ML17292A810

To Draft, Sys Analyses for Simplification of Selected Response Time Testing Requirements

ML17292A810
Person / Time
Site:	Columbia
Issue date:	02/12/1997
From:	GENERAL ELECTRIC CO.
To:
Shared Package
ML17292A808	List: GO2-97-071, Forwards Response to RAI Related to Response Time Testing. W/Related Info ML17292A809 ML17292A810
References
PROJECT-691 NEDO-32291-S01, NEDO-32291-S01-DRFT, NEDO-32291-S1, NEDO-32291-S1-DRFT, NUDOCS 9704220013
Download: ML17292A810 (68)

Text

REDO-32291 Supplement 1

DRF AXX-xxxxx

, Class I February 1997 r

BWROwaers'roup Liceasing Topical Report System Analyses For The SimpMcation Of Selects Response Time Testing Requirements WorkPerformed forthe BWROwaers'roup Response Time Testing Conunittee.

NRC Project 691 97042200iS 970415 PDR ADQCK 05000397 P

PDR,

s

NEDO-32291, Supplement 1

t B.3 GE Type HFA 120 Vac Relay Failure Nodes and Effects Analysis (FNEA) 4 B2.1.

FMEAResults Pt B222 EMEAApphbiRyy P%c'his FMEAaddresses GE Type HFA 120 Vac Relay Model Numbers HFA51, HFA71, HFA151, and HFAI71. SpecKcatly, the results and. Conclusions apply to the foHowingequipmeuL a) AnyGE Relay Model Nmnber 12HFA51A49 or 12HFA71A49A

~,

b). AnyGE Relay Model Number 12HFA51A62 c) AnyGE Relay Model Number 12HFAISIA9ar 12HFA171A9A t

I' B&.14 Boundi'ng Response 2Vmes.

The maximum undetected zesponse time af the zelays is 40 ms provided Teclmical Specificauon; >>,,'.

Functional Tests (ar ~lnvalent) are performed and include the relay, aud provided the following coudltmns aze met t

a)

The HFAmanufacturer's instructions aze followedforsetup and adjustment ofthe relay prior..

to initialoperation aud. after any zepair ormainumance and.

b) Prior to installation or af'ter any maintenance orzepair ofthe relays, the normally apen

-contacts ofthe zelays are coufzrmed to open in 20 ms or less. after power is removed Qom the.

~

COIL Provided these conditions aze met, all credible fhilure modes identified thatmctease zesponse time.

to more than 40 ms also affect normal fimctiomngofthe trip unit, and thus are detectable by tests other than RTL Even forthe identified "undetected'hiluzes,.operating history indicates that theyare nota probable failure. Even though some performance problems have been identified in the. past, all were identified by tests or actions other than zesponse time tests I

t tran..fri',<<i.

r Based on ths analysis, ltlszeasouable to conclude that any fhliure ofa sublectrelay which has

. met the conditions ofthis section that could zesult in response times greater than 40 ms is not credible BB2, Analysis 1

BODE; 'quipment Analjzed A GE. 120 Vac Type HFARelay Model No..l2HFAS1A49F was selected far.detailed. analysis.-.

'922.'Simi7arity Analjsis

~erelay analyzed is a Model No. 12HFASIA49Frelay. The "P'ufRc indicates a semi-Qush mount case design. The basic model also cmnes withno sufBx indicating sudacemount withrear:

connections and an "H"suffixindicating a surface mount case with"front"connections (actually

~

~

h 4

NEDO-32291, Supplement I on the side). The only diffezence between the "P'ase and the "ao sufBx" case or the "K'ase is the actual molding and shape ofthe exterior ofthe case. The "EP case also differs in the rooting ofthe external contact connection points. The "K'ase has the external connections forthe coatacts routed out to the side compared to the xear forthe "F'ase.

AHofthese models have the same relay, coil, contacts and xetum spring mechanisms, so the difference do notxesult in any different failure modes than those analyzed, and the conclusions ofthe analysis may be applied to all ca@ally.

Amodel 12HFA71A49Azelay is the same as the analyzed model except that itis mounted in a, "drawout" case. This diffezence, IBce forthe suf6x "EP model, affects only the external molding and shape ofthe case, and the connection details external to the basic zelay case. This model also has the same zelay, coil, contacts and zeturn spzing mechanisms as the analyzedxelay, so the diffezeaces do not zesultin any different Mmmodes than those analyzed, and, the conclusions of.,. ",

the analysis may be applied to this model as welL t,

Relay Models 12HFASIA62 are the same as zelays witha "49"foHowingthe "12HFASIA."base".:,'":"--':,".

number except that the coilzatiag is 125 Vac, 60 Hz (vs. 115 Vac, 60 Hzforthe ana?yzed'model)="-,"".'.- -.'.

The differences between this model and,the model analyzed axe limitedto the coKonly. This

-" ".-,: =-::-'. z.'"=.-.".:.

couldxesuiti slightly different failurlimitsforsome ~crating conditions,,but none that affec;.;,,"'-. ji'"'>>;-

the xespoase time of.the relay in the de-energized dizectiozx Thezefoze,,the conclusions ofthe:

analysis may be applied to these models as. welL

~

~

~

g t

Models 12HFA151A9 and 12HFA171A9A differAnnModels 12HFA51A49 and'12HFA71'A9A;,';.-~:":,.~'-'",

zespectively, only in the coil design.. For the 12HFA151 and 171 relays, the coilhas different';,",'.'aterials forlongerlife and.is, rated nominally at 120 Vac, 60 H'z (vs. 115 Vac;.60 8'z forthe:

analyzed model)

These difference could xesuit in slightlydif6ment Qulme limitsforsome operating conditions but none that affect the zesponse time ofthe zelay in the de-eneqjmA' dizectiozx., Therefoxe,: the conclusions ofthe analysis may be applied.to these models as weIL'.

~,

5M, I

Allxeiaymodels may have supplemental codes indicating the contact conGguration ofthe xelay.as;-

supplied:Gom the factory However, the coatacts are designed to be Geld.changeable, to the Geld'onfigmation can be any combinatioa ofone ormozenormaIIy open and up to Gve nozmally closed contacts (forthe:cases covered. by this analysis, at least one coatact is conGgmed.as a

. nozmally open contact). The analysis assumed the both limits,- ieallnozmally open and all'ozmally closed, so the conclusions of'the analyst apply zegazdless ofinstalled contact conGgmatzon-c' 4

Some of'theabovediscussedzelay models have been.upgzaded withimpzovedcoii:assemblies since oziginal delivery.. These diffezeaces could zesuitm sEghtly differentRihzze limitsforsome operating conditions; but none that affec the zesponse time ofthe xelay in the de-energized.

direction Therefoxe;,the conclusions ofthe analysis may be applied to these upgraded models as weIL I

B-18

NEDO-32291, Supplement 1

B929 Revinv ofOperating Experience The review ofoperating expexience documented in Ref.9.1 shows that thexe have been several types ofproblems withHFAzelays, some ofwhich had the potential to impact xesponse time of the xelay. However, most ofthose problems were detected by functional tests ofthe zelays.

Appendix D includes a mnmazy ofidentifiedHFAzelay problems that were detected based on zesponse time. Item 8 in Table D4.1 applies to the GE HFAset ofrelays and documents a total of 3 identified faBmes detected by zesponse time degtathtion.

