ML20132B781
Text
[ f
'm,
~
t UNITED STATES g ;t.
gl g NUCLEAR REGULATORY COMMISSICI.nclosura 2 F
f.%.$%:v/'f$
"~
%'4ASMNCTON. O. C. 20555 Y.r..
%.,.....f F327139
'dEMORANDUL1 FOR:
Thomas Novak, Assistant Director for Licensing
' Division of Licensing FROM:
Frank J. Miraglia, Assista'nt Diiector for Safety Assessment Division of Li. censing
SUBJECT:
GRAND GULF OPERATTNG EXPERIENCE In response to your reouest (menorandum 6f October 6,1983) the Orerating Reacters Assessment Branch (ORAB) has reviewed operating experienn.e durino the past year at the Grand Gul# facility and prepared the attached report.
The ORAB review included a survey of reported events at Grand Gule during the rest 15 nonths (i.e. the icw power license period) and a concarison o#
the event reports with reports from.two cther recently licensed RWD.s
~
(LaSalle, and Susouehanr.al filed during their Inw pnwer license cericts.
The sources of event reports included promot (telephone) notifications filed'per 10 CFR 50.72 at well as Licensee Event Reports (LEP) recuired bv the
echnic0 Specifications.
Operatino-re' actor evants briafine su' aries were r
(-
1sc examined te identify the more seignificant events.
AECD orn"jded us with
-s'ubstantial support in obtaining event rennrts.
3 in general the review revealed that high number of cronot reportable events (10 CFR 50.721 have occurred at Grand Gulf in the past year.
The rate of cceurrence of these events has been at least three -'res gre7ter than tha7 of the *.wo other recently licensed BWRs used for concarison.
The large rueber of errnot reports are concerned for the most part with inadverten* actuatirrs of engineered safety features.
According to the 50.77. reports, ecual numbers of these events have been caused by equipnent failure and errnes on the part of coerators and technicians.
d Review of operatina reactnr event briefing sur= aries indicates -hat five "sionificant" events have been reported 'cr Grand Gui# durinc !" " ear,
,hav i ciute ; low ter-,e-a'ure vessel cressuri77; ion i n d d ? * *- -
- e c ~ " ". ? "I " ~.
i ai"unctier causinc inaduer Ent EPS *r4ps. a c4csei cenera cr
-:-- - re i n c i n e n,-.
ci.uit7r. ecus malfur'iction of bo-h Transamerica n Laval diese i Ce".c-a tors, a nd e
an' opera or error w,ich resulted in 10,000 gallons of water bei
-he -c7c or vessel to the suporession pool. The number' o si9n7 Frainec Trcr kt Grand Gulf curino the low pcwer license perind is bicher thar t'a' *nr tha icfntevent5
$w.s.* n cther recentl" licensed RURs ennsicered in the review.
- L757, n fad % *"
r e v e r.: sirrificant encugh '.c be ferrr*ed ar a br M a; a'd y cue *a"2 t.-huld also be ncted that the ocrirds o, Icw 07 "~',:iCF'?e 'CF
. r e.
/
e iie anc eme,r:n--
w nuch shorter than r.-;ne Gul.
3
F
~ Thomas M. Novak FE2 2 l,._
-2
'9 durine the past year has been atypical.Sased on our review we Comparison of Grand Gulf experience with that of other SWRs indicates that the period versus 4 months for Suscuehanna and LaSalle) and that the rate of prom reportable events has been much greater then exoected.
with Region II we believe that the high rate of reported events is at leas Based on discussions have gone on during the past year.pr.rt related to the large anount o the result of design changes being implemented at the plant.This cons many events which have occurred are related to personnel errors may ct that The #a a.1ack of experience, on the part of ple'nt personnel.
indicate over the lenn term as the plant has moved closer tn Mcwever, a sudden sharp decrease in the rate did occur in Neverber 1983 may be attributed to site inactivity 'ollowing comoletien of low power te in October.
to continue this decreasino trend 'as the plant r.nves cics operation, and testing ard constr0ction activities are completed.
He have discussed the results of cur review with IE pegier !!, and they have informed 'us that our conclusions are censistant with the i
SAlp review.
Recion II will continue'to monitor clant :erforra. Ace and tak appropriate actions should problems continue to occur at a hightrate.
/
-i 8.1 E
hNG R Frank J. Mi$gli?i Ap6stant f.i' rector for Safety Assessrett Division of Licensing
Enclosure:
As Stated e
b e
qM*$
- N
OPERATING EXPERIENCE REVIEW
_AT GRAND GULF UNIT 31 INTRODUCTION
' The staff review of operating experience inclu'ded a ' survey of reported at Grand Gulf during the past 15 months (i.e. the low prver license period) and a comparison of the event reports with reports from two other recently licensed BWRs (LaSalle and Susouchanna) fil.ed during their low power license periods.
The sources of event reports include prompt (telephone) notifications filed per 10 CFR 50.72 as well as Licensee Event Reports (LER) recuired by the Technical Specifications.
Operating reactor events briefing sur: aries were als'o examined to identify the more significant events.
These briefings are regularly scheduied meetings amono NRC manacement to discuss 'recent events at operating reactors.
SURVEY OF EVENT REPORTS In the period between mid-August 1982 and September 1,1983 ISO incidents recuiring prcmpt notification were reported as recuired by 10 CFR cart 50.72 One hundred and twenty-two of these. events involved plant systems.
The remaininc 38 events involved the plant physical security syster.
This review has focused on the non-security related events.
The security rela ed ev'ents were not considered significant and were expected based on the testing and construction occurring at the plant. ' Thirty-five percent (35'.) of the non '
ecurity related events have root ctuses related to cperator and., technician
(
.ectivities (e.g. testing, troubleshoorting).
