ML18116A186

Notice Describing Reasons for Shutdown of Five Plants

ML18116A186
Person / Time
Site:	Beaver Valley, Surry, Maine Yankee, FitzPatrick
Issue date:	05/14/1979
From:	Chilk S NRC OFFICE OF THE SECRETARY (SECY)
To:
Shared Package
ML18116A184	List: ML18116A182 ML18116A186 ML18116A187
References
FRN-790514, NUDOCS 7907230752
Download: ML18116A186 (16)

Text

.... *

\\

J f-/1.!CL::AR REGUU. TORY COMMISSION ABNORJl)\\L OCCURRENCE EVENT DEFICia;CES rn PIPING DESIGN

/:;Si)-01

• Section 208 of the Energy Reorganization Act of 1974, as araended 3 requires the NRC to disseminate infonration on abnorr.1al occurrences_

(i.e., _unscheduled incidents or events v1hich the Corrmission determines

. are s*ignificant fr0,11 the standpoint of public heal th and safety)_

The

. : following inc*ident was dete11Tiined to/be an abnonnal occurrence using the criteria publish_ed in the' FEDERAL REGISTER on February 2.4, 1977 (42 FR l 0950).

• Append i~ A (Example I.~- 2) of the Po 1 icy Statement notes that a *major deficiency in desfgn, cons true ti on or opera ti on havi"ng safety.

implications (_affecting five pl~nts in this case) requi.ring immedi-ate

• remedial action can be considered an abnormal occurrence.

The following description of the event also contains the rerned\\al actions taken.

De. te and Pl ace - During design and construction, an incorrect su1;.:!@tion of earthquake loads affected the desi-gn of safety related pi*pi11-g syste~s and 2.ssoci ated pipe supports at five nuclear power pl ants.

On December 6, 1978, a Licensee Event Report from Duquesne Light Company mentioned differences between c0mputer codes used i.n analyses of force summations, but did not elaborate on then.

Then, the NRC learned of an incorrect surrming of loads in one of the codes on r'iarch 8, 1979, at a meeti_ng in Bethesda, P1aryland with Stone and Webster, an architect engineering firm and the Duquesne Light "cor.rpany (DLCL the licensee for Beaver Valley Unit 1, u

j

(

e \\

e a pn~ssurized h'ater nuclear plcnt located in Beaver Coun~y, Pennsylvc..nia.

On ~~2.;c:h 9, NRC *1earned tf'J.C.t the incorrect suiTiii1ation technique affected four other plants:

Pl ant FitzPatri ck Maine Yankee Surry 1 & 2 Location Oswego County, New York Lincoln County, Maine Surry County, Virginia Nature and Probable Cons'equences -

In October 1978, Duquesne Light Company, the 1 i censee of the Beaver Va 11 ey pl ant, was inf onned by Stone and Webster thct for loadi1'..9 conditions associated with postulated e-arthquakes, pipe supports associated with Safety Injection System piping 'doul d be overstresse-cf.* -

Stone and Webst_er (S&W) w-as reanalyzing stresses in connection with a sys tei'Tl modification re qui red by the rrnc staff to correct a design defi ci_enc,y not related to pr-otection against postulated earthquakes.

During this reanalysis effort, the S&W engineers al so came across i*nfonnati on that had been provided to them by ~esti nghouse in ~~a_y *1973 t.nat showed some check vaives in these lines were actually heavi_er than assumed i_n the ear ii er analysis.

'Sometiriie during this reana_lysis, either i"n connection *wi"th* the planned r.:odification or in ree.xaFr.ining the effect of the increased *valve we_ights, S&h' discovered a misapplication of a hand cal cul at ion method.

In correcting this ~1sapplication, S&h' fo~nd some instances of local overstress.

The j

e e correction consisted of adding a snubber and modifying or.:= support.

In doing the analysis related to making this ccrrectior., S&'.-: used two computer programs.

A new one, NUPIPE, predicted much higher* stresses than the one, PIPESTRESS, used during a 1974 as-built check of these 1ines.

On October 26, 1978, the licensee orally notified the NRC 01;-ice of Inspection and Enforcem2nt about the design error (hand calculations m2thod misapp1ication}

which re qui red correction.

