ML20082C431
ML20082C431 | |
Person / Time | |
---|---|
Site: | Catawba ![]() |
Issue date: | 07/11/1991 |
From: | Hampton J DUKE POWER CO. |
To: | NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM) |
References | |
1-C91-0199, 1-C91-199, NUDOCS 9107190105 | |
Download: ML20082C431 (10) | |
Text
.
ll l
Dub l%.r Ovnrum
!BW11 ha afaa M % karhistwh
('
ro tta :n Grur M'llUf0 h
DUKC POWER July 11, 1991 Document Control Desk U.
S. Nuclear Regulatory Commission Washington, D. C.
20555
Subject:
Catawba Nuclear Station Docket No. 50-413 IIR C91-038-1; PIR 1-C91-0199 Gentlement Attached is our Prob 1cm Investigation Report 1-C91-0199, submitted concerning PRESSURIZER SAFETY VALVE SETPOINT DRIFT DUE TO UNKNOWN CAUSE.
This incident has been determined to be non-reportable but has been investigated and is being documented as a Special Report to ensure industry awaroness of this event.
The health and safety of the public were not affected by this incident.
Very truly yours,
/
f S /$sf 8 W
/J. W. Hampton
["
Station Manager kon: REPORT.SP xc:
Mr. S. D. Ebneter M & M Nu,:lcar Consultanto Regional Administrator, Region 11 1221 Avenuco of the Anerican U. S. Nuclear Regulatory Commission New York, NY 10020 101 Marietta Stroot. NW, Suite 2900 Atlanta, GA 30323 R. E. Martin INTO Recorda Conter U. 3. Nuclear Regulatory Commir,sion Suite 1500 Office of Nuclear Reactor Regulation 1100 Circle 75 Parkway f
Washington, D. C.
20555 Atlanta, GA 30339 l
Mr. W. T. Ordera
[q NRC Resident Inspector Catawba Nuclear Station
[(
j,(/
l i
t
- 107190105 910711
.-t 7 0 0 4 6*
eon nooce oswou a s
DUKE POWER COMPANY CATAWBA NUCLEAR STATION PROBLEM INVESTIGAT10N REPORT 110.3 1-C91-0199 PRESSURIZER SAFETY VALVE SETPOINT DRIFT DUE TO UNK!10WN CAUSE ABSTRACT on May b,1991, at' 1715 hours0.0198 days <br />0.476 hours <br />0.00284 weeks <br />6.525575e-4 months <br />, with Unit 1 in Modo 6, Refueling, a Maintenanco Engineering Services (HES) engineer learned that the three Unit 1 Pressuri.,r Safety Valvos' (PSVs) (INC-001, INC-002, and 1NC-003) As-Found lift r.etpoinsa were outsido the 2485 psig, +/-1% acceptanco critoria of Technical Specification (T/) 3.4.2.2.
The valvos had been previously removed during the End of Cyclh1 Refueling Outage end sent-to Wyle Laboratories for testing.
This incident ja attributed to an unknown cause, due to a possible proceduro deficiency.
It appears that a number of factors could be contributing to sotpoint drift, involving the proceduros and practices utilized by Wyle Laboratories and Dressor, the valve manufacturer, and approved by Duke.
Several procedure enhancements are being pursued to reduce setpoint drift. A detailed investigation to datormine the causo(s) of the drift was performed.
The PSVs lustalled to replace those removed were tested uning a revised method, at tho Wyle Laboratories test f acility.
This report is being submitted as a Special Report, since the timo at which the cetpoints drifted out of the T/S allowablo rango is unknown, and no specific on-sito condition resulting in these setpoint drifts has-been identified.
l l
l l
\\
DUKE POWER COMPANY / CATAWBA NUCLEAR STATION PIR l-C91-0199/Special Report Page 2 7
1 BACKGROUND Reactor Coolant (Ells AB) (NC) System pressure is controlled by the use of the Pressurizer, where water and stenm are maintained in equilibrium using heaters
[EIIS:EHTR) and water spray.
