IR 05000275/1990031
| ML16341F941 | |
| Person / Time | |
|---|---|
| Site: | Diablo Canyon  |
| Issue date: | 01/10/1991 |
| From: | Huey F NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION V) |
| To: | |
| Shared Package | |
| ML16341F940 | List: |
| References | |
| 50-275-90-31, 50-323-90-31, NUDOCS 9101290086 | |
| Download: ML16341F941 (18) | |
Text
"
U.
S.
NUCLEAR REGULATORY COMMISSION
REGION V
Report Nos.:
50-275/90-31 and 50-312/90-31 Docket Nos.:
50-275 and 50-323 License Nos.:
Pacific Gas and Electric Company 77 Beale Street, Room 1451 San Francisco, California 94106 Facility Name:
Diablo Canyon Units1 and
Inspection at:
Diablo Canyon Site, San Luis Obispo County, California Inspection Conducted:
December 11-19, 1990 Inspector:
M. Mag Insp ctor Approved by:
uey, ie Engineering Section
'Pro a
e 1gne Contractor:
C. J.
Czajkowski, Research Engineer Department of Nuclear Energy Brookhaven National Laboratory Ins ection from December 11-19 1990 (Re ort Nos.
50-527/90-31 and Areas Ins ected:
One inspector from Region V and one research engineer from
.
roo aven a ional Laboratory conducted an inspection to evaluate those actions taken by the licensee in determining the root cause of multiple weld cracks on the Unit 2 Chemical and Volume Control System (CVCS) letdown piping.
Inspection procedures 30703 and 73753 were used as guidance during this
.
inspection.
Results:
General Conclusions Six months is an extensive period of time for not having determined the probable cause for the No.
2 cracked weld.
There appears to have been a lack of urgency on part of the licensee to
'scertain the failure modes until the fourth weld crack was discovered on December 4, 1990.
910129008b 910110 PDR ADOCK 05000275
Insufficient information had been generated to adequately determine the root cause of the socket weld cracks.
Si nificant Safet Matters:
None Summar of Violations and Deviations:
None 0 en Items Summar
One followup item was opene DETAILS 1.
Persons Contacted
- M. Angus, Assistant Plant Manager, Technical Services B. Giffin, Assistant Plant Manager, Maintenance Services
"D. Miklush, Assistant Plant Manager, Operations
"D. Gonzalez, Inservice Inspector Supervisor
"T. Bennett, Maintenance Manager
~T. Grebel, Regulatory Compliance Supervisor
+"M. Barkhuff, equality Control Manager
"P.
Lang, Senior equality Control Engineer
+"J. Griffin, Regulatory Compliance Engineer
+"K. Condron, Production Engineer
"R. Klimczak, Piping Engineer
"J. Mhaley, System Engineer
"J. Schletz, Technical and Ecological Services Engineer
+*J. Shoulders, Onsite Project Engineer
"R. Nanninga, Senior Maintenance Engineer
"K. Palmer, Maintenance Engineer
- H.'arner, equality Assurance Auditor
- M. O'onnell, Regulatory Compliance Engineer
- R. Hackman, Consultant, Failure Prevention Incorporated B. LoConte, Onsite Review Group Engineer
" Denotes those attending the exit meeting on December 12, 1990.
+ Participated in telephone discussion on December 19, 1990.
2.
Inservice Ins ection (73753):
CVCS Letdown Line Meld Cracks The purpose of the site visit was to evaluate those actions taken by the licensee in determining the root cause of multiple weld cracks on the Unit 2 CVCS letdown piping.
aO
~dk d
During the past 18 months PG&E has experience four socket weld failures on two of three Reactor Coolant System (RCS) pressure Letdown Lines in the Chemical and Volume Control System (CVCS) of Diablo Canyon Unit 2.
The general plant area of the failures is in the Unit 2 Containment Structure, elevation 91', at approximately
degrees.
The flow restricting Orifice(s) (RO) are inside a
biological shielding enclosure immediately adjacent to the containment wall.