Eouiprent problems. (mostly electrical) account for thirty-two (32y) of the events. The direct causes c#
the renaining one-third of the events are unkncwn or not acparent 'rrm the brief 50.72 reports.
Most of the events involve iraovertent actuatiens of safety systens with the plant shutdown (e.g. stancby gas treatnent system, control room fresh air system, reactor trip,, diesei cenerator ' star-).
The average monthly rate at which these events have been reported is aporoximately 10 events / month.
This rate is conpared with rates for LaSaile and Suscuehanno in Table 1 and appears to be abnormally high.
Regicn li inspectors attribute the high rate to the large amount of testing and construction goinc on at'the piant.
A review of the data by month does not reveal any norticuler trend in the incident rate.
Data for the past three norths shows a rate n' occurrence close to the average in September and October with a share cecrease in Ucva :er tc 3 events /9cnth.
The sha rp tecrease i s a t:-ibutet r s ' *.e
'-*:-' v'ty foliewirg ccm:letion of low power tests.
A stercy rea.c-cr i- ' 2
?..e -f occurrence is expected as the plant neers ccmnercial cperatirr, since design changes and associated tests are expected te be cori.c*cd.
In the period beginning Aucest i,1982 and ending July 1,19FI a tctal of 227 LE?.s were issued frem Grand Gulf.
The averace ner-h'y rite at wr' e LIES hA"?
been issued is shown in tabic 1 p.lene H'5 cc.caraMa cr es '-- tri.C ie and Suscueharra.
The Gr7nd Gulf.rato is cimiir-to the ra.es ic-L ADc Mc scue. anna.
This is in sharp contres wi-h the ;O CF ca.rt 50. 7 re: Orts
. scussed attve where the Grand Gulf rate vas significar:iy hi;.cr than tr.e nther two plants.
Review of the Grano Gulf LERs irvicates that t P ut ene ka M of the reports relate to problens sith fire orntection cyste s.
Bese rr:blem incluce nar.y instances of snnko detectnr alirr s causec b rer.cv7.1 of fire ba rriers for ccrs truction, crerses.3 cus- ' e cc-c ruc-c :
- ard, Only.irt*.sen TA-ce"-
(19 0 c' ine 277 recorted everts invrived am--c'
~ ~ - ~
-2 TABLE 1 RATE OF REPORTED EVENTS AT THREE BWR PLANTS DURING LOW POWER LICENSE PERIOD Period of Low Power License Rate of Reported E'ents v
Facility (months (Avg. No. reports / month) 50.72 LER Grand Gulf 12*
10 21 LaSalle 1 4
1 19 Suscuehanna 1 4
3 12 '.
- The study ;eriod consists of the first 12 months of the icw pcwer 'licerse period.
12 months.The actual period of the low pcwer license will be Senger than e
3-(eficiencies.
planned entry of technical specification action statene testing or construction.
REVIEW 0F SIGNIFICANT EVENTS Significant events which ha've occurred at Grand Gulf during the past y been identified through a review of issues raised at the regularly schedu briefings of NRR management on operating reactor experience.
consisted of a review of the Operating Reactor Event Briefing meeting mi The review For purposes of comparison a similar review has been perfornec for LaS Susouehanna for the periods they held low power licenses.
discussed at operatino reactor event briefings have been subjected to aEven two weeks for discussion based on the review of 10 during the two week period.
to provide a measure of the severity and extent of significant o problems.
During the Grand Gulf low pcwer license perind, five significant problem
~
Grand Gul# were reported.
was reonrted for LaSalle durinc the _ period of its low power li tre reported for Susouehanna.
No even;s The, Grand Gulf events are summar' fed below.
t Viciation ef RTNDT Heatino Limits Durine ECCS Iniection Octebe-E 1982 Durino surveillance testino with the plant in cold shuteewn a hign DC v spike occurred whicn initiated an ECCS in,iection.
inf ected and caused the reactor vessel to beccee water solic (ex encin the ".SIVs).
The resulting pressure transient violated the Technical Specifica!icn en nil-ductility reference terpera ture, RTNOT.
teac cr Protection Svstem (RPS) MG-Set Guteu Ereaker Tr'es. "av 19. 1??3
- nae.e-en: tri::irg of the RPS !G-s=*. cut:ut bre?te s 7s cc-resultinc in isolation of the instrument air system ano a reic:cr scram signa ac re:e:i-ive'.
The causes c#
catinet vertilation, and low voltroe due to vnitage swirgs wh is fed frem the alternate power supply.
To reduco the number c#
trics the licensee increased cabinet ventilatic, 'nstallec voi: Age reculators nut:ut brer.ker
.c s ce:b cut trarsmisticr. line frc. off-site. voltage ficc:uaticns, arc installec a ree ~sta-'c ciec r cal
!n acci:.cn instru-r-- t'r 5 s e.iscia:ier lays have been re-aligned to an interru..r.ble Orws sur: 1,s.
This prntiem
4.
re-occurred in January 1984 Upward voltage spikes remaining above the setpoint longer than.1 second have caused the protective MG-set output breaker to trip, resulting in de-energization of containment isolation systen logic circuits followed by isolation of the RHR system.
The licensee has been unable to identify the source of the voltage spikes.
the licensee has increased the output breaker delay time from.1 second to 1.4 seconds.
The new delay time is based on measurements of spike duration and consultation with suppliers of the electrical ecuipment.
The modificatien assures that spikes lasting less than 1.4 seconds will not result in a trip of the protective breaker.
between the licensee and Region II. Additional corrective actions are also under Inadvertent Reactor Vessel Drainace Durino Shutdown Acril 3.1983 On April 3,1983, approximately 10,000 gallons of water drained from the reactor vessel to the suppression pool through the resicual heat removal (RHR)
This drainage was caused by two RHR valves (F004 and F006) being open system.
sinultaneously.