Pressure protection is provided by three Power Operated Relief Valves {EIIS:V) and three spring loaded Pressurizer Safety
~r
- Valves (PSVs), all of which relieve from the Pressurizer to the Pressurizer Relief Tank (PRT).
Technical Specification (T/S) 3.4.2.2 requires all PSVs to be operable with a lift setpoint of 2485 psig +/-l% in Mode 1, Power _ Operation, Mode 2, startup, l
and Mode 3, Hot Standby. With one PSV inoperable, the inoperable PSV must be restored to operable status within 15 minutes, or the Unit must be in at least Mode 3 within 6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> and in at least Mode 4, Hot Shutdown, within the following 6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br />.
T/S 3.4.2.1 requires a minimum of one PSV to be operable with a lift setpoint of 2485 psig +/-1% in Mode 4 and Mode 5, Cold Shutdown. With no PSVs
-operable,--all operation: involving positive reactivity changes must be suspended, and an operable residual heat removal loop must be placed in operation in the shutdown cooling mode.
T/S 4.0.5 requires periodic testing of valves per ASME Boiler and Pressure Vessel Code,Section XI, 1983 Edition. The 1986 Edition subsection IWV, (Inservice Testing of Valves in Nuclear -Power Plants), section 3510, Safety Valve and Relief Valve Tests,. requires test frequency and method to be in accordance with ANSI /ASME OM-1-1981,-Requirements For Inservice Performance i
Testing of Nuclear Power Plant Pressure Relief Devices. This document allows a +/-3% acceptable range for th6 setpoint test, and surveillance testing to subsequent editions of codes and addenda is acceptable per 10CFR Part t
50.55a(g)(3)(v).
Catawba's PSVs are sent to.Wyle Laboratories each refueling outage for testing and refurbishment. A test procedure performed at Wyle Laboratories to repair a leak after a successful set pressure test has been r'
performed is the " jack and lap" process, in which a hydraulic jack is used to remove the spring load on the disc, enabling the disc and seat to be lapped.
LIn' September, 1990, a revision of a Special Report (reference Problem Investigation Reports (PIRs) 2-C90-0248 and 1-C90-0075), Setpoint Drift on Pressurizer Safety Valves During successive surveillance-Tests, was submitted to.the NRC. On February 23, 1990, Maintenance Engineering Services (MES) personnel were notified that a PSV previously removed from the Unit 1
' Pressurizer failed to meet the 2485 poig.+/-l% setpoint requirement.when
. tested at Wyle baboratories. On July-22, MES personnel were notified that two PSVs previously removed from theLUnit'2 Pressurizer. failed to meet the same requirement.. Past surveillance tests were reviewed, and it was found that 60%
of the PSVs-have failed to meet the 4/-l% requirement. A search of the Nuclear Plant Reliability Database System (NPRDS) identified 62 failures of
.PSVs:to be within the required range.
iw-iir 'r e mf T'ep'f1-V*-ri-N1+i-++9e
'veww'-4wwwirw--y,--,--
T--we-+ww 1r4
--mve11t=ra-9r g.amee er y qu' is le 1'r"Y
-'-ff--
W-w"wt%WF'w1 RIFE N
- TF"F-ww'w'Fr'--Wmv'M-*Tw1-w11rv N v ee--~mt*MN wTwvWWN-w-
- wi=r i
DUKE POWER COMPANY /CATAWIM HUCLEAR STATION PlR.1-C91-0199/Special Report Page 3 The ability of safety valvos to consistently lift within the +/-1% range is
~ recognized as an industry wide concern.
Input from the utility industry was solicited using the Nuclear Network and the NPRDS.
Responso at the timo indicated a 42% failure rate in not meeting the +/-14 range.
i The Catawba PSVs are manufactured by Drosser, Model 6-31749A-2-XNC019 and are totally closed, pop-type, spring loaded and self-activated with back pressure compensation. The combined capacity of the valves is greater than or equal to the maximum surga rato resulting from a complete loss of load without Heactor trip or any other control. These valves are not installed with loop seals.
