ML063170247
| ML063170247 | |
| Person / Time | |
|---|---|
| Issue date: | 12/14/2006 |
| From: | Demoss G NRC/RES/DRASP/DDOERA/OEGI |
| To: | Kauffman J NRC/RES/DRASP/DDOERA/OEGI |
| Shared Package | |
| ML063170246 | List: |
| References | |
| Download: ML063170247 (10) | |
Text
December 14, 2006 MEMORANDUM TO:
John Kauffman, Acting Chief Operating Experience and Generic Issues Branch Office of Nuclear Regulatory Research FROM:
Gary DeMoss /RA/
Operating Experience and Generic Issues Branch Office of Nuclear Regulatory Research
SUBJECT:
DISPOSITION OF ACCIDENT SEQUENCE PRECURSOR ANALYSES
REFERENCE:
Memorandum to James E. Lyons, Director, NRR/DRA and others, from Farouk Eltawila, Director, RES/DRASP, Changes to the Accident Sequence Precursor (ASP) Program, August 9, 2006, ADAMS Accession No. ML061950028 This memorandum describes the disposition of Accident Sequence Precursor (ASP) analyses of conditions involving cracking and leaking control rod drive mechanisms (CRDMs). The completion of ASP analyses documented in this memorandum is part of the overall effort which led to the completion of all ASP analyses for all events through FY 2005. Additionally, the results have been included in regular reports, such as the SECY papers for the ASP program and the Industry Trends Program.
Conditions involving primary water stress-corrosion cracking of CRDM housings initially discovered at eleven plants in FY 2001-2003 have been analyzed. The ASP team has coordinated with the Office of Nuclear Reactor Regulation (NRR) on input for several of these analyses. To complete the remainder of the analyses, we simplified the inputs for initiating event frequencies and potential sump-clogging probabilities. This simplification is justified since we determined that more detailed analysis would not result in additional insights. This simplification does not affect ASP trends and results, and the staff has included these events in the total count and trending of all precursors as reported in SECY-06-0208. Details are provided in the Enclosure, which will be the technical record of these analyses.
The risk calculations may not be as precise as historical ASP analyses. However, ASP, SDP and MD 8.3 analyses have been carefully reviewed to ensure that all analyses are precursors (conditional core damage probability >10-6) and none are significant precursors (conditional core damage probability >10-3). Consistent with the referenced memorandum, this approach to releasing and documenting analyses reduces administrative and review burdens to the NRC staff and licensees.
Enclosure:
As stated
ML063170247 Template No. RES-006 Package Accession No.: ML063170246 Publicly Available? (Y or N) N Date of Release to Public N/A Sensitive? N To receive a copy of this document, indicate in the box: C = Copy wo/encl E = Copy w/encl N = No copy OFFICE OEGIB C
NAME
- GDeMoss DATE 12/14/06 Enclosure PRECURSORS INVOLVING CRDM CRACKING Conditions involving cracking and leaking CRDMs were identified, reported and corrected in the past several years at the affected plants. A detailed engineering analysis was not performed for these events because this high cost analysis of reactor vessel heads that have been replaced is not justified. Our evaluations provides risk estimates that are adequate to meet the ASP program objective of systematically evaluating nuclear power plant events to provide a risk-informed view of plant operating experience for internal and external stakeholders and supporting the industry trends program. The analyses are listed below and summarized.
The purpose of this enclosure is to document the results of the Accident Sequence Precursor (ASP) analyses of conditions involving cracking and leaking control rod drive mechanisms (CRDMs). These conditions were identified, reported and corrected in the past several years at the affected plants. While a detailed engineering analysis was not performed for these events, the evaluations provide reasonable risk estimates for these conditions in accordance to the ASP program objective of systematically evaluating nuclear power plant events to provide a risk-informed view of plant operating experience for internal and external stakeholders and supporting the industry trends program.
Table 1 lists the originally identified events, as reported in 2004. Further analyses of LER data added an event from North Anna 1. Not listed in this table is Davis-Besse, which has been analyzed.