At the time of the event, the reactor was at atmescheric pressure with vessel water temperature approxira'ely 100*F Iccid shutdewn conoitions).
The vessel water level continued te decrease until the icw level isolation signal was received and shutdown cooling isolation valves closed to terminate the leakage.
ieselGeneratorRoomFireSeotembeh4, 1983 k
t A diesel generator encine fire was caused by a runtured fuel oif suoply line which sprayed oil on the hot exhaust manifold of the diesel.
The ciesel which caught fire was running at 25 percent load #ce testing at the time.
Two other diesel generators were not affectec by the fire.
system failed to function automatically, but was manually ertiva ec toThe water delug extinguish the fire.
The diesel generator governor and turbo chargers were damaged.
In addition some electrical ecuipment in the rnom suffereo water damace.
Inocerability of Delaval Diesel Generators October CE.1953 Or. Cctc er 23, 19?3, a Terhr.ical Scecificatica tc-'e-5
- 2. s-var a
- ;s when twc cf the tnree die.sel generators became 'rcrerab'c.
The 3,v'sice !
ciesei generator was inoperable cue to gasket failure en a lube ci' line.
The Division II diesel generator became inrre-able due to a locse tase plate nut on the turbt chr eccr which rnruited ir A tr4 r# ibe: vibrAtinn senser which tripoed the diesel.
Corrective actinn was taken to re:a'r brth diesel ge eratcrs.
Both of the diesel generators we c ~7Puf acturef
" ~-r ! A er'ca De1 7 val Inc. (IDI).
IDI cicial gereraters have
?'a-;'v Fr c ; car c'es?
a c ri.
- i t ;
1-
--^~~tr-
.., 5 - !
- ' i. ,--
n A ~3i c oc ' v e e rA * -
. The Shnrehan plar,t.
Sta#f review of the Trarsarecica Sei vA i c'esal A
,er.eratcr rcblen a-Grar,c Gulf is s tiil ongcinc.
CONCLUSIONS Sased on our review'
' during the' low power, we have concluded that operatina exnerience at Grand Gulf license period has.been a'tyoic55.
Comparison of Grand Gulf experience with that of other SWRs indicates that the period of operation with the low power license at Grand Gulf has been abnormally long (12 months versus a repor' table events has been much greater than expected. mon Based on discussions with P,egion II we believe that the high rate of reported have gone on during the past year.
the result of design changes being implemented at the plant.This construc The fact that many of the events are related to personnel errors'may indicats a lack of experience on the part of plant personnel.
The rate at which events have occurred at Grand Gulf has not decreased steadily over the long term as the plant has moved closer to commerical operation.
However, a sudden sharp decrease in the ra'te did occur in November 1983 which may be attributed to site inactivity following completion of the low power testing in October.
basis, ye believe it is reasonable,to expect the incident rate to continue On this this decreasing trend as the plant moves closer to comnercial reeration,' and testinc and construction activities cease.
Should an abnormally hich rate of ircidents re-accear, appropriate actin ~ns such es initiating a review nf sersonnel training programs and plaat procedures should be initi ted *o ide"tif" g
.the root cause of the continuing problem so that necessary correc{tive measures measures can be taken.
4
+'
e
.;d ' " *04e
,A
[,,,S v. fj NUCLEAR REGULATORY COMMISSION UNITED STATES
.s' %;^d({G:!
C Wk. s. <W Ve*ASHINGToN, D. C. 20!!5
'e FEB 2 4 lC84 MEMORANDUM FOR:
Chairman Palladino Commissioner ~Gilinsky Commissioner Roberts Commissioner Asselstine Commissioner Bernthal FROM:
William J. Dircks Executive Director for Operations SU2 JECT:
PERSONNEL ERRORS AT SELECTED OPERATING PLANTS The Office of Inspection and Enforcement and the Office for Analysis and.
Evaluat. ion'of Operational Data wer~e requusted by Commissioner Gilinsky's staf f to. provide information on the frequency of personnel errors at selected operating' plants (i.e., Grand _ Gulf, Sequoyah, and Quad Cities).
The Commission should understand that the information presented here'.is i istrictly a' staff ef. fort based cn inforhatlon available to the staff and has not been verified with the individual licenseas.
The NRC Operatior.s Center data base contains information en events that are r-c.uired to be rescrted unoer the provisions of 10 CFE 50.72.
Many differer. types of eventi are reported, inclucin; all piant rips ar.c safety system actuations.
The following characteristics of the IE data base should be kept in mind when using the information presented here:
1)
The information is called in to the NRC shortly after the event, and at that time an accurate determination of the cause czy not be availa:le.
2)
Cerrections to cri;inal reports are not reutinely cace if la sr i.forna-
-ion i.oulo incicate a oifferent event cause.
2)
Eecause the searcn capability of the system relys partially er a text search routine, some events which involve operator error tay be missed.
This search used "c,eraticnal failure" and " personnel err:r."
'rle beli eve these to be the most Trec,uently used cate; cries for labeling creratic ai i
errors.
t
' Ccntact:
F. J. :-:Ebdon, AECD 492-4450 G. Lanik, IE 492-9526
The Commissioners Table 1 provides a summary of our findings.
reported as operator arrors, personnel errors, or procedural errorsThe tabu events were judged to affect the combined units.
Some and not included as Unit 1 or Unit 2 events.
The'se are counted separately Table 1 Personnel Errors Reported to the NRC Operations Center 1983*
Personnel Errors Site Total Quad Cities, Unit 1 4
Quad Cities, Unit 2 1
Quad Cities (both) 4 9
Grand Gulf, Unit 1 27 27 Sequoyah, Unit 1 6
Sequoyah, Unit 2 3.*
Sequoyah (both) 1
_10 These reports are from calendar year 1983.
=
.In addition, KEOD searched the 'Sequgnce Codin
(
'#~' personnel action was involved in the event.
