ML20154D367
| ML20154D367 | |
| Person / Time | |
|---|---|
| Issue date: | 05/04/1988 |
| From: | Israel S NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| To: | Rosenthal J NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| References | |
| NUDOCS 8805190161 | |
| Download: ML20154D367 (5) | |
Text
.
i MAY 0 41909 MEMORAh00F FCP:
Jack F. Fcsenthal, Chief Distrit,ution:
Feacter Crerations Analysis Branch PDP '
l Divisich of Safcty Prcgrers, /E00 POAB R/F
$1srael AE00/T805 i
THRU:
Peter Lam, Chief PLem Feactor Systen $.
- 1,,,.:n : =ua. w :: m t e. %.,n rye---c ~;e sygsDJ r~.>*--.y:
s..
a nesgys n e 2i::Lua::5 w\\. ;;t 2zna s o + c' = Q* *j';.;j., ww :-% a%-
msg;M3?.i, we memm.
9fi y,r. h,w 2.r:a. };W694;Mhyagyqzi
. c a-
,f,l 3j" PERSONNEL ERRORS EQUIPMENT FAILURE
-\\
MAINTENANCE ERRORS ADMINISTRATIVE ERRORS
1 l
Exainination of the LERs in this category indicates that about half were specifically captured by licensee progransnatic reviews or follow-ups to other events.
Because serendipity plays such a large role in discovering these "hidden" deficiencies, the licensee must maintain constant vigilance and probing to ferret out latent problems that may not otherwise be detected.
Another large fraction ( one-third) of the DFl events were identified as the l
result of external stimulus, e.g., vendor reports, NRC notices and bulletins, and NRC inspectors. Many of these discovery mechar. isms are mandated by i
license recuirements and nonnat ongoing regulatory activities.
Included in this category are safety system function inspections which are indepth probes of selected systems at selected plants.
Finally, about 20 percent of the flaws were discovered by the licensee investigation of related events--either equipment failures or maintenance activities.
Topics associated with about 50 percent of the DFI events are reflected in previous AE00 studies, hRC information notices, etc., while about 15 percent I
are potential topics for new AEOD studies.
One would expect that most of the significant events would have been noted previously.
In a mature industry, older issues should have been resolved, especially those involved with design oversights. Because of changes in the AE00 screening, the trend in the number of events in this category is indeterminant although this was also a major category in the 1986 events.
The remaining causal factors noted in Figure 1 were almost equally divided.
Uncertainties introduced by the screening process blur the relative contribu-tions of each of them. Maintenance has received considerable industry attention recently and valve problems are a major contribution to this category. The high interest in valves may bias the process so that everyone is sensitive to valve events.
Potential comon mode failures are a major concern with maintenance and one significant event in this categcry dealt with a potential comon ' node condition--blockage of a drain (inadecuate maintenance) which could cause water backup and damage to redundant systems. Another event was particularly noteworthy--olugged strainers in a diesel generator fuel line is a potential coninon mode situation under different circumstances or different plant design.
Thus, equipment failures provide valuable insights to plant improvements that reliability assurance programs can harvest.
Personnel errors consisted of improper component tagout during test and mainten-i ance, improper sampling, improper steam generator tube plugging, etc. --
situations where the an operator was an active participant in the plant evolu-i tions. One event, the undetected loss of 20 percent of the primary system inventory was similar to the Three Mile Island accident in nature, not outcome.
This suggests that old lessons / issues should be revisited to minimize repeating significant events. The importance of this category may be understateo because the reporting process may not be sensitive to rarginal operator performance until it exceeds a very high threshold. Thus, important precursor events may not be observable.
pandom equipment failure is to be expected.
Pipe cracks, circuit card failures, and loss of offsite power are significant, but unavoidable events.
i
- Admibistrative errors such as missed surveillance, long-tenn misalionment of I
systems, poor procecures, mis-coordination among personnel, QA problems, etc.,
I reflect poor supervision / management. Mistakes can be minimized but not avoided, so adequate independent verification is imperative to operating a safe plant. The sr.nl1 number of events in this category provides some confi-dence that should be tempered by the possibility that poor working habits involved in benign events may be overlooked or go unreported.
Taking a broad overview of these assembled events, ele *ctrical and mechanical problems are about evenly representud as are component versus system level i
l issues. Three subcategories stuck out, each with about 10 percent of the events: valve problems, equipaent qualification problems, and ac power problems (which are station blackout issues).
CONCLUSION l
The distribution of the events in:- ' W erent subcategories does not suggest a specific area to concentrate (v. s affort to improve plant safety, except j
possibly the design, fabrication, y anstallation area. Despite the uncer-e tainties in the relative importana of the different causal factors shown in Figure 1. DFI is a dominant contributor.
Since these problems may go undetected j
by the normal plant routines unless a particularly severe event challenges a safety component / system, it is important to maintain prograrmatic reviews such as safety system function inspections and feedback programt.pnsored by the t
vendors, INPO, and NRC. The diffuse nature of the problems rWres careful screening and sensitivity to capture important precursors ai.'
'- dback the insights to the operating plants.
\\
l t