The problems were detected. in the Turbine Valve Closme fimctionin the RPS. The description does not say which specific function, but does state that the loop was "marginally'outside acceptauce criteria". The Tmbine Valve Cloazze.functions have zesponse time p~mements on.-

the order of80 ms or less. Therefore, itiszeasonable to conclude that the actual times meas~

forthe "hBed"loops were on the order of100 ms.-

The expetience data avaBablelacks some qecificdetaBs necessary to xeach absolute conclusions

'egaxding detectabiTity ofdegraded zesponse time conditions., However, dne to the very smail.'umber ofreported cases ofzelay failures detected byxesponse time testing the zelatzvely large -'-

number detected by other smveillance testing, and, the'fact that aEof'the idaztifiedfaBuzes *".."

'etected by xesponse time testing were foravery shoxtzesponse time function, itappears reasonable to conclude thatfaBures thatzesultin increases inzesponse time but that do notzesuIt in functional failmes are very IRely to xesult inonlyvery smalI increases in zespouse time.

~

~

BB2.4 Detai7ed Analysis B.32.4.1 Description The GE Type HFA, a semi-Gush mount multi-contact amdRuy zelay Model No. 12HFASIA49F, includes a fixedcoil/core assembly and.the Gxed halfofsix contacts mounted.directly to a housing, and a moving azzxratuxe/contact assembly heM against pivotpoints on the housing by an adjustable spring/screw assembly aud an axmatuxe stop mounted to the housing. Feedthroughs molded into the housing pass the elecuical connections fiomthe front side of'the housing to the zear ofthe housmg. "Pins" pzotrude out the zear side ofthe feedthzoughs to foxm the connection terminals to the zelay. Acover witha "window"covers the frontofthe assembly vIitha viewof the azmatme/contact assembly through the window The whole assembly is approximately 6S mcus wide by 7 inches high Thehousingis approximately'nches deep withthe contact pins pzotruding appzoximately 15 inches further to the rear The cover pxotzudes approximately one inch to the frontofthehousing. The relay mounts thxough a panel with the window visible fromthe fzont..

The coil connects via shoztleads to theirassociated feedtimughs.

The fzxaKhalf'ofthe contacts which connect dizectly to their associated feedthzoughs are zevexsible to formeither a normally open (¹O.) or normally closed (N.C.) contact axrangement.

Small "braids connect the moving B-19

~

~

0

NEDO-32291, Supplement 1

halfofthe contacts, those mounted on the azmatuzc/contact assembly, to their associated feedthzoughs in the housing.

The azmauue/contact assembly is a "sandwich" assembly oftwo molded plastic parts, six contacts with one end each attached to the small braid, a metal azzzeture piece and twelve springs, two for each ofthe contacts.

The molded plastic pieces are about 45 inches long and extend essentiaHy forthe widthofthe zelay. The sandwich assembly is appzoxim Wry0.7 inches thick (depth) and 12 inches high with the armature piece auached to the back ofthe assembly. The contacts pass parallel to each other between the two plastic pieces which also support the two springs foreach contact and extend approximately 1.6 inches beyond the molded plastic top ofthe sandwich, across the space occupied by the coi1/azmatuxe assembly, to the fixedcontacts. 'Ihe springs provide the closing force forthe contacts, which axe zelatively zidged, one spring foreach direction ofthe contact(N.O. or¹C.).. To the azmatuze piece attached to the back ofthe azmatuzc/contact assembly is aominaHy a triangle withthe corners cut off,aad a zectangular "tail" iu the direction. away hn the contacts.

Jn the nozmal mounting confiyuation, the tail is down aad the contacts point up The bottom ofthe axmature plate at the tailis a small "stop" bracket h

The two lower corners ofthe armature piece zeston two."chair-like'eats molded into the housing, each approximately 1/4-iach square These seats axe on either sMe ofoae end ofthe coze piece, and fozm the pivotpointforthe azmatuxe. The back ofthe,"chaiz" limits the downward travel ofthe azmamafcontactassembly while thesmall bracket on the bottom meets the pole piece to limitthe upwaxd. traveL The housing limits the travel ofthe azmatme/contact assembly to the side. When the relay is eneqpzed, the.azmatme/contactassembly is held firmlyin place bythe magnetic field The armand zetmn spring has one end connected to the housing and one eud connected via an adjustable attachment to the "tail"ofthe axmatme This spring hoMs the armature open when the relay is not energized A"stop" bracket mounted to the housingnextto the fixed contacts extends over the end ofthe azmature plate aad with an adjustable stop limits the travel ofthe armature in.the open direction. The coze pieces limitthe azmauue travel in the closed direction.

The pick-up voltage ofthe relay is set by adjusting the azmatuze/contact zeuun spring. The maximum axmatme opening distance is adjusted by adjusting the "stop". This detezzaiaes both the maximum travel in the direction,to close ¹C. contacts, and.the travel required to close the armature when the rehy is eaezgmxi The contact position between open aad close is adjusted by bending the moving contacts.

The pick-up voltage, the contacrpositioaiag, aad.the azmatme travel indirectly determine the operation time ofthezclay contacts Thexc is ao ducct adjustment ofzesponse time 8.3.2.4.2 Results of FMEA GE Type HFAxelay RVlEAzesults aze tabulated in Table B3-I aad. Table B3-2. Allcredible fhilure modes identified that inczease xespozm time more than approximately a factor of2 also affect normal functioaiag ofthe relay, and.thus azze detectable by tests other than RTI'. This B-20

I

. ~ <

..NEDO-32291, Supplement I conclusion assumes that Technical Specification Functional Tests are perfozmed and include the zelay, aad that the manufacnzrer's recommended setup and adjustment procahizes aze foHowed.

Even forthe identified "undetected" failuzes, operating history indicates that they aze not probable failures. Even though a number ofpezfozmance pzoblems have been identified in the past, all were identified by functional tests, not zesponse time tests. This result is zealnable due to the simple basic design and the high degree ofinteraction between mechatnstns that affect zespons time and those iavolved in nozmaI Smctioning ofthe relay. Based on this analysis, itis reasonable to conclude that any faiiuze or degradation ofthe zelay that couM result ia an inczease ofmoze than a factor of2 in the "de-energize" response time (drop out time) is aot credible pzovided the utility has implemented the'vendor zecommended adjustment procedures followingany zeplacemeat or zepair ofthe zelay.

B.3Z.4.3 Bounding Response Times There is ao vendor's specification forzespoase time forthe HFA 120 Vac Relay models covezed by this analysis. However, GENE's acceptance criteria forthe zelay zs 60 ms maxuzaan for closure of'a normally closed contact and 20 ms maxinann foropening of.a aozmally open contact, both measuzed from the time power is removed fzom aa energized zelay.. For the RPS and Isolation circuits, all zelays in the trip path are energized in the non-tzipped state and utilize the nozmally open contacts (contacts open and zelays de-etumgize to transmit the trip condition).

GENE'qualifzcation testing ofGEHFA-AC'zelays showed'operation times of9 ms (vs.

specification of20 ms) and 17 ms (vs. specification of60 ms) (Ref. 9Q Based on this KMEA,,

.- the worst case expected delay forthe HFA Rehy is-a factor of2 increase over nozmnaL 1

The similarity analysis concludes thatdiffezences between zelays coven@i by this analysis do not affect the detectability ofzesponse tizne degradation.. Using the specified maximum of20 ms fora nozznally open contact, which should be conservative, the worst case expected. delay ftom the HFARelay is 40 ms (nozmally open contact, zeiay changing fzom energized to de-energized)

No additional nntrgin isjudged necess tzyforthe HFARelay, pzovided the ze1ay to which the zesults ofthis analysis are applied has an initialacceptance czitezia of20 ms maximum opening time for nozmally open contacts (fzom energized state), and such criteria is applied forany zeplacemeat zelays orafterrepair or, maintenance ofazelay.

8-21

IgD 1, Supplement 1

4 Table B.3 GE Relay Type HFA Principle Design Components and Their Primary Function Item Descri tion Housing 0

Cove[

Rection

~ Provide the pivot"point'or the armature/contact assembly.