5
-Escause of the extansive amount of information from each-
- ne SCSS, it was not necessary to rely on text sear form.
Inus, if the LER text expressly stated that a " personnel error" occurra:
or if the LER implied that a personnel error occurred (e.g., "Insevertsatly operated an incorrect valve"), the information was coded into SCSS and seas captured by the subsequent search.
that could be attributed to plant personnelThe results of this searc (e.g., design errors and f abrica:icn/
manufacturing error's were excluded).
errc. " was used enien included both errors of cc missienA rather broad definition cperation of :he wrong valve) ano errors of omission (e.c.(s.c., ina ver:ent re c,ui reme nt s ).
, -issed surveillance The results of this analysis are summarized in Table 2.
I re -*+w.+e
P~
-- ^
. Tne Commissioners Table 2 Personnel Errors Reported in LERs Plant / Unit Personnel LERs Period Errors Received Quad Cities, Unit 1 1983*
7***
36 -
. Quad Cities, Unit 2
'1983*
4 20
' Grand Gulf Unit.1 1983*
58 162 Grand Gulf Unit 2 1983*
O O
Sequoyah, Unit 1 1983* -
18 85 Sequoyah, Unit 2 1983*
9 64 Sequoyah, Unit 1
. June 1, 1982-7 90 June 1, 1983**
Sequoyah, Unit 2 August 1, 1981-18 "61 August 1, 1982**
Some LER's for 1983 have not yet been received and added to the d base.
However, the period,i.s essentially the same for all units.
First year of commercial cperation.
Many o.f the personnel' e'erors r'eported to the Operations Center were '
., _. _(.)
also reported in LERs.
, _.l should not be added.
Therefore, the numbers in Tables 1 and 2 3
t Clearly from Tables 1 and 2, Grand Gul' has reported nere ;ersonnel errors t f
the cther units analyzed.
However, care should be taken in reacning ccnclusiens fec= tnis data.
As the ACRS discussed in Appendix E t:
there are many reasons for non-randemness (e.g., cutliers) in c;erational data ic7.ES-0572 (er-cies e:)
inclucing di fferences in reporting requirements, differences in reporting
-philosophies, etc.
It should be noted that many of these differences have been reduced by the recent publication of 10 CFR 50.73, " Licensee Event System and 10-CFR 50.72, "Immediate Notification! Requirement for Ocerating Nuclear 7.eacters," wnich became effective en January 1,192a.
L, a count of personnel errors does. not consider the severity of the error or itsIn acc
'J consequences.
For example, many of the errors reported by Grand Gulf were
!!L missed surveillance recuirements that did not directly. affect plant cperaticn.
L Finally, because of the time available to prepare this analysis and the size of the computer printout, we were not able to nahe copies of the prir.tcut.
cffice and have not been provided to other interested parties Cons ep te r.t i j ave not been w
l i
.~
TheCcmmissione'rs
_4.
retained by the staff.
.lf other interested parties want a copy, copies can be mace from the enclosed original, or the search strategy can be rerun on the accitional printouts produced.
t J' p:d)Ynhia l. ?nn; William J. Dircks Executive Director for OpeYations
Enclosures:
As stated cc w/ enclosures:
m j
n/
~
3, i
k
,s t
~
l h
t
A:PENDIX E C,
i ss f
STATISTICAL Ar,'Atysts c,:
LERs:
A TRIAL STUDY t
e nx,*.
w m
m
,., een e w
~
' Troducilon
' proxidately 5700 LERs Jere submitted by the licensees of U.S. commercial
. nuclear power plants during the years 1976, 1977, and 1978.
For several rgasons, the number.cf LERs varies.from unit to, unit.
These variaTiens are important, because, rightly or wrongly, they are'often' vie ed by government agencies and the public as Indications of relative safety.
While such variations may be indicative of actual differences,in safety among nuclear power units, they may have other explanations, it is therefore
- important to understand ~ all possible explanations and their contributions
?o variations in the numbers of LERs from unit to unit.
Certain cif ferences in the frecuency of submission of LERs from unit to unit will occur as a resul.t of the apparent random nature of the events being reported, cecause.of.this " randomness", it is'possible--in fact, probable--that, even among identical nuclear power plant facilities wiTh identical failure probabilities, there will be variations in The reporting rate for LERs.
In reality, however, variations beyond these due to "rando?. ness" wi l l f requent l y b e observed.
The c,easons for such,non-rancom variations include the facts that:
~
(1)
Technical Specifications and license provisions vary among nuclear power plant f acilities, because of di f f erences in reactor suppliers, architect / engineers, and constructors, and changes in designs,over e
the vears.
These variations cause 61tferences in the recortig['c
'~'
requirements among facilities.
(2)
There may be a. tendency at some facilities to re;or7 events more readily Than a? cThe.s in cases of marginal re;c.-tability.
This c:nsideraTicn pertains to events cTher than cbvices "repertable c:currences" (R0s), which all licensees must re;crt *.
Tnis I
Tendency can also change.ith time.
(3)
The c:currence of an event may af f ect the prcba:ility of future events.
As; air of a facility componer.; cr =;rovemenT or a ceticienT peccacere may significantly reduce the likelibced of an asscciatec 3
event.
Cn the cTher hand, i ne f f ecTive corrective acticn fel'lc-i ng an
]
even; may result in its repeated occurrence.
- )
The mcts cf c;eraTicn (e.g., en-line er shu cc-n; afiscTs The f equency cf varicus kinos cf inspections anc The susce;Ta:ility ci systems To re,ccm failures.