EVENT DESCRIPTION On March 26, 1991, during the Unit 1 End of Cycle 5 (EOCS) Refueling Outage, the three pSVs woro1 removed. The PSVs were sont to Wyle Laboratories for
-testing.
On May 8, 1991, at 1715 hours0.0198 days <br />0.476 hours <br />0.00284 weeks <br />6.525575e-4 months <br />,-with Unit 1 in Mode 6, Hefueling, a Maintenanco Engineering Sorvices (MES) engineer learned that the three previously removed Unit 1 Pressurizer Safety Relief Valves (INC-001, INC-002, and INC-003) As-Found lift setpoints were outside tho +/-1% acceptance criterlat valve INC-001 was at 2633 psig (6% high), valvo INC-002 was at 2626 psig (5.7% high),
and.valvo INC-003 was at 2376 poig (4.6% low) when initially tested at Wyle Laboratories. Problem Investigation Report (PIR) Nunbor 1-C91-0199 was written to: - disassenble,. inspect and evaluate the valves for cause of failure, initiate root causo analysis, reset and test the throo valvos, and perform a past operability evaluation.
On May 11, 1991, the valvo repair (jack and lap) procedure was performed on a Catawba Unit 2 Pressurizer Safety Valvo. The purpose of this experiment was t
to determine if the jack and lap prococa by itself could introduco notpoint error. The setpoint was found to be 2.20 high, indicating that the jack and lap process needed better controls.
On May 12, 1991, another Catawba Unit.2 Pressurizer Safoty Valve was tested at
'Wyle and found to have a sotpoint of 2624 ps3g (5.6% high). This valve had been sont to(Wyle in Fall, 1990, and also had not been installed between performing the jack and lap procodcro and determining the setpoint, at Wyle.
Between May 16 and May 21,-1991, the three Unit-1 PSVs were reinstalled.
Several Wyle and Dresser enhancements were.in place when those valves were at the test iacility:
valves that are leaking during or af ter steam setpoint verification testing.must be repaired and rotentod on steam setpoint verification rotest is required after the jack and lap process
+
tighter sotpoint trending controls are imposed a
4
. ~..
--i-......
m.,. -__
-,.-,._.,---__.,_-_,,-.--m._,,__,~.
..m.-
. - r,
DUKE p0WER COMPANY / CATAWBA HUC1 EAR STATION pjR 1-C91-0199/Special Report pagn 4
\\
no' ring adjustments are allowed after the final setpoint verification test the leakago acceptanco criterion is no fogging at 93% of nmueplate set pressuro
.the jack and lap proceduro requires reestablishment of the body / bonnet
+
gap within 1.0035 inches of Au-Found the method of measuring ring sottings is controlled so that tho a
frequency _and magnitudo of ring sotting adjustments will be reduced
(
These valves woro verified to be within 4/-1% of setpoint prior to installation.
CONCLUSION This incident is attributed to an unknown causo, due to a possible procedure j
deficiency.- The most probable cause of high sotpoints in adjusting a valvo with_a known-seat leak, then repairing the leak using the jack and lap process 3
without follow up steam testing.
It appears that a number of factors could be contributing to setpoint drift, involving the test procedures and practicos utilized by Wylo and Dresser, and. approved by Duke. The major proceduro enhancements Duke is pursuing are:
- 1) oliminating the jack and lap and gaseous nitrogen leak test as a final-step after a steam setpoint verification test, 2) requiring absolute leak tightness after any valid steam set pressure i
verification - no fogging allowed at 93% of set pressure, 3) tightening controls over setpoint trending for the three steam setpoint verification pops, and 4) _ eliminating ring adjustments before An-Found tests, or after a valid steam sotpoint verification test.-
Some of the most probable causos of the pSV sotpoint drift are valvo leakago during the steam setpoint test (prosaurizing the valvo's huddlo chamber and causing an apparent lower setpoint, after which the compression screw would be adjusted, resulting in an artificially high setpoint if the leak was later repaired _as is.done in the jack and lap process),' errors _ introduced during tho -
jack and lap valvo-_ disassembly process (due to difficulty in reassembling the valve exactly as.it was found), sotpoint-trending control (if a trend in setpoints is prosent, testing should continue until the trend " turns around" before a test is considered valid), and ring. adjustments being made after a setpoint test _ (those adjustments change valvo performance characteristics),
i Other possible causes include temperature effects during testing (affecting internal temperatures - the numbor'of actuations could also affect internal 3
temperatures), spring performance (springs may be affected by temperaturo, aging, relaxation during rebuild, or by varying friction factors between the
-spring and-spring washer and spindlo) seat adhesion effects (a perfectly-lapped seat, or seat corrosion, could contribute to a'high As-Found sotpoint),
r l
>.,.. ~, -.. _.,
-_.-m__. _... - _. _ _.._.....___..__._,_._,,_..-._.___..__..._....._._._.....~........:.._.-
DUKE POWER COMPANY /CATAWHA NUCLEAR STATION p1R l-C91+0199/Special Report paga 5 a
transportation / handling effects, and test process effects (steam quality, pressurization rates, valvo installation offecto, equipment calibration errors, and personnel errors could affect the test results at Wyle).
