting, and probabilistic analysis capabilities (e.g. random number generation, importance sampling). The result was program VIIPNPROF for calculation of head penetration failure probability with time.
As reported previously [34], the Westinghouse SRRA Software System has been verified by hand calculation for simple models and alternative methods for more complex models. Recently the application of this same Westinghouse SRRA methodology to the WOG sponsored pilot program for piping risk based inspection has been extensively reviewed and verified by the ASME Research Task Force on RBI Guidelines [35] and other independent NRC contractors.
Table 2-1 provides a summary of the wide range of parameters that were considered in this comprehensive benchmarking study that compared the Westinghouse calculated probabilities from the analysis (labeled SRRA) with those from the pc-PRAISE program [36]. The comparison of calculated probabilities after 40 years of operation is excellent for both small and large leaks and full breaks, including those reduced due to taking credit for leak detection.
In addition, the VIIPNPROF Program calculated probabilities of getting a given crack depth due to PWSCC were compared for four plants where sufficient head penetration information and inspection results were available. The four plants are identified in Table 2-2 along with the values of the key input parameters and calculated failure probabilities. Table 2-2 also shows the agreement between the latest available inspection results and VIIPNPROF predicted failure trends due to PWSCC.
The models and software used for the probabilistic analysis of reactor vessel head penetration i
nozzles were developed, verified, validated and controlled in accordance with the Westinghouse Quality Management System, which has been audited per ISO-9001 requirements and accepted by NRC as meeting the requirements of 10CFR50, Appendix B. The input to and output from the software were also documented, verified and controlled in accordance with the same quality management system.
l l
..J,<>-
44+d-s o
lem---
k
& 4 o,,+M AA4-Lu kk a 64
& 4 - G.
6A 14,AA.L6,wB-s-A++-44 W-AA-AEhA snau.AAm,-m m
sAnn.-a h*W-Asem-AA.LLhas....s4 Page 8 i
c l
l l
l l
l l
l l
t 1
l l
l l
l k
I l
J u
l l
Page 9 i
TABLE 2-1 l
PARAMETERS USED FOR Tile pe PRAISE BENCllMARKING STUDY Type of Parameter Low Value liigh Value l
Pipe Material Ferritic Stainless Steel Pipe Geometry 6.625" O.D.
29.0" O.D.
0.562" Wall 2.5" Wall Failure Modes Small Leak.
Full Break, Through-Wall Crack Unstable Fracture Last Pass Weld Inspection No X-Ray Radiographic l
Pressure Loading 1000 psi 2235 psi Low-Cycle 25 ksi Range 50 ksi Range Loading 10 cycles / year 20 cycles / year High-Cycle
- I ksi Range 20 ksi Range Loading 0.1 cycles / min.
1.0 cycles /sec.
l Design Limiting Stress 15 ksi 30 ksi Disabling Leak Rate 50 gpm 500 gpm Detectable Leak Rate None 3 gpm
- Note: Mechanical Vibration (Iow value of stress range and high value of frequency) for small pipe, Thermal Fatigue (high value of stress range and low value of frequency) for large pipe.
Page 10 TABLE 2-2 COMPARISON OF VIIPNPROF CALCULATED PROBABILITIES WIT 11 PLANT OBSERVATIONS Parameters Almaraz 1 D. C. Cook 2 Ringhals 2 North Anna i llours of Operation 85,400 87,000 108,400 91,000 Setup Angle ( )
42.6 50.5 38.6 Temperatm e (*F) 604.3 598.5 605.6 600.0 Yield Strength (ksi) 37.5 58 51.2 51.2 Grain Boundary Carbides 57.0 44.3 3.0 2.0
(%)
Flaw Depth / Wall 0.10 0.43 0.25 0.10 1nitiation Probability 1.1 %
41.4%
37.6 %
15.3 %
Failure Probability" 1.1%
38.1 %
34.6 %
15.3 %
1 3
0 With Reported Indications (2 with scratches) from ISI
- Calculations performed at an equivalent setup angle for the 2nd highest stress location that could be inspected.
- Defined here as the probability of reaching the specified flaw depth for the individual penetration.
Page 11 3.0 References
[1]
WCAP-13565, Rev.1," Alloy 600 Reactor Vessel Adapter Tube Cracking Safety Evaluation," February 1993 (Westinghouse Proprietary).