No _exp 1 ana ti on was provided fur the differences in stresses p_"redicted by the hm codes at that time.

Repeated NRC contacts with the 1i censee and S&W to *identify t..he reason for the differences were not effective, since withcm"t the actual computer runs to 1 ook at there was a communications problem~

Thlring a m2eti ng _held on March 8, 1979 to discuss these matters:, the Beaver Va 11 ey 1icensee informed the NRC staff that the differences in predicted piping stresses between the two computer codes were attributab1e t.o the fact that

• the SHOCK2 subroutine of the P IPESTRESS code uses an a 1 gt:b~a i c suITTTiati on of the loads calculated separately for the horizontcl and "the vertical component of earthquake motion.

The use of algebraic surrmation is only acceptable if the time phasing of these loads is knm'-ln.

The algebraic technique as used in SHOCK2 is not conservative for response spectrum modal analysis because, in such analyses, time phasing is not considered.

\\

-1

e The analytical treatment of load combinations beco~2s significant bec2.l.!se horizontal earthquake motions can produce pipir.g r.;overr.-2nt in both the horizontal an*d vertical direction. and the vertical earthquake motions can al so produce pi ping moverrent in both horizonta 1 and verti ca 1 di rec ti 011S.

For some designs the calculated piping stresses may differ significantly depending on the load summation t(;chniques used in each [;}Ode of response.

Based on the three piping syst~ms that had been reana1yzed by the

.newer code on Beaver Valley at the time of the March 8, _1979 meeting:,

stresses over a"ll owab 1 e va 1 ues were expected to be* found primarily in piping supports although significant increases in piping stresses had been observed.

NRC staff revi e1,1ers were sent to S&W' s Boston office to detennine the extent of this problem on Beaver Valley 1 and other po-t.entially affected pl ants.

In fol lowing the course of the reanalysis at the S&;.~ offices over the v.*eekend of March 10, 11 *and 12, based on the informa~ion then cvail"able, it became apparent that, \\'Jhen the NUPIPE code Wc..S used,, a number of piping systerns had calculated stresses over the allowable value for the design basis earth_qu2.ke.

Also, for a fei*t of these systems the r,,ore prob ab 1 e operating basis earthquake resulted in. stresses above the allov;able value.

In addition, the structural integrity and perfoTT.iance of pumps, valves and other essential equipment could be

< \\

(

e e

degraded.

Although results v;ere still incomplete on March 12, information av~il2ble 2t that ti~e indicated th~t high stresses were calculated in a number of syster.is important to safety~

Because the overstressing of piping and supports was predicted even for earthquakes which might occur during the lifetimes of these faci1ities, the problem took on considerable safety significance.

Sorae of the systems identified a~ that time as having oyerstressed conditions under earthquake 1 oadings were p_art of th,e reactor cool ant pressure boundary, whose fa i1 ure could c2use a 1 oss of coo 1 ant accident.

In addition, systems which wvul d be needed to shut the pl ant down safely in the event of a loss of coolant accident were ilso affetted.

Thus an earthquake, of not extremely 1ow likelihood, would have the potential both for causing an accident and for preventing safety syst~~s, designed to *cope with that accident, from operating.

A secondary concern was whether or not systems needed to provide adequate long term cooling-for the plant in the event of an earthquake without a LOCA could be assured.

Concurrent with the NRC Beaver Valley review, NRC staff records were reviewed to determine whether or not other faci1 ities liad used these

-same analysis *techniques.

Based on the review of these records and information provided by S&W, the rrnc staff concluded that four other facilities used the same techniques.

The four faciliti~s are Maine Yankee, FitzPatrick and Surry Uni*ts 1 and 2.

• -~:"""

\\.

e 6 -

e The NRC staff concluded the potential for serious adverse effects 1n the event of an earthquake was sufficiently widespread that the basic defense in depth provided by redundant safety syste.u-s may be compromised.

The NRC Director for Nuclear Reactor Regul~~ion conc1uded that the publ_ic health and safety required that the affec---u:d faci1ities be placed in a cold shutdown condition pending further onder of the Commission.

Orders to this effect were issued to the lia::nsees of the

above reactors.