Table 1. Reprinted Table 4 from SECY-05-0192. FY 2001-2005 CRDM cracking events.a, b Event Date Plant Description 12/4/00 Oconee 1 Reactor pressure vessel (RPV) head leakage due to primary water stress corrosion cracking (PWSCC) of five thermocouple nozzles and one CRDM nozzle. LER 269/00-006, LER 269/02-003, LER 269/03-002 2/18/01 Oconee 3 RPV head leakage due to PWSCC of nine CRDM nozzles.
LER 287/01-001, LER 287/00-003, LER 287/03-001 3/24/01 ANO 1 RPV head leakage due to PWSCC of one CRDM nozzle.
LER 313/01-002, LER 313/02-003 4/28/01 Oconee 2 RPV head leakage due to PWSCC of four CRDM nozzles.
LER 270/01-002, LER 270/02-002 6/21/01 Palisades RPV head leakage due to PWSCC of one CRDM nozzle.
LER 255/01-002, LER 255/01-004 10/1/01 Crystal River 3 RPV head leakage due to PWSCC of one CRDM nozzle. LER 302/01-004 10/12/01 TMI 1 RPV head leakage due to PWSCC of eight thermocouple nozzles and five CRDM nozzles. LER 289/01-002
Event Date Plant Description 10/28/01 Surry 1 RPV head leakage due to PWSCC of two CRDM nozzles.
LER 280/01-003 11/13/01 North Anna 2 RPV head leakage due to PWSCC of one CRDM nozzle.
LER 339/01-003, LER 339/02-001 4/30/03 St. Lucie 2 RPV head leakage due to PWSCC of two CRDM nozzles.
The analyses of cracking events are ongoing. The risk associated with multiple cracks at a given plant will be considered collectively in one analysis for each plant (i.e., only one precursor for each plant).
b.
The reviews and analyses for these events have not been completed and, therefore, the number of precursors due to cracking of CRDM housings may change.
One input required for these ASP analyses is the probability of ejecting a CRDM. The calculation of this probability requires the application of advanced fracture mechanics techniques and information that is not readily available to the NRC. Some initiating event probabilities have been calculated using a method that combines fracture mechanic knowledge and various correlations. In general, inspection results indicate plant operation with some period of reactor coolant system pressure boundary leakage. The duration of the various leaks cannot be definitively determined from the available data.
The risk assessment process assumes a Weibull distribution for the time that leakage started in each nozzle that has been determined to be leaking. Based on industry experience, it is assumed that any nozzle on which the annulus above the J-groove weld has become wetted by leakage will have a probability of 0.2 for a circumferential crack to initiate. A probabilistic distribution of circumferential crack growth rates is estimated on the basis of stress level and material properties for each leaking nozzle. The distributions for leakage duration and crack growth rates are combined to obtain a distribution, as a function of plant age, for the probability that a nozzle that has been leaking could develop a circumferential crack greater than 330o.
The growth of a circumferential crack to 330o is assumed to cause nozzle ejection, resulting in a medium loss-of-coolant accident (MLOCA). This distribution of ejection probability versus time is used to estimate the probability that a nozzle ejection would occur in the year prior to the inspection of interest, and that probability is the maximum increase in MLOCA frequency for the unit due to the leakage. The resulting change in the core damage frequency is calculated by multiplying this change in the MLOCA frequency by the conditional probability that a MLOCA will result in core damage at the affected unit.
It has become common practice for the agency and industry to assume that the plant age when the leakage started can be represented probabilistically by a Weibull distribution. The Weibull slope parameter (b) is assumed to be three, by consensus with industry. The Weibull shape parameter () is calculated from the total number of leaks that have occurred by a specific plant age, typically the current age when that data becomes available from an inspection.
An additional source of uncertainty in the risk calculations is the probability of the Emergency Core Cooling System (ECCS) sump plugging. This is important because the dominant cutset in each analysis is the CRDM ejection followed by a plugged sump. There is no operational
experience available to quantify sump failure probability for a CRDM ejection event. Ongoing research supporting Generic Safety Issue 191 indicates that current PRA models underestimate the sump failure probability. These analyses used the approach that was used in the Davis-Besse ASP analysis for sump failure probability calculations.