The first crack was discovered in the B line in June, 1989; a second crack was discovered in the B line in June, 1990; a third crack was discovered on the C line in October, 1990, and a,fourth crack was recently discovered in the C Line on December 4, 1990.
All of the cracks have been in the socket welds of fittings between the restricting orifices and the stop valves (8149B and C).
The first and third cracks were in the socket welds of the 90 degree elbow
k
approximately three feet downstream of the restricting orifices.
The second and fourth cracks were in the socket welds'fthe coupling downstream 'of the 90 degree elbow where the first and third cracks occurred.
The piping where all four cracks occurred is downstream of the RCS letdown isolation valves.
The nominal pipe diameter is 2 inches.
All four cracks have been tight resulting in very small leaks.
The leak rates have been less than 'technical Specification Limiting Condition for Operation 3.4.6.2.b, which limits unidentified reactor coolant system leakage to 1 gpm.
Detailed Failure Descri tion Event 1:
On June 9, 1989 the Unit 2 containment sump flow recorder indicated an increase in flow.
On June 12,.1989 the Shift Foreman (SF) noted in the log that surveillance test procedure (STP)
R-lOB showed an increase sn leakage in containment.
One June 16, 1989 the SF noted in the log'hat the iodine-134 level in Unit 2 containment was increasing.
On June 17, 1989 the SF noted that Radiation Monitor, RM-11, counts have increased by a factor of 3 over the last two weeks.
One June 19, 1989, during a routine containment walkdown, a
leak was found in the vic)nity of CVCS RO-27 in the socket weld of a pipe elbow (line S6-402-2).
The leak was flowing approximately 0. 1 ga) lons per minute (gpm), well within the TS 3.4.6.2.
allowable unidentified leakage rate of 1 gpm and/or the allowable identified leakage of 10 gpm.
Inspection of the piping in the vicinity of the cracked weld revealed a broken pipe support, 2215-33.
Inspection of the broken pipe support indicated that it.had been. broken for longer than the cracked weld had been leaking.
Pipe supports 2215-31 and 2216-190V had missing treaded fasteners.
Snubber 78-336SL was perceived to be frozen and was replaced; however, shop inspection showed it only to be stiffer than normal.
The failed weld was sent to General Electric's Vallecitos metallurgical laboratory for testing.
The testing report indicates that "... the most probable failure mode resulted from a combination of a corrosive environment, susceptible material (Type 304 stainless steel)
and most significantly initial stress concentration in the crevice area (note[i) from cyclic and residual (weld) stresses.
Without precise identification and verification of aggressive anions in the coolant, the most probable failure mechanism appears to be crack initiation in the crevice area followed by eventual socket weld failure by corrosion fatigue."
These results are not definitive and subsequently do not point to a specific root cause.
To determine the root cause of the failure, vibration measuring equipment was installed on the letdown line piping immediately down stream of the ROs during the Unit 2 refueling outage of 1989, (2R3).
The instrumentation was monitored upon the return to service following the refueling outage, however, no significant vibration data was collecte Event 2:
On June 2, 1990, the Shift Supervisor review of daily Rg sheets identified that a potential RCS leak had developed.
On June 5, 1990, a. new socket weld crack was confirmed to be located in the B
letdown line in the upstream socket weld of the 2 inch coupling installed during repairs made in June 1989.
The letdown line was isolated, the cracked weld cut out, and new piping installed as required.
This failure was not considered to be a significant event due to the fact that this line was isolated most of the time from the time of repair in 1989 and had only been in use since special instrumentation was installed during 2R3.
The installed instrumentation indicated that there had been no significant operational, transients.
The total in service time of this repaired line was approximately 4 weeks, therefore, it was perceived that this may have been a premature failure due to inadequate repairs made in 1989.
At this time the failure mode was speculation on part of the licensee and not based on a detailed failure investigation.
The failed weld material was removed to preserve the failure for, further examination.
