ML20135B355
| ML20135B355 | |
| Person / Time | |
|---|---|
| Site: | Comanche Peak  |
| Issue date: | 01/08/1997 |
| From: | Poole A OAK RIDGE NATIONAL LABORATORY |
| To: | Jackson J NRC |
| Shared Package | |
| ML20135B349 | List: |
| References | |
| TAC-M94165, TAC-M94166, NUDOCS 9703030018 | |
| Download: ML20135B355 (2) | |
Text
_..
1 An Examination of Motor-Operated Valve Failures With Application i
to Increasing the Surveillance Testing Period at Texas Utilities Company Comanche Peak, Units 1 and 2 i
i i
Nyerd by l
Oak Ridge NationalImboratory 1
i Prepared for U.S. Nuclear Regulatory en-ms. lan i
4 i
}
t I
r c)90 30S 00 (E-XA,
=
Background===
In res nse to a request by Texas Utilities Comp my to extend the time interval between surveillance testin for certain motor-operated valves that br,ve been determined to be oflow safety significance, Oak Ridg National Laboratory has analyzed historic al failures of motor-operated valves at Comanche Peak, units 1 and 2. Failure data were obtamed from the Nuclear Plant Rehability Data System (NPRDS).
Each failure narrative was reviewed and the failure characterized at ORNL by the five parameters sLown in Table 1.
Limitations If the reader is to draw accurate and useful conclusions from the data presented it is important that the limitations of the data be clearly explained and understood. There are two limitations that should be explored.
Table 1 Coding parameters for failure characterization Parameter Description Corsponent The failure was assigned to one of three affected areas - the valve, actuator, or electrical mpport.
Problem There are three potential problem types addressed in the characterization.
Problems that caused a loss of operabilityl or functionality Problems that did not cause a loss of operability or functionality Leakage related problems.
Symptom This parameter refers to the observed condition that indicated component degradation has occurred. Ty)ical symptoms include failure to open, failure to close or close completely, eakage (seat or packing), or in the case of degradatica discovered during maintenance, no symptom at all.
Cause The information entered here identifies the immediate cause of the observed symptom.
Method of Detection The method of detection category identifies the activities in progress at the time of failure discovery. One of the following five activities was assigned to each failure.
Demand - failures that occurred when the component was called on to function.
Maintenance - failures that were discovered during preventive or corrective maintenance.
Observation - assigned to failures that were detected during walkdowns or other programmatic events established to detect visible system degradation. This is most effective for detecting leakage related problems.
Testing - assigned to failures that weie discovered as a result of surveillance, mservice, post maintenance, or any other testing.
Unknown - assigned to failures where the narrative did not give a clear indication of the activities that resulted in the detection of the failure.
I The term " operability," as used here is not a detennination of the ability of the component to perform a safety function. The use of the term is limited to describing a loss of ability to perfonn a function that is part of the manufacturers design. Again, there is no anempt to determine "opurability" from the licensing standpoint.
2
i
~
The data collected do not indicate the number cf actuations for the given components. Therefore t.
absolute failure rates, defined as successful actuations per attempted actuations, cannot be calculated.
Relative failure rates may be calculated when there a significant number of failures attributed to the i
parameter in question.
}
- b. Data that establish subcomponent replacements made during :reventive maintenance are not available. For motor operated valves this means that any com>ination of motor, gears, or switches could be replaced during preventive or corrective maintenance, thus refurbishing the component, i
without the component >emg designated as a new component. 'Iberefore, a component could be sebuilt to "like new" condition, but never be raahd.
i Based on the above it is easily understood why the component " age" is an indeterminate parameter.
faWon and test intervals, flow conditions, and the operating environment are all unknowns. Also, 4
since the Mean Time Between Failures (MTBF) is defined as t z "Mean time between successive '
i i
failures of a repairable product"2 and only two components have more than one failure noted, MTBF will not be calculated.
Analysis 1
Comanche Peak is a relatively new plant, and as expected, there wera very few failures for consideratioa. A search of the NPRDS data base identified a total of.'8 fa.ilurcs imm 1990 to 1995. This data is presented in Table 2.
'Ibe list of MOV failures was compared to the list of valves for which Texas Utilities was requesting an I
extension in the surveillance interval. Of the 38 valves that are identiSed in NPRDS as having failed,3 are on the extension list see Appendix A). Three failures is insufficient in size to establish meaningful i
trends or partems in the failure record. Therefore, the failure record for the plant (38 failures) instead of failurn of valves for which an extension has been requested, will form the basis for this work.
The data show that actuator failures comprise the largest group of failures ov.erall (Fig.1), and for five of i
the six years for which there are failure data (Fig. 2). In 1990 there were more valve failures than actuator failures. B=mination of the data indicate that all three of the valve failures were classified as extemalleakage.
1 l
Table 2 - MOV Failures by component area and year i
po(nt Component 1990,1991 1992 1993 1994 1995 Valve 3
3 2
1 1
0 10 Actuator 2
5 3
6 4
3 23 Electrical 0
4 0
0 1
0 5
Total by year 5
12 5
7 6
3 38 i
i l
l l
i h
2Quality Planning and Analysis, From Product Development through Use,2nd edition, p 174,1. M. Juran and F. M. Grynn.