~ Retain one end ofthe armature/contact return spring,

~ Provide mechanical support forail ofthe components.

~ Limitthe lateral travel ofthe annatute/contact assembly.

~ Position the core assembly which in turn limits the armature/contact assembly travel,

~ Position the fixed contacts.

~ Provide a dust cover

=

~ Prevent personnel contact withelectiical contacts

~ Prevent forei ob ect en into the rela mechanism 3.

5.

Armature/Contact assembly molded plastic parts, overall assembly Moving contacts and "extensioq" braid Springs formoving contact Armature plate with attached "stop" bracket

~ Position the contacts ielative to the armature.

~ Position the contacts.

o Hold the contact springs, iLimitthe travel ofthe contacts..

=

~ Make (N.O. contacts) or interrupt (N.C. contacts) the circuit when relay is energized, and conversely when relay is de<nergized.

~ Connect to the feedthtoughs in the housing (braid)

Transfer contact "wi

" force from contact s rin s

~ Provide wiping arid seating force for the contacts iProvide mar forcontact ad stment variation iProvide closing force forthe relay contacts when relay is energized via item 3.

~ Trmsmit the opening force from the armature/contact return spring to the to the contacts when the relay is de-energized via item 3.

~ Limitthe travel ofthe armature/contact assembly when the relay is de-energized (limits travel ofcontacts) via item 8,

~ Provides the pivot point forthe armature/contact assembly

~ Lhnits the travel in two directions ofthe armaturelcontact assembly

~ Com letes the ma etic circuitforthe rela B-22

~

t NEDO-32291, Supplement 1

B.7 RPS Scram Contactor Relay Failure Nodes and EN cts Analysis (FNEA)

B.7.1 Contactor FMEAMethodology The RPS Scram Contactozs aze common to all RPS scram iasttttmeat loops. Therefore, a test of any RPS scram instrument loop tests the RPS Scram Contactors.

This FMEAdoes aot dixectly address failures ofthe RPS Scram Contactor, but rather assumes that failures can occur that affect zespoase time ofthe contactors. This FMEAfocuses on identifyingRPS scram instrument loop RTI's that aze expected to bexetained by the utility,and analyzes those RTI's to determine the bounding values ofRPS Scram Contactor response time that can xesult without detection by those xetaiaed RTX's.

SpeciQcally, this analysis ideatiQes retained RPS scram instrument loop RTI's, identiQes the acceptance criteria (m'primum time allowed) forthe RTT, aad subtracts hn that the zespoase time forloop componeats other than the RPS Scram Contactor in order to establish the response.

time forthe RPS Scram Contactor.. In order to determine the bounding mxtimum credible RPS Scram Contactor (that stillpasses the RTX},the mixtmmm credible zespoase time forthe instrument loop components other than the RPS Scram Contactoris established aad used in this analysts r '

lla This methodology'assumes that the utilityperforms the zetained RTI'as a total, loop without intermediate measurements.

In some cases utilitiespexfotm the zetained RTTin overlapping paxtial tests. In those cases, the utilitymay have a bases to show a smaHer RPS Scram Contactor boundiag xesponse time due to a lower acceptance criteria (smaller time) forthe partial loop and fewer "other component" xespoase times to subtract Rom the acceptance criteria(and.therefoxe, less uncertainty)

However, other than to acknowledge that alternate method, this analysis does not address partial loop xesponse time test appzoaches.

The conclusions ofthe analysis aze in the formofboundiag values ofxesponse times that may go undetected, and tests that are assumed to be pezfoxmed in the detezzznnation ofthe identiQed bounding values.'.7Z FMEAResults--

B.7ZZ,,- FMEAApplicability

. This HVIEAaddresses GE CR105, GE CR205, and. GE CR305 Magnetic Contactozs, and Potter R BzumQeM MDRRotary Relays when used as aa interposing xelay between two ofthe above

- Magnetic Coatactots SpeciQcally,. this analysis applies to:

. a) AnyGE CR105, GECR205, and GE CR305 Magnetic Coatactors, and b) Any 120 Vac Small ACNon-Latching type Potter 8c BrumQeld MDRseries xotazy relays, B-53

\\

0

NEDO-32291, Supplement 1-pzovided the components are applied as an RPS Sczam Contactor controlling a set ofScram Solenoid Pilot Valves (SSPVs) in one ofthe followingtwo conGguratioas:

1) One GE CR105, GE CR205, or GE CR305 Magnetic Contactor that dizectly opezates a set of SSPVs, or

2) One intedacing Potter 8h Brum6eld MDRrelay which controls a GE CR105, GE CR205, or GE CR305 Magnetic Contactor which operates a set ofSSPVs.

~ C For. this analysis, either configuration is zefezzed to as the "RPS Scram Contactoz" tr t

- B.722 ConctusionsofIiMEA

'Ihe maximum undetected zesponse time ofthe RPS Scram Gmtactor is 65 ms provided the plant pex&zms APRM upscale scram tzip RTl withan acceptance critezia of90 ms z mamtnn, and the AE'RMRTTincludes the APRM e1ectzonics and at least one intezposiag relay, not shared by other loops, between the APRMoutput and'the RPS Sczam Cot@actor.

~

J

'mvided these conditions are met, all czediblehihzze modes identified that increase zesponse time ofthe RPS Scram Coatactor to moze than 65 ms also zesuitm &iImeto meet the APRMupscale.

scram trip RTT acceptance critezia.,',;

t t

~

= '%"

~

-,C*

Based on. this analysis, itiszeasonable to conclude thatany failuze ofthe RPS Scram Coatactor..

which has met the conditions ofthis section that couMzesultin response times greater than 65 ms -', ~;, -'---

isnotczedible. '

\\'

C+t B.TB Analysis:

- ~

'J t

J

$ A

,~'

B.742, Loop Analyzed The APRMupscale scram trip (120%%uo) instrument loop fora typical BWR4, BWRS aad BWR6 were.evaluated..

'J t

".~B.7D2 Method ofAnalysis,

.Theanalysisisconductedinmultiplepartsincluding.

1) ananalysisofthetypicalloop to

, detezmine the compoaeats in the loop, 2) a review and evahzation ofavailable vendor and.

..'ualification data to detezmine the minimum 1Rely +~ouse time forcomponents in the loop other

'than the RPS Scram Contactoz"3) identi6cation ofthe RTT acceptance czitezia (zesponse time),

', and,,based on these 4) calculate. the maximum credible RPS Scram Contactorzesponse time.that',

wouMstillpass'theRTT.

't

~,

+

t B.'7BB:-.'esnipdon.

The. typical.'APRM upscale scram tzip instrmnent loop mcludes the APRMelectzoaics and output.

ze1ay (the output zelay is considered paztof the.APRM), an inter5tciag relay; and the RPS Scram

~ Contactor.

J C

r B-54

I

NEDO-32291, Supplement 1

The design ofthe APRM is siznilar for aH plants. The zesponse time delays in the APRMportion ofthe loop can be considezed in thxee parts:

1) Qlter delays at the LPRMinput amplifier (between the LPRM detectors aud the APRMprocessiag electzoaics),

2) electronic Iuocessmg delays between the LPRM amplifiezs and the input to the APRMoutput xelay, and 3) delays dae to the zesponse time ofthe APRM output xelay.

The interfacing zelay in the BWR4 generation plant is typicaJly a 120 Vac GE HFAzclay. The BWR5, BWR6 (and some later BWR4) plants generally use a Potter 4 Bzmnfieid MDRxelay, also 120 Vac.