[2]
F. Hedin and P. Gasquet, " Alloy 600 Reactor Vessel Head Penetration Cracking: An Industrial Challenge," 12* SMIRT Post Conference, August 23-25,1993, Paris, France.
[3]
Rao, G. V., and Wright, D. A., " Evaluation and Resolution of the Primary Water Stress Corrosion Cracking (PWSCC) Incidents of Alloy 600 Primary System Pressure Boundary Penetrations in Pressurized Water Reactors," Proceedings of Fontevraud II Symposium on " Contribution of Materials Investigation to the Resolution of Problems Encountered in PWR Plants," Royal Abbey of Fontevraud, France, September 10-14,1990.
[4]
A. S. O'Neill and J. F. Hall, Combustion Engineering, " Literature Survey of Cracking of Alloy 600 Components in PWR Plants," Report prepared for EPRI, January 1990.
[5]
U.S. NRC Information Notice No. 90-10, " Primary Water Stress Corrosion Cracking (PWSCC) ofInconel 600, February 23,1990.
[6]
NRC letter from William T. Russell to William Rasin of NUMARC (now NEI),
November 19,1993.
[7]
Pichon, C, Boudot, R., Benhamour, C., and Gelpi, A.," Residual Life Assessment of French PWR Vessel Head Penetrations through Metallurgical Analysis." Service Experience and Reliability Imorovement: Nuclear. Fossil. and Petrochemical Plants, PVP Vol. 288, ASME,1994.
[8]
Lagerstrom, J., Wilson, B., Persson, B., Bamford, W.H., and Bevilacqua, B,
" Experiences with Detection and Disposition ofIndications in Head Penetrations of Swedish Plants, " Services Experience and Reliability Improvement: Nuclear. Fossil. and Petrochemical Plants, PVP Vol.288, ASME,1994, pages 29 to 40.
[9]
Bamford, W. H., Fyfitch, S., Cyboron, R. D., Ammirato, F., Schreim, M., and Pathama.
R.," An Integrated Industry Approach to the Issue of Head Penetration Cracking at the USA,: Sersices Experience and Reliability Imorovement: Nuclear. Fossil. and Petrochemical Plants, PVP-Vol 288, ASME,1994, pages 11 to 19.
[10]
Rao, G. V.," Methodologies to Assess PWSCC Susceptibility of Primary Allay 600 Components in PWRs," Proceedings, Sixth International Conference on Environmental Degradation of Materials in Nuclear Power Systems, NACE, August 1993.
[11] Scott, P. M., "An Analysis of Primary Water Stress Corrosion Cracking in PWR Steam Generators," in Proceedings, Specialists Meeting on Operating Experience With Steam Generators, Brussels Belgium, September 1991, pages 5,6.
. - __ m l
Page 12
{
[12]
McIlree, A. R., Rebak, R. B., Smialowska, S., " Relationship of Stress Intensity to Crack
[
Growth Rate of Alloy 600 in Primary Water," Proceedings International Symposium Fontevraud 11, Volume 1, p. 258-267, September 10-14,1990.
[13]
Cassagne, T., Gelpi, A., " Measurements of Crack Propagation Rates on Alloy 600 Tubes i
in PWR Primary Water," in Proceeding of the 5th International Symposium on Environmental Degradation of Materials in Nuclear Power Systems-Water Reactors,"
i August 25-29,1991, Monterey, California.
l l
[14]
Personal Communication, Brian Woodman, Combustion Engineering, October 1993.
[15]
Ilunt, S. L. and Gorman, J., " Crack Predictions and Acceptance Criteria for Alloy 600 I
Ilead Penetrations" in Proceedings of the 1992 EPRI Workshop on PWSCC of Alloy 600 j
. in PWRs, December 1-3,1992, Orlando F1 (published in 1993).
J l
[16]
Personal communication - C. Amzallag to W. Bamford, Feb. 26,1997.
[17]
G. Economy, F. W. Pement, Corrosion /89 Paper 493.
i
. [18]
II. Tass et al., " Relation Between Microstructural Features and Tube Cracking Observed l
on Tube Samples of Doel 2 Steam Generator." EPRI Steam Generator Owners Group, l
SCC Contractors Workshop San Diego, CA, March 1985.
l
[19]
A. A. Stein," Development of Microstructural Correlation and a Tubing Specification for Alloy 600." Paper presented at EPRI Steam Generator Owner Group SCC Contractors
{
Workshop, San Diego, CA, March 1985.