}
Jr., McGraw-Hill Book Company,1980 3
i 4
Bectrical VaNo
" n.3
"!: s M
7
- h
)I Actt:sfor 41 %
Figure'l Distribution of failures by affected area 100 m.
E VWe 122 Actuator 80 -
7 gg Electrical
- 70 -
l 60 -
i
.i! 60 -
E 40 i
] 30 -
!!r j
l 20 -
f:i E!
l H
10-N E-:
E 4
i l
J:
o 0
a:::
1990 1991 1992 1993 1994 1995 Year of failure discovery i
Figure 2 Distribution of failures by affected area and year of failure discovery Two of the three leaks were at the body to bonnet joint. The other leak was the result of a casting defect in the valve body. It is reasonable to state then that the sigmficant failures (failures that compromised the ability of the actuator to perform its designated function) for 1990 were the result of actuator degradation.
The number of failures per year is relatively consistent with the exception of 1991. Further examination of the data indicate four of the failures are the result of a blown fuse m the circuit for one actuator. One of the four failure narratives indicated that operations y replaced the fuse without the benefit of maintenance action to determine the cause of the de on. Since the problem disappeared after maintenance action we can reasonably assume that of the four blown fuses could have been avoided by proper personnel action. Dismissing the three leakage related failures as not significant, and excluding three of the four blown fuses as failures that are related to personnel error and not degradation of the equipment results in six failures for 1991. This is more consistent with the number of failures for the other years.
4
- -..-- - - - - - -- - - - - - - - ~ ~ ~ ' ' ~ ~ ~
The data in Table 3 show the distribution of failures by symptom and year of failure. The symptoms, from a macroscopic viewpoint may 'os educed from nine to four groups to allow easier graphic
- i*
mpresentati:n Fig. 3.
l i
Table 3-Fauums by symptom sad year Symptom 1990 1991 1592 1993 1994 1995 Total Failure to open or close 2
1 3
1 1
1 9
l Failum to open or.: lose completely 0
2 1
1 2
2 8
Failure to opee 0
5 0
0 0
0 5
Externalleakage 3
2 0
0 1
0 6
Internalleakage 0
0 1
0 0
0 1
Packingleak 0
1 0
0 0
0 1
Limit switch failed to trip 0
0 0
2 0
0 2
Motor burnout 0
0 0
2 2
0 4
i Nee 0
1 0
1 0
0 2
j Total by year 5
12 5
7 6
3 38 i
i The first three can be grouped together and considered a " degraded stroke." All leakage related events
}
are combined to make the seconc category, while limit switch and motor problems are considered i
degraded subcom l
inc ividual class. ponents. 'Ihose failures that displayed no symptom prior to discovery are left as an i
10 D g aded sin *e 8-7 Ets Dwed*d W e
6-t 5-4-
n, 3-ll i
E i'.
_Il B
ll
!!l 0
1990 1991' 1992 1993 1994 1995 Year of feture *: :.sy Figure 3. ILuibution of failures by symptom and year It can be readily seen that failures that affect the ability of the actuator to complete a full stroke of the valve comprise the largest category for each year except 1990 and 1993. Again,60% of the valve failures for that year are leakage related, and not considered significant. It is a noteworthy obsemdon that almost 55% (12/22) of the " degraded stroke" failures, which make up the largest single category of failures, are directly related to either the failure or miWjustment of torque or limit switc us.
An eummination of the data by failure cause and year of failure do not reveal any single failure cause as dominant in any particular year (Table 4). 'Ibe 4 failures due to a blown fuse in 1991 were previously discussed and it was determined that 3 of the 4 failures were not attributable to equipment degradation.
Additionally, none of the 3 failures with where the cause is listed as unknown involved the same sut-:+=5+===ts, symptom of failure, or any other parameter that would relate them.
5
Table 4 Fauures by cause and year I"*
Cause of FaDum 1990 1991 1992 1993 1994 1995 TM 1
Blown fuse 0
4 0
0 0
0 4
i C==*erloose or damaged 0
0 0
1 0
0 1
Damaged pinion key 0
1 0
0 0
0 1
)
Degraded packing 0
0 0
1 0
0 1
Dirty contact 0
0 1
0 0
0 1
)
I-ymys assembly 3
0 1
0 0
0 4
Iscsiat TS setting 0
1 0
2 0
1 4
1 LS failure 0
0 0
2 0
0 2
i LS setpoint shift 1
0 1
1 2
1 6
i Mfg. defect 0
1 0
0 1
0 2
i Normal wear / aging 0
1 0
0 0
0 1
i Previous Mainten-*
0 1
0 0
0 0
1 Previous maintenance 0
0 0
0 1
0 1
Seat / disc wear 0
0 1
0 0
0 1
l Single phase in MCC 0
0 0
0 2
0 2
TS setpoint shift 1
0 0
0 0
0 1
i unknown 0
3 1
0 0
1 5
j Total by year 5
12 5
7 6
3 38 The comparison of symptom and method of detection (Table 5) indicate that 88% of the failures detected i
in demand situations are related to either a complete failure of, or a degraded stroke. For these categories (failure to open/close, failure to open/close completely, or failure to
) demand failures account for 68% of the total failures for the these categones, and 39% of the t failures for the plant.
