ML20248E633
| ML20248E633 | |
| Person / Time | |
|---|---|
| Issue date: | 09/29/1989 |
| From: | Ellen Brown NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| To: | |
| Shared Package | |
| ML20248E632 | List: |
| References | |
| TASK-AE, TASK-T918 AEOD-T918, NUDOCS 8910050366 | |
| Download: ML20248E633 (5) | |
Text
,
4 U ;;]
w.
.~
~
af,w AEOD TECHNICAL REVIEW REPORT UNITS Multiple
'TR REPORT NO.: AE0D/T918 4
-00CKETlN0.:1' Multiple DATE: September 29, 1989
' LICENSEE:
Multiple EVALUATION / CONTACT: 2E. J. Brown
- NSSS/AE:
. Multiple-u
SUBJECT:
CHECK VALVE FAILURE RATES FROM NPRDS DATA iEVENT.DATE: ' Multiple.
SUMMARY
- The NPRDS procedure to determine failure rates for one component group.was.
- utilized.to develop check valve failure rates that were requested by ACRS.:.The failure. rate in failures per million hours and the failure rate trend in equa-ition form were developed-for three groups of check valves based on inlet size.
There.was little differences in failure rate (failures per million' hours) among the groups. The trend forL each group was a gradual reduction in failure rate -
~ The failure rate information is inconclusive relative to whether over time.
further studies or actions on check valves are warranted.
DISCUSSION:
During the 349th ACRS meeting on May 4, 1989, it was requested that AEOD provide
'the NPRDS date on failure rates of check valves used in nuclear power plants (Item 9, Ref. 1). The NPRDS procedure to determine failure rates for one
~
component' group for a specified. time frame was used to develop this information.
This report discusses various issues reviewed as part of the effort to establish' the data. The response to ACRS'was by memorandum cited as Ref. 2.
The NPRDS database for check valves includes all systems, environments, and valve sizes.
It excludes incipient failures. The-program provides the number of failures'for a selected group, the failure rate over the time frame, and a
. best estimate failure rate' trend in equation' form y=mx + b where m is the slope or rate of change in failure rate and y is the failure rate in number of failures per million hours. The program that was used developed failure rate as'a function ~of calendar time. Data from January 1984 through July 1988 was used. The search conditions were as follows:
Selected Component was VALVE, Selected VALVE type was CHECK, Selected VALVE inlet sizes were: (1) all sizes, (2) 2.0 inlet and greater, and (3) less than 2.0 inch inlet.
During the process of reviewing the data, there were several concerns raised about some aspects that might adversely impact failure trends. Some specific issues considered were:
- How many plants were reporting check valve failures?
- Were only a few plants reporting in a given report period?
What are the number of failure reports per plant?
How many utilities are reporting check valve failures?
What type and distribution of failure modes are reported?
G910050366 890929 PDR ORG NEXD
-a.
It was decided to review the NPRDS printout for each event reported for a few specific months in order to obtain information about these issues.
Four j
separate months were arbitrarily selected for review. The review identified R
the following:
p Sample Month-No. Plants Reporting /No. of Utilities 1
15 plants /13 utilities 2
19 plants /14 utilities 3
17 plants /15 utilities 4
12 plants /10 utilities In general it was found that, in any single month, most plants reported only 1 or 2 check valve failures. Only one plant reported three failures (this was the highest number) in any one of the four sample months reviewed. Over the four sample months reviewed, there were check valve failures reported from 52 plants involving 34 utilities. This information suggests that several plants are reporting a small number of failures per month. The average was 1.3 failures per plant (68 failures in the four months for 52 plants).
Each failure report was reviewed to assess whether the event involved small leakage or gross or severe leakage. A large fraction of the reports identified the failure as small leakage of the type detected by routine observation or surveillance.
Approximately fifteen percent (15%) of the reports could possibly have involved severe leakage. Reports placed in this category generally included a statement that the disc was separated, a key was missing, or the disc was stuck open.
The results of the NPRDS data search are presented in Table 1 and Figure 1.
Table 1 illustrates the check valve failure information for check valves grouped as:
(1) all sizes; (2) 2.0 inch inlet and greater; and (3) less than 2.0 inch inlet.
Even though there is a wide variation in the number of reported failures for each group (1944, 1601, and 343 respectively), there is little difference in failures permillionhoursforeachgroup(5.34,5.48,(and4.77).
In addition, the difference in. rate of change in failure rate slope m) is a small negative number between -0.02 and -0.06.
The check valve group based on the 2.0 inch inlet size and greater was selected to envelop the expected range of safety systems. The failure data covers a wide range of applications including water, steam, and air systems.
Figure 1 is a line plot of the failure data for check valves with inlet size 2.0 inches and greater. For perspective relative to check valve events, the failure of main feedwater check valves et San Onof re occurred in November 1985 while the Electric Power Research Institute (EPRI) report " Application Guidelines for Check Valves in Nuclear Power Plants," EPRI NP-5479, was issued in January 1988.
These two events would corre spend to 23 and 48 months respectively on Figure 1.
The data in Figure 1 exhibit relatively large variations in failure rate on a month to month basis but the long term trend is slowly downward.
It seems plausible that licensees may have increased check valve inspections after the San Onofre event (Month 23) or issuance of the EPRI Application Guidelines (Month 48).
If this was the case, it does not appear to have resulted in discovery and reporting of an increased number of check valve failures.
i f.
l l
' FINDIPGS 1.
There is little difference in failure rate (failures per million hours)
. for the check valve size groups reviewed for this' study.
2.
The failure rate exhibits a small downward slope with time.
3.
The review of failure reports for the four sample months illustrates that many plants are reporting failures. This suggests that failure rates would not be abnormally influenced by a single plant.
CONCLUSIONS This data was assenbled and r(viewed to provide the requested inforniation to the ACRS. This effort and results are not adequate to assess whether the observed check' valve failure rate.is benign or warrants further review.
REFERENCES 1.
U.S. Nuclear Regulatory Commission, Memorandum from R.F. Fraley to V. Stello, Jr. "349th ACRS Meeting Follow-up Items," May 23, 1989.
2.
U.S. Nuclear Regulatory Commission, flemorandum f rom V. Stello, Jr. to R.F.
Fraley, " Response to 349th ACRS Meeting Follow-up Items, July 6,1989.
~
l i*
l TABLE 1 1
CHECK VALVE FAILURES FROM JANUARY 1984 THROUGH JULY 1988 Check Valve Number of Failure Rate Best Estimate Size Failure Per Million Hours Failure Rate Trend All sizes 1944 5.34 y = -0.05x + 7.03 2.0 inch inlet 1601 5.48 y = -0.06x + 7.36 and greater less than 2.0 343 4.77 y = -0.02x + 5.61 inch inlet i
4 5
8 i 4 SE i 4 2
TA R
E k
i 6 38 8 R
9 1
U ly u
L i 0J I
(
I 3-A 4
F
\\
8 9
1 1
E E
n U
i 4a R
2J V
s:
G I
L F
h t
A no V
1 i 8 M d
K C
s r
Y u
E i 1 o
2 H
H n
o C
lil i
M e
i 6 r
p s
i e
r lu a
F 0
0 9
8 6
4 3
2 1
1 l!
'