For BWR4 plaats, each instrument loop typicaHy operates two RPS Scram Coatactozs in pazaIIel, each ofwhich is typicaHy single CR105, CR205 or CR305 that is operated directly from the interfacing zelay. For BWR5 and BWR6, each instrument loop typicallyoperates either two or four RPS Scram Contactors where one is typically a single CR105, CR205 or CR305 that is opezated directly fram the interfacing zelay while the zemainder compzise a Potter 8h Bzzan&M MDRzehy operated directly fromthe inter6tcing xelay aud a single CR105, CR205 or CR305 which operates fmm the MDRrelay.

In aH cases, the RPS Sczam Contactor is common to aH RPS sczam instrument loops while the intezfacing zeIay is dedicated to the specific loop.

Response time testing ofthe APRMupscale scram trip iastannent loop includes the APRM electzonics, the interfacing relay, and the RPS Scram Contactors (aH contactors operated from the loop). The normal acceptance value forthe APRM upscale scram trip response time is 90 ms.

The design geaezaHy assmnes 40 ms forthe APRMzespoase and $0 ms forthe intedhcing xeiay and RPS Scram-Contactors B.70.4 Analysis The APRMzespoase time is the sum ofthe filterresponse delay, the electronics and the APRM output zelay. The filteris nominally a 15 ms time constant Qlter. The specific delay that xesults from this filterdepends oa the characteristics ofthe input signai, but is typically between about 10 and 20 ms. The electronics delays are short, on the ozder ofa few milliseconds. The APRM output zelay is a smaH, fairlyfast zelay, which typicallyzesponds in less than 10 ms.

The zesponse times axe likelyto zemain zelatively constant, but inparticular axe not hkely to become substantially faster than nominaL Based on a 40 ms designed zespoase time, itis assmned, based on engineering judgment, that the fastest response ofthe APRMpart ofthe channel is 20 ms, one halfofthe design allocatioa.

The Potter 8t: Bzumfield MDRzelay (smaH ACaon-latching) is specified to xelease in 5 to 18 ms.

Itisassumed forthis calculation that the release timeis 5 ms. TEe HFA 120 Vac zelay is sIeciGed to have a release time of 14 ms or less. GE qualification tests memued times oa the oxder of9"

,,ms To bound these values, the mimmum release time forthe interfacingzelay is assmned to be 5 ms. This also covers any case that uses an Agastat GP orEGP relay as the interfacing relay.

B-55

~

y

~,

NED0-32291-A R. 6. 1 Tzansmigtezs/Switches Xncluded in EPRZ Ana1ysas The EPRZ analysis scope included the majority of pressure sensor instrumentation currently installed or expected to be installed in U.S.

plants.

The pressure sensors which aze applicable to the BWR plants participating in this BWROG study are the Barton, Rosemount, and SOR transmitters/switches.

~Sensor failure modes assoc'ted with all Barton transmitters, models 763 and 764, switches, model 288/289, and',SOR switches were not found to affect sensor response time without significantly affecting calibration.

The BWROG eviewed and provided comments on the draft EPRZ analysis report prior to issuance.

All comments were addressed in the final report.

Only two failure modes and two manufacturing/handling defects were identif'ed in-Reference 1 as affecting response time without concurrently affecting sensor output.

These failure modes and defects apply only to sensors utilizing a fillfluid to transfer the process pressure to the sensing element.

Rosemount supplies the only sensors of this type identified for plants participating in this BWROG study.

The two ailuze modes are the slow

'ss of f'llfluid during pressurized operations and variable damping potent'ometer misadjustment during maintenance.

The two manufacturing and handling defects aze low sensor fillfluid from the manufacturing process and crimped capillaries from the manufacturing process,

~roper handling by the manufacture

, or damage during field installation/maintenance.

A discussion of these failure modes and effects aze included in Appendix F.

The effect of these failure modes and effects on RTT elimination can be summarized as ollows:

(1)

Slow loss of fi11 f1uid - A slow loss of fillfluid causes a gradual degradation of the measurement process by reducing the ability of the working fluid to rapidly transmit pressure changes to the sensing diaphragm.

Current response time tests aze ineffective 'n detecting the initial stages of slow fluid loss.

For sensors that are susceptible to the slow loss of fill-oil,Drift Analysis is the preferred method to detect the change 'n instrument perfonnance.

Other diagnostic techniques such as sluggish zesponse and process noise analysis may be used to supplement Drift Analysis.

When enough fluid Reference 10-13) is lost to cause a significant response time degradation, the sluggish response of the leaking sensor will be detected during =zansmittez calibration.

K-6

NED0-52291-A (2)

Variab1e damchin tentiometer misad'ustment -

Damping devices have been utilized in fast acting trip circuits to minimize the potential foz inadvertent actuations.

The use of a vaz'able damping potentiometer in the transmitter design provides a means of applying the same type of transmitter to several circuits that require additional elect onic fi3.tering capabi3.ity.

Variable damping potentiometer misadjustment can affect the response time in circuits having capacitors and resistors that control electronic response time.

Measures must be taken to ensure that the potentiometer is at the requized setting at time of installation and after major maintenance.

Therefore, no additional response ~e tests are required.

A more detailed discussion on damping filters is also found in Appendix F.

(3) Manufac~

and handlin defects -

Low sensor fillfluid during manufacturing and crimped fillcapillaries due to manufactuz'ng or mishandling during installation/maintenance were identified in Reference 1 as affecting transmitter response time.

Response

time is the only sensor characteristic affected by these manufacturing and handling defects.

Since November

1989, vendor testing has been implemented to ensure acceptable filland capillaries.

~ n addition, when low fillfluid or crimped capillaries affect response time, the degradation can be identified by pxe-installation calibration.

K.6.2 Tx'aneaitters/Switches Not Included in EFRZ Analysis The following two switch models axe not part of the FPRZ report (Reference 1 of main report) but wexe supplemen-ed by BWROG.

(K.6-2.1 Baz3csdale Pressure Switches The on y potential failure mode foz model TC9622-3 (piston with 0-ring) occurs if the switch is misapplied in process oz, range.

The 0-ring seal can swell due to pressure above its rating, and th's swelling causes the plungex pin 'co react sluggishly. This will increase the 'nstrument esponse time.

Since safety-re ated switches are carefully spec'ed and verified, this failure mode 's considezed extremely unlikely. The only electrical fai3.ure mode occurs

=n the microswitch. This will not produce a delay, but will cause failure to operate, which can be readily detec"ed during survei13.ance testing.

K-7

NED0-32291-A lThe Barksdale B1T and B2T series are Bourdon tube instruments and do not have components that can cause response time related failures.

Therefore,

>response time'esting is not required.

K.6.2.2 Barton 760 Transmitters Barton 760 is a differential pressure tzansmitter which contains a

mechanical bellows and electronic circuit similar to Model 764.

As concluded in Reference 1 of main report, response time testing is not required for the Barton 764 Model. This conclusion also applies to Model 760.

K. 7 Loop Devices K-7 ~ 1 Rosemount/GE Trip Unit Noise Suppression Filter Capacitor The WBR 2000-50 capac'toz, used as a process noise filter in an analog transmitter loop, is manufactured by Cornell Dubilier-Sangamo Components.

Zt ls a 2000 ufd aluminum electrolytic capacitor with a 50V rating.

Failure modes of the capacitor aze (1) open, (2) short, (3) increased

leakage, and (4) change in capac'ance.

When installed in pazallel with the trip unit input, short circuit and increased leakage current failures can affect analog loop accuracy.

Loop calibration procedures (end to end) performed on a periodic basis can demonstrate loop operability within the required performance requirements as long as the capacitor is in the ci"cuit during the procedure.

-pen circuit failuxes aze in the conservative direction and are not a concern with respect to response time.

apacitance change failures can 'clude (a) decxeased capacitance, which is in the conservative direction with respect to response

time, and (b) an 'ncxease of capacitance.