The aTiount of reacter c:c n-t ime, icr example, may affect the frec,vency with hich LERs are su miTTec.
a i
(
.e reference list fcilowing Chapter 4 E-1 l
.,,,,g.w.
w.....e:--
(5)
Misinterpretations by licensee er KRC personnel involved in the preparation, submission, reperting frequencies among reactor systems.and processing of LERs can a i
(6)
At some multi-unit power stations, events which involve p'la'nt' systems' or, components common to all.suc units, such as swing diesels and electrical switchyards, are filed in the'NRC data bank under-the docket number of the first unit
~
(7)
The actual presence of more safety-related deficiencies in a system at an Individual facility should, result'in more frequent submission of LERs.
Differences in the number of LERs due to this cause would be a measure of relative safety.
Although the above factors af fect the frecuency with which LERs are reported, their ef fects are often relatively small.
Frequently the variaTicns produced from these occurring on a random basis.by these ef fects are too small to be distin For example, the Point Beach 14uclear Station in 1976 had il reportable occ,urrence LERs for Uni?
,16 for Unit 2.
I and Oces thi s necessar,l.l y indicate that one or a combination of the caVses listed above produced this difference, or is it possible that a Ceviation of this magnitude could have been expected if both units had the same average prcbability of occurrence of reportable events?
('
' analysis indicates that Statistical i t and.16 Jn one. year are both consistedt wiTh average Occurrence rates in the range of cne per 20 days to one?per 37 days (10-18 per year).
In fact, the pair of numbers, 11 and 16) is the mesi ;rctable one year outcome for two units wlTh an average rate of one
- er 27 czys (13.5 per year).
In 1975, the Zicn Nuclear StaTien had 55 re:crtable cccurrence LERs for Unit I ar.d 39 fer Unit 2.
In inis case, The ceviation in the number of LERs bet-sen the t-c units is co large to be aTTributec solely to random effects.
If rancom ess alene -ere invcl'ved, Uni? I ;rciably cculd n07 have had a reporting rate less Than One per 5.2 days (70 ;er year), while Unit 2.prcbab ly coul e nct have nac a rate cresTer than one per 7.2 days (51 per year).
In fact, if both Zion units hac identical pectabilities of repcrtable events, there is nc r.cre Tnan one chance in one millicn That a deviatica this large could occur by chance GaTurally, there are cifferences between The.:cir.T Esech units.
Unit I is ;-c. ears o!:er than Unit 2.
During 1975, Uni? 2 ;r:: cec !!i mcre i
electrical enercy inan Unit l.
The results in Tnis exam;!e incicate
~
that ene shceld not necessarily conclude that cifferences in the r tes of LER sutmissien bet-een the t-o units are sig..ificanT.
At Zicn, c-ever, cne snculd expect To find that the two units reportec at significantly-cifferent rates fcr reascns cTher than rancemness.
1
.(
i E-2 w
V
,.,..y*-se=
- ^
__ _ _ __ _=
'ethodolcev
-xetheds from probability theory can be used to calculate the impact of randomness on the distribution of the number of LERs ancng identical nuclear power units.
Of ten, probability tables from reference, textbooks are sufficier.T to perform the analyses.
the more complicated analyses, Canputer simul ations are necessary for in interpreting the resulting data, it is important to note seve al basic facts:
(1)
The numerical size of expected random variations in event rates 2
increases as the average event rate increases.
Deviations of 10 or more are readily expected on a randon' basis for an average yearly rate of 100, but are unlikely for an average yearly rate of 20.
The relative size er percentage variation, however, decreases as the average rate increases.
(2)
The chance of large random variations among units increases as the-number of units being compared -increases.
For t-o units with an assumed average annual LER submission rate of 100, there is only a small chance that one rate will deviate by more 'than 20 from the
. average because of randomness.
For a comp' arisen among 30 units,
(
' hewever, there is a goed chance-that at least one will deviet,e by j
mere than 20 f, rom the annual avirage rate of 100 because of ganconness.
3 A selected set of LERs was used here to-demonstrate the application of inis methodole;y.
The sources of the LERs were the 22 EWRs ther acnievec co. er-cial c;erarica price to 1975.
Records sho tha: Tnis greve submitted a Tctal et 27 LERs fcr 30-day reportable occurrences in auxiliary prccess sysist.s curin;
!975, 1977, and 1978.
Thus, for thi s group of units, The avere'ge was a:cui c..e LER cf Tnis Type for the Three-year period, it is first assumed that all units in The grcup were identical wi?n res;ect fo their cnances of generating LERs cf this type.
Further, iT is assumed thar if a nuclear pc<er plant ex;eriences s
a repcrtable occurrence in an auxiliary precess system, The chance of ancther cccurrence is unaffected.
Throughout This study a.cisson cistrituTicn cf.
]
events is assured.
Probability theory indicates that, -hile the average is ene, it is very Onlikely that eacn incividual unit cult ex: erie.ce exact!y c..e.
In fact, tne pechability that all 22 units *: ele each re::--
inis num:er is less Than ene in Ten billion.
Tre ecs: likety result i s ina; about eight units will have no LERs, abcui eight will have cne LER, a: cut icer vill have two LERS, and abcut two will have Thras LERs.
Further, it is unlikely (55 chance) that any one unit will have six cc cre LERs.
Co?.; arisen to actua LER data she<s nine units wi th nc LERs, seven with cne LER, t.o with two LERs, era with Three LERs, T-c -itn fc.r i
=/
E-3 e
gsa e er9ampNW6P--
- c-
-gep-
-,-.mm w
gy9
_.-_-,-_,,_,y
-,,s,,__,.,,,___-_-.i.,
u_.,p-.4 v--
a-m-
w-
)
LERs, and ene with five LERs.
with the assumptions stated above.The distribution of LERs for Tne 22 E 7sistent
~
This example, does net prove, however,t tnat the 22 SWRs are identical to
-ecch other wirn regarc, To the causes,of auxiliary process systems failures.