The weld material is presently at Battelle Pacific Northwest Laboratories on hold due to the lack of a definitive contract.
Event 3:
On October 30, 1990, plant operators confirmed by observation that the upward trend indications of RCS leakage first detected on October 27, 1990, was due to CVCS letdown line leakage between RO-20 and isolation valve 8149C.
This failure was immediately considered to be a significant operational condition and Event Response Plan (ERP) 90-12 for the investigation arid repair was initiated.
A Justification for Continued Operation (JCO) 90-19, Revision 0, was written and approved.
The line was repaired by replacing the elbow immediately downstream of the restricting orifice and connecting the new elbow and subpiece to the remaining piping with a socket weld coupling for convenience of repairs.
This configuration was" identical to the repair configuration of Event 1.
Event 4:
On December 4, 1990, a cracked socket weld was confirmed on letdown line C.
This failure was immediately considered to be a significant operational condition and ERP 90-14 for the investigation and repair was initiated.
This failure has been repaired by replacement of the first elbow and approximately three feet of piping downstream of the elbow, downstream of the restricting orifice.
This repair returns the C letdown line piping to the original plant configuration.
Additional strain gauge instrumentation has been installed.
Also, PGEE and industry consultants have been called in to evaluate the evidence collected to date.
Upon initial evaluation by the consultants, operational transient testing will be performed to verify the possible cause(s)
of piping vibratio EI
Hethod of Discovery:
Leakage was discovered during the routine observation of control room indicato'rs, an RCS leakage increase noted during erformance of STP R-10, "Reactor Coolant System Leakage valuation," and visual verification by plant operators.
Operator Actions:
.Operators identifieti the leakage, performed leakage determinations, and performed 1n containment verification of location as indicated for each of the four events.
Immediate Cause:
In each case a tight through the weld material crack developed in the socket weldments of the CVCS letdown lines B and C,
immediately downstream of the restricting orifice.
c.
Documents Reviewed 1)
Purchase Order ¹551136, 8/10/90, PGEE to Battelle Pacific Northwest Laboratories.
2)
GE (Vallecitos Laboratory) Failure Analysis Report 89-010, (PG8E PO ¹526769),
8/90.
3)
Various PGKE Action Requests (supplied by utility).
4)
Various drawings and isometrics of the Unit 1 and Unit 2 letdown piping system.
5)
PGEE memo to DCPP APM/TECH SERVICES from NECS ONSITE PROJECT, ENGINEER, re:
Cracked Melds in Metallic Piping, 8/3/90.
6)
Various PGBE Mork Orders for repairs on the letdown lines.
7)
NDE reports (visual) and associated PGEE documentation.
8)
Meld metal and filler metal certifications (various).
d.-
Observations The site visit consisted of various discussions uti.lity and contractor personnel, an inspection lines (Unit 2)
and a visual examination of the failed weld.
ks of December
1990 the only analysis performed was done by (enera1 Electric first failure (6/19/89).
with cognizant of the CYCS letdown number 4 (coupling)
metallurg>cal (Vallecitos) on the An entrance meeting was held on December ll, 1990, at which the licensee outlined the events/failures to date and the 'subsequent actions taken to determine the root cause.
An exit meeting was
C
convened on December 12, 1990, with licensee personnel during which the following NRC/BNL observations were presented:
1)
It 'appeared that the six month delay (failure 2) and the one month delay (failure= 3) in obtaining a metallurgical evaluation of the specimens was an inordinately long time f'r evaluation of repetitive.type fai lures.
There appeared to be no sense of urgency on the part of the licensee to verify the generic failure mode of the failed components.
2)
The GE report (8/89) should have included some additional tests which might have"provided information (at the time) to substantiate the proposed corrosion assisted fatigue failure evaluation.
Some suggestions were:
Energy Dispersive Spectroscopy (EDS) analysis for contaminants and semi-quantitative chemical analysis of the weld metal, ferrite measurements of the weld metal, and microchardness measurements.