The vendor states that capacitance may increase by 108 with time.

These parts are alzeady specified with a -10%/+75% tolerance.

The time delay added by the capacitor should have sufficient margin to the maximum allowable to account for this possible

'czease.

With surveillance tests demonstrating loop operability, there are no failures with the capacitor which will adversely affect system response time.

K.7 ~ 2 745 Alarm Unit These alarm units aze used only in trip functions such as reactor watez c'eanup isolation.

A review of the schematic ciagzam revealed that only the input 4 '9K resistor and 10 miczofarad capacitor contribute to a delay time on the order of 50 milliseconds.'

the 'nput esistor failed to a highe" K-8

E

differences between response times for transmitters with a 41.4" 173.2" span and transmitters with a 107.5" 447.3" span (instrument range values not given).

FMEAs were performed on the sensor-types listed in table 2-1 below.

Based on the data collected from the industry for this investigation, these sensor-types represent the ma)ority of pressure sensor instrumentation currently installed or expected to be installed in safety-related systems in U.S. plants.

These sensor-types also are representative of the various sensor designs (e.g.,

bourdon tube, force-balance, capacitance, and strain gage).

Table 2-1 SENSOR TYPES INCLUDED IN FMEAs Barton 288/289 Differential Pressure Indicating Switches Barton 763 Gage Electronic Pressure Transmitter Barton 764 Differential Pressure Electronic Transmitter Foxboro N-E11DM Differential Pressure Transmitter Foxboro N-E13DM Differential Pressure Transmitter Foxboro N-E13DH Differential Pressure Transmitter Foxboro N-EllGH Gage Pressure Transmitter Foxboro N-E11GM Gage Pressure Transmitter Tobar 32PAl Absolute Pressure Transmitter Tobar 32PG1 Gage Pressure Transmitter Tobar 32DP1 Differential Pressure Transmitter Rosemount Differential Pressure Transmitter Models 1151, 1152'153, 1154 Rosemount Pressure Transmitter. Models 1151,

1152, 1153, 1154 Statham PD-3200 Differential Pressure Transmitter Statham PG-3000 Pressure Transmitter SOR Differential Pressure Switch SOR Pressure Switch The FMEA method of systems analysis was selected for the response time investigation because it provides a valid, systematic approach for identifying failure modes.

FMEAs are a semi-quantitative technique to the extent that they are not supported by explicit calculations or testing for particular failure modes.

FQ.1 fluid viscosity effects, frictional linkage forces, and capillary effects are addressed on a generic basis.

The failure modes included each physical boundary and force transmitting element, regardless of the probability for a failure mechanism.

2-4

Table 3-33 i SOR DIFFERENTIAL PRESSURE SWITCH

~

FAILURE MODES AND EFFECTS ANALYSIS TABLE NO.

NAME FAILURE MODES SYMPTOMS AND LOCAL EFFECTS INCLUDING DEPENDENT FAILURES METHOD OF DETECTION EFFECTS ON SENSOR

RESPONSE

REMARKS AND TIME OTHER EFFECTS I

High pressure Leak chamber Moisture/boron on switch body Potential increased tempera ture Potential errors due to sample line pressure drop Visual/signal comparison None None 2

Low pressure chamber Leak Moisture/boron on switch body Potential increased temperature Potential errors due to sample line pressure drop for large leak Visual/signal comparison None None 3

Diaphragm Leak Increases setpoint error as leak increases Setpoint test None None Potential increased temperature

".',"jest jiS,"j-'

Table 3-33 (Cont'd)

SOR DIFFERENTIAL PRESSURE SWITCH FAILURE MODES AND EFFECTS ANALYSIS TABLE NO.

NAME FAILURE MODES SYMPTOMS AND LOCAL EFFECTS INCLUDING DEPENDENT FAILURES METHOD OF DETECTION EFFECTS ON SENSOR

RESPONSE

REMARKS AND TIME OTHER EFFECTS 4

Piston shaft Attachment problem/defect Setpoint change or failure to operate Setpoint test None None 5

Rotary shaft Failure/defect Setpoint change or failure to operate Setpoint test None None 6

Rotary shaft bearings Excess freedom Excess friction Erratic setpoint Setpoint change or failure to operate Setpoint test Setpoint teat None None None None 7

Rotary shaf t 0-ring seals Leak to actuator spring chamber Leak to microswitch chamber Moisture/boron in spring chamber Potential setpoint change due to seal problem/flow Moisture/boron in switch chamber Potential increase in temperature, increased conductivity when switch in "open" position Inspection and setpoint test Inspection and setpoint test None None None None

Table 3-33 (Cont'd)

SOR DIFFERENTIAL PRESSURE SMITCH FAILURE MODES AND EFFECTS ANALYSIS TABLE NO.

NAME FAILURE MODES SYMPTOMS AND LOCAL EFFECTS INCLUDING DEPENDENT FAILURES METHOD OF DETECTION EFFECTS ON SENSOR

RESPONSE

REMARKS AND TIME OTHER EFFECTS 8

Microswitch At tach ment actuator linkage problem/defect Setpoint change or failure to operate Setpoint test None None 9

Microswitch Change in actuation force Electrical failure Setpoint change Spurious output or failure to operate Setpoint test Setpoint test None None None None 10 Setpoint spring Change in compression Setpoint error Setpoint test None None Change in force Setpoint error cons tant Setpbint test None None

~I

Table 3-35 SOR PRESSURE SMITCH FAILURE MODES AND EFFECTS ANALYSIS TABLE NO.

NAME FAILURE MODES SYMPTOMS AND LOCAL EFFECTS INCLUDING DEPENDENT FAILURES METHOD OF DETECTION EFFECTS ON SENSOR

RESPONSE

REMARKS AND TIME OTHER EFFECTS I

Inlet pressure Leak chamber Moisture/boron on switch body Increased temperature as process fluid leak increases Visual None None 2

Inlet diaphragm Leak Moisture/boron on switch body Increased temperature as process fluid leak increases Setpoint test None Switch failure likely Increased conductivity in switch "open" state, setpoint change as leak increases across diaphragm 3

Actuator shaft Friction Setpoint change or failure to operate Setpoint test None None

Table 3-35 (Cont'd)

SOR PRESSURE SWITCH FAILURE MODES AND EFFECTS ANALYSIS TABLE NO.

NAME FAILURE MODES SYMPTOMS AND LOCAL EFFECTS INCLUDINQ DEPENDENT FAILURES METHOD OP DETECTION EFFECTS ON SENSOR

RESPONSE

REMARKS AND TIME OTHER EPPECTS 4

Actuator spring Change in compression Setpoint error Setpoint test None None Change in force cons tant Setpoint error Setpoint test None None 5

Mlcroswitch Change in actuation force Setpoint error Setpoint test None None Electrical failure Spurious output or Setpoint test failure to actuate None None

Table 3-5

BARTON MODEL 288/289 DIFFERENTIAL PRESSURE INDICATING SWITCH;FAILURE MODES AND EFFECTS ANALYSIS TABLE NO.

NAME FAILURE MODES SYMPTOMS AN) LOCAL EFFECTS INCLUDING DEPENDENT FAILURES METHOD OF DETECTION EFFECTS ON SENSOR

RESPONSE

REMARKS AND TIME OTHER EFFECTS 1

High pressure housing and seals to transmitter housing Process fluid leak Signal drift due to increased heating of transmitter by process fluid and/or surface conductivity due to process fluid Errors due to sample line pressure drop for large leaks Electronic failure Signal drift with respect to comparable signals None None 2

High pressure bellows Leak to high pressure housing Process fluid enters high pressure bellows due to action of bellows spring force Signal drift due None to action of bellows spring Can affect low range limit due to high pressure overrange valve 3

High pressure Incorrect bellows spring spring Calibrated out or detected during initial calibration Initial calibration None None Change in spring constant Signal drift due to change in valve stem shaft spring constant Signal drift with respect to comparable signals None None

Table 3-5 (Cont'd)

BARTON MODEL 288/289 DIFFERENTIAL PRESSURE INDICATING SWITCH FAILURE MODES AND EFFECTS ANALYSIS TABLE NO.