It simply indicates that one should net expect to find signi ficant di f f erences among these units, even though some submitted as few as zero and others as many as five LERs.
is that,it provides a methodology through which signi ficantly highThe value deviations can be readily identified among a population of expected random deviations.
Anelvses For purposes of this study, the LERs from 67 nuclear power plants were reviewed.
For purposes of analyses, these were divided into FWRs (tetal = 42) and BWRs (total = 25) and each cf these groups was further separated "cicer" and " newer" power plants.
inTo In this case, " older" was arbitrarily cefined as these power plants that went into ope' ration prior To 1976 (see Table E-l).
For this group, all LERs, submitted curing calencar years.1975s through 1978 represent events That occurred during co mercial operation Cata used in *these analyses were based on the NRC computer bank and included
-secriable c:currences only.
The. Ros _were separated into those requirec be submitted on a ;rompt or two-week basis and These setmitted gh a
, arty-day basis.
These were analyzed separately since incre did a07 a;;e ar' te any correlaticn in the relative ' numbers of each type as reportac by 70 licensees at the 57' power plants.
a::crtir,; te The system To which they pertained.Lasti.y, the LERs were furtner se;arated is sacan !n Tatte E-2.
A listing cf Tnese systems The p.-imary g al in The analyses was to identif y significant caviations er variaticas in *ne numter cf LEss re:Ortec from : tant To plant and system T.o. system.
A cevistien was considered to be significant if inere l
i was a,*%
cPance Or less That it could have resulted f rom ranccT. variaticas.
C:n:lusiens
- , T's
- asis t
nese ar.alyses, The folicwing c nclusic s a t' r ctse variens
-ere me:e:
(i) ine frecuen:les of re;criable o:currence LEES a on; t'.e varicus nut!iar
' ;;.er uni s -ere significantly citieren?.
There.e e no icentifiable gecu;s of eacter units v.-hese mesters generate: The sar.e average num:er cf rs;;rtable c:currence LERs curir.; ea:h cf *.e iBree years in -te study.
E-4 s
i
,,3 e-
~ \\
2)
Censicering.the three-year perled as a whole, 5 units among the 29 older PWRs deviated significantly. from the others in terms of the total number of two-week ROs.
The numbers of LERs from Calvert Cli f f s-1, Palisades, Rancho Seco,, and Three Mile Island-l were high; Maine Yankee was low.
The remaining 24 PWRs repo,rted numbers of LERs consistent with ah average of about 20 per uni-for the period from 1976 through 1978.
l (3)
For the same 29 older PWRs,, considered year by year, the data showed that the total. number of two-week ROs, steadily decreased in each i
successive year.
The averages were ten per unit in 1976, six in'1977, and four in 1978.
Signi ficant deviations from these occurred at Calvert Cliffs-l in 1977, Palisades in 1977 and 1978, Point Beach-l in 1978, Rancho Seco in 1977, and Three Mile island-l in 1978.
All had higher than normal reporting rates.
Maine yankee had a rate in 1976 s.lgni-ficantly lower than normal.
These results indicate that the high Three year totals for the four units listed in paragraph 2 above were basically due to high reporting rates in just ene of the three years, while the rates for the other '.two years. appear to be normal.
(4)
Further analysis of the data ihowed that the high totals of t o-week ROs in four of the older PWRs were attributable' to abncemally high-numbers of LERs cencarning speci fic systems.
Calvert Cliffs-l had l
signif icantly high three-year ;;otals for electric power systems and fer reacter systems.
Palisades. reported high totals for the same.twc systs s, i n addition to engineered safety features.
Rancho *Seco repcrted a high total f or' el ect,ric. power syst ems.
Three Mile Island-l had high totals ter radiatien protection systems and fer events classed as " systems cede not applicable." Many cf The electric pc.er system LERs aere related to Off-site power systems etc et,ergency diesel generatcrs.
Reactivity centrol systems were the source cf est cf the reac;cr system LERs from Palisaces.
(3)
.t.neng the older PWRs with ncemal yearly totals for twc-week EOs, some nevertheless reported significantly higher than ncrmal totals cf LERs for specific systems.
The number of LERs in reacter systems was higher than normal at Arkansas Nuclear Cne-1, Ocenes-2 anc -3, anc c;.5. Ec:insen-2.
The numbar fcr Zicn-l -as higPer 79an termst,
'h. raciatien ;rciection systems.
LERs fer electric Oc-er syste s l
.ere hi;her than normal at Fer; Calhcun, Ocense-I anc
-3, Frairie Island-l, and Turkey Point-3.
The systems mentioned here, hcwever,
- id nc; centribute significantly to the tctal number cf LERs, s ic.ce LERs f rom engi neered sa f ety features ar.c reacter c:clant systems do.inated the t c-week Ros from older Fis.s.
As a result, cev i at ions f r o;. nerta l in the less often re;criec sys a s cic nc?
have a significant it;ac? cn the icial nut:er cf LE7.s fcr tmase I
(
- plants, i
.w e
E-5 t
7 7
r
'. 5 )
The data show that newer PWRs, af ter they achieved commercial ROs than did older PWRs.cperation, had significantly higher LER sub the older plants, engineered safety features and reactThe excepti A.s wlTh systens were responsible for a large fraction ot the LERs.
or coolant (7)
With regard to 30-day R0s, there. were no.
n totals for the three-year ' period ' It 'Is possiblethe 29 e
identify three separate subgroups among the units in this catego however, to A first' subgroup includes seven. units with an average reporting r' a of about twenty 30-day Ros for the three years.
Point Beach-l and -2, Rancho Seco, San Onofre-1, and Turkey Point 3Th and -4 for the three years.Another group'had an average of about forty-five 30-day Indi an Point-2, Maine Yankee, Oconee-I and
-3, PrairieThe 10 units. in this group were Hadc'am Neck, Isl and-l and -2, R.E. Ginna, and Three Mll e Island-l.
of 5 units with a normal reporting rate of abcut 70 for the three year A third group per iod included Arkansas Nuclear one-1, Kewanee, Palisades Surry-1 7 units with high reporting rates.and -2.