Table 5 Symptom of fauure and inethod of detection Method of Detection Symptom Demand Malatenance Observation Testing Unknown Total i
External Leakage 0
1 5
0 0
6 i
FC 4
0 0
2 0
6 J
FCC 4
1 0
2 0
7 l
FO 2
0 0
1 0
3
{
FOC 04 0
0 0
1 1
F@
5 0
0 0
0 5
Intemal Leakage 0
0 0
1 0
1
[
LS fall to trip 0
0 0
2 0
2 i
Motor burnout 2
0 0
2 0
4 None O
2 0
0 0
2 1
1 0
0 0
1 Total by method of detection 17 5
5 10' 1
JR FO Failure to open FCC Failure to close completely 1
FOC Failure to open completely Pop Failure to operate FC Failure to close a
A tabulation of cause of failure vs. method of detection shows that the limit and torque switches account for approximately 45% of the failures. Sixteen of these failures are highlighted in Table 6. The other failure involving a switch had an unknown cause. Note that no single cause accounts for more than 16%
6
- ~ ^
cf tluffailuresh71dle smand failures account for the largrst grou cf failures when method cf detection, no single cause accounts for more than 24% cf the demand failures. grouped by i
o, Table 7 shows the data tabulated by cause and symptom cf failure. Again, no single pair cf parameters constitutes a significant fraction of the failures.
i Table 6 Cause of falhare and method of detection i
Method of Detection Cause Demand Maintenance Observah T==*ia= U=k= awn Total 2
i Blown fuse 4
0 0
0 0
4 cannectorloose or da==M 0
0 0
1 0
1 Damaged pinion key 0
1 0
0 0
1 Degraded packing 0
1 0
0 0
1 Dirty contact 1
0 0
0 0
1 Improper assembly 1
1 2
0 0
4 Incorrect TS setting 0
0 0
3 1
4 LS failure 2
0 0
0 0
2 LS setpoint shift 4
0 0
2 0
6 Mfg. defect 0
0 2
0 0
2 Normal wear / aging 0
1 0
0 0
1 Previous Maintenance 0
0 1
1 0
2 Seat / disc wear 0
0 0
1 0
1 Single phase in MCC 1
0 0
1 0
2 TS setpoint shift 1
0 0
0 0
1 unknown 3
1 0
1 0
5 Total by method of detection 17 5
5 10 1
38 Table 7 Symptom and cause of failure Symptom
~
Cause IAakage FO/FC FCC/FOC fop LS Motor None Total Fall Burnout Blown fuse 0
0 0
4 0
0 0
4 Connectorloose or damaged 4
0 1
0 0
0 0
1 Damaged pinion key 0
0 0
0 0
0 1
1 Degraded packing 0
0 0
0 0
0 1
1 Dirty contact 0
1 0
0 0
0 0
1 Improper assembly 3
1 0
0 0
0 0
4 Incorrect TS setting 0
0 2
0 2
0 0
4 LS failure 0
0 0
0 0
2 0
2 LS setpoint shift 0
3 3
0.
0 0
0 6
Mfg. defect 2
0 0
0 0
0 0
2
' Normal wear / aging 1
0 0
0 0
0 0
1 Previous Maintenance 1
0 0
0 0
1 0
2 Seat / disc wear 1
0 0
0 0
0 0
1 Single phasein MCC 0
1 0
0 0
1 0
2 TS setpoint shift 0
1 0
0 0
0 0
1 unknown 0
2 2
1 0
0 0
5 Total by symptom 8
9 8
5 2
4 2
38 7
, * "' ' Although the above aummination was rather cursory in nature, ample evidence exists to questi:n the eachnical validity cf extending the inspection interval for the requested valves. The de[;ree of wear displayed at these units does not support extending the inspection interval without furt wr analysis of the failures, their causes, and actions implemented to prevent recunence.
hamination of the data indicate that these is insufficient information available to conclude that th any aging related trends for motor o wrated valves at Ca== ache Peak. It should be noted that valve components in contact with the fluk may not see aging related failures without additional service wear.
j 11 may be possible to draw some valid conclusions about the long teim iwrfonnance of these valves by carefully coruyaring data fmm other plants that involve valves with similar function, environment, use, test, and preventive maintenance practices.
a J
8
Appendix A s
} " I ' The f:llowing valves are listed on the extension list from C-=~he Peak and are also listed in the failure data as having failed beraw 1990 and 1995.
I i
Table A1 1
Unit Component ID r.tise of failure System 1
. FV4772-1 Ihnit switch adjustment Containment Spray e
i 1
FV4772-2 Packingleak Containment Spray I
HV-2492B Body / bonnet leak Amilian Feedwater 4
l Note that only FV-4772-1 had a failure that may be enneidared significant 4
h i
J e
9