NAME FAILURE MODES SYMPTOMS AND LOCAL EFFECTS INCLUDING DEPENDENT FAILURES METHOD OF DETECTION EFFECTS ON SENSOR

RESPONSE

REMARKS AND TIME OTHER EFFECTS High pressure bellows to high pressure housing seals Leak to high pressure housing Process fluid enters high pressure bellows due to action of bellows spring force Signal drift due to actions of bellows spring None None 5

Fill plug seal Small fill fluid leak to transmitter housing Substantial fillfluid leak to transmitter housing Slow collapse of bellows folds Damage to bellows Fill fluid in transmitter housing Fill fluid in transmitter housing Signal drift due to restoring action of bellows spring None None None expected due to flexibility and volume of bellows Transmitter will not calibrate and may have low range

, limit due to action of high pressure overrange valve

'I 0

Table 3-5 (Cont'd)

BARTON MODEL 288/289 DIFFERENTIAL PRESSURE INDICATING SWITCH FAILURE MODES AND EFFECTS ANALYSIS TABLE NO.

NAME FAILURE MODES SYMPTOMS AND LOCAL EFFECTS INCLUDING DEPENDENT FAILURES METHOD OF DETECTION EFFECTS OH SENSOR

RESPONSE

REMARKS AND TIME OTHER EFFECTS 6

Valve stem shaft Incorrect position Position changes Calibration offset with respect to transmitter design Signal offset Detect in initial calibration Signal drift with respect to comparable signals None None None Shaf t has thread lock to bellows and no significant net rotary force 7

High pressure bellows overrange valve Does not stop valve shaft or isolate bellows on overpressure Potential bellows damage Signal offset with respect to comparable signals None Offset and range may be changed by significant overpressure 8

Low pressure bellows overrange valve Does not stop valve stem or isolate bellows on reverse overpressure Potential bellows damage Signal offset with respect to comparable signal None Offset and range may be changed by significant overpressure

Table 3-5 (Cont'd)

BARTON MODEL 288/289 DIFFERENTIAL PRESSURE INDICATING SWITCH FAILURE MODES AND EFFECTS ANALYSIS TABLE NO.

NAME FAILURE MODES SYMPTOMS AND LOCAL EFFECTS INCLUDING DEPENDENT FAILURES METHOD OF DETECTION EFFECTS ON SENSOR

RESPONSE

TIME REMARKS AND OTHER EFFECTS 9

Silicone fill fluid Increased viscosity Increased response time

Response

time test Yes No identified mechanism for gross viscosity increase; temperature effects must be acceptable for application Small magnitude of bellows/shaft motion and shaft clearness provide low sensitivity to viscosity lO Torque tube drive arm Loosening Offset and/or non-linear response Signal offset or calibration linearity error or erratic response None None

Table 3-5 (Cont'd)

BARTON MODEL 288/289 DIFFERENTIAL PRESSURE INDICATING SWITCH FAILURE MODES AND EFFECTS ANALYSIS TABLE NO.

NAME rll Torque tube FAILURE MODES Small fill fluid leak to transmitter housing SYMPTOMS AND LOCAL EFFECTS INCLUDING DEPENDENT FAILURES Slow collapse of bellows folds METHOD OF DETECTION Fill fluid in transmitter housing None None expected due to flexibility and volume of bellows EFFECTS ON SENSOR

RESPONSE

REMARKS AND TIME OTHER EFFECTS Substantial fillfluid leak to transmitter housing Damage to bellows Fill fluid in transmitter housing Signal drift due to restoring action of bellows spring None Transmitter will not calibrate and may have low range limit due to action of high pressure overrange valve 12 Low pressure housing and seals to transmitter housing Process fluid leak Signal drift due to increased heating of transmitter by process fluid and/or surface conductivity due to process fluid Errors due to sample air pressure drop for large leaks Electronic failure Signal drift with respect to comparable signals None None

Table 3-5 (Cont'd)

BARTON MODEL 288/289 DIFFERENTIAL PRESSURE INDICATING SWITCH FAILURE MODES AND EFFECTS ANALYSIS TABLE NO.

NAME FAILURE MODES SYMPTOMS AND LOCAL EFFECTS IHCLUDIHG DEPENDENT FAILURES MET))OD OF DETECTION EFFECTS ON SENSOR

RESPONSE

REMARKS AHD TIME OTHER EFFECTS 13 Low pressure bellows Leak to low pressure housing Fill fluid enters low pressure housing due to action of bellows spring force Signal drift due to actions of low pressure bellows spxing None Can affect low range limit due to low pressure overrange valve 14 Low pressure Incorrect bellows spring spring Calibrated out or detected dux'ing initial calibration Initial calibration None None Change in spring constant Signal drift due to change in valve stem shaft spring constant Signal drift with respect to comparable signals None None 15 Low pressure bellows to low pressure housing seals Leak to low pressure housing Fill fluid enters low pressure housing due to action of bellows spring force Signal drift due to actions of low pressure bellows spring None None 16 Actuating cern Loosening on torque tube Wear Setpoint error Setpoint error Setpoint test-Setpoint test None None None None

Table 3-5 (Cont'd)

BARTON MODEL 288/289 DIFFERENTIAL PRESSURE INDICATING SWITCH FAILURE MODES AND EFFECTS ANALYSIS TABLE NO.

NAME FAILURE MODES SYMPTOMS AND LOCAL EFFECTS INCLUDING DEPENDENT FAILURES METHOD OF DETECTION EFFECTS ON SENSOR

RESPONSE

REMARKS AND TIME OTHER EFFECTS 17 Plunger screw Change in position Setpoint error Setpoint test None None 18 Switch link Loosening/

friction/wear Erratic setpoint ad) us tment Setpoint adjust None None 19 Switch ad)ust lever Loosening/

friction/wear Erratic setpoint ad)ustment Setpoint ad)ust None None 20 Microswitch Change in actuation force Setpoint change Setpoint test None None 21 Points drive linkage Friction Setpoint error Setpoint test None None NOTE: Applications with capillary tubes need to be tested to verify that a tube crimp has not degraded response time.

MSIV Low Pressure One sided u er toierance bounds

- Barksdale 81T By: WLLDate: 4/14/97 4/i~9 35 DI:=

7 7

10 25 25 6

4 21 21 6

D2:=

3 7

7 6

5 I

I 9

Data.'= stack(DI,D2)

Sy: WLLDate: 4/14/97 Establish mean and standard deviation using standard Mathcad functions:

/t+

0//SJSP'otation as follows:

n = number of data points Mean = mean of the data s = standard deviation n.'= rows(Data) n= 31 Mean.'= mean(Data)

Mean = 12.7097 s '.= stdev(Data) s = 11.9141 n

n-1 The following analysis establishes the 95%/95% one sided upper tolerance interval. The tolerance interval is obtained from the matrix Toiss using MathCad's linterp function.