S igni ficant dev-l ations from these groups cc, curred and in Cook-1, Fort Calhoun, Mll istone-2
'These were Calvert Cliffs-1, D.C'.,
Yankee Rowe, and Zion-l and -2.
- It Engineering reactors are in thPs categcry.is interesting to t
i, These are Calvert Cli f f s-1,
'cri Calheen, and Mllistene-2.
In addition, tnis category ivcludes~
all These are O.C. Cock-l and Zion-l and -2.three of the older PWRs (5)
The data shew that the cne year totals for thirty-day RCs in olcer F#Rs were similar to the thres year tcTals in That cefinit'e subgrou;s can te icentified, in general, a unit that was in a Ice or nigner repcrting subgrcup in ene year remained in tne sar.e subgroup in larar The exceptiens were Yankee Rcwe, which was in a higher re-years.
- ceting subgroup in 1977, but in lo.er reperting sutgrou
- s in tne c?her two years, and Surry-l and
-2, which were in a Ic.er re;crting se:gecep during tne first two years but in t'.e higher subgrev: in 1973.
Several significanT ccrrelatiens were fcunt.
These vr.its nich ren:e: to remain in the ic<est re:crting sut; c :s avertre!ess :n-
- aase: t. air re;crtir.; rates f er thirty-cay RCs f rom Tne sum of their inicty-day and tec-week ROs, nc.ever, year ic ysar.
r e.ma i n e c essenticily ccastant in time, since the tec -eek RO t Tal cecreased durin; the Tnree year period.
steacily Large units cf 1000 'We er cre re; cried higher num:ers of 30-day RCs, exce:
. hen e -lanT facter for tr.e year -as icw (less inan ens-thirc),
t.ater Co-tustion
(
l l
E-6
?
Engineering units (not including Maine Yankee) also submitted higher numbers of LERs for thirty-day Ros, except when the plant availability factor was low (less than one-half).
i l
i
-(g)
Newer PWRs reported thirty-day Ros at ratps cons,istent with the higher reporting subgroups among older PWRs.
L
'(10) The systems most responsible for the higher LER submission., rates for thirty-day Ros in Combustion Engineering units were auxiliary process systems, electric power systems, instrumentation systems, and steam t
and power conversion systems.
These ' units usually deviated from the normal reporting rate for these systems.
In large units the systems involving a higher than normal number, of thirty-day R0s were auxillary process systems, engineered safety features, instrumentation systems, and radiation protection systems.
l (11) With regard to two-week Ros among the 22 older SWRs, eight units deviated f rom the normal reperting rate,d,uring the Three year period.
These were Dresden-2, Duane Arnold, E.I. Hatch-1, Fitzpatrick, and Peach 5cttem-2 and -3, with h1gher rates than rcr.al anc Dresden-l and LaCresse with lower rates than normal.
The remaining units reported an average rate of about twenty-four two-week Ros fer the
(
three yea'r period.
The ' rate remained constant at about eig.)t per year.
?,
(12) E.I. Hatch-i reported two-week R0s at a ccmparatively hich rate for each of the three years.
The numb.er of reports pertaining to nearly every system deviated f rom normal repcrting rates for these systems.
(II) Ouane Arnold reported twc-week R0s a7 a ccmparatively high rate in l
1976 and 19et.
she systems with highe Than normal num:ers of repcrts were related to elec*ric pcwer.
Fcr Fitzpatrice, the nust.er cf twc-week RCs for 1975 was high.-
7his unit also had a high r. umber of RCs in instrumentation systems.
Fcr Peach Ecite -2 anc
-3, The r.umter of twc-week Ros fer 1976 and 1977 was high.
Unit 2 had an aincemally high number cf ROs fcr reacter coolant systems and steam l>
and pcwer ccnyersion systems.
Unit 3 re; cried a high number in en;lneere,d saf ety f eatures and f er c ner auxiliary syste s.
Oresden-3 re:cete: a nigher-than-ncrmal number cf LI:.s in ;977.
- Furiner,
--it unit re;crtec an abncemally high number et R0s in electric pc.ee systems.
Nine. Mile Fcint-l reperieo higher-Than-ncrmal totals of LE:s concerninc instre entation systems.
Quad Cl?ies-l re:criec a high Incidence of Two-week Ros in stes?. arc ;Cver CCnversion systems.
i 1(
E-7
~
a (14)
A.meng the three newer BWRs, only Browns Fe commercial Operation.
l
'(15) Two BWR units, Fitzpatrick and Brunswick-l, repceted abncemally high numbers of thirty-day R0s in nearly every system.
As an ex' tension to the above,1ERs,, pertaining.to'iet point dri ff we[e analy' zed using as a data source the computer _ bank,at the Nuclear Safety Information Center'(see Appendif D'lli).
These analyses showed that there was no significant deviation in the total annual LER submlTtal rate for setpoint dri f t among ol der BWRs or among older PWRs.
The average rate for SWRs, however, older PWRs reported rates higher than normalwas appecximately five times as Six for the three-year period.
These were Zion-l It ~is interestino to note that three of these are Cstbustio Fort Calhoun, Millstene-2, Palisades, and Kevance.
, Engineering units.
Among newer PWRs, four units reported at high' rates in
'1973.
These were J.M. Farley-1, Indian Point-3, North Anna-l, and Sal em.
Three older BWRs reported set point drif t even'ts at adncrmally hich rates f or 'the, entire three year period. - These were Duane Arnold, Brunswick-2, and Nine Mile Point-l.
