ML18045A836
| ML18045A836 | |
| Person / Time | |
|---|---|
| Site: | Palisades  |
| Issue date: | 10/28/1980 |
| From: | Hoffman D CONSUMERS ENERGY CO. (FORMERLY CONSUMERS POWER CO.) |
| To: | Crutchfield D Office of Nuclear Reactor Regulation |
| References | |
| NUDOCS 8010310353 | |
| Download: ML18045A836 (4) | |
Text
consumers Power company General Offlcee: 212 West Michigan Avenue, Jackson, Michigan 49201 * (617) 788:.0660 October 28, 1980 Director, Nuclear Reactor Regulation Att Mr Dennis M Crutchfield, Chief Operating Reactors Branch No. 5 US NuclearRegulatory Commission Washington, DC ~0555 DOCKET 50-255 - LICENSE.DPR PALISADES PLANT - RESPONSE TO LOSS OF OFF-SITE*POWER SURVEY
. On Janua_ry 31 ~ 1980, Consumers Power Company responded.to an NRC survey questionnaire dated September 25, 1979, identifying that. between Ji.ine 6, 1971 and July 22, 1978, the Palisades Plant experienced94 partial losses of off-site power and four total losses of off-site power.
Thes.e losses were thert categorized into. generic causes. (Table 3 of June 19, 1980 *letter -from NRC to CP Co);. Palisades was.then averaged in with 48 other plants and each cause was given an indust:ry average which.fa described as target. failure rates.
The statistical analysis then shows that Palisades had* *five causes in which the failure rate exceeded the target value (target violations).
The Loss of Off-Site Power ~ Survey Status Report dated June 19; 1980, was prepared by Raymond Scholl,.Jr from the Systematic Evaluation Progr~ "9ranch of *the Division of Licensing.
Mr R Scholl was contacted by telephone on August 20, 1980 by Mr M R Wade and Mr S R Frost of CP. co* and discussed his steps in data reduction and analysis' and his conclusions and recommendations.
This. analysis showed Palisades had five different power.loss causes where the failure rate exceeded the target value.
As pointed out by Mr R Scholl, Consumers Power Company responded to the September 25, 1979 questionnaire in a more thorough manner than did most of the other respondents. *Our responses were.very comprehensive and, therefore, resulted in more events being reported.
Inasmuch as we were compared with the other respondents, we naturally had. some above-average event. r_ates.
Another cause of the high event rate was the fact that Palisades presently has.six transmission circuits as opposed to only two or three for most of the other plants.
This naturally results in more*loss of circuit events being reported.
The statistical analysis failed-to give credit for extra transmissiou circuits.. Mr R Scholi admitted that: this was a major.shortcoming in his results an'd indicated.that his program required somemodificatiorts in this../JolS area.
s 1/0 80 1oa 1.o.353-
-~
.,-JC Mr D M CJ:'.utchfield~l;i.ief Palisades Plant October 28, 1980 2
Before analyzing the individual target violations, we have several.
recommendations for improving the survey and its results:
- It would *be very appropriate to base.. fault data on a per circuit mile per year statistic rather than total number of faults.
These statistics would become ~uch *more * ~eaningful.
- A short li.st of caus*es should be used (ie, faults *-.line, faults - equipment
- failure, protective* relaying failure, construction errors, othe~). This' list should be provided to the utilities and they should be responsible for assigning causes to their event's... This would help in' the comparison of events hetwe~n utilities; Subsets could be assigned to the bc;sie lis't to
- further separate type of faults* or causes of f1mlts if desired; Some thought should. be giveI1. to.establishing a minimum 'time threshold for reoccurring faults so that they could'be combined int9o~e'long event" instead of multiple short events~*
Calise.4 is entirely. attribut~d to human error factors except circuit failures.
Circuit *failures should be incl.uded in* a different category.
A breakdown of the number. of everits per specific..,,.calis~ i;;hows there were 31
- events resul tin:g *from Caus.e 4, circuit breaker trips* and human errors*.
- . Several ft;tctors need to be looked at when.evaluating this number:
- *Twelve. even~s resulted _from a los.s of carrier *sfgrial **for a fault on the I&M
- ** '.?45 kV.system; *.cause 4.*is human error only.*
- Loss of. carrier signal i~ not human error;
- A separate category. is needed for los~ of carrier signal.
- Nineteen events on six transmission circuits would be the same as* only six events. ori two transmissicm cii~uits.
The artalys is shows that eight events resulted from Cau~e 5, *ground.fault.
A review of these events revealed numerous. errors. in~:.. the data.reductfori and
- Failu_re Map printout (Table 4. of _the* June 19, *.* 1980 Survey Report): * *
- Consum'ers Power Company sU:binftted the completeciquestiorinaireshowing 94 part.:i.al *.circuit lpsses.. The Fal.lure Map shows a total.of 96 ~ * 'Other diSciepancies in' the :Fai).ure. Map include:
(a) 'five eventS.. in Gause. o, LOP.-A.
. printou,t !is.ts.only two.events; '(b) eight events in Gause 5; LOP-A printout;
- *.(June.19, 1980 -Survey R~port) only has seven* events;. (c) eleven events in**.