[20]
A. R. McIlree, "Results of Reannealing Studies of Trojan, Doel 2, Ringhals 2 and 3, Ginna and Indian Point 3 Steam Generator Tubing" Paper as in Ref.18.
[21]
G. P. Airey, " Optimization of. Metallurgical Variables to Improve Corrosion Resistance ofInconel Alloy 600," Palo Alto, CA, Electric Power Research Institute, EPRI NP 3051, July 1983.
[22]
"Intergranular Stress Corrosion Cracking in Steam Generator Tubing, Testing of Alloy 690 and Alloy 600 Tubes," Norring, Engstrom and Norberg, in ThirdInternational Symposium on Environmental Degradation ofMaterials in Nuclear Power Systems -
l Water Reactors - Proceedings, The Metallurgical Society,1988 l
l
[23]
Z. Szklarska-Smialowska," Factors influencing IGSCC of Alloy 600 in Primary and Secondary Waters of PWR Steam Generators" Proceedings of the 5*ntemational Symposium on " Environmental Degradation of Materials in Nuclear Power Systems '
j Water Reactors." Nace Meeting. Edited by D. Cubicciotti, August 1989, p. 6-1.
[24]
R. Bandy and D. Van Rooyen," Stress Corrosion Cracking ofInconel Alloy 600 in liigh Temperature Water - An Update" Corrosion, Vol. 40, No. 8, page 425 (1984).
4 4
Page 13
[25]
Letter, J. A. Begley (APTECH) to B. A. Bishop," Review of the Westinghouse Structural Reliability, Model for PWSCC of RV Head Penetrations," June 23,1997.
[26]
" Evaluation of Leaking Alloy 600 Nozzles and Remaining Life Prediction for Similar Nozzles in PWR Primary System Application," Hall, Magee, Woodman and Melton, in Service Experience and Reliability Improvement, ASME PVP-Vol. 288,1994
[27]
"The Status of Laboratory Evaluations in 400 C Steam of the Stress Corrosion of Alloy 600 Steam Generator Tubing," Gold, Fletcher and Jacko in Proceedings of2nd International Topical Afecting on Nuclear Power Plant Thermal Hydraulics and Operations,1986
[28]
WCAP-13525, Rev.1, R V Closure Head Penetration Alloy 600 PWSCC (Phase 2), Bali et a'., December 1992 (Westinghouse Proprietary)
[29]
WCAP-13493, Reactor Vessel Closure Head Penetration Key Parameters Comparison, Duran, Kim and Pezze, September 1992 (Westinghouse Proprietary)
[30]
G. V. Rao, " Development of Surface Replication Technology for the field assessment of Alloy 600 micro-structures in Primary Loop Penetrations," WCAP-13746, June,1993 (Westinghouse Proprietary).
[31]
G. V. Rao and T. R. Leax, Microstructural Correlations with Material Certification Data in Several Commercial Heats of Alloy 600 Reactor Vessel Head Penetration Materials -
WCAP-13876, Rev.1, 1997 (Westinghouse Proprietary).
[32]
WCAP 13929, Rev. 2, Crack Growth and Alicrostructural Characteri:ation ofAlloy 600 Head Penetration Afaterials, Bamford, Foster and Rao, November 1996 (Westinghouse Proprietary).
[33]
Newman, J.C. Jr. And Raju, I.S. " Stress Intensity Factors for Intemal Surface Cracks in Cylindrical Pressure Vessels" Transactions ASME. Journal of Pressure Vessel Technology, Volume 102,1980, pp. 342-346.
[34]
WCAP-14572, Westinghouse Owners Group Application of Risk-Based Methods to Piping inservice Inspection Topical Report, pp. E-1 to E-6, March 1996.
[35]
Risk-Based Inspection - Development ofGuidelines, Volume 1, General Document, ASME Research Task Force on Risk-Based Inspection Guidelines Report CRTD-Vol.
20-1 (or NUREG/GR-005, Vol.1), American Society of Mechanical Engineers,1991
[36]
NUREGICR-5864, Theoreticaland User's Afanualforpc-PRAISE, A Probabilistic Fracture Afechanics Computer Codefor Piping Reliability Analysis, Harris and Dedhia, July 1992