The CRDM cracking issue has been addressed by other NRC programs and the CRDM cracking issue is being addressed by Generic Safety Issue 191. There would not be useful insights obtained from a more rigorous probabilistic analysis. Therefore, the approach used for the ASP analyses is consistent with the overall objectives of the ASP program to put these events in the proper risk perspective relative to other events.
Sensitivity analyses show that for the plausible range of CRDM cracking probabilities and sump failure probabilities, these events will be greater than 10-6 but less than 10-3, which means that they are all accident precursor events (>10-6) but will not be significant precursors (>10-3) as defined in the ASP program. No events were rejected for low risk. These scoping analyses are sufficient to meet the mission of the ASP program, but not sufficient for publishing a precise risk estimate. Table 2 summarizes the initiating event probabilities calculated and summarizes the analysis results.
Table 2. FY 2001-2005 CRDM Cracking Event Analysis Results.
Plant IE Prob.
IE Source Status Oconee 1 0.020 2, 3 CDP estimate from a scoping ASP analysis package, included sump-related deficiency.
Oconee 3 0.016 2, 3 CDP estimate from a scopng ASP analysis package, included sump-related deficiency.
ANO 1 0.013 6
CDP is based on average initiating event probability & SPAR model calculations.
Oconee 2 0.006 2, 3 CDP estimate would be about 1/3 of Oconee 1 or 3, above, included sump-related deficiency.
Palisades 0.013 6
CDP is based on average initiating event probability & SPAR model calculations.
Crystal River 3
0.0006 6
CDP is based on risk estimates at similar B&W plants and SPAR model calculations.
TMI 1 0.012 6
CDP is based on risk estimates at similar B&W plants and SPAR model calculations.
Surry 1 0.013 5
CDP is based SPAR model calculations.
North Anna 2 0.035 4
CDP estimate from a scoping ASP analysis package.
North Anna 1 0.007 4
CDP estimate from a scoping ASP analysis package.
St. Lucie 2 0.013 6
CDP is based on average initiating event probability & SPAR model calculations. Dual sump plant reduces risk.
Sources for initiating event (IE) probability:
- 1.
Caruso, M. memorandum to Cheok, M. C., Reevaluation of Increase in Medium LOCA Frequency Attributable to Circumferential Cracking Potential in Leaking CRDM Nozzles at the Davis-Besse Nuclear Power Plant, September 7, 2004. (Adams Accession No. ML0425104015).
2.
Tschiltz, M. D. memorandum to Lesser, M., Risk Assessment to Support Significance Determination for Leaking Nozzles Detected at Oconee Units, September 30, 2004 (Adams Accession No. ML0427301630).
- 3. to Tschiltz memo, Risk Assessment to Evaluate Significance Determinations for Findings from CRDM Nozzle Inspections at Oconee Units 1, 2 and 3, September 30, 2004 (Adams Accession No. ML0427302470).
4.
Long, S. M., Email to Lesser, M., Risk Assessment for North Anna CRDM Leaks, June 9, 2005 (Predecisional).
5.
Long, S. M., Email to Lesser, M., Risk Assessment to Support Closeout of Surry Unit 1 LER on CRDM Nozzle Cracking, October 12, 2005 (Adams Accession No. ML052860015) 6.
OEGIB Calculation files, which use sources 1 through 5.
LER
SUMMARY
LER# 255/01-004, Palisades Nuclear Plant, 6/21/01 A pressure boundary leak was discovered from the CRDM #21 upper housing assembly (Austenitic SS). The leak was observed to be near the number three weld. Nondestructive testing revealed that an axial through-wall crack propagated through the weld. Destructive tests revealed axial and circumferential indications. Nondestructive tests identified recordable indications in 42 of 45 CRDM nozzles at the same weld locations. The source of these cracks and indications was deemed to be transgranular stress corrosion cracking. No specific geometry of the through-wall crack was given, but a leak rate of 0.1- 0.3 g.p.m. was observed.