3)
More PGEE metallurgical involvement in placing purchase orders to outside contractors (for failure analysis)
should be considered.
The involvement would include:
type of testing to be done (minimum)
methods of cutting/sectioninp specimens, e.g.,
wet or dry?
PGEE cutting out of "in situ specimens prior to analysis 4)
The utility should consider placement of monitoring devices on Unit I CVCS letdown lines to develop a base line and comparative information study of Unit 1 versus Unit 2.
This monitoring should take into account "lessons learned" on Unit 2 regarding placement of and types of devices required.
5)
A great deal of work has been done to date, but. has not resulted in a definitive root cause for the problems.
A follow-up telephone discussion among the licensee, NRC Region V, the site senior NRC resident Inspector, and the BNL representative took place on December 19, 1990.
The atems discussed included the inspectors reasons for additional tests and PGEE metallurgical.
involvement.
The bases for the additional metallurgical involvement stemmed from the fact that the licensee Purchase Order to Battelle Pacific Northwest Laboratories (PNL) (¹551136 8/10/90) only stated that a Failure Analysis was to be done on Coupling 402.
The PNL statement of work stated in part, "...It is assumed that the first components to be examined will be a'failed pipe coupling..."
The licensee did not indicate that the failure analysis would include the elbow welded to the same line.
The welded elbow had seen the same number of cycles as the coupling but no plans were formalized to examine it.
Also, the licensee,dsd not express any plans to perform EDS on the specimens, nor any tests other than metallographic and fractographic sectioning.
Additionally, no mention was made of the comp')ementary elbow welded to the coupling on failure number 4 (12/4/90). It appears that this elbow was lett
in situ and will not be examined at all as part of this investigation.
e.
Conclusions 1)
Potential Contributor Cause(s)
PGIEE investigations performed to date have identified the following potential sources of vibration forcing function(s)
that may have contributed to cyclic stress to the piping:
1.
Failure of the relief valve (RV) 8117,on August 1, 1986, November 8, 1987, and July 18, 1989 (Refer to NRC DCO-90-MM-N013 for details of Crosby relief valve failures).
2.
Low operating backpressure as a result of the pressure channel PI-135 out of tolerance in June 1988 (Refer to AR A0113485).
3.
Operational related occurrences of low backpressure that have been measured by installed test equipment.
2)
Investi ative Action to Determine Root Failure Cause 1.
PG8E has placed additional instrumentation on the affected lines to measure vibrational levels, internal pressure and strain near the failure locations.
2.
PGImE has contracted with qualified metallurgical laboratories to perform an indepth analysis of the failed weld material.
3.
PGlIE has contracted two independent consultants to review all evidence gathered to date and provide their preliminary evaluation.
4.
Unit 2 will conduct operational transient testing of the letdown lines consistent with the recommendations of the PGLE and industry consultants called to evaluate these events.
3)
Root Cause The root cause of the failures is unknown.
Further investigation of the failed parts by contract metallurgists will be conducted on all removed welds.
Additional vibration, displacement, pressure and strain gauge data is being gathered.
Controlled plant operational transient testing will be conducted following preliminary evaluation after'll PG8E and industry consultants have reviewed the failure evidence to date.
Current instrumentation data indicates that this section of piping does not have high
h
)
IK
I
vibration levels which may result in fatigue failure of the piping or supports.
Monitoring of the line will continue until the root cause can be identified and verified.
The conclusions drawn by the inspectors was that the licensee had not yet generated sufficient information to adequately determine the
'oot cause of. the multiple socket weld failures.
The results of this effort will be examined further by Region Y as an Open Item (50-323/90-31-01).
3.
Exit Meetin (30703)
The inspectors met with the licensee management representatives denoted in paragraph 1 on December 12, 1990.
The scope of the inspection and the inspector s findings were discussed as described in this report.
Also, on December 19, 1990 a telephone conversation was held between Region V, BNL, and license personnel (denoted in paragraph 1) to discuss the metallurq)cal test results of the cracked socket weld discovered on June 19, 1989.