Or. Hay 20 and 21, 1991, a meeting was held at Wyle baboratories between Wyle, Drosser, and Duke personnel as a result of a concern in using the current test method at Wylo. Concerns were whether nuclear safety systems could operato within their pressure protection boundaries. High notpoint drifts found during the testing phase could be increaand by variables due to site conditiona.- A detailed discussion of seve ral items was held:
i Jack and Lap _ process controls - Dresser had already enhanced the process to ensure a precise body /bonnot gap, and to ensure that a proper-i alignment is maintained, Dresser is to investigate further procedure onhancements. A steam setpoint verification test will be required after the jack and lop _ process. Also, Dressor is to determine if any spindio j
stretch occurs during the jacking process, which could result in net point variance. Wyle is to supply Jack and Lap verification test results (on a Crosby valvo) to Duke, if available.
Temperature effects - Higher valve temperatures result in lower setpoints.
Dressor recommended that Duke install two thermocouples (120'F apart) to detect any uneven heating of the valvo body and bonnet.
The on-site temperature conditions of the pressurizer Safety Valves will be determined and input will be supplied to Wyle. Dressor in to-supply any test results of the effects of thermal variations on valve sotpoint to Duke. The minimum hold time between actuations will be increased at Wyle, to allow improved temperaturo stabilization. Also, Duke will ensure:that Wyle~ uses _two thermocouplesLon each flange to ensure even heating. Duke has conducted a preliminary investigation of opring versus flange temperature, and found that the variancon were minimal.
Dresser will further investigate the affect of temperature on upring performanco. The effect of temperature on setpoint will be better understood when the results of this invostigation are available.
Setpoint trending control -.Wyle procedures will require all throo setpoint verification pops to be within-10 pai if trending occurn with no turnaround.
Transportation / Handling Effects _- It was concluded that-transportation-and handling aspects wero_not of sufficient concern to warrant an immodlate investigation. Dresser cited cases in which valves had been dropped or returned after many years,-and still had as-found cetpointa within 41/-14.
l L
l
DUKE POWER COMPAMY/ CATAWBA NUCLEAR STATION PIR 1-C91-0199/Special Report l' age 6 Seat Leakage Effects - Seat leakage occurring during the steam set pressure verification test can cause an error when the leaking valve is cepaired (Jack and Lapped), without a follow-t.p steam test.
Valves leaking during or after the steam setpoint verification test are to be repaired and tested on steam. Wyle is to investigate a means to document the amount of seat leakage on steam, using descriptive words (such as zero leakage, fogging, droplets,. visible, or audible). Duke will review the present Wyle Test Procedure 1028, Rev. B, and supply recommended revisions to Wyle, and Wyle will incorporate Duke comments.
Spring' Performance - The spring manufacturer confirmed that the Catawba-and Oconee Pressurizer Safety valve springs are not stressed to the point that a " set" is developed.