LER# 269/00-006, Oconee Nuclear Station Unit 1, 12/4/00 During an RPV head inspection, a small amount of boric acid deposits were found around the bases of five (of eight) thermocouple nozzles and the CRDM #21 nozzle. Video inspection revealed that the boric acid deposits came from the nozzles themselves. Eddy-current tests determined axial crack indications on the ID of the eight T/C nozzles. The cracks ranged from 0.163-0.28 inches for six of the nozzles, and two nozzles had cracks all the way the J weld.
CRDM #21 nozzle testing revealed three very small, rounded, pinhole indications with diameters <0.05 inches. Further testing found a long (0.75 inches) linear indication oriented transversely to the direction of the weld. These cracks were attributed to PWSCC.
LER# 270/01-002, Oconee Nuclear Station Unit 2, 4/28/01 During an RPV head inspection, a small amount of boric acid deposits were found around the bases four CRDM nozzles (4, 6, 8, and10). After PT testing, several axial cracks were found on the four CRDM nozzles near the toe of the fillet weld and propagated radially into the nozzle material and axially on the OD surface. Eddy-current testing revealed two axial flaws on CRDM
- 16 nozzle and craze cracking on the four CRDM nozzle ID surfaces. Ultrasonic testing revealed some axial cracks and a short circumferential crack (0.07 inches in depth x 1.26 inches in length) on CRDM #18 nozzle. Additional crack geometry was not included in the LER.
LER# 287/01-001, Oconee Nuclear Station Unit 3, 2/18/01 During an RPV head inspection, a small amount of boric acid deposits were found around the bases of nine CRDM nozzles (3, 7, 11, 23, 28, 34, 50, 56 and 63). After PT testing, several deep axial cracks were found on the nine CRDM nozzles near the toe of the fillet weld and propagated radially into the nozzle material and axially on the OD surface. Eddy-current testing revealed additional cracks on several nozzles. Ultrasonic testing confirmed several deep (some through-wall) cracks in all nine CRDM nozzles. Forty-seven flaws were found in the nine leaking CRDM nozzles. Nineteen flaws were OD initiated, axial cracks (not through-wall), three cracks were ID initiated cracks (not through-wall), sixteen flaws were OD initiated through-wall cracks, and nine circumferential (one ID flaw and nine OD flaws) cracks. Two of the circumferential cracks were through-wall (nozzle 50 & 56).
LER# 289/01-002, Three Mile Island, Unit 1, 10/12/01 During an RPV head inspection, boric acid deposits were found around the bases of all eight thermocouple nozzles and twelve (11, 20, 29, 32, 35, 37, 41, 44, 48, 51, 64, 65) CRDM nozzles.
The T/C nozzles were determined by visual inspection to be pressure boundary leakage points (no other tests were done). PT tests determined four CRDM (35, 37, 44, 64) nozzles had flaws (both axial and circumferential). UT tests identified seven CRDM nozzles with axial flaws and no circumferential flaws. The nozzles with flaws were identified: 11, 29, 35, 44, 51, 64, and 65.
The flaws size varied, but the largest flaw was measured 0.43 inches in depth and 2.06 inches in length. The conclusion gathered from the various tests performed indicated that five (29, 35, 37, 44, and 64) CRDM nozzles displayed pressure boundary leakage. The flaws were attributed to PWSCC.
LER# 368/00-001, Arkansas Nuclear One, Unit 2, 7/30/00 During a routine inspection, boric acid deposits were found around twelve pressurizer heater sleeves and one RCS hot leg RTD sleeve. All sleeves were determined by visual inspection to be pressure boundary leakage points. Three pressurizer sleeves were Eddy-current tested.
These tests did not examine the full sleeve, however, each sleeve was identified as having a single, through-wall, axial flaw (no circumferential flaws). The flaws were located near the J weld and ID initiated. The longest flaw was 0.43 inches. The flaws were attributed to PWSCC.