Tol 95.'=

4 5.14 5

42 6

3.71 7

34 8

3.19 9

3.03 10 2.91 12 2.74 15 2.57 20 2.4 25 2.29 30 2.22 40 2.13 60 2.02 TF:= linterp Tol95, Tol95,n TF =2.211 One-sided upper tolerance bound T ttpper Mean+ s TF T Qp

= 39.0518

Establish normali lot:

By: WLLDate: 4/1 4/97 g/jsjpp-Residuals Residuals

.'= Data Mean Standard Residuals s

OrdRes:=csorr(StsndsrdR d~s,1) cRowOR:=1..mws(OrdRes) cRowOR-l 2

cRotsrOR OrdRes '.= auynent(OrdRes,Prob)

~R rows(ordRes) x:=0 z~n.'=met(normal(0, t,x)- Prattle,x) m.'= 1 intercept:=0 line~R'.=m OrdRes~g t+ intercept

• cRotttoR 0

cRotttOR Od t

LS 2

MSlV Low Pressure One sided u er tolerance bounds

-8 H

d I

B1T Cnl~k: Av( 4,*

~'~ aspic)

By: WLLDate: 4/14/97 CH~ ~

9//5'PV 4

35 25 25 16 16 Dl:=

10 101 25 25 6

4 21 21 6

DZ:=

42 3

3 3

7 7

6 5

Data '.= stack(D1,D2)

0

Establish mean and standard deviation using standard Mathcad functions:

Notation as follows:

By: WLLDate: 4/1 4/9?

1/~5pp-n = number of data points Mean = mean ofthe data s = standard deviation n.'= rows(Data) n = 32 Mean.'= mean(Data)

Mean = 15.4688 s.'= stdev(Data) s = 19.5184 n-1 The following analysis establishes the 95%/95% one sided upper tolerance interval. The tolerance interval is obtained from the matrix Tolas using MathCad's Iinterp function.

Tol 95.'=

4 5.14 5

42 6

3.71 7

3.4 8

3.19 9

3.03 10 2.91 12 2.74 15 2.57 20 2.4 25 2.29 30 2.22 40 2.13 60 2.02 TF:= lintetp Tol95, Tol95,n TF ~2.202 One-sided upper tolerance bound T upp~.'=Mean+ s TF T=58.4482

Establish normali lot:

Sy: WLLDate: 4/1 4/97 4//A>&

Residuals Residuals;= Data - Mean Standard Resides

'.=

s OrdRcs:=csort(StmdardR~d~s,!)

cRowOR:=!..mws(OrrRm!

I cRowOR-Prob 2

OrdRes:= augment(OrdRes,Prob)

~~R'owg~es) z

=mat(oormrd(0, t,x!Prob,s) m:= 1 intercept: = 0 line

'.= m OrdRes~

+ intercept

• cRowOR cRotzrOR 0

cRotvOR, I

I

~

~

Main Steam Line Hi h Flow One sided u er tolerance bounds By: WLLDate: 4/14/97 Cue~

t7l-z q]/spP Barton 288A-Data (44:At( 4~~

I

~wc)

Dl:=

19 20 60 72 31 26 66 70 74 12 12 63 70 63 112 70 97 47.5 D2:=

65 31 90 27 30 73 45 72 92 158 40 6

100 ll~ ~(,~ n g ~i aan B'ing q f Sc] 5~ (

2.4I S"fW r1 35 Zk5+6'. 1~7 S

SC.,W Data '.= stack(D I,D2)

~QAA/

(+~<( g

'5 ~5 ~ - Ki~

cA Establish mean and standard deviation using standard Mathcad functions:

Notation as follows:

n = number ofdata points Mean = mean ofthe data s = standard deviation n.'= rows(Data) n=35 Mean.'= mean(Data) n S a<.'= g (Dana Mean) n Mean = 61.3286 s '.= stdev(Data)

~

~ n-1 S sq ~ 56801.4714 s ~ 40.8734 File: SMSHF.MCO Page:1

~ v4, ~ ii 0

By: WLLDate: 4/14/97

~5 g/izQw The following analysis establishes the 95%/95/o one sided upper tolerance intewal. The tolerance intenial is obtained from the matrix Tol~ using MathCad's linterp function.

Tol 95

.'=

4 5.14 5

4.2 6

3.71 7

3.4 8

3.19 9

3.03 10 2.91 12 2.74 15

?.57 20 2.4 25 2.29 30 2.22 40 2.13 60 2.02 TF:= linterp Tol95, Tol95,n c7>

TF =2.175 One-sided upper tolerance bound T upper

= Mean+ s TF T

= 150.2281 File:SMSHF2v1CD Page: 2

Establish normali lot:

By: WLLDate: 4/1 4/97 R-x lAPw Residuals Residuals,'= Data Mean Standard R~;d~ '=

s OrdRes:=osorr(S dsrdR;d~,i) eRowOR:=I..rows(ordRes) eRowOR-I 2

Prob OrdRes:= augnent(OrdRes, Prob) rows(OrdRes) z~

.--roor(oonnsl(0,1,x) Probes,z) m '.= 1 mtereept:= 0 ime

.=m.OrdRes

+ mtercept 1

• dbneOR cRo D'OR 0

I cRowOR,1 File: SMSHF.MCD Page: 3

Main Steam Line Hi h Flow One sided v er tolerance boUnds By: WLLDate: 4/14/97 Pw g/a+~

Barton 288A-Data 19 20 D1:=

72 31 26 70 74 12 12 63 70 63 112 70 97 47.5 65 9

60 31 90 27 D2:=

30 73 45 72 92 40 6

100 Data '.= stack(D I,D2)

Establish mean and standard deviation using standard Mathcad functions:

Notation as follows:

n = number of data paints Mean = mean ofthe data s = standard deviation n,'= rows(Data) n = 33 Mean = 54.197 s.'= stdev(Data).

n

~n Mean.'= mean(Data) n sq'ata.

Mean S sq 2

4/ 96 s

28.S032 File: SMSHF.MCD Page:

1

Sy: WLLDate: 4/1 4/97 IfrXP+

The following analysis establishes the 95%/95% one sided upper tolerance interval. The tolerance interval is obtained from the matrix Tolss using MathCad's linterp function.

4 5.14 5

42 6

3.71 7

3.4 8

3.19 9

3.03 10 2.91 Tol 95'= 12?74 15 2.57 20 2.4 25 229 30 2.22 40 2.13 60 2.02 TF:= linterp Tol95, Tol95, n TF = 2.193 One-sided upper tolerance bound Tup~

.'e Mean+ s TF T uppm 117 3624 File: SMSHF.MCD Page: 2

Establish normali lot:

Sy: WLLDate: 4/14/97 rh

<P~pa Residuals:= Data Mean Standard Residuais

-'= Residuals s

Ordmtw:=snort(Slsndsrd R~d~s, tj oRowOR:= t..mws(Orden) cRowOR-I cRotNOR OrdRes '.= augment(OrdRes, Prob) rows(OrdRes) x:=0 z~z.'-- root(normsl(0, t,x) Pmb~, xj m '.= l intercept:= 0 line

.'=m OrdRes

intercept I cRottttOR o

cRowOR

-I.S I

0.5 0

Od I

1.5 2

2.5 cRorsOR. I File: SMSHF.MCD Page: 3

~,

RAel Der@

PRO&

SU(-IL.I spam~~

5Uzv, XI~&is Sheet1 Primary Containment Isolation On Rx Lvl low MS-LS<1A

, low2:

Quant 7.4.3.2.3.2 Sensor Ts(sec)

Date 6/11/94

Response

0.685 Date 4/25/92 0.675 MS-LS<1B 7.4.3.2.3.35 5/29/93 0.78 6/4/91 0.605 MS-LS<1C MS-LS-61D SOR 103AS-BB203-NX-JJTTXG.

7.4.3.2.3.36 7.4.3.2.3.37 5/11/93 7/23/94 0.5 0.432 6/5/91 0.53 6/9/92 0.554 Prim Cont Isol On MSL Press Low MS-PS-15A MS-PS-15B MS-PS-15C MS-PS-15D Barksdale B1T-M12SS-GE.