Six older SWRs reported at abnormally low rates.,
Tnese were Big Rock Foint, Menticello.
Browns Ferry -l,
-2, and
-3, Lacrosse, and
(
' Ccmmenta v i
This portien cf, the study has clearly de=cnstrated the potential of statistical usefulness analyses in the evaluation of LERs submitted by 1icensees.
Sucn analyses make 17 pcssible to cisTinguish ceviations in T.ie num:ers of LERs which wculd be expected on the basis of cancemness fec: these inaT-almest certainly culd nct.
The itenTi ficatica of areas for pcssible f urTherlatter can be used as a means for The investicaticos.
Wnile The ceviaticns notec in this study co not necessarily imply safety-relatec
- rc0lems, they shculd ncneiheless be purseed in creer To cetermine The true implications.
It <culd pretably be desirable to'com; uteri:s these analyses fcr a t u omaTic
- rc:sssing of re;ce?s as they are'ic;;ec lr.To -.s LER data tase.
Utiliza-Ti:n cf The data base in This mar.r.er -cui mais 17 ;cssi:;e Tc ceTe:7 signifi. Cant deviaTicns frCm normal.
Furtner, an automate system ccult be pr ogr a?.me d to cbtain detail beyond the system level, in creer To identify repcrting rate ceviaticas fcr relevant setsystems a-d co ;ccanis.
E-8
r-q Table E-l Number of Reportable Occurrence I.ERs from Commercial Nuclear Power Plants (1976-1978)
GROUP 1:
Dider PWRs (commercial operation prior to 1976) Total = 29 Nuclear Reoortable* Occurrences.'
Nuclear Reoortab'le Occurrences Power Plant 30 day 2-week Power Plant 30-cay 2-week Arkansas Nuclear One-I 71 17 Point aaach-l 15 30 Calvert Cilffs-l 169 35 Point Beach-2 18 20 0.C. Cook-l 147 20 Prairie Island-l SI 17 Fort Calhoun 109 24 P
irie Island-2 36 16 H.B. Robinson-2 53 26 Rancho Seco 17 40 Maddam Neck 41 1.9 ^
R.E. Ginna 44 24 4
(
a
. indian Point-2 57 26 San Onofre-I 19 11 3
Kewanee 75 19 Surry-l 79 19 Maine Yankee 47 6
Surry-2 71 Millstone-2 118 21 Three Mile Islanc-1 44
'41 Cconee-1 42 34 Turkey Point-3 2a li Ocenee-2 21 26 Turkey Point-4 20 15 Oconee-3 41 21 Yankee 0.cwe 99 13 Palisaces 64 55 Zica 1 155 25 Zion 2 I22 15 Average 65.5 22.7 f
E-9 e
.-ae
- W * * *""*'T*I~""
I
r-o 7able E-l. Continued GROUP 11:
I4 ewer PWRs (commercial operation af ter January I,1976) Total = 13 Nuclear Reoortable Occurrences
' Nuclear
'Re'ocriabla Occurrenc Power PIant 30-cay 2-week Power P1 ant 30-cay 2-week es Arkansas.
Nuclear one-2 21 7
Indian Point-3 85 15 Beaver Valley-l 216 27 J.M. Farley-l 138 23 Calvert Cliffs-2 135 25 North Anna-l 93 29 Crystal River-3 154 32 5,t. L uc ie-I (23 22 0.C. Cook-2 95
. 7. '
SaIem-1
~
II8 65 Davls-Besse-I 220 32 Three Mlle Island-2 42 17 f
1~
Trojan
[63 N..
44 Average il6.5 25.5 e
i 4._-
E-10
.i
, *= - g >w we v.
e
. o 1
sale E-l'Centinued 5ROUP 111:
Older BWRs.(commercial cperation pri'er to 1976) Tctal = 22 Nuclear Reccrtable Occurrences Nuclear ReoNrtab Fe Occurrences Power Pl. ant 30-day 2-week Power Plant' 30-day
_2-week
.u,
Big Rock Point 105 3.0 taCrossv 27 10 Ercwns Fercy 55 26 MEl.1 stone-l.
80 27 3rewns Ferry 33 13 Montl.cetier 65 30 Brunswick-2' 261; 3'4.
Nine Mlle Point-l.
93 27 Cooper 122.
IE Oys'thr Creek-1 56 35 Dresden-1 70 ICP Peach Bottom-2 1.46
, 56
. ? enden-2 153 5'l* -
Peach.Sottom-3 1.07 5'6
- (.
sr'es de n-3 105 29 P11 grim-1 (03 25 Duane Arnold 120 88[
Quad Cities-l 94 31 E.1. Hatch-l' 94 I62' Quad Cities-2 75 14 Fitzpatrick ISI 41 verment Yankee 96 IS Average 102.0 35.0 t
9 E-Il j
e.
Table E-l. Cor.tinued GROUP IV: - Newer BK:.s (commercial operation af ter January 1 1976) Tctal = 3 Nuclear Reoortable Occurrences Power. Plant 30-day 2-week
- Browns Ferry-3 58 12 Brunswick-l 21I s
E.l. Hatch-2 65
, 12 nyerage ttt 3 11.0 a
9
+
e e
4 e-L...;
m O e
i 9
4 1
e 6
I i '
?
\\-
i E-l2 1
i J
f Table E-2 System Codes for LERs System System 1.
Auxillary Process Systems 8.
Other Major Systems l
2.
Auxiliary Water systems l
9.
Radiation Protection ~ Systems t
s 1
.3.
Electric Power Systems 10.
Radicactive Waste Management System:
4 Engineered Safety Features 11.
Reactor Systems 3.
Fuel Storage and Handling Systems 12.
Instrumentation and Controi Systems 13.'
Steam and Power Conversion Sys w s 7.
Other Auxiliary Systems I4 System Code Not Applicable a
/
E-13
---