Dresser is to investigate the need for design modifications or maintenance procedure enhancements to ensure that valve spring / spring washer sliding surfaces are not affecting valve
-performance due to increased friction factors due to corrosion or lubricant degradation. Dresser in to test the spring removed from Catawba valve S/H Bs02871, including verification of spring free length, tilt under load, spring rato, and repeatability and performance at various temperatures.
Seat Adhesion Effects - According to Dresser, a perfectly lapped seat can provide enouf surface tension to affect valve setpoint.
This 10 prevented by the 1. dishing compound and the cast iron lap used.
Corrosion has not been a problem with materials used by Dresser.
Riug Setting Effects - While ring settings affect valve performance,
-Electric power Research Institute (EPRI) test results show that almost 90% of Dresser's vt.lves achieve full lift _without overpressurization (accumulation). tlistorically, ring setting variations have resulted in As-Left cettings not agreeing with'As-Found settings due to past methods of measuring internal dimensions of disassembled valve parts and calculating the ring positions.
This'resulted in a stack-up of-tolerances. On March 7, 1990, Dresser modified their procedure to achieve much greater consistency.
Wyle is to revise their testing l
procedure to prohibit ring position changes after the setpoint verification test.
Dresser has implemented changes to better control the method of measuring ring' settings.
Test Process Effects - Pressurization rates can affect lift setpoint.
Dresser is to determine if the pressurization rate of 100 to 200 psi /sec (150 to 200 psi /sec is normal)_is appropriate. Also, Wyle is to
- ~'
' implement procedure changes to place tighter controls over the trending j
of successive valve actuations.
C -Test Equipment Effects - Calibration of all equipment.is under Wyle's Quality Assurance. program, traceable to the National Bureau of Standards.
DUKE POWER COMPANY / CATAWBA NUCLEAR STATION l
p1R 1-C91-0199/Special Report Page 7 l
Valve Reassently Effects - Relaxation of a spring after valvo disassembly (at Wyle), for an extended period of time, requires restabilization of the spring when it is recompressed and setpoint adjusted.
Valve Installation Effects - Wyle does not une a specific torquing procedure.
A concern is that valvo inlet flange loading may be uneven, possibly distorting the valve alignment. Wyle is to modif y the testing procedure to ensure even loading on the inlet flange of the valve when installing on the test header.
other Miscellaneous items - Other items included the effects of multiple a
safety valve actuations on the temperature of upper valve internals, and the issuance of three information notices on ring adjustments by Dresser.
pSV setpoint drift is a recurring problem at Catawba (reference p1Rs 1-C90-0075 and 2-C90-0248), and an industry wide concern.
Duke is pursuing enhancements to the repair and testing procedures and practicos at the Wyle Laboratories testing facility. Also, Duke plans to investigate safety valve setpoint repeatability under the new enhancements by retecting two valves, two months after the initial test.
Future discussions among Wyle, Dresser and Duke personnel are planned. Duke will continue to provide technical cupport toward the implementation of these enhancements.
This report is being submitted as a Special Report, since the time at which the uetpoints drifted out of the T/s allowable range is unknown, and no specific on-site condition resulting in these setpoint drifts has been identified.
CORRECTIVE ACTION SUBSEQUENT 1)
Further PSV testing was performed at Wyle Laboratories.
2)
A meeting was held between Duke, Wyle and Dresser personnel at Wyle Laboratories in which several factors were discussed in detail and action items were developed to deal with setpoint drift.
Several possible cauces of setpoint drift were identified.
3)
Three pSVs were reinstalled on the Unit 1 pressuriner.
PLANNED 1)
Duke will continue to prcvide technical support toward the implementation of the previously mentioned vendor /manuf acturcir enhancements.
4 DUKE POWER COMPANY / CATAWBA NUCLEAR STATION pIR 1+C91-0199/Special Repor*.
Page B 2)
The on-site temperature conditions of the PSVs will be determined, and results will be forwarded to Wyle baboratories.