Cause 15, LOP-A printout has -12.
That makes four errors in the. trans lat ion
.of* da.ta from one table to _the next by the computer.. If the program is unable to.simply transfer numbers f.rom one location.to another, the entire program.may have errbrs*and.all of the results aie void.
- As shown above~* there are only seven e'\\(imts.resulting* from Cause 5:
A closer examination shows* two of these event.s (1/11/74 and 6/14/77) were reported as unkno~ *causes. (Cau~e O}.and one event (6/28/77) was reported as*
caused by lightning; Cause. 22: *In this case,.three* out of seven events were incorrectly categori::zed;. therefore,.*there are otjly.four. events instead bf seven:.
Mr D M Crutchfield, Chief 3
Palisades Plant October 28, 1980 Four events on six transmission lines would. be the same as those plants that have only one event on two transmission lines.
The analysis shows 13 events resul~ing from Cause 11, lightning, with redundant lines out of service.
Here is another obvious error in data reduction.
LOP-A printout lists Palisades as having six transmission circuits throughout the entire reporting period.
In reality, only four circuits existed prior to the summer of 1973.
Our questionnaire response lists only four circuits in service.
This data was mistakenly translated to mean that_ two circuits were out.of service.
These 13 events. should be transferred to Cause 22, lightning.
- Adding the two categories together results in 28 lightning r*elated events.
Again,*- having multiple lin.es tend to reduce, the magnitude of this number.
The transmission system is also in a very lightning prone location.
Storms
.traveling east over La~e Michigan do not have-any points to discharge on until they* reach the shore where many o.f *our lines are_ located..
The analysis shows two events resulting from Cause 14, overload.
- In this case of data redtictien, Mr R Scho.11 assumed that when an instantaneous elem.e11t overreached,. the circuit was. overloaded~ These events were actually caused-by a type Of relay WhiCh measures the_ Ctirr.ent fo a transmission iin~ to :detect* ground* faults*.. *on Mai-ch 29, 1974, *:.several lines were out of service which caused the relay to overreach for this* multiple
. contingency condition and trip th.e circuit breaker.
These. events should be trans'ferred -to Cause 5.
Consum~rs Power reported that on March 29, 1974, this overreac.h trip*.
occurred_ three tiines..
The descriptions in our submittals.were identical but one of.these events was credited to Cause 15, winter storm, instead of Cause 14..,
It is important to note that at no* time has.. any transmission line tripped due to_overload,
- Finally, the_ analysis-shows e_ight events resul tirig. from *cause *29 ~
- construction.
It should be pointed.out that our submittal-described six of these events as having unknown causes and stated that construction work was being performed in the switchyard.
One-event occurred on January 13, 1972 and the other five occurred on January.14, 1912...
No-construction related outages have occurred since-August 11, 1973.
Looking-at the dates of -these-events, we do not believe that any detectable trends exist.
Again;. it should be noted that Palisades had four transmission lines in s.ervice at the time. compared to on'ly two lines for many of the other plants.
)
.:;.i Mr D M Crutchfield.lief Palisades Plant October 28, 1980 4
- It is statistically.correct to include construction as a 'cause but it is not correct to compare a plant which has experienced extensive construction to plants without any co11Struction..
In summary, during our phone conversation, August 20, 1980, it was revealed that this survey is intended to produce two basic results: *(a) validation of the Rasmussen Report (WASH-1400) with regard to probabalistic failure rates
' and (b) identification of activities or equipment which need to be modified...
The first attempt to.validate.the Rasmussen Report shows the report to be in~orrect*by a factor of 10.
Based on the errors made for Palisades arid the NRC staff's inability to correctly program the computer to transfer numbers properly, we.suggest that a real good look be taken*at.what was done and then have an independent reviewer check the work before submitting.any results about the Rasmussen Report.
We do not feel.that Palisades exhibits any* abnormal trends for any of the causes that were identified.
The fact that Palisades and a few other plants have more than* two or three transmission circuits should be factored into*the analysis. *.If this is done, the number of target violations will be reduced considerably or at least distributed more evenly among the plants.*
To evaluate the need for Palisades to make modifications; two factors are importan:t.
First,.there.are no correctable trends to modify Un.less a method of. subduing lightning can be found.. Second~ our suhmi ttal. shows that the event rate has.steadily decreased*in. recent years (except lightning);
therefore,. the analysl.s should include a means of weighing the impor.tance of
- an*event by.its date of* occurrance.
If these were done, our. construction*
related' outages.* would become relatively insigni~icant.
We also discussed with Mr R Scholl the choice of words in *the survey results and conclusions..
We pointed out that the use of*the words "violations" and "failures" are totally inappropriate.
None of our reported partial losses of off-site power violates any NRC regulations or our Technical Specifications.
It i.s also incorrect to.use the word "failure.., to describe an event.
The proper operation o~a.relay or protective circuit breaker does not*constitute a failure..
David P Nuclear Licensing Administrator CC Director, Region III, USNRC.
NRC Resident. Inspector-Palisades