The Orders.provide that within 20 days each licensee EUSt respond with respect to:

(1) why the licensee should not reanalyze the facility pnping systems for seismic loads on the piping system and a:ny other affected safety" systems using an appropriate piping analysis computer code which does not combine loads algebraically, (2) why the licensee should not make any modifications ito the facility piping systems indicated by the reanalysis,., and

.(3) why facility operation should not continue to be suspended until completion of the reanalysis and any required mDdifications_

All of the plants are now in a cold shutdown condition.

(Surry Unit 2 h'cS c 1 :=ady in an extended outage for s teem generator rep 1 a cement. )

--*~*.

-:~;.

r 7 -

Cause or Causes - The uncertainty in the calculated piping stresses and support loadings in safety-related piping syster.1s at th~ fiv2 plants is attributable to the incorrect. application. of the algebraic summation technique in the SHOCK.2 subroutine of the PIPESTRESS computer code, proprietary to Stone and l~ebster.

• Actions Taken to Prevent Recurrence Licensee/Architect Enoineer - Identification of all safety related systerrs that have been analyzed with a pi ping computer code i nv~ l ving a program deficiency is underway.

Computer inputs are being checked to assure that -

all reanalyzed piping will reflect the as-built condition at each plant.

Piping analyses-- are being rerun 'and piping and supports ex:a:ec'ri:';"lg allowable stresses will be identified.

Modifications will be rrade as necessary.

NRC "" The NRC ordered each of the uti 1 iti es of the five i d:=nti-fied nuclear power plants to shut down their plants within 48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br />.

The utilities are to remain shutdown pending further order of the Com;ni ss ion.

The NRC is in contact with the 1icensees and the architect engine-er on actions being taken.

Piping stress computer codes to be used for reanalysis of the piping will be tested with NRC established benchmark problems.

AlsO, an. independent.audit of selected piping runs will be conducted by NRC consultants to verify the piping stress reanalysis.

(

\\

,l

~

e e

8 -

In cddi ti on to revi ei'li ng the 1 i censees I corrective actions, the

.;RC ~s r2vi2;,;ing any generic implications ct other fc.cil~ties_

The J-i;:!.C's Office of Inspection and Enforcement issued Information Notice (IN).

No. 79-06, which described the event, on March 23, 1979, to a11 ho1ders of reactor operating 1 i censes and construction permits.

On *Apri 1 14,. 1979.,

the NRC's Office of Inspection and Enforcement issued Bu1letin No_ 79-07 to applicable licensees which identified actions to be tc.ken_

This includes "r

i den ti fi cation of the methods of analyses used, how they were verified, safety systems ~ffected, 'and a plan of action to assure-plant safety.

As of May 9,* 1979, the NRC has received responses to Bulletin No. 79-07 from a~l l l i cens_?es of operating.reactors except for Three Mi1e Isl and Units 1 and 2 which are shut down.

The NRC staff is reviewing these responses on a high priority basis.

Adpitional actions wi11 be taken as appropriate.

Cor:::mi s s i on Dated at Wcshington, D. C.

. \\

Nl!CLEAR REGULATORY COMMISSION ABNOR1~AL OCCURRENCE EV ENT DEFICia~CIES IN PIPING *oESIGN Section 208 of the Energy Reorganization Act of 1974, as amended, requires the NRC to disseminate information on abnonna1 occurrences (i.e., _unscheduled incidents or events which the Commission detennines

. _are significant fr0t'll the standpoint of public hea 1th and safety). The

. : following inc"ident was detennined to be an abnonnal occurrence using the

.. criteria puolish_ed in the' FEDERAL REGISTER on February 24, 1977 (~2 FR 10~50).

• Appendi~ A (Example I.~.2J of the Policy Statement notes that a* major deficiency in desfgn, construction or operation havtng safety.

impli.cations ta.ffecting five plants in this caseJ requi.ring. immediate remedial_ action can be considered an abnormal occurrence.

Th*e fpllowing description of the event al so contains the remedi_-a 1 actions taken.

Date and Pl ace - Duri_ng design and construction, an incorrect summation of earthquake loads affected the desi-gn of safety related pi'pi'l'lg systems and associated pipe supports at five nuclear power plants.