SAFETY ANALYSIS Design Engineering has analyzed the effects of increased safety valve setpoint drift. The PSVs are designed to prevent NC System overpressurization during
-transients and accidents. Limiting transients involve a mismatch between primary heat source and secondary heat sink, resulting in a rapid Pressurizer insurgo. Limiting transients include uncontrolled rod withdrawal, turbino (EIISITRB) trip, and rod ejection. The impact of a higher PSV lift setpoint would be to decrease the margins between peak NC pressuro and the limit. Thr) recent Unit 1 Cycle 5 and current Unit 2 Cycle 4 were evaluated to assess their conditions with respect to possible PSV settings.
l Unit 1, Cycle 5
]
The following parameters were usedt two PSVs with +6% drift, and one with no j
drift,_along with 2% accumulation; conservative initial and boundary-conditions for plant-systems; Reactor physica parameters bounding Unit 1, Cycle 5.
It was found that the peak pressure for either the turbine trip or uncontrolled rod withdrawal transient, under these parameters, was less than 2750 psia.
Since the rod. ejection is a condition-IV ovent, the acceptance l
criterion is 3000 psi. The PSVs would be full open at 2700 psia. The impact of a 46% setpoint drfit on the rod ejection accident peak pressura margin is insignificant._ The consequences of a complete loss of feedwater event are typically evaluated by assessing the capability of the plant to maintain core cooling by feed-and-bleed. This capability is shown in a Catawba-design type
-of plant by using the Pressurizer Power Operated Relief Valves (PORVs) for the blood path. The Pressurizer code safety-valves are not relied on for this i
function.
There is no impact on feed and bleed capability (which normally credits the 3 Pressurizer PORVs), due to__PSV drift if the PORVs_are available for this beyond-design-basis event (the situation with no PORVs is low in probability). For the Anticipated Transient Without Scram (ATWS) transient, the PSVs would go full open (similar to rod ejection); therefore, there would
-be no impact due.to the setpoint drift.
t Unit 2, Cycle 4 The following parameters were used:
three PSVs at 461 drift, with 2%
4 accumulation; conservativa initial and boundary conditions for plant systems; Reactor physica parameters bounding Cycle 4 for the current core burnup.
It was found that the peak pressure for either the turbine trip or uncontrolled rod withdrawal transienti under these conditions, was less than 2750 paia.
l' The conclusions regarding rod ejection, food and bleed capability, and the ATWS transient described in the Unit 1, Cycle 5 analysis are also applicatie
-for. Unit 2, Cycle _4_.
i.
I l
_______._m
.m.
DUKE POMER COMPANY / CATAWBA NUCLEAR DTATION PlR l-C91-0199/Special Report Page 9 As a result of the Unit 1, Cycle 5 analysis it was concluded that the impact of the as-found setpoints did not present a situation where the overpressuro limit could have been violated if the limiting design basis transient had occurred. As a result of the Unit 2, Cycle 4 analyuis, assuming +fA drif t for all three PSVs, it was concluded that the situation does not present the potential for exceeding the overprossure limit for the remainder of the fuel cycle.
Design Engineering aleo provided the following:
Analycos applicable to both McGuiro and Catawba Huclear Stations have been
- performed to provido juntification for increasing the Technical Specification drtit allowance from 41/-lt to +3/-2%.
A Technical Specification propoced revision will bo prepared for Catawba, based on those analyses.
The reculta-of-these analyses show-that the-43/-2% drift allowance can be accommodated based on success in meeting the overpressure acceptance criteria for all l
affected FSAR Chaptor 15 transients. The analynos of 43/-2 sotpoint drift aro l
applicable for the start of the present Unit 1 cycle through its next refueling outage.
Tho 43/-2% analysis satisfies the planned actions addressing a +/-3% Safety Evaluation in the Special Report submitted on April 4,1990 (PIR l-C90-0075).
This previous report was submitted with 10CFR Part 21 reportability concerns.
Baned on the results of the current investigation, it does not appear that a Part 21 reportability concern exists.
The health and safety of the public were not affected by this incident.
l b
e we<ry,--
-c-,e
,,,,..wwwm w.,
.....,..,.-mm--mm.m.,_-.-__4,-
m*
- - - _.,_.m-m--,,--- -, - -- _
-...--..,-n-.~.._
.ew.m
-.-.ms--.
. - -....