7.4.3.2.3.5 7.4.3.2.3.5 7.4.3.2.3.21 7.4.3.2.3.21 (Sufficient Data Not Available.)

(

. sensor data substituted.)

(Sufficient Data Not Available.)

sensor data substituted.)

Prim Cont Isol On MSL Flow Hi MS-DPISNA, 9A, 810A, 11A MS-DPIS-8B, etc MS-DPISCC, etc MS-DPIS-8D, etc Barton 288A.

7.4.3.2.3.6 7.4.3.2.3.18 7.4.3.2.3.19 7.4.3.2.3.20 (Sufficient Data Not Available.)

(

sensor data substituted.)

RPS Trip On Stm Dme Pres Hi MS-PS-23A MS-PS-23B MS-PS-23C MS-PS-23D 7.4.3.1.3.9 7.4.3.1.3.10 7.4.3.1.3.1 1

7.4.3.1.3.12 5/4/93 5/1 7/93 5/7/94 5/7/94 Note A 5/16/91 0.04 0.22 5/27/92 0.181 5/26/92 0.050*

0.040*

0.072 4/28/91 0.063 SOR 29N6-B45-NX-C1A-JJ7TX12.

RPS Trip On Rx Low Lvl 3 "Ts+ TL Note A: Test done by Not a valid t MS-LIS-24A MS-LIS-24B MS-LIS-24C MS-LIS-24D Barton 288A.

7.4.3.1.3.13 7.4.3.1.3.14 7.4.3.1.3.15 7.4.3.1.3.16 5/5/93 5/18/93 5/9/94 5/9/94 PER 1.11 6/8/93 PER2.053 6/7/93 0.48 4/29/92 0.64 4/30/92 0.015 0.083 0.02 0.19 Other observations onthis particular set of procedur es cc Trtt criteria:

7.4.3.2.3.6 7.4.3.2.3.18 7.4.3.2.3.19 7.4.3.2.3.20 Per SOR RTI For A 29N6-B45=<1 00msec.

Note: "XXX"means not able to find.

6/2/92 6/7/89 6/1/92 0.5sec 1.0sec 0.5 sec 6/7/94 5/15/93 5/15/93 6/4/94 1.0 sec 0.5sec 0.5 sec 1.0sec Page 1

Sheeti Supplemental SOR Pressure Switch Rtt data:

Ts MS-LS-61 B MS-LS-61C 7.4.3.2.3.25 7.4.3.2.3.26 5/30/93 5/31/93 0.415 0.14 Supplemental ITT Ba MS-LIS-24A rton Pressure Switch 7.4.3.2.3.16 Rtt data:

5/26/94 Ts 0.52 MS-LIS-24C MS-LIS-24B 7.4.3.2.3.46 7.4.3.2.3.45 6/6/93 6/8/93 0.308 0.2 More Directly Applicable Data For The Barksdale Pressure Switches:

Quant Sensor Plus Logic Train Response:

TL Ts Criteria Criteria Prim Cont Isol On MSL Press Low MS-PS-1 5A MS-PS-15B 7.4.3.2.3.5 Ts(msec):

5/21/91 6/3/92 6/12/89 13 TS "0" 1 3 TS IfPII 26 (Ts=24)

(sec) 0.95 (msec) 50 MS-PS-1 5C S-PS-15D arksdale B1T-M12SS-GE.

7.4.3.2.3.2 5/22/90 6/4/92 5/21/91 20 TS="0" 37 TS="0" 5 TS IIPII Page 2

Sheet1 Ts Criteria Logic Train TL (ms).

TL Criteria Date Date 0.95 4/25/92 6/4/91 6/5/91 7/23/94 30 6/11/94 20 5/29/93 30 5/11/93 25 6/9/92 25 10 20 20 50 0.95 Sensor Data Utilized.)

Sensor Data Utilized.)

Sensor Data Utilized.)

Sensor Data Utilized.)

50 0.45 6/2/92 5/1 5/93 6/7/89 6/4/94 17 14 30 20 6/7/94 5/15/93 6/1/92 26 24 12 50 0.5 4/28/91 5/16/91 5n/e4 5/7/94 xxx 33 30 35 5/4/93 5/17/93 5/27/92 5/26/92 XXX 30 21 32 50 the slow ramp only.

st, data ignored.

1 5/5/93 6/7/93 4/1 9/92 5/9/94 30 25 30 40 6/8/93 5/18/93 5/9/94 4/30/92 15 25 30 30 50 0

Page 3

RTT Statistical Anal sis Results Summa Sensor Data EPNI Function Statistical Data Source mean sec S

sec Tupper sec Ts sec Assumed Sensor RTT Data Source Comments MS-LS-61A,B,C,D Rx Low Level MS-PS-15A,B,C,D MSLine Low Pressure MS-DPIS MS High Flow MS-PS-23A,B,C,D RPS Trip on High Pressure MS-LIS-24A,B,C,D RPS Tri on Low Level SS Surv, 2 bench tests Barksdale B1T Barton 288A SS Surv, 4 bench tests Barton 288A 10

.5666

.117 0.907 32 0.015 0.02

.0584 35 0.061 0.041 0.150 11 0.083 0.06 0.2533 25 0A56 0.066 0.607 0.95 0.95 0.45 0.50 1.0 LCS 1.0 sec (sensor) minus 0.05 secs (logic from RPS LCS 1.0 sec (sensor) minus 0.05 secs (logic from RPS)

LCS 0.5 secs (sensor) minus 0.05 secs (logic from RPS)

GE Design Spec 23A1 877AA GE Design Spec.

23A1 877AA Acceptable fitto normal curve.

One outlier removed.

Normality plot indicates that the distribution is skewed.

However, there is a large margin between T~and T, indicating acce tabilit of the results.

Two Outliers removed.

Normality plot indicates that the distribution is slightly skewed.

However, there is a large margin between T~ and T, indicating acce tabilit of the results.

Two outiiers removed.

Normality plot indicates that the distribution is skewed.

However, there is a large margin between T~ and T, indicating acce tabilit of the results Good fitto normal curve Results Summa

-Lo ic Data EPNI Function Statistical Data Source mean (sec) s (sec)

Tupper (sec)

Tx(logic)

(sec)

Assumed Logic RTT Data Source Comments MS-I S-61A,B,C,D Rx Low Level and MS-DPIS-8A,B;C,D MS High Flow MS-PS-15A,B,C,D MSLine Low Pressure SS Surv tests NA 15 0.022 0.006 NA NA NA 0.038 NA 0.05 (Includes Solenoid) 0.05 (includes Solenoid Use Data from RPS FSAR Table 7.2-5 Use Data from RPS FSAR Table 7.2-5 Normality plot indicates that the distribution has a high peakedness indicating the normality assumption is conservative.

In addition, there is a large margin between Tu~, and Txixxii.i indicatin acce tabilit of the results.

Valid data not available MS-PS-23A,B,C,D RPS Trip on High Pressure and MS-LIS-24A,B,C,D RPS Tri Rx Low Level Notes:

ext Page SS Surv tests 14 0.029 0.006 0.045 0.05 (Includes Solenoid)

FSAR Table 7.2-5 Large central peak.

Normal distribution consewative.

n = number of data points s = standard deviation T~ = Upper one sided 95/95 tolerance bound T, = response time assumed for instrument sensor T~~i,~ = response time assumed for logic strings (relays). The time assumed for RPS and MSlVisolation logic includes the actuated solenoid valve.

The results in this table conservatively do not take credit for outlier removal.

Outliers are identified only to assess ifthe distribution is normally distributed.