LER# 339/01-003, LER339/02-001 North Anna, various dates Unit 1 Inspection, 2001 Outage: Based on new industry experience documented in NRC Information Notice 2001-05, NRC Bulletin 2001-01, the Licensee performed a bare metal visual inspection of the Unit 1 reactor pressure vessel during a refueling outage in the fall of 2001.
The inspection included performing eddy current inspection under the head to detect small surface connected flaws. The bare metal visual of the penetration to bare metal head surface, using a combination of robotic cameras and a boroscope, initially reported 34 of the 65 penetrations with relevant indications of boric acid. The visual inspection was inconclusive with respect to CRDM nozzle leaks because there were boric acid streaks and spatter residue on the head surface due to a history of conoseal leakage above the head and an active conoseal leak during the previous operating cycle. Further review of the tapes of the inspection resulted in acceptance of 7 of the 34 penetrations. Based on a combination of the available inspection techniques at the time (OD and ID eddy current (ET) examinations, ID ultrasonic test (UT) examinations, and dye penetrant test (PT) examinations), and flaw growth evaluations, the Licensee concluded that there were no through-wall leaks on the other 27 penetrations and the head was acceptable for another cycle of operation.
Since the visual inspection of Penetration 50 initially characterized the penetration as having evidence of significant leakage, all available NDE techniques were used to determine if cracks were present in the J-groove weld or if through-wall flaws existed. The surface of the J-Groove weld and the penetration OD surface below the J-Groove weld were ET examined. No indications were identified on the penetration OD below the weld, and one non-recordable indication was identified on the surface of the J-Groove weld. ET of the ID tube surface covering 98% the weld area did not detect any indications. PT inspection of the J-Groove weld surface identified one non-service induced flaw on the surface of the weld and several other indications at the toe of the weld on the clad side. The non-service induced flaw was near the penetration tube side of the weld and was removed by grinding. The indications at the toe of
the weld were considered to be in the clad and not relevant to identification of flaws or IGSCC in the weld surface or the penetration material. (Similar PT indications were identified in J-Groove welds of three other penetrations.) A partial UT of the penetration tube ID at the weld was performed from 196 degrees to 293 degrees. A portion of the thermal sleeve was removed to gain additional access of the penetration above the centering ring for UT and ET. The penetration ID was UT and ET inspected and no indications were identified. Based on the NDE performed on Penetration 50, the Licensee concluded that no evidence of through-wall flaws existed and therefore, the boric acid identified at penetration 50 was from an external source.
After Unit 1 was returned to service, an inspection of the Unit 2 RPVH identified apparent leakage at three penetrations. PT inspection of the J-Groove welds for these three penetrations revealed indications at the toe of the welds similar to those identified in Unit 1 penetrations, including penetration 50. Removal of a boat sample and metallurgical analysis of the indications in one of the Unit 2 welds (penetration 62) determined that the PT indications were caused by hot-short cracking in the J-Groove weld butter layers, which resulted in through-wall leaks. Based on the Unit 1 PT indications being similar to the Unit 2 indications, it was concluded that the Unit 1 indications were also caused by hot-short cracking and could result in through-wall leakage. Therefore, Justification for Continued Operation (JCO) C02-02 was issued for Unit 1.
The 2001 inspection of Penetration 50 identified flaw indications (PT indications at the toe of the J-Groove weld) that were not repaired.
Unit 2 Inspection, 2001 Outage: Following issuance of NRC Order EA 03-009, the licensee conducted a bare metal visual inspection of the unit 2 reactor vessel head in the fall of 2001.
White boric acid crystals were seen on the head surface at 3 of the 65 nozzles where they exited the bore holes in the reactor vessel head. These nozzles (numbers 51, 62 and 63) were subsequently inspected with dye penetrant, and eddy current on the wetted surface of the welds and with ultrasonic testing of the tubes. Each of the 3 welds showed crack-like indications on the wetted surface, and each of the 3 tubes showed shallow axial (numbers 51 and 63) or craze crack (number 62) indications on the inner diameter surface. [Review of the documents in the attached reference list did not provide information sufficient to determine whether the UT inspections were directed to the portion of the nozzle tubes within 1.5-inches above the J-groove weld, where risk-significant circumferential cracking had previously been observed at Oconee.] A boat sample was removed from penetration number 62, and subsequent analysis indicated that the cracking probably originated from hot cracking during fabrication, with PWSCC contributing as a secondary factor. The welds were subsequently repaired with a temper-bead weld overlay, and the unit was returned to service.