On December 6, 1978, a Licensee Event Report from Duquesne Light Company 1nentioned differen_ces between computer codes used in analyses of force SUDlllations, but did not elaborate on than. Then, the NRC learned of an incorrect sul'mling of loads in one of the codes on March 8, 1979, at a meeti_ng in Bethesda, Maryland with Stone and Webster, an architect engineeri_ng firm and the Duquesne Light °Company (DLCl, the licensee for Beaver Valley Unit l,

7590 a pressurized water nuclear plant located in Beaver County, Pennsylvania.

On March 9, NRC *learned that the incorrect sumation technique affected four other plants:

Plant FitzPatrick Maine Yankee Surry l & 2 Location Oswego County, New York Lincoln County, Maine Surry County, Virginia Nature and Probable Cons~quences - In October 1978, Duquesne Light Company, the licensee of the Beaver Valley plant, was informed by Stone and Webster that for loadi~.9 conditions associated with postulated earthquakes, pipe supports associated with Safety Injection System piping would be overstressed.

Stone and Webst_er (S&W) was reanalyzinQ stresses_ in connection with a system modification r~quired by the NRC staff to correc.t a design defici_-ency not related-to protection against postulated earthquakes. During this reanalysis effort,.the S&'W engineers also came across _tn-formation tha~

had been provided to them by Westinghouse i"n Ma.,y *1978 that showed some check valves in these lines were actually heavi:er than assumed i_TI the earlier analysis.

Sometime during this reana_lysis, ei_the.r -i'n connection wi"th* the planned modification or in reexamining the effect of the.increased *valve we_ights>

S&W discovered a misapplication of a hand calculation method.

ln correcting this misapplication, SMl found some instances of local overstress.

The

f. 7590-01.

e 3 -

correction consisted of adding a snubber and modifying on: support.

In doing the analysis related to making this correction, S&'.{ used two computer programs.

A new one, NUPIPE, predicted much higher* stresses than the one, PIPESTRESS, used during a 1974 as-built check of these lines.

On October 26, 1978, the licensee orally notified the NRC Office of Inspection and Enforcement about the design error (hand calculations method misapplication) which required correction.

No.explanation was provided for the differences in stresses p.redicted by the two codes at that time.

Repeated NRC contacts with the licensee and S&W to *identify the reason for the differences were not effective, since without the actual computer runs to look at there was a communications problem.

Du)ring a meeting ~eld on March 8, 1979 to discuss these matters, the Beaver Valley 1 i censee informed the NRC staff that the differences in predicted piping stresses between the two computer codes were attributable to the fact that the SHOCK2 subroutine of the PIPESTRESS code uses an algebraic surnrration of the loads calculated separately for the horizontal anci the vertical component of earthquake motion.

The use of algebraic surrmation is only acceptable if the ti*me phasing of these loads is known.

The algebraic technique as used in SHOCK2 is not conservative for response spectrum modal analysis because>

in such analyses, time phasing is not considered.

590-01 The analytical treatment of load combinations becomes significant because horizontal earthquake moti ans can produce piping, r.mvement in beth

• the horizontal and vertical direction and the vertical earthquake motions can also produce piping movement in both horizontal and-vertical directions.

For some designs the calculated piping stresses may differ significantly depending on the load summation techniques used in each mode of response *

. Based on the three piping systems that had been rea~alyzed by the

newer code on Beaver Valley at the time of the March 8, _1979 meeting, stresses over a*llowable values were expected to be*found *primarily in piping supports although significant increases in piping stresses had been observed..

.NRC staff reviewers were sent to S-&W' s Boston office to determine the extent of this problem on Beaver Valley 1 and other potentially affected plants.

In following the course of the reanalysis at the S~ offices over the weekend of March 10, 11 *and 12 1 based on the information then available, it became apparent that, when the NUPIPE code was used, a

• number of piping systems had calculated stresses over the allowable.

value for the design basis earth_quake.

Also, for a few of these systems the more probable operating basis earthquake resulted in stresses above the allowable value.

In addition, the structural integrity and pe*rformance of pumps, valves and other essential equipment could be

7590-01 e degraded.

Although results were still_ incomplete on March 12, infonnation available at that time indicated that high stresses were calculated in a number of systems important to safety.