Unit 2 Inspection, 2002 Outage: A bare metal visual inspection during the September 2002 refueling outage indicated that 2 nozzles (numbers 51 and 63) were leaking again, and that 4 more nozzles were probably leaking (numbers 10, 21, 31 and 35). An additional 21 nozzles were masked by boric acid from leaks above the head. A more extensive inspection was conducted with eddy current on the welds of 59 0f the 65 nozzles, revealing crack-like indications on 57 of them. The remaining 6 welds (including numbers 51, 62 and 63, which were repaired in the previous outage) were inspected with dye penetrant, and all were found to have rejectable indications. The licensee removed the thermal sleeves from 35 nozzles to facilitate ultrasonic inspection of the tubes from the inner diameter surface. None of the inspected tubes had evidence of through-wall cracks. A boat sample was removed from the repair of nozzle 51. Evaluation indicted that the repair did not completely cover the original weld material, and that the flaws were caused by hot cracking during the welding process.
Unit 1, 2003 Outage: On March 4, 2003, while in a refueling outage during which the reactor vessel head was being replaced, a visual inspection of nozzle 50 was performed to follow-up on previous inspection results from the 2001 outage. It was concluded that leakage had occurred at this nozzle during the previous operating cycle, based on the presence of a small boric acid deposit on the down-hill side of the nozzle-to-head junction. Only visual inspection was performed with no verification with other NDE techniques. At the time of the inspection, the Licensee had already made the decision to replace the Unit 1 head during the outage in process.
LER #302/01-004, Crystal River 3, 10/01/01 While performing a visual inspection of the reactor vessel head (RVH), FPC personnel identified one potential leaking CRDM nozzle (nozzle #32). The RVH inspection was performed to satisfy a commitment made by FPC in response to NRC Bulletin 2001-01, "Circumferential Cracking of Reactor Pressure Vessel Head Penetration Nozzles." Ultrasonic testing (UT) examination of RVH nozzle #32 identified the leakage path as two (2) axially oriented cracks that were through-wall. The cracks were caused by primary water stress corrosion cracking.
Eight (8) additional CRDM nozzles were examined using UT. No evidence of cracking was observed in the additional CRDM nozzles inspected. CR3 is developing a long-term strategy to deal with the CRDM nozzle cracking issue. No previous similar CR-3 occurrences have been reported to the NRC.
LER 280/01-003, Surry Unit 1, 10/28/01 With Unit 1 at Refueling Shutdown conditions, a "bare head" visual inspection of the reactor pressure vessel (RPV) head was performed as requested by NRC Bulletin 2001-01. Liquid penetrant examinations revealed surface indications in the J-groove weld of two control rod drive mechanism (CRDM) penetrations (Nos. 27 and 40). Based on the nature of these indications, it was determined that repairs were required. Subsequent examinations revealed unacceptable indications on four additional penetrations (Nos. 18, 47, 65, and 69) that also required repair. The six penetrations were modified and repaired, and Unit 1 was returned to power operation.
LER 389/03-002, St. Lucie Unit 2, 4/30/03 On April 30, 2003, St. Lucie was in a refueling outage and the reactor pressure vessel head surface and associated penetration nozzles were being inspected in accordance with NRC Order EA-03-009. Although the head surface visual examination revealed no evidence of reactor coolant system boundary leakage, ultrasonic testing revealed an axial flaw in each of the nozzles for reactor pressure vessel head penetrations 18 and 72. No other indications were found during the inspection activities. The most likely cause of the nozzle flaws was attributed to primary water stress corrosion cracking. The subject nozzles were replaced during the refueling outage. St. Lucie will continue to perform reactor pressure vessel head surface and associated penetration nozzle inspections in accordance with the requirements of the NRC Order.