• Because the overstressing of piping and supports was predicted even for earthquakes which ~ight occur during the lifetimes of these facilities, the problem took on considerable safety significance. Some of the systems identified a:~ that time as having overstressed conditions under earthquake loadings were p~rt of th~ reactor coolant pressure boundary, whose failure could cause a loss of coolant accident.

In addition, systems which would be needed to shut the plant down safely in the event of a. loss of coolant accident were also affected.

Thus an _earthquake, of not extremely low likelihood, would have the potential both for causing an accident and for preventing safety systems, designed to *cope witn that accident, from operating.

A secondary concern was whether or not systems needed to provide adequate long term cooling *for the plant in the event of an earthquake without a LOCA could be assured.

Concurrent with the NRC Beaver Valley review, NRC staff records were reviewed to determine whether or not other facilities ~ad used these

-same analysis *techniques. Based on the review of these records and infonnation provided by S&W, the NRC staff concluded that four other facilities used the same techniques.

The four faci1iti*es are 1*1aine Yankee, FitzPatrick and Surry Uni-ts 1 and 2.

i )90-01 The NRC staff concluded the potenti a 1 for serious adiverse effects in the event of an earthquake was suffici~ntly widespr~ad that the basic defense in depth provided by redundant safety systems may be_

compromised.

The NRC Director for Nuclear Reactor Regulation concluded

• that the publ_ic health and safety required that the affected facilities be placed in a cold shutdown condition pending further o;roer of the Cormnission *. Orders to this effect were issued to the 1i:e::ensees of the

above reactors.

The Orders* provide that within 20 days each licensee EJSt respond with respect to:

(1) why the licensee should not reanalyze the facility piping systems for seismic loads on the piping system and. any other affected safety systems using an appropriate priping analysis computer code which does not combine load$

algebraically, (2) why the licensee should not make any modifications ito the facility piping systems indicated by the reanalysis.,. and

{3)

• why facility operation should not continue to be s.1:1S.pended unti 1 comp 1 eti on of the reanalysis and any required\\ modi fi cations.

A11 *of the plants are now in a cold shutdown condition.

(Surry Unit 2 was al !'"eady in an extended outage for stea'm generator replacement.. )

~ :

_ 7590-01 e

7 -

Cause or Causes - The uncertainty in the calculated piping stresses and support loadings in safety-related piping systems at the five plants_

is attributable to the incorrect application of the algebraic summation technique in the SHOCK2 subroutine of the PIPESTRESS computer code, proprietary to Stone and Webster *.

• Actions Taken to Prevent Recurrence Licensee/A.rchitect Enoineer "". Identification of all safety related systems that have been analyzed with a piping computer code invo,lving a program deficiency is underway.

Computer inputs are being checked to assure that -

all r~analyzed piping will reflect the as-built condition at each plant.

Piping analyses~are being rerun'and piping and supports exceediag allowable stresses will be ioentified. Modifications will be made as necessary.

NRC - The NRC ordered each of the utilities of the five identified nuclear power plants to shut down their plants within 48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br />.

The uti 1 iti es are to remain shutdown pending further order of the Comnission.

The NRC is in contact with the 1 i censees and the architect engineer on actions being taken.

Piping stress computer codes to be used for.

• reanalysis Of the piping will be tested with NRC established benchmark problems.

Also', an. independent.audit. of selected piping runs will be

• conducted by NRC consultants to verify the piping stress reanalysis.

.J

7590-01

~ -. In addition to reviewing the licensees* corrective actions, the i1RC is reviewing any generic implications at other facilities.

The J-i?..C's Office of Inspection and Enforcement issued Information Notice (IN) No. 79-06, which descrJbed the event, on March 23, 1979, to all holders of reactor operating licenses and construction permits.* On April 14, 1979, the NRC' s Office of Inspection and Enforcement issued Bulletin No_ 79-07 to applicable licensees which identified actions to be taken.

This includes

• identification of the methods of analyses used, how they were verified, safety systems ~ffected, 'and a plan of action to assure-p1ant safety.

As of May 9,* 1979, the NRC has received responses to Bunetin No. 79-07 from a:.11 1 i cens?es of operating.reactors except for Three Mile Island

. Units l and 2 which are shut down *. The NRC staff is reviewing these responses on a high priority basis. Adoitional actions will be taken as appropriate.

CoI!Jl'li s s i on Dated at Washington, D. C.